desarrollo de metodologÍa analÍtica para la …

DESARROLLO DE METODOLOGÍA ANALÍTICA

PARA LA DETERMINACIÓN DE FILTROS SOLARES

EN MUESTRAS AMBIENTALES Y COMPUESTOS

RELACIONADOS EN ALIMENTOS ENVASADOS

NOELIA NEGREIRA FERROL

Memoria para optar al grado de Doctora en Química

Santiago de Compostela, Diciembre 2010

UNIVERSIDAD DE SANTIAGO DE COMPOSTELA

Facultad de Química Departamento de Química Analítica, Nutrición y Bromatología Instituto de Investigación y Análisis Alimentario (IIAA)

D. Antonio Moreda Piñeiro, Profesor Titular de Universidad y Director

del Departamento de Química Analítica, Nutrición y Bromatología de la

Universidad de Santiago de Compostela,

Informa:

Que Dña. Noelia Negreira Ferrol presenta el trabajo “DESARROLLO DE

METODOLOGÍA ANALÍTICA PARA LA DETERMINACIÓN DE FILTROS

SOLARES EN MUESTRAS AMBIENTALES Y COMPUESTOS

RELACIONADOS EN ALIMENTOS ENVASADOS” que ha realizado en este

departamento bajo la dirección de D. Isaac Rodríguez Pereiro y Dña. Elisa

Rubí Cano, Profesores Titulares de Universidad, para optar al grado de

Doctora en Química.

Y para que así conste, firmo el presente informe en Santiago de

Compostela, 3 de diciembre de 2010.

D. Antonio Moreda Piñeiro

D. Isaac Rodríguez Pereiro y Dña. Elisa Rubí Cano, Profesores Titulares

de Universidad del Departamento de Química Analítica, Nutrición y

Bromatología de la Universidad de Santiago de Compostela,

Autorizan:

A la licenciada Dña. Noelia Negreira Ferrol a la presentación del trabajo

recogido en la memoria titulada “DESARROLLO DE METODOLOGÍA

ANALÍTICA PARA LA DETERMINACIÓN DE FILTROS SOLARES EN

MUESTRAS AMBIENTALES Y COMPUESTOS RELACIONADOS EN

ALIMENTOS ENVASADOS” que ha realizado bajo su dirección en el

Departamento de Química Analítica, Nutrición y Bromatología de la Facultad

de Química de la Universidad de Santiago de Compostela para optar al grado

de Doctora en Química

Y, para que así conste, firmamos el presente informe en Santiago de

Compostela, 3 de diciembre de 2010.

D. Isaac Rodríguez Pereiro Dña. Elisa Rubí Cano

AGRADECIMIENTOS

Al finalizar un trabajo tan lleno de dificultades, como lo es una Tesis

Doctoral para la que se requiere de mucho esfuerzo y dedicación por parte de la

doctoranda y de los directores, resulta inevitable mostrar mis más sinceros

agradecimientos a aquellas personas que de una forma desinteresada han

contribuido a la elaboración de esta Tesis y sin las cuales, no hubiera sido

posible su finalización.

En primer lugar, quiero dar las gracias a mis directores, D. Isaac

Rodríguez Pereiro y Dña. Elisa Rubí Cano, por haberme dado la oportunidad

de realizar este trabajo, por poner vuestro apoyo y confianza en mí. En especial,

a Isaac por tus ideas, orientación, respuestas ante inquietudes surgidas, en

definitiva, por tu incalculable aportación al desarrollo de esta tesis y mi

formación como investigadora.

Quisiera también agradecer al Departamento de Química Analítica,

Nutrición y Bromatología, al Instituto de Investigación y Análisis Alimentarios

(IIAA), y en especial a D. Rafael Cela, director del grupo de investigación de

Cromatografía de Gases y Quimiometría, por haber puesto a mi disposición

todos los medios técnicos y materiales para la realización de esta Tesis.

Asimismo, debo agradecer al Ministerio de Educación por la beca FPU

disfrutada y a los fondos autonómicos, estatales y europeos que han financiado

mi trabajo (2010-02, DE2009-0020, DGICT CTQ2009-08377, CTQ2006-03334 y

PGIDIT06PXIB237039PR).

A mis compañeros del IIAA, tanto a los que ya se han ido (Pablo, Lucía y

María F.) como a los quedan (en especial, Iria y Paula) y, sobre todo, a los que

están empezando desearles mucha suerte y paciencia.

A Jorge, por ser la persona con la que compartí más momentos dentro y

fuera del laboratorio durante esta etapa. Tu compañía hizo que me olvidara de

los ratos amargos y tu alegría me dieron la fuerza necesaria para seguir

adelante. Gracias por tu comprensión, tu paciencia y tu cariño. Gracias por esos

momentos inolvidables.

A mis amigas, María, Bea, Lupe y por supuesto, Isabel, gracias por vuestra

compañía, por ayudarme a sobrellevar los malos momentos, y por ofrecerme

amistad y diversión. A Carmen Trillo, por escucharme y aconsejarme en

momentos de angustia, por brindarme tu apoyo y tus ánimos.

Y, por supuesto, el agradecimiento más profundo a mi familia, en especial

a mis padres y a mi hermana, por vuestro apoyo, cariño, consejos, por estar

siempre a mi lado y por hacerme feliz.

En resumen, agradezco a todas y cada una de las personas que han vivido

conmigo la realización de este trabajo, con sus altos y bajos, y en especial, a

aquellas que han sido mi soporte en momentos de angustia y desesperación.

Más que el brillo de la victoria,

nos conmueve la entereza ante la adversidad

(Octavio Paz)

ÍNDICE

Índice

ÍNDICE

I. JUSTIFICACIÓN Y OBJETIVOS ...........................................................1

II. INTRODUCCIÓN ..................................................................................... 5

A. Filtros solares.......................................................................................................... 7

1. ASPECTOS GENERALES ................................................................................. 7

1.1. Definición y clasificación .................................................................. 7

1.2. Filtros solares permitidos por la Unión Europea. ........................ 9

1.3. Filtros solares considerados en este estudio ................................ 15

2. DISTRIBUCIÓN MEDIOAMBIENTAL ........................................................ 21

2.1. Muestras de agua ............................................................................ 22

2.1.1. Niveles de filtros solares en agua ............................................ 22

2.1.2. Toxicidad y reacciones de transformación de los filtros

solares en el medio acuoso ............................................................... 29

2.2. Lodos y sedimentos ........................................................................ 32

2.3. Biota ................................................................................................... 36

3. METODOLOGÍA ANALÍTICA ..................................................................... 38

3.1. Preparación de muestra .................................................................. 38

3.1.1. Muestras de agua ....................................................................... 38

3.1.1.1. Extracción en fase sólida (SPE) .................................... 38

3.1.1.2. Técnicas basadas en la microextracción en fase sólida 46

3.1.1.2.1. Microextracción en fase sólida (SPME) ................... 46

3.1.1.2.2. Microextracción con barras agitadoras (SBSE) ......... 52

3.1.1.2.3. Microextracción mediante sorbentes empaquetados

(MEPS) .................................................................... 55

3.1.1.3. Técnicas basadas en la microextracción en fase

líquida……………………………………………………...57

3.1.1.3.1. Microextracción con gota suspendida (SDME) ........ 57

3.1.1.3.2. Microextracción líquido-líquido con membranas .... 58

3.1.1.3.3. Microextracción en fase líquida con fibra hueca (HF-

LPME) ..................................................................... 58

Índice

3.1.1.3.4. Microextracción líquido-líquido dispersiva

(DLLME)…………………………………………..59

3.1.2. Muestras sólidas ........................................................................ 67

3.1.2.1. Dispersión de la matriz en fase sólida (MSPD) ........... 67

3.1.2.2. Extracción con disolventes presurizados (PLE) .......... 70

3.1.2.3. Otras técnicas de extracción ........................................ 76

3.1.3. Muestras de biota ...................................................................... 78

3.2. Técnicas de determinación ............................................................. 81

3.2.1. Cromatografía de gases acoplada a espectrometría de masas

simple (GC-MS) y en tándem (GC-MS/MS) ................................. 82

3.2.2. Cromatografía líquida acoplada a espectrometría de masas

(LC-MS) ............................................................................................... 91

B. Fotoiniciadores .................................................................................................... 97

1. ASPECTOS GENERALES ............................................................................... 97

2. ESTRUCTURA Y PROPIEDADES ................................................................ 98

3. PRESENCIA EN ALIMENTOS ...................................................................... 99

4. METODOLOGÍA ANALÍTICA PARA LA DETERMINACIÓN DE

FOTOINICIADORES EN ALIMENTOS ................................................ 101

III. METODOLOGÍA DESARROLLADA .......................................................... 105

A. Filtros solares .................................................................................................... 107

1. MUESTRAS ACUOSAS ................................................................................ 107

1.1. Introducción ................................................................................... 107

1.2. Esquemas de los métodos desarrollados para muestras

acuosas.. .......................................................................................... 110

1.3. Publicación: “Dispersive liquid-liquid microextraction followed by gas

chromatography-mass spectrometry for the rapid and sensitive determination of UV

filters in environmental water samples” ....................................................... 117

1.4. Publicación: “Silicone discs as disposable enrichment probes for gas

chromatography-mass spectrometry determination of UV filters in water

samples”……. ........................................................................... ………145

Índice

1.5. Publicación: “Sensitive determination of salicylate and benzophenone type UV

filters in water samples using solid-phase microextraction, derivatization and gas

chromatography tandem mass spectrometry” ............................................... 173

1.6. Publicación: “Solid-phase extraction followed by liquid chromatography tandem

mass spectrometry for the determination of hydroxylated benzophenone UV absorbers

in environmental water samples" ............................................................... 201

1.7. Publicación: “Study of some UV filters stability in chlorinated water and

identification of halogenated by-products by gas chromatography-mass spectrometry”

............................................................................................................ 225

2. MUESTRAS SÓLIDAS .................................................................................. 251

2.1. Introducción ................................................................................... 253


sólidas….......................................................................................... 255

2.3. Publicación: “Optimization of pressurized liquid extraction and purification

conditions for gas chromatography-mass spectrometry determination of UV filters in

sludge” .................................................................................................. 257

2.4. Publicación: “Determination of selected UV filters in indoor dust using matrix

solid-phase dispersion and gas chromatography tandem mass spectrometry” ....... 281

B. Fotoiniciadores .................................................................................................. 305

1. FOTOINICIADORES EN ALIMENTOS ..................................................... 307

1.1. Introducción ................................................................................... 307

1.2. Esquema del método desarrollado para fotoiniciadores en

alimentos......................................................................................... 308

1.3. Publicación: “Solid-phase microextraction followed by gas chromatography mass

spectrometry for the determination of ink photo-initiators in packed milk” .......... 309

IV. CONCLUSIONES ................................................................................. 337

V. BIBLIOGRAFÍA .................................................................................... 345

VI. ABREVIATURAS Y ACRÓNIMOS .................................................. 357

VII. Anexo: OTRAS PUBLICACIONES ................................................... 363

Justificación y objetivos

3

I. JUSTIFICACIÓN Y OBJETIVOS

Actualmente, se utilizan un gran número de compuestos químicos como

aditivos en alimentos, productos de cuidado personal, textiles, etc. En los

últimos años, el interés por los posibles efectos de estos compuestos, conocidos

como contaminantes emergentes, se ha incrementado de forma notable. Dentro

de este grupo se engloban los filtros solares, también denominados filtros UV.

Estos compuestos constituyen los principios activos de protectores solares, cuyo

consumo ha aumentado notablemente en los últimos años [Richardson, 2006]

debido a la concienciación de los efectos nocivos de las radiaciones ultravioletas

sobre la salud. Además, los filtros solares aparecen también en la formulación

de otros cosméticos de uso diario como cremas de belleza, lociones para la piel,

barras de labios, sprays para el pelo, tintes, champú e incluso en ropa y

plásticos [Richardson, 2006] [Salvador, 2005-A; Salvador, 2005-B] [Chisvert, 2001-A]

[Schlumpf, 2004]. Las aplicaciones anteriores provocan la introducción, directa o

indirecta, de filtros UV en el medio acuático. Sus efectos medioambientales son

todavía desconocidos, sin embargo es sabido que algunos de ellos presentan

actividad estrogénica [Díaz-Cruz, 2009] e incluso tendencia a acumularse en

sedimentos y lodos.

Otra familia de compuestos considerada en este trabajo es la

correspondiente a los llamados fotoiniciadores, usados como aditivos en las

tintas de envases alimentarios con el fin de acelerar el secado de las impresiones

realizadas sobre los mismos. Desde un punto de vista estructural, la mayoría de

los fotoiniciadores son derivados de la benzofenona o del ácido p-

aminobenzoico, empleados como filtros UV en productos de cuidado personal.

Recientemente, algunos fotoiniciadores han sido detectados en diversos

alimentos, entre ellos leche, causando alarma social y pérdidas económicas

derivadas de la destrucción de estos alimentos.

Justificación y objetivos

4

El objetivo general de esta Tesis Doctoral ha sido la optimización de

nuevas metodologías para la determinación de filtros solares en muestras

ambientales y de fotoiniciadores en leche. En todos los casos se ha minimizado,

en la medida de lo posible, el consumo de disolventes y la generación de

residuos peligrosos, intentando mejorar las prestaciones analíticas (límites de

cuantificación, exactitud, coste, tiempo de respuesta, etc.) de los métodos ya

disponibles en la bibliografía.

La determinación de filtros solares se llevó a cabo en matrices acuosas

(agua de río, piscina y residual) y sólidas (polvo y lodos). Los métodos

analíticos propuestos se basan en el empleo de técnicas de preparación de

muestra adecuadas a cada matriz. Entre ellas se encuentran la extracción en fase

sólida (SPE), la microextracción en fase sólida (SPME), la absorción sobre

siliconas, y la microextracción líquido-líquido dispersiva (DLLME) para

muestras de agua. En el caso de las matrices sólidas se desarrollaron métodos

de dispersión de la matriz en fase sólida (MSPD) y extracción con fluidos

presurizados (PLE). Las técnicas de determinación empleadas fueron la

cromatografía de gases y la cromatografía de líquidos en combinación con

espectrometría de masas. Una vez optimizadas y validadas las condiciones de

trabajo, los métodos fueron aplicados al estudio de la distribución de filtros

solares en muestras de agua superficial y residual, lodos de estaciones

depuradoras y polvo procedente de atmósferas interiores. También se han

realizado estudios de laboratorio para evaluar la reactividad de ciertos filtros

UV cuando entran en contacto con aguas cloradas.

Por último, un capítulo de esta Tesis se ha dedicado a la optimización y

validación de un método rápido y robusto para la determinación de

fotoiniciadores en muestras de leche. En este caso, se combinó la utilización de

la microextracción en fase sólida (SPME) con cromatografía de gases acoplada a

espectrometría de masas para la determinación selectiva de los compuestos

anteriores evitando la utilización de disolventes orgánicos.

Introducción-Filtros solares

7

II. INTRODUCCIÓN

A. Filtros solares

1. ASPECTOS GENERALES

1.1. Definición y clasificación

Los protectores solares son preparados farmacéuticos de aplicación tópica

que tienen la propiedad de absorber o reflejar la radiación ultravioleta de origen

solar, o de fuentes artificiales, atenuando la acción perjudicial de los rayos

solares. Generalmente, los compuestos activos en protectores solares protegen

contra radiación tipo UVB (290-320 nm) y algunos contra UVA (320-400 nm), la

cual puede penetrar en profundidad en la piel y, después de una larga

exposición, dañarla fuertemente [Chisvert, 2001-A; Chisvert, 2001-B] [Salvador,

2005-A; Salvador, 2005-C]. Actualmente, hay una tendencia a aumentar el factor

de protección lo que, generalmente, significa mayor concentración de filtros UV

en los protectores solares. Frecuentemente, se usan varios filtros solares para

cubrir un amplio rango de longitudes de onda [Balmer, 2005].

La luz solar provoca daño cutáneo porque las radiaciones ultravioleta

(UV) son absorbidas por el ADN, ARN, proteínas, lípidos de membranas y

orgánulos celulares presentes en la epidermis y dermis, incluyendo el sistema

vascular. Los efectos son acumulativos y están en relación con la duración,

frecuencia e intensidad de radiación. El efecto inmediato es la inflamación y el

tardío, cáncer de piel. El 95% de radiaciones que inciden sobre nuestra piel son

infrarrojos (longitudes de onda superiores a 760 nm), y luz visible (400-760 nm).

Sólo el 5% es radiación UV, de la cual el 2% corresponde a UVB y el 98% a UVA

que, a su vez puede dividirse en UVA largos o UVA-I (340-400 nm) y UVA

cortos o UVA-II (320-340 nm). La denominada radiación UVC (con longitudes

de onda inferiores a 290 nm) no llega a nuestra piel ya que es absorbida por la


8

capa de ozono, aunque comenzó a tomar importancia debido a la progresiva

disminución de ésta en los últimos años.

Atendiendo a su naturaleza química, los filtros solares pueden clasificarse

en:

Filtros solares de naturaleza inorgánica. Debido al tamaño y a la

uniformidad de las partículas que los componen, son fotoestables y

seguros. En concentraciones altas pueden sufrir aglomeración

presentando un aspecto blanquecino. Actúan mediante atenuación de la

radiación UV, resultado de la combinación de mecanismos de reflexión,

dispersión y absorción. Los ejemplos más represantativos de este tipo de

compuestos son:

- Óxido de zinc: es un óxido metálico empleado históricamente

como protector de la piel. Es seguro para la aplicación en piel

inflamada y con afectación de la barrera cutánea. Protege de

radiaciones UVB, UVA-II y parcialmente de UVA-I.

- Dióxido de titanio: es un óxido metálico, casi inerte. Protege

frente a UVB y UVA-II.

Filtros solares de naturaleza orgánica. Son los más empleados y se

caracterizan por poseer estructuras aromáticas simples o múltiples, a

menudo substituidas con grupos hidrofóbicos para mejorar sus

propiedades. Estos compuestos son aplicados superficialmente sobre la

piel y su penetración subcutánea es limitada [Straub, 2002]. Sin embargo,

estudios recientes han demostrado su presencia en fluidos biológicos,

tales como orina [Ye, 2005] [Vidal, 2007] [Kawaguchi, 2008-B; Kawaguchi,

2009] [Ito, 2009] [Knisue, 2010], plasma [Sarveiya, 2004] y semen [León,

2010]. Pueden dividirse en dos grupos según absorban radiación UVA o

UVB y su combinación, junto con los filtros de naturaleza inorgánica,


9

permite obtener preparados que cubren un amplio rango de longitudes

de onda y aumentan la eficacia de protección solar.

1.2. Filtros solares permitidos por la Unión Europea.

En general, las concentraciones máximas permitidas de filtros UV en

productos de uso tópico (protectores solares) de acuerdo a la legislación

Europea están entre el 5 y el 10% [Chisvert, 2002-A; Chisvert, 2002-B] [Schakel,

2004]. Existen otras sustancias que podrían ser usadas como filtros solares, que

se encuentran autorizadas por la legislación americana (FDA) o la japonesa,

pero no todas son permitidas por la Unión Europea. Además, compuestos cuyo

uso está permitido actualmente posiblemente se retiren del mercado a medio

plazo. Por ejemplo, el ácido p-aminobenzoico (PABA) causa problemas

dermatológicos [Chisvert, 2001-A] y aún así, sigue permitido en un 15% por

FDA (Estados Unidos) y en un 5% por la EU [Wang, 2007]. Muchos otros son

objeto de estudio entre ellos el ácido 2-hidroxi-4-metoxibenzofenona-5-

sulfónico (BP-4), la 2-hidroxi-4-metoxibenzofenona (BP-3), el

butilmetoxidibenzoilmetano, el 2-etilhexil-4-dimetilaminobenzoato (EHPABA),

el 2-etilthexil-p-metoxicinamato (EHMC), el homosalato (HMS) y el 2-

etilhexilsalicilato (EHS) [Chisvert, 2001-A]. Por último, es necesario destacar que

la legislación comentada anteriormente es sólo aplicable a protectores solares,

pero no a otros productos de cuidado personal, tales como cremas faciales,

colonias, etc. que pueden contener otros compuestos como filtros de la

radiación ultravioleta.

A continuación, se muestra una tabla con la lista de filtros UV que pueden

contener los protectores solares comercializados en la Unión Europea, así como

la concentración máxima autorizada y el tipo de radiación UV que absorben,

Tabla 1.


10

Tabla 1: Filtros solares permitidos por la legislación europea1.

Nombre químico Nombre común

(acrónimo)

Tipo

absorción

UV

Concentración

máxima

autorizada(%)

Filtros de naturaleza inorgánica

Dióxido de Titanio Dióxido de Titanio (TiO2) A, B 25

Filtros de naturaleza orgánica

Derivados de la benzofenona

2-hidroxi-4-metoxibenzofenona Benzofenona-3 (BP-3),

Oxibenzona A, B 10

Ácido 2-hidroxi-4-

metoxibenzofenona-5-sulfónico y

su sal sódica

Benzofenona-4

(BP-4) A, B

5

(expresada

como ácido)

Ácido p-aminobenzoico y sus derivados

Ácido 4-aminobenzoico PABA B 5

2-etilhexil-4-

dimetilaminobenzoato

2-etilhexil-4-

dimetilaminobenzoato

(EHPABA)

B 8

Etil-4-aminobenzoato etoxilado PEG-25 PABA B 10

Ácido benzoico,2-(4-

(dietilamino)-2-hidroxibenzoil-,

hexiléster

Dietilamino hidroxibenzoil

benzoato de hexil A 10

Salicilatos

2-etilhexilo salicilato Etilhexil salicilato (EHS) B 5

3,3,5-trimetilciclohexilo salicilato Homosalato

(HMS) B 10

Metoxicinamatos

4-metoxicinamato de 2-etilhexilo

o metoxicinamato de octilo

Etilhexil metoxicinamato

(EHMC) B 10

4-metoxicinamato de isoamilo Isoamil metoxicinamato

(IAMC) B 10

1 Anexo VI: REGULATION (EU) 1223/2009 del Parlamento Europeo del 30 de

noviembre 2009 relativa a productos cosméticos.


11

Tabla 1 cont.: Filtros solares permitidos por la legislación europea.


(acrónimo)

Tipo absorción

UV

Concentración máxima

autorizada(%) Derivados del alcanfor

3-Bencilideno alcanfor 3-Bencilideno

alcanfor (3-BC) B 2

3-(4’-Metilbencilideno)-d-1 alcanfor 4-metilbencilideno

alcanfor (4-MBC) B 4

Ácido α-(2-oxoborn-3-iliden)-toluen-4-

sulfónico y sus sales

Ácido bencilideno

alcanfor sulfónico

(BCS)

B

6

(expresada

como ácido)

Sulfato de metilo de N,N,N-trimetil[(oxo-

2-bornilideno-3)metil]-6-anilinio

Alcanfor

benzalconio

metosulfato (CBM)

B 6

Ácido 3,3’-(1,4-Fenilendimetileno)bis(7,7-

dimetil-2-oxobiciclo-[2,2,1]hept-1-

ilmetanosulfónico) y sus sales

Ácido tereftaliden

dialcanfor sulfónico

(TDS)

A

10

(expresada

como ácido)

Polímero de N-{(2 y 4)-[(2-oxoborn-3-

iliden)metil]bencil} acrilamida

Poliacrilamidometil

bencilideno alcanfor

(PBC)

B 6

Derivados de la triazina

2,4,6-Trianilina-(p-carbo-2’-etilhexil-1’oxi)-

1,3,5-triazina

Etilhexil triazona

(ET) B 5

Benzoato de bis (2-etilhexil) 4,4’-[[6-[[[(1,1-

dimetiletil)amino]carbonil]fenil]amino]-

1,3,5-triazina-2,4-diil]diimino]bis-

Dioctil butamido

triazona

(DBT)

B 10

2,2’-[6-(4-Metoxifenil)-1,3,5- triazina-2,4-

diil]bis[5- [(2etilhexil)oxi]fenol

Bis-etilhexiloxifenol

metoxifenil triazina

(EMT)

A, B 10

Derivados del benzotriazol

2-(2H-benzotriazol-2-il)-4-metil-6-[2-metil-

3-(1,3,3,3-tetrametil-1-(((trimetilsilil)oxi)-

disiloxanil)propil], fenol

Drometrizol

trisiloxano (DRT) A, B 15

2,2’-metilen-bis-[4-(1,1,3,3-tetrametilbutil)-

6-(2H-benzotriazol-2-il) fenol

Metilen

bisbenzotriazolil

tetrametilbutilfenol

(MBP)

A, B 10


12

Tabla 1 cont.: Filtros solares permitidos por la legislación europea.


(acrónimo)

Tipo

absorción

UV

Concentración

máxima

autorizada(%)

Derivados del benzoimidazol

Ácido 2-Fenilbencimidazol-5-

sulfónico y sus sales de potasio, de

sodio y de trietanolamina

Ácido fenilbencimidazol

sulfónico (PBS) B

8

(expresada

como ácido)

Sal monosódica del ácido 2,2’-bis-

(1,4-fenilen)1H-benzimidazol-4,6-

disulfónico

Ácido

fenildibenzimidazol

tetrasulfónico (PDT)

A

10

(expresada

como ácido)

Otros

1-(4-tert-butil-fenil)-3-(4-

metoxifenil)propano-1,3-diona

Butil

metoxidibenzoilmetano

(BDM)

A 5

Éster 2-etilhexil del ácido 2-ciano-

3,3-difenilacrílico

Octocrileno

(OCR) B

10

(expresada

como ácido)

Dimetilcodietilbencilmalonato Polisilicona-15 (P-15) B 10

A continuación, se muestran las estructuras químicas para algunos filtros

solares, permitidos por la Unión europea, y para los que apenas existe

metodología para su determinación en muestras medioambientales (Fig.1). No

se incluyen los compuestos considerados en este estudio ya que sus

propiedades se presentan en el siguiente apartado de esta memoria.

Ácido p-aminobenzoico y sus derivados:

Z

PABA x+y+z=25PEG-25 PABA

Dietilamino hidroxibenzoil benzoato de hexil


13

Derivados del alcanfor:

3-Bencilideno alcanfor(3-BC)

Poliacrilamidometil bencilideno alcanfor

(PBC)

Ácido bencilidenoalcanfor sulfónico

(BCS)

Alcanfor benzalconio metosulfato

(CBM)

Ácido tereftaliden dialcanfor sulfónico

(TDS)

Derivados de la triazina:

Etilhexil triazona(ET)

Bis-etilhexiloxifenol

metoxifenil triazina(EMT)

Doctil butamidotriazona

(DBT)

R1=R2=R3

R1=R2

R3

R3

R1=R2


14

Derivados del benzotriazol:

Drometrizol trisiloxano (DRT)

Metilen bisbenzotriazoliltetrametilbutilfenol

(MBP)

2

Derivados del benzoimidazol:

Ácido fenildibencimidazoltetrasulfónico

(PDT)

Ácido fenilbencimidazolsulfónico

(PBS)

Otros:

Butil metoxidibenzoilmetano

(BDM)

Figura 1: Estructuras químicas de distintas familias de filtros solares [Salvador, 2005-

A].


15

1.3. Filtros solares considerados en este estudio

A continuación, se presentan de forma detallada las características físico-

químicas más relevantes de los filtros solares considerados en este estudio,

además de otros compuestos con estructuras químicas muy semejantes2.

a) La familia de las benzofenonas se caracteriza por presentar la misma

estructura base, con sustituyentes hidroxi y metoxi en diferentes

posiciones de los anillos aromáticos. En la Unión Europea (UE) sólo está

permitido el uso de 2-hidroxi-4-metoxibenzofenona, también

denominada benzofenona-3 (BP-3 o Bz-3), y el ácido 2-hidroxi-4-

metoxibenzofenona-5-sulfónico (BP-4) en protectores solares, pero en

otros países como en Japón también se permite el uso de 2,4-

dihidroxibenzofenona o 2,4-dihidroxifenilmetanona (BP-1), 2,2’,4,4’-

tetrahidroxibenzofenona (BP-2) y 2,2’-dihidroxi-4,4’-metoxibenzofenona

(BP-6) [Shaath, 2007]. Además, la BP-1 es el principal metabolito de la BP-

3 detectado en muestras de orina [Gonzalez, 2008] [Díaz-Cruz, 2008]. Otras

benzofenonas, tales como la 2,2’-dihidroxi-4-metoxibenzofenona (BP-8)

son incluidas como fotoestabilizadores en muchos otros productos de

cuidado personal, barnices, ropa y plásticos para envasado de alimentos

[Richardson, 2008] [Jeon, 2006]. A continuación se presentan las

estructuras químicas de las benzofenonas consideradas en este estudio

(Fig. 2).

2 Calculated using Advanced Chemistry Development (ACD/Labs) Software

V8.14 for Solaris (© 1994-2009 ACD/Labs).


16

BP‐4 BP‐8

BP‐3BP‐2BP‐1

BP‐6

Figura 2: Estructuras de las benzofenonas.

Las propiedades físico-químicas de las benzofenonas permitidas por la

UE (BP-3 y BP-4) y de las demás benzofenonas consideradas en esta memoria

(BP-1, BP-2, BP-6 y BP-8) se recogen en las siguientes tablas (Tabla 2 y Tabla 3). A

excepción de la BP-4, se trata de compuestos ligeramente ácidos (pKas entre 7 y

7,6 unidades) y de moderada polaridad. Por su parte, la BP-4 es altamente

soluble en agua y presenta un carácter fuertemente ácido.

Tabla 2: Propiedades físico-químicas de las benzofenonas permitidas por la UE.

Propiedad Compuesto

BP-3 BP-4

Nº CAS 131-57-7 4065-45-6

Peso molecular (g mol-1) 228,2 308,3

Densidad (g cm-3) 1,20 ± 0,06 1,45 ± 0,06

pKa 7,6 ± 0,4

-0,7 ± 0,5

Entalpía de vaporización (KJ mol-1) 64 ± 3 -

logKow 3,6 ± 0,4 0,9 ± 0,5

Presión de vapor (mTorr) 0,0526 -

Punto de ebullición (ºC) 370 ± 27 491

Punto de fusión (ºC) 141 ± 17 145-190

Solubilidad molar (mol L-1), pH 1 9,2E-4 0,11

Solubilidad molar (mol L-1), pH 7 1,2E-3 3,24

Solubilidad molar (mol L-1), pH 10 0,21 3,24

-, dato no disponible


17

Tabla 3: Propiedades físico-químicas de BP-1, BP-2, BP-6 y BP-8.

Propiedad Compuesto

BP-1 BP-2 BP-6 BP-8

Nº CAS 131-56-6 131-55-5 131-54-4 131-53-3

Peso molecular (g mol-1) 214,2 246,2 274,3 244,2

Densidad (g cm-3) 1,30 ± 0,06 1,53 ± 0,06 1,29 ± 0,06 1,30 ± 0,06

pKa 7,5 ± 0,4 7,0 ± 0,4 7,0 ± 0,4 7,0 ± 0,4

Entalpía de vaporización (KJ mol-1) 67 ± 3 84 ± 3 72 ± 3 65 ± 3

logKow 3,2 ± 0,4 3,1 ± 0,4 4,1 ± 0,0 3,9 ± 0,4

Presión de vapor (mTorr) 0,012 6,69E-9 2,49E-5 0,0373

Punto de ebullición (ºC) 391 ± 27 531 ± 25 439 ± 45 375 ± 0,0

Punto de fusión (ºC) 215 ± 17 289 ± 20 164 ± 22 146 ± 19

Solubilidad molar (mol L-1), pH 1 3,0E-3 4,0E-3 2,4E-4 9,5E-4

Solubilidad molar (mol L-1), pH 7 3,0E-3 0,013 7,0E-4 1,8E-3

Solubilidad molar (mol L-1), pH 10 4,67 4,06 3,15 4,09

b) El derivado del ácido p-aminobenzoico, más conocido como PABA,

considerado a lo largo de este estudio fue su éster: 2-etilhexil-4-

dimetilaminobenzoato (EHPABA), empleado como filtro solar. A

continuación se presentan sus estructuras (Fig. 3) y sus propiedades

físico-químicas (Tabla 4).

PABA EHPABA

Figura 3: Estructuras del PABA y del EHPABA.


18

Tabla 4: Propiedades físico-químicas del PABA y del EHPABA.

Propiedad Compuesto

PABA EHPABA

Nº CAS 150-13-0 21245-02-3

Peso molecular (g mol-1) 137,1 277,4

Densidad (g cm-3) 1,32 ± 0,06 0,99 ± 0,06

pKa (grupo carboxilo) 4,9 ± 0,1 -

pKa (grupo amino) 2,5 ± 0,1 2,4 ± 0,1

Entalpía de vaporización (KJ mol-1) 62 ± 3 63 ± 3

logKow 0,8 ± 0,2 6,1 ± 0,2

Presión de vapor (mTorr) 0,0345 0,00457

Punto de ebullición (ºC) 340 ± 25 383 ± 25

Punto de fusión (ºC) 159 ± 23 122 ± 14

Solubilidad molar (mol L-1), pH 1 1,61 1,8E-4




c) Los salicilatos considerados en este estudio fueron el 2-etilhexilsalicilato

(EHS), el homosalato (HMS) y el bencilsalicilato (BzS) que, aunque su

uso no está permitido en protectores solares, aparece en la formulación

de muchos cosméticos y perfumes. A continación se muestran sus

estructuras (Fig. 4) y sus propiedades físico-químicas (Tabla 5). Al igual

que las benzofenonas poseen un ligero carácter ácido; sin embargo, su

solubilidad en medio acuoso es muy inferior.

OH

O

O

EHS HMS BzSOH

O

O

Figura 4: Estructuras del EHS, HMS y BzS.


19

Tabla 5: Propiedades físico-químicas del EHS, HMS y BzS.

Propiedad Compuesto

EHS HMS BzS

Nº CAS 118-60-5 118-56-9 118-58-1

Peso molecular (g mol-1) 250,3 262,3 228,2

Densidad (g cm-3) 1,04 ± 0,06 1,1 ± 0,1 1,22 ± 0,06

pKa 8,1 ± 0,3 8,1 ± 0,3 8,1 ± 0,3

Entalpía de vaporización (KJ mol-1) 57 ± 3 61 ± 3 58 ± 3

logKow 5,8 ± 0,2 5,8 ± 0,3 4,0 ± 0,3

Presión de vapor (mTorr) 0,0807 0,0417 0,075

Punto de ebullición (ºC) 332 ± 15 341 ± 15 320 ± 0

Punto de fusión (ºC) 128 ± 13 132 ± 13 164 ± 14

Solubilidad molar (mol L-1), pH 1-6 1,1E-4 7,0E-5 7,6E-4

Solubilidad molar (mol L-1), pH 10 7,9E-3 5,4E-3 0,056

d) Los metoxicinamatos considerados fueron el isoamil-p-metoxicinamato

(IAMC) y el 2-etilhexil-p-metoxicinamato (EHMC) ambos permitidos por

la UE en concentraciones máximas individuales del 10%. Otros dos

filtros solares considerados en esta tesis son el octocrileno (OCR), de

estructura similar a los cinamatos y el 3-(4-Metilbencilideno alcanfor),

también conocido como Eusolex 6300 o por las siglas 4-MBC. Sus

estructuras químicas (Figura 5) y propiedades físico-químicas (Tabla 6) se

recogen a continación.

OCR4-MBC

IAMC EHMC

Figura 5: Estructuras del IAMC, EHMC, 4-MBC y OCR.


20

Tabla 6: Propiedades físico-químicas de los metoxicinamatos (IAMC y EHMC), OCR y

4-MBC.

Propiedad Compuesto

IAMC EHMC 4-MBC OCR

Nº CAS 71617-10-2 5466-77-3 36861-47-9 6197-30-4

Peso molecular (g mol-1) 248,3 290,4 254,4 361,5

Densidad (g cm-3) 1,03 ± 0,06 1,00 ± 0,06 1,06 ± 0,06 1,06 ± 0,06

Entalpía de vaporización (KJ mol-1) 61 ± 3 66 ± 3 62 ± 3 74 ± 3

logKow 4,1 ± 0,2 5,7 ± 0,2 4,9 ± 0,3 7,5 ± 0,8

Presión de vapor (mTorr) 0,0189 0,0889E-2 9,99E-3 2,56E-6

Punto de ebullición (ºC) 363 ± 17 405 ± 20 372 ± 22 478 ± 33

Punto de fusión (ºC) 152 ± 16 172 ± 16 167 ± 13 239 ± 12

Solubilidad molar (mol L-1), pH 1-10 2,4E-4 1,7E-5 1,4E-5 5,5E-7


21

2. DISTRIBUCIÓN MEDIOAMBIENTAL

Los filtros solares son usados en una amplia variedad de productos de

cuidado personal y compuestos farmacéuticos, por lo que entran en el medio

ambiente acuático indirectamente a través de las aguas residuales domésticas, y

directamente a través de las piscinas, playas, ríos y los efluentes de las plantas

de tratamiento de aguas residuales, donde no son eliminados de manera

completa [Straub, 2002] [Richardson, 2006].

Debido al elevado carácter lipofílico de algunos de estos compuestos

pueden acumularse en biota y matrices sólidas, alcanzando niveles similares a

los de PCBs y DDT [Heneweer, 2005]. Estudios recientes sobre la distribución

medioambiental de los filtros UV incluyen su determinación en agua de lagos

[Rodil, 2009-C] [Moeder, 2010] [Haunschmidt, 2010], ríos [Liu, 2010], efluentes e

influentes de plantas de tratamiento de aguas residuales urbanas [Balmer, 2005]

[Rodil, 2008-A] [Richardson, 2006]; así como, en sedimentos [Jeon, 2006] [Rodil,

2008-C] y lodos [Plagellat, 2006] [Rodil, 2009-D] [Nieto, 2009] [Wick, 2010]. Incluso

se encontraron residuos de estos analitos en pescado [Balmer, 2005] [Zenker,

2008] [Fent, 2010-B]. En esta tesis, además del desarrollo de metodología de

preparación de muestra para la determinación de filtros solares en algunas de

las matrices anteriores, se aportan datos relativos a su presencia en aguas

superficiales y residuales, muestras de lodo y polvo de atmósferas interiores.


22

2.1. Muestras de agua

2.1.1. Niveles de filtros solares en agua

Las concentraciones de filtros UV encontrados en agua varían según la

época del año, siendo mayores en los meses de verano, cuando el uso de cremas

para la protección solar es máximo. No obstante, se han detectado filtros solares

en muestras tomadas en otoño e invierno [Negreira, 2009-A], lo que demuestra

que estos compuestos son empleados en muchos productos de cuidado

personal, y no sólo en protectores solares. En muestras de agua superficial, los

niveles detectados varían desde los 5 ng L-1 en río [Rodil, 2008-B] hasta los 4000

ng L-1 en lago [Rodil, 2009-C]. En agua residual tratada, los niveles van desde 1

ng L-1 [Trenholm, 2008] hasta 4000 ng L-1 [Kasprzyk-Hordern, 2008]; en cambio en

agua residual sin tratar, las concentraciones son mayores abarcando desde los

25 ng L-1 para la BP-2 [Kasprzyk-Hordern, 2008] hasta los 20000 ng L-1 para el

EHMC [Kupper, 2006] [Balmer, 2005]. La diferencia de concentraciones entre

efluente e influente proporciona una primera estimación de la eficacia de

eliminación de estos compuestos en las plantas de tratamiento de agua residual.

De todos modos, estos valores deben considerarse con suma precaución ya que

en la mayoría de los estudios se hace referencia sólo a las concentraciones en

disolución, sin tener en cuenta la fracción adsorbida en el lodo. A continuación,

se muestra la revisión bibliográfica relativa a la distribución de filtros solares en

muestras de agua.

Balmer y col. [Balmer, 2005] fueron pioneros en la determinación de filtros

solares de naturaleza orgánica en aguas residuales y superficiales de lagos

suizos. Los análisis de aguas superficiales realizados en el año 2002 ofrecen

valores de 28 ng L-1 para 4-MBC y 35 ng L-1 para BP-3, mientras que en el año

1998 los valores fueron superiores a 82 ng L-1 para 4-MBC y 125 ng L-1 para BP-

3. Las concentraciones en aguas residuales sin tratar se encontraron en el rango

de 600 a 6500 ng L-1 para 4-MBC, de 700 a 7800 ng L-1 para BP-3, de 500 a 19000


23

ng L-1 para EHMC y de 200 a 12000 ng L-1 para OCR. En aguas residuales

tratadas, los valores fueron: de 60 a 2700 ng L-1 para 4-MBC, de 10 a 700 ng L-1

para BP-3, de 10 a 100 ng L-1 para EHMC y de 10 a 270 ng L-1 para OCR. Las

concentraciones de filtros solares en efluentes de plantas de tratamiento de

aguas residuales fueron considerablemente más bajas que las correspondientes

en influentes, indicando una eliminación significativa de los mismos. Los

porcentajes medios de eliminación fueron de 18-82% para 4-MBC, 68-99% para

BP-3, 88-99% para OCR y 97-99% para EHMC. Los análisis de muestras

tomadas en diferentes días indicaron que la eliminación variaba de día a día,

resultado de condiciones cambiantes en las estaciones depuradoras (descarga

de agua, tormenta, tiempo de residencia).

Li y col. [Li, 2007] cuantificaron BP-3, 4-MBC, EHMC y OCR en diferentes

unidades de una estación depuradora de agua residual en China que recibe

efluente de otra planta de la misma área y, una vez allí, sufre el tratamiento

terciario. La planta estudiada consta de los siguientes tratamientos:

coagulación-floculación, microfiltración y ozonización. Las concentraciones

encontradas en el influente fueron de 97 a 722 ng L-1 para BP-3, de 475 a 2128 ng

L-1 para 4-MBC, de 54 a 116 ng L-1 para EHMC y de 34 a 153 ng L-1 para OCR.

Las eficacias totales de eliminación en la planta fueron de 28-31% para BP-3, 37-

40% para 4-MBC, 40-43% para EHMC y 36-38% para OCR [Li, 2007]. Estos

porcentajes son relativamente más bajos que los obtenidos por Balmer y col.

[Balmer, 2005].

Poiger y col. [Poiger, 2004] determinaron concentraciones de filtros solares

a diferentes profundidades en los lagos Zurich y Hüttnersee antes, durante y

después del verano. Las concentraciones medidas fueron mayores en verano y

más altas en el lago Hüttnersee que en el lago Zurich. Los valores encontrados

en el lago Zurich estuvieron en el rango de 2 a 22 ng L-1 para 4-MBC, de 2 a 25

ng L-1 para EHMC y de 2 a 4 ng L-1 para BP-3. Los valores encontrados en


24

Hürtnersee se situaron en el rango de 5 a 125 ng L-1 para BP-3, de 2 a 82 ng L-1

para 4-MBC, de 2 a 27 ng L-1 para OCR y de 2 a 19 ng L-1 para EHMC.

Giokas y col. [Giokas, 2004] cuantificaron residuos de filtros solares en

diferentes muestras de agua. En agua de mar encontraron 2 ng L-1 de BP-3; en

agua de piscina, 4 ng L-1 de BP-3, 7 ng L-1 de 4-MBC y 5 ng L-1 de EHMC.

Lambropoulou y col. [Lambropoulou, 2002] también cuantificaron filtros

solares en el ambiente acuático. Las concentraciones obtenidas en agua de

piscina para BP-3 fueron de 2400 a 3300 ng L-1 y de 2100 ng L-1 para el PABA.

En residuos de ducha, los resultados fueron de 8200 a 9900 ng L-1 para BP-3 y de

5300 a 6200 ng L-1 para PABA.

Rodil y col. [Rodil, 2008-A] determinaron, en diferentes tipos de muestras

acuosas, los filtros solares: BP-4, BP-3, 4-MBC, IAMC, OCR y EHPABA. Los

valores encontrados en tres muestras diferentes de influente, tomadas en

estaciones depuradoras situadas en el Noroeste de España, fueron de 237 a 1481

ng L-1 para BP-4, de 31 a 168 ng L-1 para BP-3, 122 ng L-1 para 4-MBC y 36 ng L-1

para OCR. En los correspondientes efluentes se encontraron valores de 376 a

1947 ng L-1 para BP-4, 16 ng L-1 para BP-3, de 23 a 122 ng L-1 para 4-MBC, 20 ng

L-1 para OCR y 59 ng L-1 para IAMC. En aguas de río encontraron valores de

849 ng L-1 para BP-4 y 27 ng L-1 para BP-3. En dos muestras diferentes de agua

de mar sólo se encontró BP-4 en concentraciones de 38 a 138 ng L-1. De su

estudio se deduce que la BP-4 es uno de los filtros solares más difíciles de

eliminar en las estaciones depuradoras de aguas residuales, y también uno de

los más móviles en el medio acuático.

Kupper y col. [Kupper, 2006] encontraron concentraciones en influente

(agua residual sin tratar) para EHMC y OCR de 20070 y 1680 ng L-1,

respectivamente.


25

Cuderman y col. [Cuderman, 2007] investigaron la presencia de diferentes

filtros UV en 21 muestras de aguas recreacionales de Eslovenia encontrando

mayoritariamente BP-3 a niveles de 114 ng L-1 en agua de río, en el rango de 32

a 85 ng L-1 en lagos y en el rango de 103 a 400 ng L-1 en piscinas. Los demás

filtros solares no fueron encontrados a niveles superiores al límite de detección

del método.

Kasprzyk-Hordern y col. [Kasprzyk-Hordern, 2008] determinaron 4 filtros

solares de la familia de las benzofenonas (BP-1, BP-2, BP-3 y BP-4) con

concentraciones entre 4 ng L-1 para BP-2 y 227 ng L-1 para BP-4 en muestras de

agua de río, de 1 ng L-1 (BP-2) a 4309 ng L-1 (BP-4) en efluente y de 25 ng L-1 (BP-

2) a 5790 ng L-1 (BP-4) en influente.

Moeder y col. [Moeder, 2010] analizaron muestras de agua procedentes de

una planta de tratamiento de Liepzig y de un lago próximo a una zona de

recreo, utilizando diferentes métodos para la determinación de filtros solares.

Las concentraciones obtenidas fueron de 113 ng L-1 para 4-MBC, 385 ng L-1 para

BP-3, 362 ng L-1 para EHMC y de 440 ng L-1 para OCR en el agua de la planta de

tratamiento. Para la muestra de agua del lago se obtuvieron concentraciones de

2472 ng L-1 para 4-MBC, 62 ng L-1 para BP-3, 3686 ng L-1 para OCR y 150 ng L-1

para EHMC.

Haunschmidt y col. [Haunschmidt, 2010] tomaron muestras de agua en

lagos en los que se practican actividades recreativas, obteniendo

concentraciones de 32 a 40 ng L-1 para BP-3 y de 1400 a 1710 ng L-1 para OCR,

mientras que EHS, HMS, 4-MBC, EHPABA no fueron detectados.

Pedrouzo y col. [Pedrouzo, 2009] determinaron BP-1, BP-8, BP-3, OCR y

EHPABA en muestras de agua residual procedentes de Cataluña. Los valores

en efluente variaron de 20 a 100 ng L-1 para BP-3, 19 ng L-1 para EHPABA, 11 ng

L-1 para BP-1 y en influente, de 11 a 286 ng L-1 para BP-3, 103 ng L-1 para


26

EHPABA y de 47 a 155 ng L-1 para BP-1. En un trabajo posterior [Pedrouzo,

2010], los mismos autores determinaron BP-8, BP-3, OCR y EHPABA en

muestras de agua superficial y residuales. En agua de río sólo detectaron BP-3

con valores desde 6 a 28 ng L-1. En efluente, las concentraciones encontradas

fueron de 25 y 55 ng L-1 para EHPABA y BP-8, respectivamente. En cambio, en

influente todos los compuestos fueron detectados con concentraciones en el

rango de 59 a 185 ng L-1 para BP-8, de 75 a 127 ng L-1 para BP-3, 129 ng L-1 para

OCR y de 55 ng L-1 para EHPABA.

Tarazona y col. [Tarazona, 2010] desarrollaron un método para la

determinación de benzofenonas hidroxiladas en 3 muestras de agua de mar. Las

concentraciones encontradas fueron de 280 ng L-1 para BP-1 y de 1340 a 3300 ng

L-1 para BP-3.

Wick y col. [Wick, 2010] determinaron 4 benzofenonas (BP-1, BP-2, BP-3 y

BP-4) en muestras acuosas. Las concentraciones medidas fueron de 2 ng L-1 (BP-

1) a 1980 ng L-1 (BP-4) en agua superficial, de 12 ng L-1 (BP-1) a 572 ng L-1 (BP-4)

en efluente y de 35 ng L-1 (BP-2) a 5130 ng L-1 (BP-4) en influente.

Liu y col. [Liu, 2010] determinaron 4 filtros solares (EHS, BP-3, 4-MBC y

OCR) en agua de río. Las concentraciones encontradas oscilaron entre 8 ng L-1

para EHS hasta 59 ng L-1 para la BP-3.

A continuación, se muestra una Tabla resumen (Tabla 7) donde se

recogen las concentraciones de filtros solares encontrados en la bibliografía para

diferentes muestras de agua superficial (agua de río, lago, mar y piscina) y agua

residual tratada (efluente) y sin tratar (influente). En aquellos trabajos en los

que se procesan varias muestras del mismo tipo, se indican las concentraciones

mínimas y máximas encontradas.


27

Tabla 7: Niveles de filtros solares en muestras acuosas.

Tipo de

agua Compuestos detectados Concentración (ng L-1) Referencia

Río

BP-1 47 [Jeon, 2006]

BP-3 23 [Kawaguchi, 2006]

BP-3 114 [Cuderman, 2007]

BP-3 14 [Kawaguchi, 2008-A]

BP-3 85 [Okanouchi,2008]

BP-4, BP-3 27 (BP-3)-849 (BP-4) [Rodil, 2008-A]

EHPABA 5 [Rodil, 2008-B]

BP-1, BP-2, BP-3, BP-4 4 (BP-2)-227 (BP-4) [Kasprzyk-Hordern,

2008]

BP-3 6-28 [Pedrouzo, 2010]

BP-1, BP-2, BP-3, BP-4 1 (BP-1)-51 (BP-4) [Wick, 2010]

EHS, BP-3, 4-MBC, OCR 8 (EHS)-59 (BP-3) [Liu, 2010]

Arroyo BP-1, BP-2, BP-3, BP-4 2 (BP-1)-1980 (BP-4) [Wick, 2010]

Lago

4-MBC, EHMC, BP-3 2-125 (BP-3) [Poiger, 2004]

4-MBC, BP-3 28 (4-MBC)-125 (BP-3) [Balmer, 2005]

EHS, HMS, BP-3,

4-MBC, IAMC, EHMC,

EHPABA, OCR

2 (EHPABA)-250 (OCR) [Rodil, 2008-B]

EHS, BP-3, 4-MBC, IAMC,

EHMC, EHPABA, OCR 40 (BP-3)-4381 (OCR) [Rodil, 2009-C]

BP-3 32-85 [Cuderman, 2007]

4-MBC, BP-3, EHMC, OCR 62 (BP-3)-3686 (OCR) [Moeder, 2010]

BP-3, OCR 32 (BP-3)-1710 (OCR) [Haunschmidt, 2010]

Mar

BP-3 2 [Giokas, 2004]

BP-4 38-138 [Rodil, 2008-A]

BP-3, BP-1 280 (BP-1)-3300 (BP-3) [Tarazona, 2010]

Piscina

BP-3, PABA 2100 (PABA)-3300 (BP-3) [Lambropoulou, 2002]

4-MBC, EHMC, BP-3 4 (BP-3)-7 (4-MBC) [Giokas, 2004]

BP-3 103-400 [Cuderman, 2007]

IAMC 700 [Vidal, 2010]


28

Tabla 7 cont.: Niveles de filtros solares en muestras acuosas.

Tipo de

agua Compuestos detectados Concentración (ng L-1) Referencia

Residual

tratada

(efluente)

4-MBC, BP-3, EHMC, OCR 10 (OCR)-2700 (4-MBC) [Balmer, 2005]

BP-3 1-13 [Trenholm, 2008]

BP-4, BP-3, 4-MBC,

OCR, IAMC 16 (BP-3)-1947 (BP-4) [Rodil, 2008-A]

EHS, HMS, BP-3,

4-MBC, IAMC, EHMC,

EHPABA, OCR

2 (EHPABA)-54 (BP-3) [Rodil, 2008-B]

BP-3, 4-MBC, OCR,

EHMC, EHS 32 (EHS)-899 (EHMC) [Rodil, 2009-E]

BP-3, 4-MBC, OCR 18 (BP-3)-179(OCR) [Rodil, 2009-C]

BP-1, BP-3, EHPABA 11 (BP-1)-100 (BP-3) [Pedrouzo, 2009]

BP-8, EHPABA 25 (EHPABA)-55 (BP-8) [Pedrouzo, 2010]


2008]

4-MBC, BP-3, EHMC, OCR 113 (4-MBC)-440 (OCR) [Moeder, 2010]


Residual

sin tratar

(influente)

4-MBC, BP-3, EHMC, OCR 200 (OCR)-19000

(EHMC) [Balmer, 2005]

EHMC, OCR 1680 (OCR)-20070

(EHMC) [Kupper, 2006]

4-MBC, BP-3, EHMC, OCR 34 (OCR)-2128 (4-MBC) [Li, 2007]


2008]

BP-3 300-2300 [Trenholm, 2008]

BP-4, BP-3, 4-MBC, OCR 31 (BP-3)-1481 (BP-4) [Rodil, 2008-A]

BP-3, 4-MBC, OCR,

IAMC, EHMC, EHS 69 (IAMC)-899 (EHMC) [Rodil, 2009-E]

BP-3, IAMC, 4-MBC,

OCR, EHMC, EHS 66 (IAMC)-5322 (OCR) [Rodil, 2009-C]

BP-1, BP-3, EHPABA 11 (BP-3)-286 (BP-3) [Pedrouzo, 2009]

BP-1, BP-3, OCR, EHPABA 55 (EHPABA)-185 (BP-1) [Pedrouzo, 2010]



29

En general, BP-3, 4-MBC, EHMC y OCR son los filtros solares

considerados en la mayoría de los estudios. Además, en los últimos años se

incluyen la BP-4 y los metabolitos de la BP-3. En lo relativo a su

comportamiento en estaciones depuradoras, BP-4 y 4-MBC son las especies que

presentan los porcentajes de eliminación más bajos, considerando sus

concentraciones en disolución a la entrada y salida de las plantas depuradoras.

En esta Tesis doctoral se aportan datos de distribución de filtros solares en río,

piscina y agua residual [Negreira, 2009-A; Negreira, 2009-B; Negreira, 2010-A],

siendo el valor más alto encontrado de 26000 ng L-1 para el OCR en agua de

piscina [Negreira, 2010-A].

2.1.2. Toxicidad y reacciones de transformación de los

filtros solares en el medio acuoso

La presencia de filtros UV en el medio acuático se ha relacionado con

alteraciones encontradas en los peces tales como: inducción de vitelogenina

(proteína de la yema del huevo, importante para la reproducción), alteraciones

de las gónadas, disminución de la fertilidad y de la tasa reproductiva y la

feminización de las características sexuales de los machos [Díaz-Cruz, 2009].

Existen estudios in vivo e in vitro que evidencian este problema y sugieren un

riesgo potencial para la salud humana. Además estos compuestos son

bioacumulativos y capaces de penetrar la barrera cutánea. Así se ha detectado

la presencia de BP-3 y EHMC en leche humana [Hany, 1995].

Algunos filtros solares presentan mayor o menor actividad disruptora

como el EHMC [Kunz, 2006-A; Kunz, 2006-B] [Seidlová-Wuttke, 2006] [Straub,

2002] [Díaz-Cruz, 2009], el 4-MBC [Maerkel, 2005] [Seidlová-Wuttke, 2006]

[Schlumpf, 2008-A; Schlumpf, 2008-B] [Díaz-Cruz, 2009], la BP-4 [Fent, 2010-A], la

BP-3 [Díaz-Cruz, 2009], la BP-1 [Kunz, 2006-B] [Fent, 2010-A] y la BP-2 [Kunz,

2006-B] [Fent, 2010-A]; de los cuales, el EHMC, 4-MBC y BP-3 ya han sido

englobados en el grupo de los productos químicos considerados como

estrogénicos [Díaz-Cruz, 2009].


30

Los estudios de toxicidad in vivo se han realizado administrando filtros

UV en la comida de roedores a la generación de los padres (antes del

apareamiento, durante el embarazo y la lactancia) y a las crías hasta la edad

adulta. Los machos recién nacidos, expuestos a filtros UV, exhibían alteraciones

en el crecimiento de la próstata y las hembras en la expresión de los genes del

útero [Díaz-Cruz, 2009] [Schlumpf, 2008]. También se apreciaron interacciones

del 4-MBC con el tiroides, con el mARN y los niveles de proteínas en el útero,

próstata y cerebro de los adultos [Schlumpf, 2008]. Además, el PABA y sus

derivados, pueden incrementar la citotoxicidad bacteriana e interactuarían con

el ADN tras la radiación UV potenciando la fotocarcinogénesis [Debuys, 2000].

Una línea de investigación de gran actualidad, en relación con la presencia

de contaminantes emergentes en el medio acuático, se centra en evaluar sus

posibles reacciones de transformación, en la identificación de los productos que

se generan (by-products) y en la estimación de su estabilidad, persistencia y

toxicidad en el medio acuático [Richardson, 2003]. Un porcentaje importante de

estas reacciones son de tipo fotoquímico o redox. Las segundas, normalmente

implican procesos de oxidación de los analitos por diversos agentes

introducidos de forma intencionada, o no, en el medio acuático. Entre ellos, el

más habitual es el cloro libre (combinación de ácido hipocloroso e hipoclorito

sódico). En algunos casos los productos de cuidado personal reaccionan con el

cloro libre produciendo especies volátiles fácilmente eliminables. Este tipo de

reacciones pueden tener interés a la hora de proponer tratamientos terciarios,

que mejoren el porcentaje de eliminación de estos compuestos en las estaciones

depuradoras de aguas residuales [Huber, 2005]. Sin embargo, en ocasiones, los

subproductos formados en los procesos de oxidación pueden resultar más

persistentes y tóxicos que los analitos de partida; además, algunas de estas

reacciones pueden ser muy favorables desde el punto de vista cinético [Gallard,

2002].


31

La estabilidad de los filtros solares en presencia de luz y de agentes

oxidantes, empleados en procesos de desinfección de agua potable o residual,

es un aspecto poco estudiado en la bibliografía; no obstante, atendiendo a la

presencia de grupos fenólicos o amino en su estructura parece factible que

puedan experimentar reacciones de fotodescomposición y/o oxidación. Se han

encontrado dos trabajos en relación con su fotodegradación [Sakkas, 2003] [Rodil,

2009-A]. Sakkas y col. [Sakkas, 2003] se centraron en el éster octilado del PABA,

describiendo la formación de productos de metilación del grupo amino e

hidroxilaciones en el anillo aromático debido a reacciones de tipo fotoquímico.

Las reacciones siguieron una cinética de primer orden, y la vida media del filtro

UV se encontró entre 27 y 39 horas, considerando radiación solar natural y

distintos tipos de agua. Rodil y col. [Rodil, 2009-A] estudiaron la fotoestabilidad

de 6 filtros solares (OCR, BP-3, 4-MBC, IAMC, EHMC y EHPABA) en

disolución acuosa y demostraron que EHMC, IAMC y EHPABA se degradaban

siguiendo cinéticas de primer orden, con vidas medias comprendidas entre 20 y

59 horas. Estos autores consiguieron identificar dos derivados de la

fotodegradación del EHPABA como consecuencia de la pérdida de los grupos

metilo unidos al átomo de nitrógeno, sin embargo, no fueron capaces de

detectar ningún fotoproducto del IAMC ni del EHMC.

En cuanto a la estabilidad de los filtros solares en presencia de agentes

oxidantes, Sakkas y col. [Sakkas, 2003] han descrito la presencia de derivados

clorados del octil PABA en agua de piscina. Puesto que estos compuestos no

tienen ninguna aplicación comercial conocida, se supone que son el resultado

de la reacción entre el cloro existente en el agua y el éster del PABA, presente en

los protectores solares que usan los bañistas. En cualquier caso, en el artículo

anterior, no se aportan datos sobre las vidas medias ni sobre mecanismos de

formación de estas especies.

En esta Tesis doctoral, se estudió la estabilidad en agua clorada de tres

filtros UV: EHPABA, EHS y BP-3. Se demostró que la BP-3 y el EHPABA


32

presentaban una elevada reactividad en muestras de agua conteniendo

concentraciones de cloro libre similares a las utilizadas en agua de grifo o de

piscina [Negreira, 2008].

2.2. Lodos y sedimentos

Muchos filtros solares presentan elevados coeficientes de partición

octanol/agua (Kow), por lo que es de esperar su acumulación en sedimentos,

lodos y materia particulada. Los niveles detectados en sedimentos son bajos. En

general, no superan los 16 ng g-1 [Ricking, 2003] [Jeon, 2006] con excepción de

Rodil y col. [Rodil, 2008-C] que encontraron valores de 14 a 34 ng g-1 para

EHMC y de 61 a 93 ng g-1 para OCR. Sin embargo, los valores encontrados en

lodos son mucho más elevados, destacando la presencia de OCR (el más

lipofílico de los filtros solares determinado en muestras medioambientales,

logKow 7,5) que alcanza valores de 18000 ng g-1 [Plagellat, 2006]. La determinación

de filtros solares en lodos de depuradoras es un aspecto de interés para calcular

el riesgo de re-introducir estos compuestos en el medio terrestre debido al uso

de lodos como fertilizantes en agricultura, así como para calcular sus

porcentajes reales de eliminación en las estaciones depuradoras.

Plagellat y col. [Plagellat, 2006] determinaron filtros solares en lodos de

estaciones depuradoras de aguas residuales urbanas en Suiza, encontrando

concentraciones en invierno y verano de 1730 y 1820 ng g-1 para 4-MBC, 105 y

115 ng g-1 para EHMC, 4270 y 5410 ng g-1 para OCR, respectivamente.

Nieto y col. [Nieto, 2009] analizaron muestras de lodos de depuradora que

fueron previamente homogeneizadas, congeladas y liofilizadas utilizando el

sistema de secado por congelación. Se calculó la concentración de filtros solares

en distintas épocas del año obteniéndose valores para BP-3 entre 10 y 20 ng g-1,

entre 700 y 1842 ng g-1 para OCR y de 132 a 170 ng g-1 para EHPABA, referidos


33

a peso seco. En el estudio anterior no se contempla la determinación de 4-MBC

ni de EHMC.

Rodil y col. [Rodil, 2009-D] determinaron los niveles de varios filtros UV en

una muestra de lodos de una estación depuradora en Alemania, que recibe el

agua residual de una población de 100.000 habitantes, y en un material de

referencia (MR) de lodos obtenidos del IRMN (Centro Común de Investigación

de la Comisión Europea, Geel, Bélgica). En el primer caso, las mayores

concentraciones correspondieron a OCR, EHS y HMS; mientras que en el MR,

los niveles máximos corresponden a 4-MBC seguido de OCR.

Wick y col. [Wick 2010] determinaron 4 compuestos de la familia de las

benzofenonas (BP-1, BP-2, BP-3 y BP-4) en muestras de lodo recogidas en una

planta de tratamiento de aguas residuales provenientes de una población de

320.000 habitantes en Alemania. Las concentraciones encontradas fueron bajas

desde 5 ng g-1 para la BP-1 hasta 132 ng g-1 para la BP-3.

A continuación, se muestran dos tablas resumen donde se recogen las

concentraciones de filtros solares encontrados en muestras de sedimentos (Tabla

8) y lodos (Tabla 9). En conjunto, la información disponible es muy limitada en

comparación con el estudio de su distribución en muestras de agua.


34

Tabla 8: Niveles de filtros solares en sedimentos.

Tipo de muestra Localización Compuestos

determinados

Concentración

(ng g-1) Referencia

Sedimento

( río) Alemania

BP

EHMC

0,5-3

4 [Ricking, 2003]

Sedimento Korea

BP

BP-3

BP-1

BP-8

2-10

nd

nd

0,5-2

[Jeon, 2006]

Suelo Korea

BP

BP-3

BP-1

BP-8

0,03-17

0,03-4

nq

0,5-4

[Jeon, 2006]

Sedimento

(lago) Alemania

EHMC

OCR

14-34

61-93 [Rodil, 2008-C]

n.d., no detectado

n.q., no cuantificado


35

Tabla 9: Niveles de filtros solares en lodos.

Tipo de muestra Localización

Compuestos determinados

Concentración (ng g-1) Referencia

Lodo de aguas residuales

Suiza 4-MBC EHMC OCR

150-4980 10-390 320-18740

[Plagellat, 2006]


España

BP-3 OCR EHPABA BP-1 BP-8

10-20 700-1842 132-170 nd nd

[Nieto, 2009]

Lodo IRMN

-

BP-3 IAMC 4-MBC OCR EHPABA EHMC EHS HMS

6,6 5,0 3893 2479 1,4 127 49 22

[Rodil, 2009-D]


Alemania (Leipzig)

BP-3 IAMC 4-MBC OCR EHPABA EHMC EHS HMS

29 20 73 585 1,9 35 280 331

[Rodil, 2009-D]


Alemania

BP-1 BP-2 BP-3 BP-4

5,1 11 132 29

[Wick, 2010]

n.d., no detectado

En la bibliografía, no aparecen datos relativos a la distribución de filtros

solares en atmósferas interiores. No obstante, en esta tesis doctoral [Negreira,

2009-C] se analizaron varias muestras de polvo procedentes de viviendas y de

vehículos.


36

2.3. Biota

Balmer y col. [Balmer, 2005] determinaron algunos filtros solares de

naturaleza orgánica en diferentes tipos de pescado de agua dulce encontrando

valores de 44 a 166 ng g-1 para 4-MBC, de 66 a 123 ng g-1 para BP-3, 64 a 72 ng g-

1 para EHMC y 25 ng g-1 para OCR, referidos a materia grasa. En la misma

línea, Buser y col. [Buser, 2006] también determinaron 4-MBC y OCR en pescado

de ríos suizos. Las concentraciones encontradas se situaron en el rango de 50 a

2400 ng g-1 en pescado de río y de 20 a 170 ng g-1 en pescado de lago, ambos

datos referidos a materia grasa.

Meinerling y col. [Meinerling, 2006] determinaron BP-3, 4-MBC, OCR y

EHMC en pescado y encontraron concentraciones en el rango de 3 a 21 ng g-1

para la BP-3, 3 ng g-1 para 4-MBC, 4 ng g-1 para OCR y 6 ng g-1 para EHMC,

referidas a músculo de pescado liofilizado.

Zenker y col. [Zenker, 2008] validaron un método para la determinación de

7 filtros solares (BP-1, BP-2, BP-3, BP-4, EHMC, Et-PABA y 4-MBC) en muestras

de pescado recogidas en diferentes puntos del río Glatt y sus afluentes, cerca de

Zürich, Suiza. Sólo detectaron la presencia de EHMC con una concentración de

42 ng g-1 en la muestra perteneciente a la salida de un lago y 142 ng g-1 a la

salida de una planta de agua residual, ambas referidas a contenido graso. En la

misma línea, Fent y col. [Fent, 2010-B] también encontraron EHMC en pescado

perteneciente al mismo río con concentraciones en el intervalo de 49 a 129 ng g-1

referidas a contenido graso.

A continuación, se muestra una tabla resumen (Tabla 10) donde se

recogen las distintas concentraciones de filtros solares en pescado. En general,

la distribución y acumulación de filtros solares en biota, ha sido estudiada en

áreas geográficas muy concretas y los datos disponibles parecen todavía

insuficientes para evaluar su potencial de bioacumulación en las cadenas

tróficas.


37

Tabla 10: Niveles de filtros solares encontrados en pescado.

Tipo de pescado Localización Compuestos

detectados

Concentración

(ng g-1) Referencia

Pescado blanco

(Coregonus sp.)

Rutilo

(Rutilus rutilus)

Perca

(Perca fluviatilis)

Suiza

(lago)

4-MBC

BP-3

EHMC

OCR

44-166

66-123

64-72

25

[Balmer, 2005]

Trucha marrón

(Salmo trutta fario)

Suiza

(río)

4-MBC

OCR

50-2400

20-170 [Buser, 2006]

Trucha marrón

(S. trutta fario)

Suiza

(lago)

4-MBC

OCR

<20-170

nd [Buser, 2006]

Trucha arco iris Alemania

(río)

BP-3

4-MBC

OCR

EHMC

3-21

3

4

6

[Meinerling,

2006]

Bagre o cacho

(Leuciscus cephalus)

Lengüeta

(Barbus barbus)

Suiza

(río) EHMC

42-142 [Zenker, 2008]

49-129 [Fent, 2010-B]

n.d., no detectado


38

3. METODOLOGÍA ANALÍTICA

3.1. Preparación de muestra

La preparación de muestra es una etapa crítica en cualquier

procedimiento analítico. Suele incluir una serie de pasos como son: la extracción

de los analitos desde la matriz de la muestra, la concentración de los analitos

hasta un nivel medible, la eliminación de especies interferentes o la conversión

química de los analitos a otras formas más fácilmente detectables [Cela, 2002].

En esta Tesis Doctoral se determinaron varios filtros solares en distintas

matrices tanto sólidas (polvo y lodo), como acuosas, empleando diferentes

técnicas de preparación de muestra cuyo fundamento se describe a

continuación. También se presenta una revisión bibliográfica de las

metodologías usadas para la determinación de los compuestos objeto de

estudio.

3.1.1. Muestras de agua

Los métodos de preparación de muestras acuosas utilizados en esta Tesis

Doctoral han sido: extracción en fase sólida (SPE), microextracción en fase

sólida (SPME), microextracción líquido-líquido dispersiva (DLLME) y

microextracción en fase sólida con siliconas en formatos no comerciales.

3.1.1.1. Extracción en fase sólida (SPE)

La SPE es una técnica de uso extendido que permite transferir los analitos

de la muestra líquida a una fase sólida (retención) y recuperarlos

cuantitativamente con un disolvente adecuado (elución). En este proceso se

logra separar los analitos de los compuestos interferentes (purificación) y

concentrarlos en un volumen pequeño de disolvente. La SPE es la técnica de

referencia para la concentración de filtros solares en muestras acuosas,


39

encontrando numerosas aplicaciones en la bibliografía que se detallan a

continuación.

Poiger y col. [Poiger, 2004] cuantificaron cinco filtros solares en agua de río

y de baño. La preparación de la muestra fue llevada a cabo mediante SPE. El

volumen de muestra usado fue de 1 litro con adición de bencilcinamato, como

surrogado interno, a un nivel de 100 ng L-1. Las extracciones fueron llevadas a

cabo con columnas de vidrio reutilizables que contenían un adsorbente

macroporoso de poliestireno. Posteriormente, los analitos fueron cuantificados

por GC-MS, empleando un espectrómetro de masas equipado con un

analizador de sector magnético, bajo ionización de impacto electrónico (EI, 70

eV, 200 ºC). La columna usada fue de tipo apolar. Los límites de detección

alcanzados fueron de 2 ng L-1 y las recuperaciones obtenidas variaron entre 77%

para 4-MBC, 90% para EHMC, 64% para BP-3 y 57% para OCR.

Giokas y col. [Giokas, 2004] utilizaron la SPE para la extracción de filtros

solares en diferentes muestras de agua (piscina, residuos de ducha de hoteles).

La etapa de concentración se llevó a cabo sobre discos de C18 de 500 mg (47

mm de diámetro). El volumen de muestra fue de 500 mL. La elución se realizó

con acetato de etilo:diclorometano (1:1, v/v) recogiendo dos alícuotas de 5 mL

que fueron evaporadas y reconstituidas con 0,010 mL de hexano para la

posterior determinación por GC-MS, y con 0,050 mL de metanol para su

inyección en cromatografía líquida. La inyección de 3 µL de extracto en GC-MS

proporcionó recuperaciones en el rango de 95 a 99%, y límites de cuantificación

de 0,9 a 1,4 ng L-1. La determinación mediante cromatografía líquida, con

detector UV-VIS, alcanzó límites de cuantificación entre 8 y 24 ng L-1.

Sakkas y col. [Sakkas, 2003] usaron SPE para estudiar la presencia de

EHPABA en diferentes muestras acuosas. Las muestras (5 mL) fueron pasadas a

través de cartuchos C18. En la etapa de elución se empleó

diclorometano:acetato de etilo en proporción 1:1 recogiendo dos alícuotas de 3


40

mL que, después de combinadas, fueron evaporadas y redisueltas con 0,1 mL

de metanol. La determinación se realizó mediante cromatografía líquida. Las

recuperaciones obtenidas fueron de 98% para agua de mar, 101% para agua de

piscina y de 98% para agua destilada y el límite de detección se situó en 300 ng

L-1.

Balmer y col. [Balmer, 2005] determinaron tres filtros solares (BP-3, 4-MBC,

EHMC) en aguas residuales, superficiales y en pescado de lagos suizos. La

preparación de la muestra se realizó mediante SPE, empleando columnas de

vidrio reutilizables que contenían aproximadamente 10 mL de un polímero

adsorbente y macroporoso de poliestireno divinilbenzeno. El volumen de

muestra fue de 300 mL y el flujo a través de los cartuchos se ajustó a 10 mL min-

1. La elución se llevó a cabo con metanol:diclorometano y la limpieza posterior

de los extractos se realizó con sílica. Seguidamente, los extractos fueron

concentrados con corriente de nitrógeno y analizados mediante GC-MS. Las

recuperaciones se encontraron en el rango de 78 a 129% y los límites de

detección fueron de 2 ng L-1 para aguas superficiales y 10 ng L-1 para aguas

residuales.

Cuderman y col. [Cuderman, 2007] usaron la SPE como método de

concentración de diferentes filtros solares (HMS, 4-MBC, BP-3, OCR y EHMC)

en diversas muestras acuosas. Los cartuchos empleados fueron Strata X (60 mg).

El volumen de muestra fue de 500 mL ajustado a pH 3. Una vez finalizada la

etapa de concentración, se lavó el cartucho con 1,5 mL de agua que contenía 1%

de metanol para eliminar impurezas. Después del secado, se eluyeron los

analitos con acetato de etilo:diclorometano, en proporción 1:1 (v/v). Los

extractos (3 x 0,5 mL) fueron combinados y evaporados a sequedad bajo

corriente de nitrógeno, antes de su redisolución con 0,4 mL de tolueno. Las

recuperaciones obtenidas se encontraron en el rango de 82 a 98% para agua

desionizada y de 50 a 98% en agua superficial.


41

Li y col. [Li, 2007] determinaron BP-3, 4-MBC, EHMC y OCR en muestras

de agua residual usando SPE como método de concentración. Volúmenes de 1

litro de muestra fueron acidificados a pH 3 y se les adicionó un 1% de metanol,

v/v. Los cartuchos empleados fueron de C18, acondicionados previamente con

5 mL de acetato de etilo:diclorometano en proporción 1:1 (v/v), 5 mL de

metanol y 5 mL de agua desionizada. Los cartuchos se secaron durante 5

minutos aplicando vacío y se eluyeron con alícuotas de 5 ml de acetato de

etilo:diclorometano (1:1, v/v). Los extractos fueron evaporados y redisueltos

con 1 mL de n-hexano. La determinación posterior se llevó a cabo mediante GC-

MS. Las recuperaciones obtenidas se encontraron en el rango de 67 a 118% y el

límite de detección fue de 10 ng L-1.

Rodil y col. [Rodil, 2008-A] presentaron un nuevo método analítico basado

en SPE y LC-(ESI)MS/MS para la determinación de seis filtros solares (BP-4,

BP-3, 4-MBC, IAMC, EHPABA y OCR) en muestras acuosas. Los adsorbentes

considerados fueron Oasis HLB (60 y 200 mg), Sep-Pak Plus C18 (aprox. 360

mg) y Bond Elut Plexa (60 y 200 mg). De ellos, los que aportaron mejores

recuperaciones fueron los Oasis HLB (60 mg). Estos fueron acondicionados con

3 mL de metanol, 3 mL de agua pura y 3 mL de disolución tampón. El método

optimizado consistió en adicionar 20 mL de una disolución de tampón (2%

metanol, 50 mM tri-n-butilamina ajustada a pH 4,5 con ácido fórmico) a una

muestra de 200 mL (pH 2) de agua con objeto de obtener el par iónico de la BP-

4, facilitando así su retención en el cartucho de fase reversa. Una vez pasada la

muestra a través de los cartuchos se lavaron con 3 mL de tampón y 3 mL de

Milli-Q y, posteriormente, se secaron durante 30 min con nitrógeno. La elución

se hizo con metanol (3 x 10 mL) y los extractos combinados (30 mL aprox.)

fueron concentrados y ajustados a un volumen final de 1 mL con metanol:agua

(1:1). Se estudiaron también otros disolventes de elución como acetonitrilo y

acetona con recuperaciones más bajas. Esta metodología ofrece límites de

detección en el rango de 7 a 46 ng L-1 y recuperaciones en el rango de 63 a 108%.

El mismo método es propuesto para la determinación de filtros solares junto


42

con herbicidas, organofosforados usados como plastificantes y retardantes de

llama, y otros compuestos farmacéuticos pero, en este caso se emplearon los

cartuchos Oasis HLB de 200 mg [Rodil, 2009-B].

Trenholm y col. [Trenholm, 2008] desarrollaron un método para la

determinación de 24 contaminantes emergentes, entre ellos la BP, la BP-3 y el

alcanfor en aguas residuales empleando SPE, como técnica de concentración, y

GC-MS/MS, como técnica de determinación para el alcanfor y BP y LC-MS/MS

para la BP-3. Para la determinación por GC-MS/MS, 500 mL de muestra fueron

concentrados a través de cartuchos HLB (200 mg) previamente acondicionados

con 5 mL de diclorometano, 5 mL de metil-t-butil éter (MTBE), 5 mL de metanol

y 5 mL de agua ultrapura. Los surrogados adicionados fueron 0,25 µg de [13C6]-

o-fenilfenol, BHT-d24, dibutilftalato-d4 y 1 µg de [13C6]-vanilina. Después de

pasar la muestra a través de los cartuchos a un flujo de 15 mL/min, estos se

lavaron con 5 mL de agua ultrapura y se secaron bajo corriente de nitrógeno

durante 30 min. La elución se llevó a cabo con 5 mL de metanol:MTBE 10:90

(v/v), seguido de 5 mL de diclorometano. Los eluatos fueron combinados,

concentrados y reconstituidos con 0,5 mL de isooctano. El procedimiento de

SPE desarrollado para LC-MS/MS fue muy similar al de GC salvo algunas

diferencias: los cartuchos fueron acondicionados con 5 mL de MTBE, 5 mL de

metanol y 5 mL de agua. Los surrogados adicionados fueron 0,02 µg de [13C6]-

triclosán, [13C6]-simazina, [13C6]-atrazine y 0,1 µg de [13C6]-bisfenol A y [13C6]-o-

fenilfenol. La elución de los cartuchos se llevó a cabo con 5 mL de

metanol:MTBE 10:90 (v/v) y estos se concentraron a 500 µL. El método

proporciona límites en el rango de 1 a 50 ng L-1 en agua y recuperaciones entre

67 y 87%.

Kasprzyk-Hordern y col. [Kasprzyk-Hordern, 2008] desarrollaron un método

para la determinación de una gran variedad de compuestos, entre los cuales se

encuentran BP-1, BP-2, BP-3 y BP-4, en muestras de agua superficial y residual.

Los adsorbentes de SPE probados fueron Oasis HLB, MCX, MAX, WCX, WAX


43

(60 mg), Chormabond C18 (200 mg), Isolute ENV+ (100 mg) e Isolute HCX (200

mg). De todos ellos, los cartuchos Oasis MCX resultaron ser los más eficientes.

El volumen de muestra empleado fue de 1000 mL para agua superficial y 250

mL para agua residual. Las muestras se acidificaron a pH 2 y la elución tuvo

lugar con 2 mL de metanol y 2 mL de metanol con un 5% de amoníaco. El

extracto resultante se concentró y se reconstituyó con 0,5 mL de fase móvil,

para su posterior determinación mediante UPLC-(ESI)MS-MS. Los límites de

cuantificación del método se situaron en el rango de 0,1 ng L-1 (agua superficial)

a 35 ng L-1 (agua residual sin tratar) y las recuperaciones obtenidas variaron

desde 17 a 117%.

Pedrouzo y col. [Pedrouzo, 2009] presentan un método basado en SPE y

UPLC-(ESI)MS/MS para la determinación de varios productos de cuidado

personal, entre ellos 5 filtros solares (BP-1, BP-3, BP-8, OCR y EHPABA), en

muestras acuosas. La SPE fue llevada a cabo con 2 adsorbentes poliméricos:

Oasis HLB (500 mg) para agua residual y Bond Elut Plexa (200 mg) para agua

de río. Ambos adsorbentes fueron acondicionados con 5 mL de metanol y 2 mL

de milli-Q. Los volúmenes de muestra extraídos fueron de 100 mL para

influente, 250 mL para efluente y 500 mL para agua de río. Las muestras fueron

pasadas a través de los cartuchos a un flujo de 10-15 mL min-1. A continuación

se lavaron con agua conteniendo un 15% de metanol y se secaron durante 5

min, antes de ser eluidos con 5 mL de metanol y 5 mL de diclorometano. Los

extractos fueron concentrados a 3-4 mL y posteriormente diluidos a 5 mL con

Milli-Q para inyectar un volumen de 50 µL en el sistema cromatográfico. Las

recuperaciones obtenidas para los filtros solares en 500 mL de agua de río

oscilaron entre 46 y 97%. Para aguas residuales, las recuperaciones estuvieron

en el rango de 20 a 71% (250 mL de efluente) y de 27 a 86% (100 mL de

influente). Los límites de detección estuvieron en el rango de 1 a 4 ng L-1 para

río, de 3 a 10 ng L-1 para efluente y de 5 a 20 ng L-1 para influente.


44

Oliveira y col. [Oliveira, 2010] proponen un método automático para la

determinación de 3 filtros solares: BP-3, EHMC y HMS en muestras de agua de

piscina y de mar. La nueva metodología consiste en una SPE-on line con el

sistema cromatográfico, en este caso LC-UV/VIS. El sistema microanalítico

propuesto, usando multijeringas como unidad de propulsión, combina todas las

etapas de la SPE incluyendo la renovación del adsorbente para prevenir la

contaminación cruzada entre muestras, y el ajuste, después de la extracción, de

la composición del eluato para prevenir el ensanchamiento de la banda de

inyección en cabeza de columna. Con objeto de acelerar la separación en LC, se

usó una columna monolítica de C18 y se llevó a cabo una elución en modo

isocrático. Las recuperaciones logradas estuvieron en el rango de 51 a 140% y

los límites de detección variaron entren 0,81 y 3,2 µg L-1. El interés del método

anteriormente descrito se basa en su completo grado de automatización; sin

embargo, los límites de detección alcanzados son demasiado elevados para su

aplicación rutinaria a muestras de aguas superficiales.

A continuación, se muestra una tabla resumen con las aplicaciones más

significativas de SPE a la determinación de filtros UV en muestras de agua,

Tabla 11.


45

Tabla 11: Resumen de aplicaciones de SPE como técnica de concentración para la

determinación de filtros solares en muestras de agua.

Matriz Adsorbente Eluyente (volumen, mL) R.

(%) Referencia

Mar y piscina C18 Acetato de etilo:diclorometano,

1:1 (2 x 3 mL)

98-

101 [Sakkas, 2003]

Río y baño Poliestireno Metanol (5 mL),

diclorometano (10 mL)

57-

90 [Poiger, 2004]

Piscina,

residuos de

duchas

Discos C18

(500 mg)

Acetato de etilo:diclorometano,

1:1 (2 x 5 mL)

95-

99 [Giokas, 2004]

Residual Poliestireno-

divinilbenzeno Metanol:diclorometano

78-

129 [Balmer, 2005]

Río, lago,

piscina, mar

Strata X

(60 mg)


1:1 (3 x 0,5 mL)

50-

98

[Cuderman,

2007]

Residual C18

(200 mg)


1:1 (5 mL)

67-

118 [Li, 2007]

Residual HLB

(200 mg)

Metanol:MTBE, 10:90 (5 mL);

Diclorometano (5 mL)

67-

87

[Trenholm,

2008]

Superficial y

residual

Oasis MCX

(60 mg)

Metanol (2 mL), Metanol-

5%hidróxido amónico (2 mL)

17-

117

[Kasprzyk-

Hordern, 2008]

Río, mar

y residual

Oasis HLB

(60 mg) Metanol (3 x 10 mL)

63-

108 [Rodil, 2008-A]

Grifo, ría

y residual

Oasis HLB

(200 mg) Metanol (3 x 10 mL)

56-

132 [Rodil, 2009-B]

Residual

Río

Oasis HLB

(500 mg)

Bond Elut Plexa

(200 mg)

Metanol (5 mL),

diclorometano (5 mL)

46-

97

20-

86

[Pedrouzo,

2009]

Aunque SPE permite llevar a cabo en una única etapa la extracción y

concentración de los analitos, la cantidad de muestra a procesar es más elevada

que en el caso de las técnicas de microextracción, además el coste de los

adsorbentes empleados es considerable. Las técnicas más modernas de

preparación de muestra tienden hacia la simplificación y miniaturización del


46

proceso analítico, reduciendo tiempo, esfuerzo, cantidad de muestra y consumo

de disolventes orgánicos [Pawliszyn, 2002] [Kloskowski, 2007]. Como resultado de

la miniaturización de la LLE (extracción líquido-líquido) y de la SPE, surgieron

diferentes metodologías de microextracción. La mayoría de ellas han sido

aplicadas a la concentración de filtros solares en muestras de agua. A

continuación se revisan estas aplicaciones, prestando especial interés a las

modalidades empleadas en este estudio.

3.1.1.2. Técnicas basadas en la microextracción en fase sólida

Como resultado de la minituarización de la SPE, surgieron varias

técnicas de microextracción en fase sólida: la microextracción en fase sólida

(SPME) [Belardi, 1989] y otras más recientes como: la microextracción con barras

agitadoras (SBSE) [Baltussen, 1990] y la microextracción mediante sorbentes

empaquetados (MEPS) [Mohamed, 2010]. Además, en esta Tesis Doctoral se

propone el uso de siliconas de grado técnico, en formato no comercial, como

sustituyente de las barras agitadoras recubiertas con polidimetilsiloxano

(PDMS) empleadas en SBSE, para la extracción de filtros solares en muestras

acuosas. A continuación, se describe el fundamento de las técnicas empleadas

en esta memoria (SPME y siliconas), y se revisan sus aplicaciones, así como las

de SBSE y MEPS, a la determinación de filtros solares.

3.1.1.2.1. Microextracción en fase sólida (SPME)

La microextracción en fase sólida (SPME) es una técnica de preparación de

la muestra desarrollada en 1989 por Belardi y Pawliszyn [Belardi, 1989]

[Pawliszyn, 2002] y que se basa en la utilización de una fibra de sílice fundida,

recubierta con una fase estacionaria adsorbente/absorbente de naturaleza

polimérica. El dispositivo empleado puede verse en la figura siguiente (Fig. 6).

Los analitos presentes en la muestra, por lo general, no se extraen


47

cuantitativamente sobre la fibra, sino que simplemente se establece un

equilibrio entre ambas fases. La SPME no necesita de la utilización de

disolventes orgánicos, requiere una manipulación mínima de la muestra y es

aplicable tanto a compuestos volátiles como semivolátiles [Cámara, 2002] [Cela,

2002].

Cuerpo de la jeringa

Émbolo

Guía del émbolo

Muelle

Aguja de acero

Fibra de sílice

Séptum

Férrula

Pieza de acero fijación fibra

Cuerpo de la jeringa

Émbolo

Guía del émbolo

Muelle

Aguja de acero

Fibra de sílice

Séptum

Férrula


Muelle

Aguja de acero

Fibra de sílice

Séptum

Férrula


Muelle

Aguja de acero

Fibra de sílice

Séptum

Férrula


Figura 6. Esquema de un dispositivo comercial de SPME

La concentración de las muestras mediante SPME consta de dos etapas:

1 ª Extracción o muestreo. La fibra se pone en contacto con la muestra, o en

espacio de cabeza, situada en un vial cerrado con un séptum y una cápsula. Se

deja un tiempo determinado para que los analitos se absorban/adsorban en la

fibra, produciéndose el reparto de los mismos entre la muestra y el

recubrimiento de la fibra. A continuación, la fibra se retrae en la aguja

protectora.

2 ª Desorción. Inmediatamente después, la fibra se introduce en un inyector

de un instrumento analítico (GC o HPLC), donde los analitos son desorbidos

térmicamente o por disolución en la fase móvil, según la técnica de

determinación empleada. Esta etapa se lleva a cabo en 1-2 minutos.

Los modos de extracción empleados en SPME son los siguientes:

Extracción directa. En este modo, la fibra se introduce directamente

en la muestra líquida y los analitos son transportados directamente


48

desde la matriz de la muestra a la fase extractante. Está

recomendada para compuestos poco volátiles, en muestras con un

nivel bajo o moderado de interferencias.

Extracción en espacio de cabeza (HS). La fibra se expone al espacio de

cabeza de la muestra, de manera que los analitos son transportados

primero al espacio de cabeza y luego son concentrados en la fibra.

Así se protege la fase extractante de compuestos de alto peso

molecular y poco volátiles que no son de interés.

Extracción indirecta a través de una membrana protectora. Con esta

membrana se evita el deterioro de la fibra cuando se extraen

muestras complejas. Su utilización es adecuada cuando se requiere

determinar analitos con volatilidades demasiado bajas para su

extracción en el modo de espacio de cabeza. Se trata de la

modalidad menos utilizada dada su lenta cinética de extracción y la

posible competencia entre la membrana y el recubrimiento de la

fibra por los analitos.

A continuación, se presenta un esquema de SPME para los modos de

extracción directo y espacio de cabeza (Fig. 7).

Figura 7: Esquema del proceso de SPME-GC: (a) Inmersión directa; (b) Espacio de

cabeza; (c) Desorción térmica en GC.

La SPME es un proceso de equilibrio entre múltiples fases. Normalmente,

para simplificar la descripción del mismo, sólo se consideran tres fases: el

recubrimiento de la fibra, la fase gaseosa o espacio de cabeza y una matriz

homogénea como el agua pura. Mediante un sencillo balance de masas se


49

puede relacionar la cantidad total de analito en el vial de SPME con la masa

extraída por la fibra y la remanente en el espacio de cabeza y en la matriz, como

se indica en la siguiente ecuación:

ffsshhso VCVCVCVC (1)ffsshhso VCVCVCVC ffsshhso VCVCVCVC (1)

Co: Concentración inicial del analito en la matriz.

Cf∞: Concentración del analito en la fase estacionaria al alcanzar el

equilibrio.

Ch∞: Concentración del analito en el espacio de cabeza al alcanzar el

equilibrio.

Cs∞: Concentración del analito en la matriz al alcanzar el equilibrio.

Vf: Volumen del recubrimiento en la fibra de SPME.

Vh: Volumen del espacio de cabeza.

Vs: Volumen de la matriz.

Los coeficientes de reparto entre las tres fases son:

h

ffh C

CK

s

hhs C

CK

s

ffs C

CK(2) (3) (4)

Así, la cantidad de analito en el recubrimiento de la fibra (n = CfVf) se

puede expresar como:

shhsfhsfh

sofhsfh

VVKVKK

VCVKKn

(5)

Cuando se alcanza el equilibrio:

hsfhfs KKK (6)hsfhfs KKK (6)

Y, substituyendo en la ecuación anterior:

shhsffs

soffs

VVKVK

VCVKn

(7)

Cuando no existe espacio de cabeza, es decir, la muestra ocupa por

completo el vial, el término KhsVs es despreciable:

sffs

soffs

VVK

VCVKn

(8)


50

Si además estipulamos que el volumen de la muesta acuosa (Vs) es mucho

mayor que el volumen de recubrimento de la fibra (Vf), se puede cumplir que

KfsVf << Vs, por eso:

n = KfsVfCo (9)

De acuerdo con la expresión anterior, una vez alcanzado el equilibrio, la

cantidad de analito extraído es independiente del volumen de la muestra.

La constante Kfs controla la eficacia y la selectividad del proceso de

extracción. Su valor es función de las características del recubrimiento, de la

muestra y de las propiedades físico-químicas de cada analito en cuestión.

Actualmente, se comercializan varios tipos de fases estacionarias con diferentes

espesores y polaridades, que muestran mayor o menor afinidad por diferentes

analitos. En la Tabla 12 se indican algunas de las fases más empleadas. En breve,

está previsto la comercialización de fibras recubiertas con C18, más adecuadas

para su desorción con disolventes orgánicos que las hasta ahora disponibles.

Tabla 12: Recubrimentos más utilizados en SPME.

Fase estacionaria Espesor (µm) Polaridad Tª Máxima (ºC)

PDMS

100

Apolar

280

30 280

7 340

PDMS/DVB 65 Semipolar 270

PA 85 Polar 320

CAR/PDMS 75 Semipolar 320

DVB/CAR/PDMS 50/30 Semipolar 270

PEG 60 Polar 250

Carbopack 15 Semipolar 340

Los trabajos que proponen el uso de SPME para la concentración de filtros

solares son los siguientes:

Felix y col. [Felix, 1998] aplicaron la SPME, en combinación GC-MS, a la

determinación de BP-3 y sus metabolitos (BP-1, BP-8) en orina. Compararon tres

fibras diferentes: una de 30 µm de polidimetilsiloxano (PDMS), de 85 µm de


51

poliacrilato (PA) y de 65 µm de carbowax-divinilbenceno (CW-DVB). La fibra

de CW-DVB (actualmente descatalogada) fue la que proporcionó una mayor

eficacia de extracción. Los parámetros evaluados fueron: el rango lineal, que se

encontró entre 10 y 500 ng mL-1, y los límites de cuantificación que fueron de

1667 ng L-1, 10000 ng L-1 y 3704 ng L-1 para BP-3, BP-1 y BP-8 respectivamente.

Lambropoulou y col. [Lambropoulou, 2002] desarrollaron un método para la

determinación de BP-3 y EHPABA en agua de baño y de piscina. La

determinación se llevó a cabo usando cromatografía de gases con detector de

ionización de llama y también, espectrometría de masas. Las fibras más eficaces

fueron PDMS de 100 µm y PA de 85 µm. La SPME fue realizada en modo

directo y en espacio de cabeza. Las recuperaciones relativas (respecto a agua

ultrapura) obtenidas fueron de 82% a 98% con límites de cuantificación de 1000

a 4100 ng L-1 para GC-FID y 730 a 4430 ng L-1 para GC-MS.

Evidentemente, los valores anteriores son claramente insuficientes para

la determinación de filtros UV en muestras de agua superficial. En esta tesis, se

ha profundizado en la aplicabilidad de SPME a la determinación de estos

compuestos en muestras de agua, centrando el estudio en compuestos fenólicos

del grupo de los salicilatos y las benzofenonas, combinando extracción y

derivatización con objeto de mejorar las características analíticas del método

resultante [Negreira, 2009-A].

Posteriormente, Liu y col. [Liu, 2010] optimizaron un método de SPME

para la determinación de cuatro filtros UV (BP-3, EHS, 4-MBC y OCR), junto

con fragancias, en agua de río. La fibra empleada fue la de PDMS y ésta fue

sumergida directamente en 3 mL de muestra ajustada a pH 7 y conteniendo un

10% de NaCl. El muestreo se realizó a 24ºC durante 90 min., posteriormente la

fibra se desorbió en el inyector de un sistema GC-MS durante 7 min. Los límites

de detección y de cuantificación estuvieron en el rango de 0,2 a 2,0 ng L-1 y de

0,7 a 6,7 ng L-1, respectivamente.


52

3.1.1.2.2. Microextracción con barras agitadoras (SBSE)

La extracción en fase sólida con barras agitadoras (stir-bar soptive

extraction, SBSE) tiene el mismo fundamento que la SPME. La principal

diferencia entre ambas técnicas es el diseño del sistema de extracción y la

cantidad de sorbente utilizado. La barra agitadora (Twister) está recubierta con

PDMS y, dado que su superficie y volumen son superiores a los de la fibra de

SPME, la cantidad de recubrimiento es superior; en consecuencia, la eficacia de

extracción es también mayor que la de la SPME. La muestra líquida se agita con

la barra durante un cierto tiempo, tras el cual ésta se retira y se desorbe. La

desorción puede realizarse con un disolvente orgánico o, para alcanzar límites

de cuantificación más bajos, se puede llevar a cabo térmicamente en

combinación con GC-MS. En este caso, los tiempos de desorción típicos son de

10 minutos, por lo que los analitos deben reenfocarse en cabeza de columna,

antes de su separación. Esto se lleva a cabo, habitualmente, con un sistema de

enfriamiento criogénico. Por otro lado, a diferencia de SPME, con SBSE

normalmente se alcanzan recuperaciones cuantitativas para compuestos de

polaridades medias y bajas.

Fundamento teórico

La base teórica de SBSE es exactamente la misma que la de SPME. La

principal diferencia es que se utiliza un volumen de polímero mucho mayor

que en caso de las fibras convencionales de SPME, entre 20 y 150 µL de PDMS

dependiendo de las aplicaciones, frente a los 0,5 µL de una fibra de PDMS

comercial. Así, se puede aumentar el rendimiento del proceso.

La eficacia de extracción, E, puede calcularse como:

100100000

VC

VC

n

nE fff (10)


53

Siendo nf la cantidad de analito retenida sobre el polímero, en

condiciones de equilibrio, y n0 la cantidad total de compuesto presente en la

muestra.

100

ssff

ff

VCVC

VCE (11)

fsfsf

s

fs

sf

f

KKV

V

K

VV

VE

1

1100

1

1100 (12)

La eficacia de extracción está relacionada con el parámetro β, cociente

entre el volumen de muestra y de PDMS, y la constante de partición Kfs. En

SBSE, es habitual usar las Kow como estimado de los valores de Kfs.

Aplicaciones

Los primeros trabajos que aplican SBSE para la extracción y

concentración de filtros solares están enfocados a la determinación de

benzofenonas, entre las que figuran la BP, BP-1 y BP-3 en muestras de orina

[Kawaguchi, 2008-B] y de agua [Kawaguchi, 2008-A; Kawaguchi, 2006]. El método

propuesto consiste en introducir una barra cubierta con PDMS en 10 mL de

muestra y agitar durante 120 min a temperatura ambiente (25ºC). A

continuación, el Twister se pasa al sistema de desorción térmica conectado on-

line con GC-MS [Kawaguchi, 2008-A; Kawaguchi, 2006]. En uno de esos trabajos,

realizan la derivatización in-situ de las benzofenonas con anhídrido acético,

antes de la desorción térmica (TD) en GC-MS [Kawaguchi, 2008-A]. Los límites

de detección alcanzados se recogen en la Tabla 13.

Rodil y col. [Rodil, 2008-B] aplicaron SBSE en combinación con TD y GC-MS

a la determinación de filtros solares (EHS, HMS, IAMC, 4-MBC, BP-3, EHMC,

EHPABA, OCR) en muestras acuosas. La barra agitadora cubierta con

polidimetilsiloxano (PDMS) fue introducida en 20 mL de agua a pH 2 (10% de


54

metanol) y agitada a 1000 rpm, durante 180 min a temperatura ambiente. La

desorción se llevó a cabo a 250 ºC durante 15 min. Los límites de cuantificación

se situaron en el rango de 0,6 a 26 ng L-1, los coeficientes de determinación en el

rango de 0,9941 a 0,9999, la eficacia de extracción fue superior al 63% y las

desviaciones estándar relativas inferiores al 16%.

Pedrouzo y col. [Pedrouzo, 2010] usaron también SBSE y cromatografía

líquida (UPLC-(ESI)MS-MS) para la determinación de 4 filtros solares (BP-8, BP-

3, OCR y EHPABA) en muestras de agua. Tomando un volumen de muestra de

50 mL, y considerando 1 mL de acetonitrilo para la desorción de los analitos, se

alcanzaron eficacias de extracción entre 31 y 97% y LODs comprendidos entre

2,5 y 10 ng L-1.

En esta Tesis doctoral, se propone, como substituyente de las barras

agitadoras, el uso de siliconas en un formato no comercial, para la extracción y

concentración de filtros solares en muestras de agua. Es una modalidad de

microextracción en fase sólida con fundamento idéntico a la SPME

convencional, utilizando un absorbente líquido, en la que, en lugar de utilizar el

dispositivo comercial de SPME (Supelco) o de SBSE (Gerstel), se emplean

siliconas de grado técnico, en distintos formatos: láminas, tubo, cuerda… El

polímero, en estos formatos, puede ser cortado por el usuario en unas

dimensiones adecuadas para optimizar el rendimiento de la extracción para

cada analito, volumen de muestra y volumen de disolvente empleado en la

etapa de desorción.

Debido a su bajo coste (normalmente inferior a 0,1 €), la fase extractante

(silicona) puede ser utilizada como material desechable, con lo que se evitan los

problemas de contaminación cruzada, o el “efecto memoria” de las fibras de

SPME y los Twisters usados en SBSE.


55

Los primeros trabajos en los que se utilizaron las siliconas, en formato de

cuerda, fueron publicados por el grupo de investigación de Popp y col.

[Montero, 2004] [Popp, 2004] para la determinación de PCBs y PAHS en muestras

de agua. Posteriormente, se han desarrollado aplicaciones a otros compuestos

[Paschke, 2006] [Bicchi, 2007] [Shellin, 2007] [van Pinxteren, 2010]. Sin embargo, no

existen trabajos en los que usen siliconas de grado técnico, en formatos no

comerciales, para la determinación de filtros solares, salvo el presentado en esta

Tesis.

3.1.1.2.3. Microextracción mediante sorbentes empaquetados (MEPS)

Moeder y col. [Moeder, 2010] describieron el uso de MEPS para la extracción

automatizada de 4 filtros solares (BP-3, 4-MBC, EHMC y OCR) en muestras de

agua. En esta técnica, el material adsorbente (C18) se encuentra empaquetado

entre la aguja y el cuerpo de una jeringa de 100 µL, instalada en el autosampler

del equipo de GC-MS. La muestra (0,8 mL) se aspiró a través del material

adsorbente (1 mg de C18) y a continuación se pasó a desecho. Una vez secado el

adsorbente mediante la aspiración de aire, los analitos se desorben con dos

fracciones de 25 µL de acetato de etilo que se tranfirieron directamente al

inyector de grandes volúmenes del sistema GC-MS. Los LODs alcanzados

estuvieron en el rango de 34 a 87 ng L-1 y las recuperaciones obtenidas variaron

de 61 a 114%.

En la Tabla 13 se resumen las condiciones experimentales, las eficacias de

extracción, y los LOQ de las técnicas de microextracción en fase sólida

empleadas en la bibliografía para la determinación de filtros solares.


56

Tabla 13: Resumen de aplicaciones de las técnicas de microextracción en fase sólida a la

determinación de filtros UV.

Matriz Condiciones EF.

(%) Det.

LOQ

(ng L-1) Referencia

SPME

Orina CW-DVB, Vm: 4 mL - GC-MS 1667-

10000 [Felix, 1998]

Agua de piscina

y baño

PDMS, PA,

directa (0% NaCl)

y HS (30% NaCl),

Vm: 5 mL, 45min

- GC-MS 730-

4430

[Lambropoulou,

2002]

Agua de río

PDMS, inmersión,

Vm: 3 mL (pH 7, 10%

NaCl), 24ºC, 90 min

- GC-MS 0,7-6,7 [Liu, 2010]

SBSE

Agua de río SBSE (24 µL),

10 mL, 120 min

98-

115

106-

128

TD-GC-

MS

2-5

2-10

[Kawaguchi,

2006]

[Kawaguchi,

2008-A]

Agua de río,

lago y residual

SBSE (24 µL), 20 mL (pH

2, 10%metanol), 180 min

63-

122

TD-GC-

MS 0,6-26 [Rodil, 2008-B]

Agua de río,

efluente e

influente

SBSE (24 µL),

50 mL (pH 5), 180 min

Desorción: 1 mL ACN,

30ºC, 15 min

31-

97

UPLC-

MS/MS 8-33 [Pedrouzo, 2010]

MEPS

Agua de lago

y efluente

Vm: 0,8 mL

Desorción: 2 x 25 µL,

acetato de etilo

61-

114

PTV-GC-

MS 113-290 [Moeder, 2010]


57

3.1.1.3. Técnicas basadas en la microextracción en fase líquida

Tanto las modalidades de microextracción líquido-líquido directa,

(microextracción con gota suspendida, SDME y microextracción líquido-líquido

dispersiva, DLLME), como aquellas en las que existe una membrana entre

ambas fases (microextracción líquido-líquido con membranas no porosas,

MALLE y microextracción en fase líquida con fibra hueca, HF-LPME) han sido

evaluadas para la concentración de filtros solares en matrices acuosas. A

continuación, se resumen las aplicaciones de éstas técnicas a la determinación

de filtros solares, prestando especial atención a la DLLME, que es la modalidad

utilizada en esta Tesis Doctoral [Negreira, 2010-A].

3.1.1.3.1. Microextracción con gota suspendida (SDME)

Vidal y col. [Vidal, 2007] presentaron la primera aplicación de SDME a la

determinación de BP-3 en muestras de orina. En SDME, la fase extractante es

una gota de unos pocos microlitros de un disolvente inmiscible con la muestra,

suspendida en el extremo de una microjeringa, y normalmente expuesta

directamente a la muestra. Como alternativa, para analitos volátiles, la gota

puede ser suspendida en espacio de cabeza del vial que contiene la muestra. En

la aplicación desarrollada por Vidal y col. [Vidal, 2007] emplearon un líquido

iónico (IL) en lugar de un disolvente orgánico. Después del proceso de

microextracción, la fase extractante fue inyectada en un equipo de HPLC. Las

condiciones experimentales óptimas encontradas fueron: NaCl 13% (p/v), 25

min de tiempo de extracción, 900 rpm como velocidad de agitación y pH 2. El

volumen de gota fue de 5 µl y el de muestra de 10 mL. El método se usó para

determinar las concentraciones de BP-3 presentes en orina humana, después de

la aplicación de protectores solares que contienen este compuesto. El límite de

detección fue de 1300 ng L-1 y la repetibilidad del método, expresada como RSD

fue del 6% (n=8). Posteriormente, la técnica de SDME fue usada también para la


58

determinación de benzofenonas en muestras de aguas por Okanouchi y col.

[Okanouchi, 2008] y Vidal y col. [Vidal, 2010].

3.1.1.3.2. Microextracción líquido-líquido con membranas

La microextracción líquido-líquido con membranas no porosas, conocida

por las siglas MALLE correspondientes a “non-porous membrane-assisted liquid-

liquid extraction”, fue aplicada por Rodil y col. [Rodil, 2009-C] para la

determinación de filtros solares (BP-3, IAMC, 4-MBC, OCR, EHPABA, EHMC,

EHS y HMS) en muestras de agua mediante LC-(APPI)MS/MS. Utilizaron

membranas de 2 cm de polietileno de baja densidad (LDPE), rellenas con 100 µL

de propanol. La membrana fue sumergida en 15 mL de muestra, conteniendo

un 10% de metanol, con agitación durante 120 min a 40 ºC. A continuación, la

fase orgánica fue retirada de la membrana y transferida a un inserto. El método

optimizado proporcionó recuperaciones de 60% (BP-3) a 104% (EHS) y límites

de cuantificación entre 3 ng L-1 (EHPABA) y 53 ng L-1 (EHMC).

3.1.1.3.3. Microextracción en fase líquida con fibra hueca (HF-LPME)

Kawaguchi y col. [Kawaguchi, 2010] describieron una aplicación de la

denominada microextracción en fase líquida con fibra hueca (HF-LPME), para

la determinación de benzofenonas, entre ellas BP y BP-3, en orina humana. En

su trabajo, emplearon una fibra porosa de polipropileno que impregnaron con

tolueno y conectaron a una jeringa. La punta de la aguja junto con la fibra hueca

se sumerge en la muestra y después de la extracción, agitando durante 15 min a

temperatura ambiente, se inyectan 2 µL de extracto en el sistema GC-MS. Se

obtuvieron recuperaciones de 89 y 99% y LODs de 10 y 5 ng L-1 para BP y BP-3,

respectivamente.


59

3.1.1.3.4. Microextracción líquido-líquido dispersiva (DLLME)

La microextracción líquido-líquido dispersiva (DLLME) fue desarrollada

en 2006 por Assadi y col. [Rezaee, 2006] para la extracción de compuestos

orgánicos e inorgánicos de muestras acuosas. DLLME presenta principios

comunes a LLE pero evita el elevado consumo de disolventes orgánicos que

ésta conlleva y permite la extracción y concentración simultánea de los analitos.

DLLME se basa en el uso de un sistema ternario de disolventes, constituido por

la fase acuosa (la muestra de la que los analitos pretenden ser extraídos) y una

mezcla de dos disolventes orgánicos: uno miscible con agua, que funciona como

agente dispersante, y otro altamente inmiscible con agua y miscible con el

agente dispersante. Este último se denomina extractante y debe poseer una

densidad diferente a la de la mezcla muestra: dispersante. En la modalidad más

habitual de DLLME se usan disolventes halogenados de elevada densidad

como extractantes.

La mezcla dispersante-extractante se pone en contacto con la fase acuosa

en el interior de un tubo cónico formándose una emulsión. De esta forma se

incrementa al máximo la superficie de contacto entre las fases, favoreciendo así

la rápida transferencia del analito entre la muestra y el extractante.

Posteriormente, se procede a la centrifugación y se obtiene la separación de

fases, por un lado, el agente extractante, que se deposita en el fondo del tubo

cónico debido a su mayor densidad y que contendrá disueltos los analitos y,

por otro lado, la muestra con el agente dispersante (Fig. 8).


60

1ª 2ª

Figura 8: Etapas en DLLME: 1ª Formación de una emulsión de la mezcla dispersante-

extractante-muestra; 2ª Separación de la fase extractante en el fondo del vial cónico.

Fundamento termodinámico

El coeficiente de distribución, K, se define como el cociente entre la

concentración de analito en la fase orgánica y en la fase acuosa. La DLLME sólo

es aplicable para analitos en forma neutra, que tengan un elevado carácter

hidrofóbico, y es poco viable para la extracción de especies hidrofílicas (K<500).

No obstante, para compuestos ácidos o básicos, pueden realizarse

modificaciones del pH del medio, con el fin de desplazar el equilibrio ácido-

base hacia la forma neutra y con ello, incrementar su afinidad por la fase

orgánica.

Los parámetros que, normalmente, se utilizan a la hora de caracterizar la

eficacia de la DLLME son el factor de enriquecimiento (EF) y la recuperación

(R). El factor de enriquecimiento se define como el cociente entre la

concentración del analito en la fase sedimentada (Csed) y la concentración inicial

en la muestra (C0) (Ecuación 13). La recuperación se calcula multiplicando el

factor de enriquecimiento por el cociente entre el volumen de la fase

sedimentada (Vsed) y el de la muestra (V0) (Ecuación 14).

0C

CEF sed (13)

1000

V

VEFR sed (14)


61

Parámetros que afectan a la eficacia de la extracción

Selección del disolvente extractante

En DLLME, el agente extractante debe satisfacer las siguientes

condiciones indispensables:

- tener una elevada densidad y baja solubilidad en agua, lo que

posibilitará el depósito de la fase sedimentada en el fondo del vial

cónico mediante centrifugación;

- tener capacidad para extraer los compuestos de interés;

- un buen comportamiento cromatográfico, ya que va a ser

inyectado directamente en el sistema de cromatografía de gases.

Los disolventes más utilizados como extractantes son los hidrocarburos

halogenados, tales como cloroformo CHCl3, tetracloruro de carbono CCl4,

tricloroetano CH3Cl3, clorobenzeno ClBz, etc, aunque también hay aplicaciones

en las que se ha empleado el disulfuro de carbono CS2 [Rahnana, 2007] y los

líquidos iónicos [Zhou, 2008]. Estos últimos suelen utilizarse cuando el método

de determinación es la cromatografía líquida. La cantidad de extractante que se

utiliza suele ser inferior a 200 µL, con lo que la fase sedimentada que se obtiene

es de tamaño muy pequeño, consiguiéndose así unos elevados factores de

enriquecimiento. En la Tabla 14, se muestran algunas propiedades de los

disolventes más utilizados como extractantes en DLLME.

Tabla 14: Propiedades físico-químicas de los extractantes más utilizados en DLLME.

Propiedad Disolvente

CCl4 ClBz CH3CCl3 CHCl3 CS2

Densidad (g mL-1) 1,59 1,10 1,34 1,48 1,26

Solubilidad en agua (10-3M) 2,0 0,79 7,3 16 5

Log Kow 2,89 2,81 2,10 1,76 1,94

Selección del disolvente dispersante

El disolvente que se utiliza como dispersante debe ser miscible con el

agente extractante y con la muestra acuosa, lo que permite la formación de una


62

emulsión cuando la mezcla dispersante-extractante es añadida sobre la muestra,

este fenómeno provoca un gran aumento de la superficie de contacto entre la

muestra y el extractante, favoreciendo el paso de los analitos a la fase orgánica e

incrementando, por lo tanto, la eficacia y la cinética de extracción. Los

disolventes más utilizados para este fin son etanol, metanol, acetona,

acetonitrilo y tetrahidrofurano.

Volúmenes de extractante y dispersante

La cantidad de agente extractante utilizada tiene una gran influencia sobre

el factor de enriquecimiento. Un aumento en el volumen añadido provoca un

aumento en el tamaño de la fase sedimentada, lo que conlleva una disminución

de la concentración del analito en esta fase (Csed). Según la ecuación 13, esta

disminución producirá una disminución en el factor de enriquecimiento (EF),

ya que la concentración inicial C0 permanece constante. La cantidad óptima será

aquella que genere un elevado factor de enriquecimiento, y que sea lo

suficientemente elevada como para que la fase sedimentada pueda ser

fácilmente manipulable. Normalmente, se emplean volúmenes entre 20 y 100

µL.

Por otra parte, el volumen de agente dispersante afecta principalmente a

la formación de la emulsión. Cuanto mayor sea el grado de dispersión, mejor

será el contacto entre fases, y mayor será la eficacia de la extracción. Este

volumen puede afectar también, aunque en menor medida, al tamaño de fase

sedimentada, por lo que ambas contribuciones deben tenerse en cuenta. Suelen

seleccionarse volúmenes comprendidos entre 0,5 y 1,5 mL para muestras de 10

mL.

Efecto de la centrifugación

La centrifugación es necesaria para que la fase orgánica se deposite en el

fondo del tubo de fondo cónico. Sin embargo, no se han encontrado evidencias

de que sean necesarios tiempos elevados de centrifugación, por lo que la

mayoría de los autores utilizan valores inferiores a 5 minutos.


63

Tiempo de extracción

En DLLME, se entiende por tiempo de extracción el que transcurre entre la

adición de la mezcla extractante-dispersante sobre la muestra y la

centrifugación. Este parámetro apenas tiene influencia sobre la eficacia de la

extracción ya que la transferencia de los analitos entre la muestra y el

extractante ocurre de manera inmediata tras la dispersión y el equilibrio entre

fases se alcanza rápidamente [Montes, 2009]. Esta característica representa una

de las ventajas más relevantes de la DLLME frente a otras técnicas de

microextracción.

Fuerza iónica y pH

La adición de sal sobre la muestra suele disminuir la solubilidad de los

analitos en esta fase, favoreciendo su paso hacia la fase orgánica (extractante) y

aumentando, por tanto, la eficacia de la extracción. Por otro lado, la adición de

sal provoca un aumento del volumen de fase sedimentada, con lo que

disminuye la concentración de los analitos en esta fase y por lo tanto, el factor

de enriquecimiento se verá afectado negativamente. Por todo ello, es necesario

mostrar especial atención a ambos efectos y elegir la fuerza iónica adecuada.

La variación del pH es especialmente importante cuando se trata de

analitos con características ácidas o básicas, ya que ajustando el pH puede

desplazarse el equilibrio de los mismos hacia su forma neutra consiguiendo la

extracción de especies que a priori, no podrían ser extraídas.

Aplicaciones

A pesar de que la DLLME es una técnica bastante reciente (año 2006), se

han desarrrollado numerosas aplicaciones en estos últimos años [Rezaee, 2010].

Siguiendo el trabajo inicial de Assadi y col. [Rezaee, 2006], para la determinación

de PAHS en agua, se usó la DLLME para la determinación de otras familias de

compuestos: pesticidas organofosforados [Berijani, 2006], trihalometanos

[Rahnama, 2007], clorobencenos [Kozani, 2007], bifenilos policlorados [Rezaei,


64

2008], ésteres del ácido ftálico [Farahani, 2007] y retardantes de llama

organofosforados [García-López, 2007].

La DLLME también permite la incorporación de una etapa adicional de

derivatización para aquellos compuestos que no son adecuados para su

determinación directa mediante cromatografía de gases. Así, Montes y col.

[Montes, 2009] realizaron de manera simultánea la derivatización, extracción y

concentración de triclosán en muestras de agua incluyendo el agente

derivatizante (MTBSTFA) en la mezcla dispersante (acetona)-extractante

(tricloroetano).

Se debe resaltar también la eficacia de la DLLME como etapa de

purificación y concentración para extractos obtenidos mediante otras técnicas

como la SPE. Normalmente, el extracto orgánico se emplea como agente

dispersante que, combinado con un agente extractante y añadido sobre la

disolución acuosa, da lugar a fases sedimentadas libres de interferencias. La

combinación SPE-DLLME fue aplicada por Assadi y col. a la extracción y

derivatización de clorofenoles en muestras acuosas [Fattahi, 2007] y

posteriormente por otros autores para la determinación de éteres difenil

polibromados [Liu, 2009] y herbicidas [Zhao, 2009] en agua.

Por último, decir que la técnica de DLLME también ha experimentado

adaptaciones. Regueiro y col. [Regueiro, 2008] proponen una técnica de

microextracción, conocida como microextracción–emulsificación asistida por

ultrasonidos (USAEME), para la determinación de fragancias y ftalatos en

muestras de agua. Esta técnica consiste en la formación de la emulsión con un

agente extractante (cloroformo) en una matriz acuosa por acción de

ultrasonidos, prescindiendo del agente dispersante. Durante este proceso tiene

lugar la transferencia de los analitos desde la muestra acuosa a las microgotas

de disolvente orgánico que se encuentran dispersas en la misma. Ambas fases

son posteriormente separadas mediante centrifugación y el extracto orgánico

resultante es recogido para su análisis. Otras tendencias claras en DLLME son la


65

utilización de disolventes no halogenados como extractantes y el uso de micelas

como dispersantes [Herrera-Herrera, 2010].

Tan sólo se descirben dos aplicaciones de la DLLME a la determinación de

filtros solares. Ambas se centran en el análisis de muestras acuosas y emplean

GC-MS como técnica de determinación. Tarazona y col. [Tarazona, 2010] la

aplican a la determinación de cuatro benzofenonas (BP-3, BP-8, BP-1 y 2,3,4-

trihidroxibenzofenona) en agua de mar. Utilizan 1 mL de acetona y 60 µL de

cloroformo como agentes dispersante y extractante, respectivamente. El

volumen de muestra fue de 5 mL, ajustada a pH 4 y conteniendo NaCl (10%).

Los extractos fueron evaporados y reconstituidos con N,O-bis-

(trimetilsilil)trifluoroacetamida (BSTFA) para su derivatización antes de la

inyección en el sistema GC-MS.

La otra aplicación es la desarrollada en esta Tesis para la determinación

de EHS, HMS, BzS, IAMC, EHMC, 4-MBC, BP-3, EHPABA y OCR en muestras

de agua de río, de piscina y residual [Negreira, 2010-A].

En la Tabla 15 se resumen las condiciones experimentales, las

recuperaciones y los LOQs de las técnicas de microextracción en fase líquida

empleadas en la bibliografía para la determinación de filtros solares.


66

Tabla 15: Resumen de las aplicaciones de las técnicas de microextracción en fase líquida

a la determinación de filtros UV.

Matriz Condiciones EF.

(%) Det.

LOQ

(ng L-1) Referencia

SDME

Orina Vm: 10 mL (pH 2, 13% NaCl),

5 µL [C6MIM][PF6], agit. 25 min - LC-UV 4400 [Vidal, 2007]

Agua de río Vm: 2 mL, 3 µL tolueno,

agit. 15 min

99-

100 GC-MS 50

[Okanouchi,

2008]

Agua Vm: 20 mL (pH 2, 1% etanol),

10 µL [C6MIM][PF6], agit. 37 min 8-98 LC-UV

200-

10000 [Vidal, 2010]

MALLE

Lago y

residual

Vm: 15 mL (1,5 mL metanol)

100 µL propanol, 40ºC, 120 min

60-

104

LC-APPI-

MS/MS 3-53 [Rodil, 2009]

HF-LPME

Orina

HF con tolueno, sumergida

en 1 mL de muestra,

15 min con agitación

89-

99 GC-MS 17-33

[Kawaguchi,

2009]

DLLME

Agua de

mar

Vm: 5 mL (pH 4, 0,5 g NaCl)

Disolvente: 1 mL acetona

(60 µL cloroformo)

65-

169 GC-MS 108-110

[Tarazona,

2010]

Como colofón a la revisión realizada en relación con la determinación de

filtros solares en muestras de agua, es preciso destacar que para determinados

compuestos (ej. EHMC, OCR y, en ocasiones, BP-3) los LOQs alcanzados vienen

condicionados por la señal de los blancos [Balmer, 2005] [Rodil, 2008-A] [Rodil,

2008-B] y no por la eficacia de la técnica de extracción o la sensibilidad de la

técnica de determinación.


67

3.1.2. Muestras sólidas

La extracción y recuperación de compuestos orgánicos de matrices

sólidas tales como sedimentos, suelos y lodos puede resumirse en cuatro etapas:

- Desorción de los analitos de los sitios activos de la matriz.

- Difusión de los analitos a través de la matriz.

- Solubilización de los analitos en el extractante.

- Recolección de los extractos con los analitos de interés.

La interacción entre los analitos y las muestras sólidas suele ser mucho

más intensa que en el caso de las matrices líquidas, de modo que es necesario

emplear técnicas de extracción más enérgicas, lo cual disminuye la selectividad

del proceso y, normalmente, hace necesario la consideración de etapas

posteriores de limpieza.

Entre las técnicas empleadas en esta Tesis Doctoral para la determinación

de filtros solares en matrices sólidas ambientales se encuentran: la dispersión de

la matriz en fase sólida (MSPD) y la extracción con disolventes presurizados

(PLE), cuyos fundamentos y aplicaciones se presentan a continuación.

3.1.2.1. Dispersión de la matriz en fase sólida (MSPD)

La dispersión de la matriz en fase sólida (MSPD) fue desarrollada en

1989 por Barker [Barker, 1989]. La MSPD se basa en homogeneizar y dispersar

una pequeña cantidad de muestra con ayuda de un mortero sobre un soporte

sólido. A continuación, la matriz homogeneizada se introduce en un cartucho,

normalmente de propileno, y los analitos son eluidos con un disolvente

apropiado. MSPD permite integrar la purificación en la misma etapa que la

extracción, introduciendo un co-adsorbente en el fondo del cartucho que

retenga los compuestos interferentes. La distribución de los analitos entre la

muestra dispersada y el disolvente de elución depende de las constantes que

regulan la partición (fase estacionaria líquida) o el equilibrio de adsorción (fase


68

estacionaria sólida). La eficacia y la selectividad en MSPD depende de varios

factores entre los cuales destacan la selección del material dispersante, la

utilización de co-adsorbentes, el tipo de disolvente y la secuencia de elución.

Naturaleza del soporte sólido y de la fase enlazada.

Los materiales derivados de la sílica son los más empleados para la

disrupción de la matriz en MSPD, ya que presentan la ventaja de poseer grupos

silanoles no enlazados, tanto en la superficie de las partículas como en los

poros, que interaccionan con el agua de la muestra, actuando a su vez como

agentes desecantes en el caso de matrices semi-sólidas. Dentro de los

adsorbentes denominados de fase reversa, el más empleado es el octadecilsilano

(C18), frente a otros como el C8 y el C30. Las partículas de sílica actúan como

dispersantes, mientras que el C18 solubiliza los componentes de la matriz sobre

su superficie. Con este tipo de fases es posible obtener extractos relativamente

libres de grasas para muestras de músculo [Kubala-Drinic, 2003] [Le Boulaire,

1997], hígado [Crescenzi, 2001] [Horne, 1998], riñón [Ruiz, 2005] y pescado con

alto contenido lipídico [Tolls, 1999] [García-Reyes, 2007] [Canosa, 2008], usando

acetonitrilo como disolvente. En la actualidad, las fases reversas basadas en

cadenas hidrocarbonadas se están sustituyendo por amino‐propil‐sílica o por

aminas primarias‐secundarias (PSA) [García-Reyes, 2007] [Ferrer, 2005] [Cunha,

2007] para reducir todavía más el contenido lipídico en los extractos

procedentes de muestras biológicas de origen animal. El Florisil (MgSiO3), la

alúmina (Al2O3) y la sílica (SiO2) son adsorbentes denominados de fase normal

utilizados también para la extracción de pesticidas, herbicidas y contaminantes

prioritarios en matrices biológicas [Gómez-Ariza, 2002] [Martínez, 2005] y frutas

[Hu, 2006] mediante MSPD. Además, se han usado como dispersantes de

muestras ambientales (lodos de depuradora, sedimentos, polvo de

aspiradora,…) para la extracción de contaminantes tanto prioritarios como

emergentes en MSPD [Pena, 2007] [Blanco, 2006] [Shen, 2006]. Estos adsorbentes

pueden ser utilizados tal y como se encuentran comercialmente o modificados

mediante la adición de agua, ácidos o bases en función de las características y el


69

comportamiento que queramos conferirles [Carro, 2005].

Naturaleza del co‐adsorbente

Generalmente debe ser de distinta naturaleza que el sólido utilizado para

dispersar la muestra. El co-adsorbente actúa reteniendo las interferencias,

normalmente las de mayor contenido lipídico [Pensado, 2005], o también puede

destruirlas cuando presenta algún tipo de modificación química [Carro, 2005]

[Canosa, 2008].

Naturaleza de la matriz de la muestra

Los componentes de la matriz dispersada se mueven a través de la fase

cromatográfica, contenida en el cartucho de MSPD, lo que teóricamente, hace

posible un fraccionamiento en función de la naturaleza de estos compuestos, así

como de las sustancias interferentes.

Tipo de disolvente y secuencia de elución

Al igual que en cromatografía, o en SPE, la polaridad del disolvente es de

gran importancia a la hora de determinar qué analitos eluyen del cartucho de

MSPD y en qué orden lo hacen. La correcta elección del disolvente y el diseño

del perfil de elución permite obtener extractos libres de impurezas en base a la

retención de las mismas en la fase estacionaria [Pensado, 2005], o mediante una

primera elución para retirarlas del cartucho de MSPD [Canosa, 2007] [García,

2007], previamente a la extracción de los analitos.

En esta Tesis doctoral, MSPD ha sido la técnica elegida para la extracción

de filtros solares en muestras de polvo [Negreira, 2009-C]. De acuerdo con la

revisión bibliográfica realizada, este trabajo constituye la primera referencia

describiendo la presencia de filtros UV en atmósferas interiores, así como una

de las primeras aplicaciones de MSPD a la determinación de estos compuestos

en matrices ambientales. En trabajos previos del grupo de investigación se ha

demostrado la aplicabilidad de MSPD a la determinación de productos de


70

cuidado personal [Canosa, 2007] y retardantes de llama [García, 2007] en

muestras de polvo.

3.1.2.2. Extracción con disolventes presurizados (PLE)

La extracción con disolventes presurizados (PLE), también denominada

extracción acelerada con disolventes (ASE), es una técnica de preparación de

muestra que combina el uso de temperaturas (50-200 ºC) y presiones (1500-2000

psi) elevadas, para extraer rápida y eficazmente los analitos de matrices sólidas

o semisólidas [Garrido-López, 2005]. Los parámetros fundamentales que afectan

a la eficacia de extracción, y que deben tenerse en cuenta a la hora de

desarrollar un método de PLE, son los siguientes:

Temperatura

Tiene que ser suficientemente elevada como para aumentar las

recuperaciones y favorecer la cinética de extracción, sin degradar a los

compuestos objeto de estudio [ConchaGraña, 2004]. Normalmente, es superior

al punto de ebullición del disolvente, pero ligeramente inferior a su punto

crítico. Al aumentar la temperatura, el disolvente disminuye su viscosidad y

penetra con mayor facilidad en los poros de la matriz, favoreciendo la difusión

de los analitos, es decir la cinética de extracción. De este modo, la eficacia de

extracción se incrementa, minimizando el volumen de disolvente empleado

[Richter, 1996].

Disolvente

Debe ser capaz de solubilizar los analitos sin arrastrar el resto de los

componentes de la matriz. Las mezclas de disolventes de diferentes polaridades

pueden ser usadas para la extracción de un amplio rango de familias de

compuestos. La mayoría de disolventes pueden emplearse en PLE, incluidos

agua y mezclas acuosas tamponadas. No se recomiendan ácidos fuertes debido

a su reacción con el acero, pero se pueden usar ácidos débiles, como ácido


71

acético o fosfórico, añadidos a agua o a disolventes polares en porcentajes de

hasta un 10% (v/v).

Presión

Ésta debe ser elevada para mantener el disolvente en estado líquido a la

temperatura de trabajo. Sin embargo, no es un factor que afecte en gran medida

a la eficacia de extracción, de modo que puede ser fijado de antemano [Camel,

2001]. El modo más usual de hacer la extracción es el modo estático, en el que el

disolvente es introducido en la celda y ésta se mantiene a presión constante un

tiempo determinado. Tras esta etapa, la celda se vacía recogiendo todo el

extracto en un vial colector. En el modo dinámico, el disolvente está pasando

continuamente a un flujo constante a través de la celda presurizada. En este

caso, la extracción es más efectiva, pero presenta el inconveniente de

incrementar el volumen del extracto [Camel, 2001] y además, el equipo necesita

una válvula restrictora que permita operar en ese modo.

Tiempo

El aumento del tiempo estático a elevadas temperaturas favorece la

difusión de los analitos al disolvente de extracción evitando su retención en la

matriz.

Número de ciclos

El uso de varios ciclos de extracción estáticos fue desarrollado para

introducir fracciones nuevas de disolvente durante el proceso de extracción. El

uso de varios ciclos de extracción es útil para muestras difíciles de penetrar o

que presenten muy alta concentración de analito.

Porcentaje de flush

Después de cada ciclo de extracción en modo estático, se hace pasar un

volumen de disolvente a través de la celda, expresado en porcentaje (% flush)

de su volumen interior, para arrastrar posibles trazas de los analitos que


72

pudiesen quedar en ella. Ese volumen, en el caso de que se usen dos o más

ciclos de extracción, es dividido entre el número de los mismos [Richter, 1996].

Equipamiento para PLE

La técnica de PLE fue comercializada en 1995 por la compañía Dionex

[Fidalgo-Used, 2007] y el modelo más utilizado es el extractor ASE 200, cuya

fotografía se muestra en la Fig. 9.

Figura 9: Fotografía del equipo de extracción ASE 200

Independientemente del modelo y la casa comercial, los equipos de PLE

constan de:

- una bomba para impulsar el disolvente,

- un horno donde se introducen las celdas de acero para

mantenerlas a la temperatura seleccionada,

- un vial colector en donde se recoge el extracto líquido,

- nitrógeno para purgar la celda una vez terminado el proceso de

extracción.

El esquema de un equipo de extracción con disolventes presurizados se

muestra en la siguiente figura, Fig. 10.


73

Figura 10. Equipo de extracción con disolventes presurizados.

Etapas del proceso de extracción

Las etapas principales del proceso de extracción son:

- Preparación de muestra. Con esta etapa se pretende reducir el

tamaño de partícula para aumentar la superficie de contacto entre

la muestra y el disolvente, evitando también la agregación de las

partículas de la muestra. Para ello, se utilizan agentes dispersantes

como la arena o la tierra de diatomeas. La muestra debe secarse

para evitar la presencia de agua que dificulte la penetración de

disolvente en sus poros, sobre todo cuando se trata de extraer

analitos poco polares con disolventes apolares. Para ello, se

aconseja la liofilización de la muestra, el secado en horno o la

adición de agentes desecantes.

- Preparación de celda. Se colocan filtros de celulosa y/o fibra de

vidrio en los extremos de la celda con el fin de evitar la obturación

de los conductos del extractor de PLE. La celda se rellena con la

muestra y con una matriz inerte (tierra de diatomeas, arena…)

para ocupar el volumen muerto. Además se pueden introducir co-

adsorbentes para purificar el extracto en la misma etapa.

- Extracción estática. La celda se introduce en el horno a la

temperatura de extracción, se rellena con disolvente y se presuriza.

También se puede precalentar la celda previamente y mantenerla a

esa temperatura un tiempo determinado. Después de recoger el


74

disolvente con los analitos en el vial colector, se bombea disolvente

fresco a través de los conectores y celda (% de flush). Finalmente, se

purga la celda con nitrógeno para arrastrar el disolvente que

permanece impregnando la muestra, el dispersante y, en su caso, el

co-adsorbente en la celda.

Aplicaciones

PLE es equivalente a otras metodologías de extracción, como Soxhlet,

siendo utilizada incluso como método de referencia por la Agencia de

Protección Medioambiental estadounidense (EPA) en uno de sus métodos

(método 3545A)3 para la extracción de compuestos volátiles y semivolátiles en

matrices medioambientales sólidas. La aplicación de PLE a la extracción de

filtros solares en muestras sólidas se ha centrado en sedimentos y lodos de

depuradora. En relación con la primera matriz, Rodil y col. [Rodil, 2008-C]

extrajeron EHS, HMS, IAMC, EHMC, 4-MBC, BP-3, EHPABA y OCR

mezclando 4 g de muestra con 1 g de sulfato sódico anhidro que introdujeron

en una celda conteniendo 2 g de sílica gel y la misma cantidad de cobre en

polvo. La extracción se llevó a cabo en 4 ciclos de 5 min, a 160 ºC y 100 bar

usando acetato de etilo:hexano (80:20) como disolvente de extracción. Después

de concentrar el extracto, añadieron N,O-bis(trimetilsilil)trifluoroacetamida

(BSTFA) para la derivatización de los salicilatos y la BP-3, antes de su

determinación mediante GC-MS. Las recuperaciones obtenidas oscilaron del 73

al 128% y los LODs variaron entre 2 y 20 ng g-1 sin derivatizar y de 2 a 5 ng g-1

derivatizando los analitos.

Nieto y col. [Nieto, 2009] también emplearon PLE como técnica de

extracción y purificación para la determinación de filtros solares mediante

UPLC-MS/MS en lodos de depuradora. La celda fue cargada con 1 g de óxido

de aluminio, 1 g de muestra y, otra vez, óxido de aluminio para rellenar el

3 U.S.Environmental Protection Agency (EPA), disponible en:

http://www.epa.gov/ (acceso en diciembre de 2009).


75

volumen muerto restante. Las condiciones de la extracción fueron de 2 ciclos de

5 min con metanol, seguidos de 2 ciclos con agua (pH 7): metanol (1:1) durante

un tiempo de 5 min, a 100ºC y 140 bar. Las recuperaciones oscilaron desde el

30% (BP-1) al 108% (EHPABA).

Rodil y col. [Rodil, 2009-D] desarrollaron un método para la

determinación de filtros UV en muestras de lodos mediante LC-(APPI)MS/MS.

Su principal novedad radica en el uso de membranas poliméricas no porosas en

combinación con PLE. La muestra (0,5 g lodo) y 1 mL de disolvente de

extracción se encerraron en una membrana preparada con polietileno de baja

densidad (LDPE) y se introdujeron en un equipo convencional de PLE. La

extracción se lleva a cabo a 70ºC, 4 ciclos de 5 min con acetato de etilo:hexano,

(1:3). El extracto evaporado se reconstituyó con metanol: agua (1:1) antes de su

inyección en LC-MS. La principal ventaja de este proceso es la reducción de

tiempo y disolvente al combinar extracción y limpieza en un único paso. Sin

embargo, la extracción no fue cuantitativa, siendo necesario el uso de adiciones

estándar sobre la muestra para poder determinar los niveles de analitos en

muestras de lodos.

Wick y col. [Wick, 2010] mezclaron 0,2 g de lodo liofilizado con arena de

mar purificada. La extracción se llevó a cabo en 4 ciclos de 10 min a 80 ºC con

agua:metanol (1:1). Los autores emplearon una etapa posterior de purificación

mediante SPE para la cual, el extracto resultante se diluyó a 800 mL con agua

ajustada a pH 6 antes de ser pasado a través de un cartucho Oasis HLB (200

mg). La elución se realizó con 4 x 2 mL de una mezcla de metanol:acetona

(60:40) y la determinación tuvo lugar por LC-MS/MS, obteniéndose valores de

recuperación comprendidos entre 18% y 196% y LOQs entre 2,5 y 25 ng L-1. El

método desarrollado sólo incluye 4 benzofenonas (BP-1, BP-2, BP-3 y BP-4),

junto con otros compuestos de la familia de los benzotriazoles y, probablemente

no proporciona recuperaciones cuantitativas para compuestos más lipofílicos

como 4-MBC, EHMC y OCR.


76

En esta tesis doctoral también se desarrolló una metodología para la

determinación de filtros solares en lodos de depuradora empleando PLE como

técnica de extracción. Los objetivos del estudio realizado ha sido la obtención de

recuperaciones cuantitativas con un nivel de selectividad adecuado para la

determinación de los compuestos mediante GC-MS.

3.1.2.3. Otras técnicas de extracción

La preparación de muestra para la determinación de filtros solares en

sedimentos y, sobre todo, en lodos implica la combinación de metodologías

eficaces de extracción, con un grado adecuado de selectividad, teniendo en

cuenta la complejidad de la matriz de partida (sobre todo en el caso de lodos), y

la técnica empleada en la etapa de determinación. A continuación, se resume la

bibliografía encontrada para la determinación de filtros solares en sedimentos,

suelos y lodos, excluyendo los trabajos de PLE que han sido citados en la

página anterior.

Jeon y col. [Jeon, 2006] determinaron filtros UV en suelo extrayendo 10 g

de muestra, mezclada con sulfato sódico (10 g), con 20 mL de metanol, durante

20 min. A continuación, el extracto se concentra (ca. 3 mL) se agita junto con 1

mL de una disolución acuosa de NaCl (5%) y 5 mL de acetato de etilo y se

congela a -30ºC para la separación de la fase orgánica que se evapora para ser

derivatizada e inyectada en GC-MS. Las recuperaciones obtenidas variaron de

60% a 115% y el LOQ alcanzado fue de 500 ng g-1 para todos los compuestos.

Plagellat y col. [Plagellat, 2006] determinaron tres filtros solares (EHMC,

OCR y 4-MBC) en muestras de lodos de depuradora utilizando extracción

líquido-líquido. Para ello, los lodos espesados, y sin liofilizar, se mezclaron con

3 g de NaCl y 20 mL de pentano:acetona (1:1, v/v), se agitaron durante 30 min y

luego se llevaron a cabo otras 2 extracciones sucesivas con 20 mL pentano:dietil


77

éter (1:1, v/v) y dietil éter:diclorometano (4:1, v/v). Después de centrifugar, las

diferentes fracciones se concentraron a sequedad, se redisolvió el extracto con 1

mL de hexano y se transfirió a una columna con 5 g de sílica gel para llevar a

cabo el clean-up. Los analitos fueron recogidos con 50 mL de hexano:dietil éter

(9:1, v/v) después de descartar 2 fracciones anteriores de 20 mL de hexano y 20

mL de hexano:dietil éter. Para su inyección en GC-EI-SIM-MS fue necesario la

evaporación del extracto y su reconstitución con 1 mL de acetato de etilo. Las

recuperaciones se situaron entre 88 y 101% y los LODs ente 2 y 6 ng g-1; sin

embargo, el método es difícil de automatizar y consume del orden de 150 mL de

disolventes orgánicos por muestra.

En la siguiente tabla, se hace un resumen de las condiciones de

extracción y purificación utilizadas en la determinación de filtros solares en

muestras de sedimentos y lodos, Tabla 16.

Tabla 16 : Resumen de la metodología analítica para la determinación de filtros solares

en sedimentos y lodo de depuradora.

Muestra Analitos Técnica

extracción Purificación

Técnica

Det. Referencia

Sedimento

BP, BP-1, BP-3, BP-8 LLE - GC-MS [Jeon, 2006]

EHS, HMS, 4-MBC,

BP-3, IAMC, EHMC,

EHPABA, OCR

PLE (extracción y

purificación) GC-MS

[Rodil,

2008-C]

Lodo

4-MBC, OCR, EHMC LLE Sílica gel GC-MS [Plagellat,

2006]

EHS, HMS, 4-MBC,

BP-3, IAMC, EHMC,

EHPABA, OCR

PLE

Difusión

membrana no

porosa

LC-

MS/MS

[Rodil,

2009-D]

BP-1, BP-3, BP-8,

OCR, EHPABA

PLE (extracción y

purificación)

UHPLC-

MS/MS

[Nieto,

2009]

BP-1, BP-2,BP-3, BP-4 PLE SPE LC-

MS/MS

[Wick,

2010]


78

3.1.3. Muestras de biota

En el caso de biota (Ej. músculo o órganos de pescado) la etapa más

importante en la preparación de muestra es la separación de los analitos y los

lípidos, sobre todo en el caso de emplear cromatografía de gases como técnica

de determinación. Para llevar a cabo ese fraccionamiento, se suele utilizar la

cromatografía de exclusión por tamaños (SEC). Aunque efectiva, SEC implica el

consumo de ingentes volúmenes de disolventes orgánicos.

Balmer y col. [Balmer, 2005] analizaron muestras de pescado para la

determinación de filtros solares. Estas muestras fueron homogeneizadas con

sulfato sódico y se extrajeron con diclorometano:ciclohexano (1:1). Para el

proceso de clean-up se utilizó cromatografía de exclusión por tamaños (SEC)

empleando una columna Biobeads S-X3 y diclorometano:ciclohexano (35:65)

como fase móvil. Para la determinación se utilizó GC-MS obteniendo

recuperaciones de 93% a 115% y LODs de 3 a 380 ng g-1. La misma metodología

fue empleada por Buser y col. [Buser, 2006] en un estudio posterior para evaluar

la acumulación de filtros UV en pescado de ríos y lagos en Suiza.

Meinerling y col. [Meinerling, 2006] determinaron BP-3, 4-MBC, OCR y

EHMC en tejidos de pescado. La extracción se llevó a cabo mediante Soxhlet

empleando 200 mL de n-hexano:acetona (9:1, v/v) durante aprox. 3 h. El

extracto frío se redujo a sequedad usando un rotavapor y fue disuelto en 20 mL

de n-hexano:acetona (9:1, v/v). A continuación se purificó empleando SEC y

SPE. La cuantificación fue desarrollada por LC-MS. Las recuperaciones

obtenidas se encontraron en el rango de 86% a 108%. El límite de cuantificación

determinado para cada analito fue de 8 ng g-1 de muestra fresca.

Zenker y col. [Zenker, 2008] desarrollaron una metodología para el análisis

de músculo de pescado agitándolo vigorosamente con una mezcla de acetato de

etilo, n-heptano y agua (1:1:1). Después de centrifugar, el sobrenadante fue


79

concentrado y disuelto con etanol para posteriormente llevar a cabo la limpieza

mediante HPLC con una columna en fase reversa. La elución de los analitos se

realizó con una mezcla de metanol y agua (70:30) recogiendo dos fracciones: la

primera contenía los compuestos más polares (BP-1, BP-2, BP-8, BP-4, Et-PABA)

que fueron inyectados en LC-MS, y en la segunda se recuperaban los más

lipofílicos (BP-3, EHMC, 4-MBC) que fueron determinados mediante GC-MS.

Las recuperaciones obtenidas variaron desde 76% a 99% y los LODs de 11 a 36

ng g-1 para BP-3, EHMC y 4-MBC y de 78 a 205 ng g-1 para el resto de

compuestos. Fent y col. [Fent, 2010-B] aplicaron el mismo método para el

análisis de pescado en Suiza con resultados similares.

Mottaleb y col. [Mottaleb, 2009] mezclaron tejido de pescado con 10 mL de

acetona, concentraron y reconstituyeron el extracto con hexano:acetona (65:35,

v/v). A continuación, este extracto se purificó empleando un adsorbente en fase

normal (sílica) y SEC. La determinación se llevó a cabo utilizando GC-MS/MS

obteniéndose recuperaciones de 57% a 87% y LODs de 16 a 120 ng g-1. Para

muestras con bajo contenido lipídico, los autores prescinden de la purificación

mediante SEC, obteniendo en este caso recuperaciones entre 98% y 101%.

Kwon y col. [Kwon, 2009] emplearon LLE y SPE para extraer y purificar

muestras de hígado de pescado antes de la determinación de BP-3, junto con

otros contaminantes emergentes, mediante LC-MS/MS con ionización mediante

electrospray en modo negativo y positivo. El extracto primario en n-hexano fue

llevado a sequedad, reconstituido con acetonitrilo y diluido con 50 mL de agua.

Esta disolución se concentró con un cartucho de SPE (Oasis HLB) que

posteriormente fue eluido con metanol. La determinación mediante LC-MS

proporcionó un LOD de 8 ng g-1 con recuperaciones entre 72% y 77% para BP-3.

En la siguiente tabla, se hace un resumen de las condiciones de

extracción y purificación para la determinación de filtros solares en muestras de

biota, Tabla 17. En general, los métodos propuestos en la bibliografía para esta


80

matriz son multietapa y presentan un bajo grado de automatización, así como

un elevado consumo de disolventes orgánicos.

Tabla 17: Resumen de la metodología analítica para la determinación de filtros solares

en pescado.

Analitos Técnica

extracción Purificación

Técnica

Det. Referencia

BP-3, 4-MBC, EHMC, OCR ESL SEC GC-MS

[Balmer, 2005]

4-MBC, OCR [Buser, 2006]

BP-3, 4-MBC, OCR, EHMC Soxhlet SEC + SPE LC-MS [Meinerling, 2006]

BP-1, BP-2, BP-8, BP-4, Et-PABA ESL RP-HPLC

LC-MS [Zenker, 2008],

[Fent, 2010-B] BP-3, 4-MBC, EHMC GC-MS

BP, 4-MBC, OCR ESL SEC + SPE GC-MS [Mottaleb, 2009]

BP-3 ESL SPE LC-MS [Kwon, 2009]

ESL, extracción sólido-líquido

RP, fase reversa


81

3.2. Técnicas de determinación

La determinación de niveles traza de filtros solares en muestras

ambientales complejas se basa normalmente en la utilización de cromatografía

de gases (GC) o líquidos (LC) combinadas con espectrometría de masas simple

(MS) o en tándem (MS/MS). Otras técnicas consideradas para la determinación

de estos compuestos en productos de cuidado personal, ej. LC con detección

UV-visible, no ofrecen prestaciones adecuadas en términos de sensibilidad y

selectividad para poder abordar el estudio de matrices ambientales.

La elección de la técnica de separación (GC o LC) viene condicionada

fundamentalmente por los pesos moleculares y los grupos funcionales

presentes en la estructura de los analitos considerados. A modo de ejemplo, la

BP-3 y sus metabolitos (ej. BP-1), presentan grupos fenólicos en su estructura lo

que provoca un ensanchamiento de los correspondientes picos en GC cuando se

aborda su determinación directa. Este mismo comentario es también válido

para los salicilatos. Otros filtros UV que contienen grupos más polares en sus

estructuras, tales como la BP-4, son sólo cuantificables mediante LC. La

afirmación anterior es también aplicable a filtros de mayor tamaño molecular,

tales como los derivados de la triazona y del benzotriazol, así como las

polisiliconas.

Puesto que LC se emplea normalmente en combinación con MS/MS es

necesario tener en cuenta, no sólo la capacidad de separación de la etapa

cromatográfica sino también el rendimiento de la etapa de ionización, que en

ocasiones, controla los LOQs alcanzables. A modo de ejemplo, los filtros de la

familia de salicilatos presentan eficacias de ionización muy bajas en sistemas

equipados con interfases de electrospray (ESI) [Rodil, 2009-E].

A continuación se comentan las aplicaciones de las técnicas anteriores a

la determinación de filtros UV en matrices medioambientales.


82

3.2.1. Cromatografía de gases acoplada a espectrometría de

masas simple (GC-MS) y en tándem (GC-MS/MS)

Sin duda, GC ha sido la técnica más empleada para la determinación de

filtros solares en muestras medioambientales dada su disponibilidad en la

mayoría de laboratorios. Las diferencias básicas entre los estudios realizados

radican en el tipo de analizador de masas empleado, las características de la

columna cromatográfica, el modo de inyección y la utilización o no de

reacciones de derivatización para mejorar la detectabilidad de ciertos analitos.

A continuación, se describe de forma resumida el fundamento del acoplamiento

GC-MS, así como GC-MS/MS, para luego comentar algunas aplicaciones

representativas a la determinación de filtros UV en matrices medioambientales.

El acoplamento GC-MS data de la década de los 70, y revolucionó el

análisis de mezclas complejas de compuestos orgánicos debido a que combina

el elevado poder de resolución que proporciona la cromatografía de gases, con

la alta sensibilidad y la información estructural de la espectrometría de masas.

Mediante esta técnica se obtienen registros tridimensionales, es decir, para cada

tiempo de retención, obtenemos un espectro de masas de las especies que

emergen de la columna cromatográfica.

En GC, los compuestos en estado vapor se separan en la columna

cromatográfica en función de su distribución entre la fase estacionaria y la fase

móvil. Debido al bajo flujo de fase móvil empleado en las columnas capilares, es

posible la conexión directa de la columna cromatográfica con la fuente de

ionización del espectrómetro de masas. Posteriormente, los iones pasan al

analizador de masas, el cual está sometido a un alto vacío para evitar la colisión

y reconstrucción de los fragmentos cargados generados tras la ionización. Una

vez son seleccionados por el analizador de masas, se registran las intensidades

correspondientes a cada relación m/z.

Para la determinación de filtros UV mediante GC-MS se hace uso de la

ionización por impacto electrónico (EI). Las fuentes de EI consisten en un

filamento de wolframio, que emite electrones acelerados hacia un ánodo. Estos


83

electrones chocan con las moléculas del efluente cromatográfico dado que sus

trayectorias son perpendiculares. A continuación, los iones generados pasan al

analizador de masas. En esta Tesis Doctoral se emplearon sistemas GC-MS

equipados con dos tipos de analizadores de masas diferentes (trampa de iones y

cuadrupolo):

- Trampa de iones

Una trampa de iones, Fig. 11, consta de tres electrodos que forman una

cavidad en la que tiene lugar el proceso completo de ionización, fragmentación,

almacenamiento y separación de los iones. Al electrodo central se le aplica un

potencial de radiofrecuencia (RF) que crea un campo eléctrico hiperbólico

tridimensional, en el que los iones son atrapados en órbitas estables. A medida

que se aumenta el voltaje de RF, las trayectorias de los iones se hacen inestables

en el sentido en que aumenta su relación masa/carga, y son expulsados de la

trampa hacia el multiplicador de electrones.

Para evitar las reacciones ión-ión e ión-molécula, causa de espectros mal

resueltos, se aplica un voltaje adicional de RF (voltaje de modulación axial) que

aumenta la resolución entre masas. Mediante este potencial, los iones ocupan

mayor volumen en el interior de la trampa.

Figura 11: Esquema de una trampa de iones.


84

- Cuadrupolo

El analizador de masas tipo cuadrupolo, Fig. 12, consta de dos pares de

cilindros o barras a cada uno de los cuales se le aplica una combinación de

potenciales, de radiofrecuencia (RF) y de corriente continua (DC), que se van

variando de forma que sólo los iones con determinada relación m/z sean

capaces de atravesar completamente el filtro de masas. Los potenciales

aplicados a los dos pares de cilindros o barras son iguales pero de signo

opuesto.

Figura 12: Esquema de un cuadrupolo. Una de las ventajas de estos analizadores es la posibilidad de trabajar en

modo SIM (selected ion monitoring), cuando el objetivo principal del análisis es

maximizar la sensibilidad. Por su parte, las trampas de iones permiten trabajar

en modo MS/MS sin coste adicional de la instrumentación, frente al empleo de

triples cuadrupolos cuando se pretende trabajar con espectrometría de masas en

tándem.

En las tablas siguientes (Tablas 18 y 19) se recogen las aplicaciones más

relevantes en las que se describe el uso de GC-MS para la determinación de

filtros solares en muestras medioambientales.


85

Tab

la 1

8: R

esum

en d

e la

s ap

licac

ione

s de

GC

-MS

para

la d

eter

min

ació

n de

filt

ros

sola

res

en m

uest

ras

acuo

sas.

Mat

riz

acu

osa

An

alit

os

Téc

nic

a d

e ex

trac

ción

C

olu

mn

a em

ple

ada

LO

D (n

g L

-1)

Rec

up

. (%

) R

efer

enci

a

Pis

cina

y b

año

BP

-3, E

HP

AB

A

SPM

E

DB

-5-M

S

(30

m x

0,3

2 m

m, 1

µm

)

0,2-

1 82

-99

[Lam

brop

oulo

u, 2

002]

SP

E

2 93

-97

Río

y b

año

BP

-3, E

HS,

EH

MC

,

4-M

BC

, OC

R

SPE

SE

54

(25

m x

0,3

2 m

m)

2 57

-90

[Poi

ger,

200

4]

Pis

cina

y d

uch

a B

P-3

, 4-M

BC

SP

E

DB

-5-M

S

(30

m x

0,2

5 m

m, 0

,25

µm)

0,2-

0,4

95-9

9 [G

ioka

s, 2

004]

Río

y b

año

BP

-3, E

HM

C, 4

-MB

C

LL

E

DB

-5-M

S

(30

m x

0,2

5 m

m, 0

,25

µm)

2-30

95

-102

[G

ioka

s, 2

005]

Sup

erfi

cial

y re

sid

ual

BP-

3, E

HM

C,

4-M

BC

, OC

R

SPE

SE54

(25

m x

0,3

2 m

m, 0

,25

µm)

BG

B-5

(30

m x

0,2

5 m

m, 0

,25

µm)

2 (s

upe

rfic

ial)

10 (r

esid

ual

)

78-1

29

[Bal

mer

, 200

5]

Res

idua

l E

HM

C, 4

-MB

C, O

CR

L

LE

D

B-5

-MS

(50

m x

0,2

0 m

m, 0

,33

µm)

3-14

75

-91

[Ku

pper

, 200

6]

Río

, lag

o

y re

sid

ual

B

P-3,

BP

-1, B

P-8

L

LE

U

ltra

2

(30

m x

0,2

mm

, 0,3

3µm

) 5

76-1

13

[Jeo

n, 2

006]

Río

B

P-3

SB

SE

DB

-5-M

S

(30

m x

0,2

5 m

m, 0

,25

µm)

1 11

0-11

4 [K

awag

uch

i, 20

06]

Río

, mar

y pi

scin

a

BP-

3, H

MS,

EH

MC

,

4-M

BC

, EH

PA

BA

, OC

R

SPE

H

P-5

-MS

(30

m x

0,2

5mm

, 0,2

5 µm

) 13

-266

50

-98

[Cu

der

man

, 200

7]


86

Tab

la 1

8 co

nt: R

esum

en d

e la

s ap

licac

ione

s de

GC

-MS

para

la d

eter

min

ació

n de

filt

ros

sola

res

en m

uest

ras

acuo

sas.

Mat

riz

acu

osa

An

alit

os

Téc

nic

a d

e ex

trac

ción

C

olu

mn

a em

ple

ada

LO

D (n

g L

-1)

Rec

up

. (%

) R

efer

enci

a

Res

idua

l B

P-3

, EH

MC

, 4-M

BC

, OC

R

SPE

D

B-5

-MS

(30

m x

0,2

5 m

m, 0

,25

µm)

10

67-1

18

[Li,

2007

]

Río

B

P-3

, BP-

1 SB

SE

DB

-5-M

S

(30

m x

0,2

5 m

m, 0

,25

µm)

0,5-

1 10

2-12

8 [K

agaw

uch

i, 20

08-A

]

Río

B

P-3

L

PM

E

DB

-5-M

S

(30

m x

0,2

5 m

m, 0

,25

µm)

10

100

[Oka

nou

chi,

2008

]

Río

, lag

o y

resi

du

al

BP

-3, E

HS,

HM

S, 4

-MB

C,

IAM

C, E

HM

C, E

HP

AB

A, O

CR

SB

SE

HP-

5-M

S

(30

m x

0,2

5 m

m, 0

,25

µm)

0,2-

16

63-1

22

[Rod

il, 2

008-

B]

Lag

o 4-

MB

C, O

CR

SB

SE

HP

-5m

s

(30

m x

0,2

5 m

m, 0

,25

µm)

0,3-

4 50

-100

[H

auns

chm

idt,

2010

]

Río

E

HS,

BP

-3, 4

-MB

C, O

CR

SP

ME

R

tx-5

-MS

(30

m x

0,2

5 m

m, 0

,25

µm)

0,2-

2 72

,5-1

14

[Liu

, 201

0]

Mar

B

P-1,

BP

-3, B

P-8

D

LL

ME

T

R-5

-MS

(30

m x

0,2

5 m

m, 0

,25

µm)

32-3

3 65

-169

[T

araz

ona,

201

0]

Lag

o y

eflu

ente

B

P-3

, 4-M

BC

, EH

MC

, OC

R

ME

PS

HP-

5-M

S

(30

m x

0,2

5 m

m, 0

,25

µm)

34-8

7 61

-114

[M

oed

er, 2

010]


87

Tab

la 1

9: R

esum

en d

e la

s ap

licac

ione

s de

GC

-MS

para

la d

eter

min

ació

n de

filt

ros

sola

res

en m

uest

ras

sólid

as.

Mat

riz

An

alit

os

Téc

nic

a d

e ex

trac

ción

C

olu

mn

a em

ple

ada

LO

D (n

g g-

1 )

Rec

up

. (%

) R

efer

enci

a

Sed

imen

to

BP

-3, B

P-1

, BP

-8

LL

E

Ult

ra 2

(30

m x

0,2

mm

, 0,3

3µm

) 0,

1 60

-84

[Jeo

n, 2

006]

Sed

imen

to

BP

-3, E

HS,

HM

S, 4

-MB

C, I

AM

C,

EH

MC

, EH

PAB

A, O

CR

P

LE

H

P-5

MS

(30

m x

250

mm

, 0,2

5 µm

) 2-

6 73

-128

[R

odil,

200

8-C

]

Lod

o E

HM

C, 4

-MB

C, O

CR

L

LE

D

B-5

(60

m x

0,2

5 m

m, 0

,25

µm)

3-6

88-1

01

[Pla

gella

t, 20

06]

Pes

cad

o B

P-3

, 4-M

BC

, EH

MC

, OC

R

ESL

, SE

C

SE54

(25

m x

0,3

2 m

m, 0

,25

µm)

BG

B-5

(30

m x

0,2

5 m

m, 0

,25

µm)

3-38

0 93

-115

[B

alm

er, 2

005]

Pesc

ado

4-M

BC

, OC

R

ESL

, SE

C

- 3-

60

- [B

user

, 200

6]

Pes

cad

o B

P-3

, 4-M

BC

, EH

MC

E

SL

OPT

IMA

-5-M

S

(50

m x

0,2

mm

, 0,5

µm

)

11-3

6 72

-102

[Z

enke

r, 2

008]

6-50

70

-105

[F

ent,

2010

-B]

Pes

cad

o 4-

MB

C, O

CR

E

SL, S

EC

V

F-5

(30

m x

0,2

5mm

, 0,2

5 µm

)

5-17

98

-99

[Mot

tale

b, 2

009]

36-1

20

57-7

9 [M

otta

leb,

200

9]


88

En la determinación de compuestos polares mediante GC es habitual

recurrir al uso de reacciones de derivatización con objeto de: mejorar la

estabilidad térmica de los analitos, la resolución entre picos y, en ocasiones,

también la detectabilidad. Entre las reacciones de derivatización más frecuentes

utilizadas en combinación con cromatografía de gases figuran las sililaciones y

las acetilaciones. Las primeras consisten en la sustitución de un hidrógeno

activo (perteneciente a un grupo hidroxilo, carboxílico, amida, etc.) por un

grupo sililo, así se reduce la polaridad del analito y la posibilidad de formación

de enlaces de hidrógeno intermoleculares, además aumenta su volatilidad y

estabilidad térmica. Los agentes sililantes más utilizados son el bis‐(trimetilsilil)

trifluoroacetamida (BSTFA), el N‐metil‐N‐(trimetilsilil)trifuoroacetamida

(MSTFA) y el N‐(tert‐butildimetilsilil)‐N‐metiltrifluoroacetamida (MTBSTFA)

(Fig. 13). Las sililaciones transcurren en medio anhidro, por lo que la

derivatización va a realizarse después del proceso de extracción.

Figura 13: Estructura de los principales agentes sililantes utilizados para la

derivatización de los compuestos estudiados.

Por su parte, las reacciones de acetilación, usando anhídrido acético

como derivatizante, son adecuadas para disminuir la polaridad de especies

fenólicas y pueden llevarse a cabo tanto en medio acuoso como orgánico,

empleando en ambos casos una base como catalizador del proceso.

A continuación se resumen las aplicaciones en las que se ha usado GC en

combinación con reacciones de derivatización para la determinación de filtros

solares en matrices ambientales.

Jeon y col. [Jeon, 2006] determinaron diferentes benzofenonas por GC-MS

usando MSTFA como agente derivatizante. Se alcanzaron recuperaciones de


89

76% a 113% en muestras acuosas y de 60% a 84% en muestras de suelo. Los

límites de detección fueron de 5 ng L-1 para las muestras acuosas y de 0,1 ng g-1

para el suelo y los de cuantificación de 25 ng L-1 para agua y de 0,5 ng g-1 para

suelo.

Cuderman y col. [Cuderman, 2007] usaron el MSTFA como agente

derivatizante en la determinación de HMS y BP-3 en muestras acuosas para ser

analizadas por GC-MS. Los límites de detección obtenidos se encontraron en el

rango de 13 a 266 ng L-1.

Rodil y col. [Rodil, 2008-C] emplean BSTFA para la derivatización de EHS,

HMS y BP-3 después de su extracción de muestras de sedimentos y antes de la

determinación mediante GC-MS. Las recuperaciones obtenidas variaron del

73% al 128% y los LOD oscilaron entre 2 y 20 ng g-1 sin derivatizar y de 2 a 6 ng

g-1 derivatizando.

Tarazona y col. [Tarazona, 2010] utilizan BSTFA para la derivatización de

benzofenonas (BP-1, BP-3 y BP-8) en extractos de agua de mar antes de su

determinación mediante GC-MS. Los límites de cuantificación alcanzados

fueron de 108 a 110 ng L-1.

Kawaguchi y col. [Kawaguchi, 2008-A] desarrollaron un procedimiento para

la determinación de BP-1 y BP-3 en muestras de agua consistente en la

acetilación de los compuestos con anhídrido acético en medio básico, seguido

de su concentración y extracción con SBSE y posterior determinación mediante

GC-MS. La acetilación permitió mejorar la detectabilidad de los compuestos y

también la eficacia de extracción sobre los “Twister”. Ito y col. [Ito, 2009],

aplicaron la misma estrategia a la determinación de benzofenonas en muestras

de orina.


90

En la siguiente tabla, se indican las condiciones de derivatización y los

agentes derivatizantes empleados para la determinación de algunos filtros UV

considerados en esta Tesis doctoral, Tabla 20.

Tabla 20: Resumen de aplicaciones empleando agentes sililantes y anhídrido acético para

la derivatización de benzofenonas y salicilatos.

Analitos Matriz Condiciones de derivatización Autor

Derivatización durante la extracción

Anhídrido acético

BP-1, BP-3 Agua de río 100 µL + 1 mL carbonato potásico

(1M)

[Kawaguchi, 2008-

A]

BP-1, BP-3 Orina 20 µL + 50 µL carbonato potásico

(1 M) [Ito, 2009]

Derivatización después de la extracción

MSTFA

BP-1, BP-3, BP-

8

Agua río, lago y

suelo 50 µL, 80ºC, 30 min [Jeon, 2006]

HMS, BP-3 Agua residual

y recreacional 100 µL, 60ºC, 60 min [Cuderman, 2007]

BSTFA

BP-3, BP-1, BP-

8 Agua de mar 60 µL, 75 ºC, 30 min [Tarazona, 2010]

EHS, HMS,

BP-3 Sedimentos 50 µL, Tªamb, 1 h [Rodil, 2008-C]

En esta tesis se ha considerado la derivatización de varios filtros solares

(salicilatos y benzofenonas), usando N-(tert-butildimetilsilil)-N-

metiltrifluoroacetamida (MTBSTFA) en combinación con SPE [Negreira, 2008] y

N-metil-N-(trimetilsilil)trifluoroacetamida (MSTFA) cuando los analitos habían

sido concentrados previamente en una fibra de SPME [Negreira, 2009-A]. El

MTBSTFA presenta un grupo tert-butilo con gran impedimento estérico,

protegiendo el enlace silicio-oxígeno del ataque hidrolítico del agua y de ahí


91

que los derivados formados sean muy estables. Por su parte, el MSTFA presenta

una mayor reactividad pero da lugar a derivados menos estables.

3.2.2. Cromatografía líquida acoplada a espectrometría de

masas (LC-MS)

La cromatografía líquida de alta resolución (HPLC) nace en la década de

los sesenta y desde entonces ha desempeñado un papel fundamental en los

laboratorios analíticos debido a su enorme aplicabilidad a compuestos no

volátiles, polares, termosensibles, de alto peso molecular (cromatografía de

exclusión por tamaños), e incluso, iónicos (cromatografía de intercambio iónico)

[Rouessac, 2003] [Cela, 2002].

Su éxito se debe a la posibilidad de actuar de forma muy precisa sobre la

selectividad de la separación a través de la elección de la columna y de la

composición del eluyente, es decir, a sacar partido de las interacciones

analito/fase móvil/fase estacionaria. El acoplamiento LC-MS se empieza a

estudiar en los años 70, centrándose los veinte años posteriores en

compatibilizar las condiciones de operación de ambas técnicas y en la

innovación tecnológica de diferentes interfases. Cada interfase aplica una

aproximación diferente para resolver los dos problemas principales que plantea

este tipo de conexión:

- Eliminar la gran cantidad de gas y vapor procedente de la fase

móvil, antes de entrar a la región de alto vacío del espectrómetro de

masas.

- Transformar las moléculas en disolución en iones en fase gaseosa,

sin que se produzca su degradación térmica.

Actualmente, los fabricantes de instrumentación de LC-MS han optado

por ofertar sus equipos con varias interfases. Los más populares se incluyen

dentro de la modalidad de ionización a presión atmosférica (API), denominadas


92

electrospray (ESI), ionización química a presión atmosférica (APCI) e ionización

fotoinducida a presión atmosférica (APPI).

Las aplicaciones de LC-MS a la determinación de filtros solares en

muestras complejas suele basarse en la utilización de separaciones en fase

reversa, combinadas con un espectrómetro de masas equipado con un triple

cuadrupolo [Trenholm, 2008] [Kasprzyk-Hordern, 2008] [Rodil, 2008-B; Rodil, 2009-

C; Rodil, 2009-D] [Pedrouzo, 2009; Pedrouzo, 2010] [Nieto, 2009] [Wick, 2010]. Una

dificultad importante a la hora de generalizar la aplicación de los sistemas LC-

MS a la determinación de filtros UV es que no existe un único sistema de

ionización que proporcione eficacias elevadas para los analitos más

ampliamente estudiados en matrices ambientales. Rodil y col. [Rodil, 2009-E]

han comparado las interfases de ESI y APPI, observando que APPI ofrece

mejores límites de detección para los filtros solares menos polares y más

lipofílicos: 4-MBC, OCR, EHPABA y los salicilatos. Estos últimos presentan

muy bajo rendimiento por ESI; mientras que las benzofenonas (BP-3 y BP-4) son

ionizadas de forma más efectiva empleando ESI como fuente de ionización. Por

otro lado, Wick y col. [Wick, 2010] comparan ESI y APCI para la determinación

de benzofenonas. Aunque la eficacia de ionización en APCI se ve menos

afectada por otros componentes de la matriz, el mejor rendimiento es

proporcionado por ESI.

Otra limitación para la determinación simultánea de un número amplio

de compuestos mediante LC-MS es la mayor o menor tendencia de los

diferentes analitos a ionizarse en modo positivo o negativo. A modo de

ejemplo, la eficacia de ionización de la BP-3 en ESI es mucho mayor en modo

positivo que en negativo; por el contrario, sus metabolitos y la BP-4 presentan

una mayor tendencia a ionizarse en modo negativo [Rodil, 2008-A] [Nieto, 2009]

[Zenker, 2008] [Pedrouzo, 2009; Pedrouzo, 2010]. Dentro de esta tesis se ha

desarrollado un método para la determinación simultánea de BP-4, BP-3 y otras

benzofenonas hidroxiladas, relacionadas estructuralmente con la BP-3 (BP-1,


93

BP-2, BP-6 y BP-8), mediante LC-MS/MS empleando de forma simultánea los

modos de ionización positivo y negativo en ESI [Negreira, 2009-B].

A continuación se resumen en tablas las características de los métodos de LC-

MS y LC-MS/MS aplicados a la determinación de filtros solares en muestras

ambientales acuosas (Tabla 21) y sólidas (Tabla 22).


94

Tab

la 2

1: R

esum

en d

e la

s ap

licac

ione

s de

LC

-MS

y LC

-MS/

MS

a la

det

erm

inac

ión

de fi

ltro

s so

lare

s en

mue

stra

s ac

uosa

s.

Mat

riz

acu

osa

An

alit

os

Téc

nic

a

det

ecci

ón

Inte

rfas

e C

olu

mn

a em

ple

ada

Fase

móv

il

(mod

ific

ador

)

LO

D

(ng

L-1

) R

efer

enci

a

Río

, mar

y

resi

dua

l

BP

-4

LC

-

MS/

MS

ESI

(-)

Sym

met

rySc

hiel

d R

P-1

8

(150

x 2

,1 m

m, 3

,5 µ

m)

Agu

a:M

etan

ol

(5 m

M a

ceta

to a

món

ico)

7-46

1-30

[Rod

il, 2

008-

A]

[Rod

il, 2

009-

B]

BP

-3, 4

-MB

C,

IAM

C, E

HM

C,

EH

PA

BA

, OC

R

ESI

(+)

Res

idua

l B

P-3

LC

-

MS/

MS

ESI

(+)

Sine

rgy

Max

-RP

(250

x 4

,6 m

m)

Agu

a:M

etan

ol

(0,1

% á

cid

o fó

rmic

o)

1,0

[Tre

nhol

m, 2

008]

Supe

rfic

ial

y re

sid

ual

BP

-1, B

P-2

,

BP

-3, B

P-4

LC

-

MS/

MS

ESI

(-)

AC

QU

ITY

UP

LC

BE

H C

18

(100

x 1

mm

, 1,7

µm

)

Agu

a:M

etan

ol

(0,5

% H

Ac,

0,5

% N

H3)

0,

1-5

[Kas

przy

k-H

ord

ern,

200

8]

Lag

o y

resi

dua

l

BP-

3, B

P-4

, 4-M

BC

,

HM

S, IA

MC

, EH

MC

,

OC

R, E

HPA

BA

LC

-

MS/

MS

AP

PI

C8

Ecl

ipse

XD

B

(150

x 4

,6 m

m, 5

µm

) A

gua:

Met

anol

0,

4-16

[R

odil,

200

9-C

]

Río

y

resi

dua

l

BP

-1, B

P-8

BP

-3, E

HPA

BA

, OC

R

UP

LC

-

MS/

MS

ESI

(-)

ESI

(+)

C18

Ecl

ipse

XD

B

(50

x 4,

6 m

m, 1

,8 µ

m)

Agu

a:M

etan

ol

(áci

do

acét

ico,

pH

2,8

)

1-4

(río

)

1-10

(efl

u.)

5-20

(inf

lu.)

[Ped

rouz

o, 2

009]

2,5

(río

)

5-10

(res

.) [P

edro

uzo,

201

0]


95

Tab

la 2

2: R

esum

en d

e la

s ap

licac

ione

s de

LC

-MS

y LC

-MS/

MS

para

la d

eter

min

ació

n de

filt

ros

sola

res

en m

atri

ces

sólid

as.

Mat

riz

An

alit

os

Téc

nic

a

det

ecci

ón

Inte

rfas

e C

olu

mn

a em

ple

ada

Fase

móv

il

(mod

ific

ador

)

LO

D

(ng

g-1 )

R

efer

enci

a

Pes

cad

o B

P-3

, EH

MC

,

4-M

BC

, OC

R

LC

-

MS/

MS

ESI

(+)

Perf

ectS

il 12

0 O

DS-

2

(125

x 3

mm

)

Agu

a:M

etan

ol

(0,1

% á

cid

o ac

étic

o)

2,4

[Mei

nerl

ing,

200

6]

Pes

cad

o

BP

-3, 4

-MB

C, E

HM

C

BP-

1, B

P-2

, BP

-4

LC

-

MS/

MS

ESI

(+)

ESI

(-)

Zor

bax

SB-C

18

(150

x 3

mm

, 3,5

µm

)

Agu

a:A

ceto

nitr

ilo

(0,1

% á

cid

o fó

rmic

o)

78-2

05

[Zen

ker,

200

8]

6-50

[F

ent,

2010

-B]

Pes

cad

o B

P-3

L

C-M

S E

SI (+

) A

lltim

a

(250

x 2

,1 m

m, 5

µm

)

Agu

a:A

ceto

nitr

ilo

(0,1

% á

cid

o fó

rmic

o)

2 [K

won

, 200

9]

Lod

o B

P-3

, EH

S, H

MS,

IA

MC

, EH

MC

, OC

R,

4-M

BC

, EH

PA

BA

,

LC

-

MS/

MS

AP

PI

Ecl

ipse

XD

B C

8,

(150

x 4

,6 m

m, 5

µm

) A

gua:

Met

anol

0,

3-18

[R

odil,

200

9-D

]

Lod

o B

P-3

, OC

R, E

HP

AB

A

UP

LC

-

MS/

MS

ESI

(+)

Zor

bax

(50

× 4,

6 m

m, 1

,8 µ

m)

Agu

a:M

etan

ol

(áci

do

acét

ico)

1,

5-3,

5 [N

ieto

, 200

9]

BP

-1, B

P-8

E

SI (-

)

Lod

o

BP

-3

LC

-

MS/

MS

ESI

(+)

Syne

rgi F

usi

on-R

P 80

(150

× 3

mm

, 4 µ

m)

Agu

a (1

0 m

M fo

rmat

o

amón

ico)

:Ace

toni

trilo

(0,1

% á

cid

o fó

rmic

o)

0,75

-7,5

[W

ick,

201

0]

BP-

1, B

P-2

, BP

-4

ESI

(-)

Introducción-Fotoiniciadores

97

B. Fotoiniciadores

1. ASPECTOS GENERALES

El segundo grupo de compuestos considerados en este trabajo son los

llamados fotoiniciadores, compuestos químicos que se añaden a las tintas

aplicadas sobre envases alimentarios con el fin de acelerar el proceso de secado

de las impresiones realizadas sobre los mismos. En el año 2005, en Italia, se

produjo una alarma social al detectar restos de un fotoiniciador de esos

componentes químicos en leche infantil distribuida en envases tipo brick. Éste

fue la Isopropil tioxantona (ITX). Esta preocupación condujo a que las

autoridades retiraran del mercado más de dos millones de litros de leche en

Italia, España, Francia y Portugal.

En el caso anterior, parece que el problema de contaminación se originó

al almacenar los envases impresos enrollados en grandes bobinas. Durante este

proceso, la cara externa impresa entra en contacto con la cara interna del

tetrabrick; de esa manera, parte del ITX utilizado en el sistema de impresión

puede depositarse en el interior del envase, desde donde se produjo su

posterior migración a la leche.

El mecanismo de actuación de los fotoiniciadores se basa en su

descomposición en presencia de luz ultravioleta, lo que genera radicales libres

que activan la polimerización de los componentes de la tinta. Tradicionalmente,

las tintas incluían en su formulación disolventes orgánicos o agua, que tenían

que ser eliminados mediante un proceso de secado relativamente lento.

Actualmente, la mayoría de los procesos de secado están basados en la

exposición a radiación ultravioleta. Este proceso es iniciado por un

fotoiniciador, que es altamente propenso a absorber un fotón de luz y crear

especies activas (radicales, cationes o aniones) que inician y completan dicho

proceso de polimerización y secado de la tinta [Sanches-Silva, 2008-B] [Bradley,

2006].


98

2. ESTRUCTURA Y PROPIEDADES

Generalmente, los fotoiniciadores presentan uno o dos anillos aromáticos

en sus moléculas con estructuras parecidas a los filtros UV empleados en

protectores solares. De hecho, el 2-etil-4-dimetilaminobenzoato (EDMAB) es un

derivado del PABA. Los analitos incluidos en este estudio han sido:

benzofenona (BP), 1-hidroxiciclohexil-fenilcetona (CPK), 4-metilbenzofenona (4-

MBP), 2-isopropiltioxantona (ITX), 2,2’-dimetoxi-2-fenilacetofenona (DMPA),

EDMAB y el EHPABA, usado también como filtro solar y de estructura muy

similar al EDMAB. A continuación, se muestran sus estructuras químicas

(Figura 14) y sus propiedades físico-químicas más relevantes (Tabla 23). Los

datos correspondientes al EHPABA han sido presentados en páginas anteriores

de esta memoria.

4-MBPCPK

ITX DMPA

BP

EDMAB

Figura 14: Estructuras de los fotoiniciadores objeto de estudio.


99

Tabla 23: Propiedades físico-químicas de los fotoiniciadores estudiados.

Propiedad Analito

BP CPK 4-MBP ITX DMPA EDMAB

Nº CAS 119-61-9 947-19-3 134-84-9 5495-84-1 24650-42-8 10287-53-3

Peso molecular

(g mol-1) 182,20 204,26 196,24 254,35 256,30 193,24

Densidad

(g cm-3) 1,09 ± 0,06 1,14 ± 0,06 1,07 ± 0,06 1,20 ± 0,06 1,12 ± 0,06 1,06 ± 0,06

pKa - 13,2 ± 0,2 - - - 2,6 ± 0,1

Entalpía de

vaporización (KJ mol-1) 55 ± 3 61 ± 3 57 ± 3 65 ± 3 62 ± 3 54 ± 3

logKow 3,2 ± 0,3 2,3 ± 0,3 3,6 ± 0,3 5,3 ± 0,3 4,8 ± 0,5 3,1 ± 0,2

Presión de vapor

(mTorr) 0,823 0,0368 0,194 0,043 0,0106 1,43

Punto de ebullición

(ºC) 305,4 ± 0,0 339 ± 25 328 ± 11 399 ± 32 371 ± 42 297 ± 23

Punto de fusión

(ºC) 124 ± 14 144 ± 16 141 ± 14 217 ± 11 169 ± 14 115 ± 14

Solubilidad molar

(mol L-1), pH 1 7,1E-4 2,4E-3 3,4E-4 2,0E-6 2,4E-4 0,037

Solubilidad molar

(mol L-1), pH 4-10 7,1E-4 2,4E-3 3,4E-4 2,0E-6 2,4E-4 1,1E-3

3. PRESENCIA EN ALIMENTOS

Después de la alarma social provocada por la detección de ITX en leche,

aparecieron publicaciones en las que determinan este compuesto no sólo en

leche [Sagratini, 2008] [Gil-Vergara, 2007] [Bagnati, 2007] [Sun, 2007] [Allegrone,

2008] [Sanches-Silva, 2008-A] [Shen, 2009] sino también en otros alimentos como

yogures [Sun, 2007], zumo [Sagratini, 2008] e incluso té [Sun, 2007] y vino

[Sagratini, 2008]. El ITX aparece como una mezcla de dos isómeros pero el más

frecuentes es el 2-ITX encontrando niveles de hasta 346 ± 102 µg L-1 en leches

distribuidas en envases tipo brick [Bagnati, 2007].

Además del ITX, se encontraron trazas de otros fotoiniciadores como

EHPABA en leche [Gil-Vergara, 2007] [Sagratini, 2008] [Sanches-Silva, 2008-A] y


100

zumo [Sagratini, 2008]; BP en leche [Sagratini, 2008] [Shen, 2009], zumo

[Sagratini, 2008] y vino [Sagratini, 2008] y CPK en vino [Sagratini, 2008]. En la

siguiente tabla (Tabla 24), se recogen las concentraciones encontradas para los

compuestos anteriores en alimentos envasados.

Tabla 24: Niveles de fotoiniciadores en muestras de leche, zumo, yogur, té y vino.

Tipo de

muestra

Lugar del

estudio

Compuestos

detectados

Concentración

(µg L-1) Referencia

Leche España ITX

EHPABA

2,5-189

8-101 [Gil-Vergara, 2007]

Leche Italia

2-ITX

4-ITX

249-346

17-9

[Bagnati, 2007]

Leche

China ITX

<0,5-4,78

[Sun, 2007] Zumo <0,5-84,30

Yogur <0,5

Té <0,5

Leche Italia ITX 4-53 [Allegrone, 2008]

Leche

Italia

EHPABA

BP

0,13-0,8

5,25-39

[Sagratini, 2008]

Zumo

EHPABA

BP

2-ITX

0,14-0,8

5-90

0,2

Vino

BP

2-ITX

CPK

5,5-217

0,2-0,24

1,2

Leche España EHPABA

ITX

78-214

37-213 [Sanches-Silva, 2008-A]

Leche China

2-ITX 0,81-8,87

[Shen, 2009] BP 2,84-18,35

EHPABA 0,32


101

4. METODOLOGÍA ANALÍTICA PARA LA DETERMINACIÓN DE

FOTOINICIADORES EN ALIMENTOS

Dentro de este apartado se presentan las características básicas de los

métodos propuestos en la bibliografía para la determinación de fotoiniciadores

en muestras de alimentos envasados. La mayoría de los trabajos recopilados en

la bibliografía se centran en alimentos líquidos (leche, zumos, bebidas

alcoholicas), semisólidos (yogures, margarinas) y sólidos (fundamentalmente

leches infantiles en polvo). De forma genérica, se estima que el contenido

lípidico o etanólico (caso del vino) de los alimentos anteriores podría favorecer

la liberación de los fotoiniciadores que pudiesen estar adheridos a la cara

interna de los envases multicapa, debido a problemas de contaminación

durante su almacenamiento, o a una potencial migración desde la cara externa

impresa, hacia la interna en contacto con el alimento. El interés por la

determinación de fotoiniciadores en leche se debe por un lado a la frecuente

distribución de este alimento en envases tipo brick y a su consideración como

nutriente imprescindible en la dieta de bebés y niños.

La complejidad de las muestras de alimentos hace necesaria una etapa

previa para separar los analitos de otros componentes de la matriz y, en la

mayoría de los casos, un paso adicional de limpieza para mejorar la selectividad

de la extracción antes de la etapa de determinación. La técnica más utilizada

para la extracción de fotoiniciadores en leche [Morlock, 2006] [Bagnati, 2007]

[Sanches-Silva, 2008-A] y otras matrices, como yogur [Morlock, 2006] [Benetti,

2008] es la LLE. En la mayoría de los trabajos se emplea acetonitrilo como

disolvente de extracción, permitiendo la recuperación de los compuestos con un

bajo nivel de lípidos, posibilitando la inyección del extracto en un equipo de

HPLC [Bagnati, 2007] [Sanches-Silva, 2008-B] [Benetti, 2008]. Otros autores pasan

los extractos a través de cartuchos de sílica [Sagratini, 2008], C18 (1 g, 6 mL)

[Allegrone, 2008] o Oasis HLB (60 mg, 3 mL) [Sun, 2007] [Gallart-Ayala, 2008]

[Shen, 2009] antes de su inyección en el sistema cromatográfico. En algunos

trabajos, Morlock y col. [Morlock, 2006] y Gil-Vergara y col. [Gil-Vergara, 2007]


102

usan PLE para la extracción de ITX y EHDAB de muestras de leche y yogur. El

disolvente de extracción fue ciclohexano:acetato de etilo (1:1) [Morlock, 2006] y

acetato de etilo [Gil-Vergara, 2007]. En ambos casos, la temperatura de

extracción fue de 100 ºC y la técnica de determinación fue LC-MS. La

determinación mediante GC-MS sólo se llevó a cabo en dos trabajos [Sagratini,

2008] [Allegrone, 2008] y en ambos casos emplean SPE para la extracción y

limpieza de los extractos.

A continuación se presenta una tabla resumen donde se recogen los

analitos considerados, las técnicas empleadas y los LOQ alcanzados en la

bibliografía en relación a la determinación de fotoionizasores en alimentos

(Tabla 25 y Tabla 25 cont.).

En la mayoría de las aplicaciones se consumen volúmenes considerables

de disolvente orgánico además de necesitar una etapa previa de limpieza antes

de su inyección en GC-MS [Allegrone, 2008]. En esta Tesis Doctoral se evaluó,

por primera vez, la SPME como técnica de extracción y concentración de siete

fotoiniciadores: BP, CPK, 4-MBP, EDMAB, EHPABA, 2,2-DMPA y ITX en

muestras de leche [Negreira, 2010-B].


103

T

abla

25:

Res

umen

de

met

odol

ogía

s em

plea

das

para

la d

eter

min

ació

n de

foto

inic

iado

res

en a

limen

tos.

M

atri

z A

nal

itos

E

xtra

cció

n

Pu

rifi

caci

ón

T. D

et.

LO

D

(µg

L-1

)

Ref

eren

cia

Lec

he,

yogu

r IT

X

PL

E: 4

mL

o 4

g m

uest

ra,

4 g

hyd

rom

atri

x,

cicl

ohex

ano:

ace

tato

de

etilo

(1:1

), 10

0 ºC

- L

C-(

ESI

)-M

S 0,

13

[M

orlo

ck, 2

006]

Mar

gari

na,

soja

IT

X

1 g

de

mue

stra

,

1 m

L a

ceto

nitr

ilo, 4

0ºC

, 30

min

-

LC

-(E

SI)-

MS

1

[Mor

lock

, 200

6]

Lec

he,

yogu

r,

té, z

umo

2-IT

X

4-IT

X

10 g

de

mu

estr

a,

100

mL

ace

toni

trilo

: agu

a (1

:1),

1% r

eact

ivo

Car

rez,

agit

ació

n y

cent

rifu

gaci

ón

SPE

: Oas

is H

LB

(60

mg)

,

4 m

L a

ceto

nitr

ilo

LC

-(E

SI)-

MS/

MS

0,15

[Sun

, 200

7]

Lec

he

2-IT

X

4-IT

X

100

µL d

e m

uest

ra,

300

µL a

ceto

nitr

ilo,

agit

ació

n y

cent

rifu

gaci

ón

- L

C-(

ESI

)-

MS/

MS

0,75

[Bag

nati

, 200

7]

Lec

he

infa

ntil

ITX

, BP

,

CP

K, D

MP

A,

EH

DA

B

10 m

L d

e m

uest

ra,

1,5

mL

NH

3 y 2

x 2

0 m

L h

exan

o -

LC

-MS

20- 3

0

[San

ches

-Silv

a, 2

008-

A]

Lec

he

2-IT

X

4-IT

X

EH

DA

B

PL

E: 1

mL

de

mu

estr

a,

2 g

aren

a, 9

g N

a 2SO

4,

acet

ato

de

etilo

a 1

00ºC

- L

C-M

S/M

S 0,

1-0,

3

[G

il-V

erga

ra, 2

007]

-, no

dis

poni

ble


104

Tab

la 2

5 co

nt.:

Res

umen

de

met

odol

ogía

s em

plea

das

para

la d

eter

min

ació

n de

foto

inic

iado

res

en a

limen

tos.

M

atri

z A

nal

itos

E

xtra

cció

n

Pu

rifi

caci

ón

T. D

et.

LO

D

(µg

L-1

) R

efer

enci

a

Zu

mo

de

nara

nja

ITX

, EH

PAB

A

DM

PA

, CP

K

10 m

L d

e m

uest

ra, c

once

ntra

n a

5 m

L

y ex

trae

n co

n 5

mL

ace

toni

trilo

-

LC

-UV

20

-30

[San

ches

-Silv

a, 2

008-

B]

Com

ida

infa

ntil,

zum

os, l

eche

2-IT

X

4-IT

X

2,5

g d

e m

ues

tra,

10 m

L a

ceto

nitr

ilo, r

eact

ivo

Car

rez,

agit

ar y

cen

trif

ugar

SPE

: Oas

is H

LB

(60

mg)

,

6 m

L a

ceto

nitr

ilo

LC

-(E

SI)-

MS/

MS

0,00

2-

0,01

3

[Gal

lart

-Aya

la, 2

008]

Lec

he, z

umos

,

vino

2-IT

X, E

HPA

BA

,

EH

DA

B, B

P, C

PK

EL

L: 2

5 m

L d

e m

ues

tra

3 x

30 m

L n

-hex

ano

SPE

: Síli

ca (1

g),

2 x

2 m

L h

exan

o:

acet

ato

de e

tilo

(30:

70)

LC

-(E

SI)-

MS/

MS

2-10

0

[Sag

rati

ni, 2

008]

GC

-MS

0,2-

1

Yog

ur

2-IT

X

4-IT

X

10 g

de

mue

stra

,

40 m

L a

ceto

nitr

ilo,

agit

ació

n y

cent

rifu

gaci

ón

-

LC

-(E

SI)-

MS

- [B

enet

ti, 2

008]

Lec

he

ITX

SPE

: 1 m

L d

e m

ues

tra

+ 9

mL

agu

a:

met

anol

(8:1

), ca

rtu

chos

Isol

ute

C18

(1 g

), 5

mL

ace

toni

trilo

- G

C-

MS/

MS

0,1

[Alle

gron

e, 2

008]

Lec

he

BP

, IT

X, C

PK,

EH

PA

BA

2 g

de

mue

stra

, 10

mL

ace

toni

trilo

,

soni

caci

ón y

cen

trif

uga

ción

SPE

: Oas

is H

LB

(60

mg)

,

6 m

L a

ceto

nitr

ilo

LC

-(E

SI)-

MS/

MS

0,1-

0,74

[S

hen,

200

9]

-, no

dis

pon

ible

Metodología desarrollada-Muestras acuosas

107

III. METODOLOGÍA DESARROLLADA

A. Filtros solares

1. MUESTRAS ACUOSAS

1.1. Introducción

La SPE es la técnica más utilizada para la concentración de filtros solares

en muestras acuosas; sin embargo, requiere el consumo de volúmenes elevados

de muestra y moderados de disolventes orgánicos. Las técnicas de

microextracción permiten subsanar los inconvenientes anteriores,

proporcionando una excelente sensibilidad. No obstante, algunas de la

modalidades empleadas, tales como SPME, SBSE, MEPS y ciertas versiones de

LPME, presentan cinéticas de extracción lentas, utilizan dispositivos frágiles

con un coste considerable y/o pueden requerir importantes adaptaciones en los

sistemas cromatográficos para transferir los analitos desde el dispositivo de

extracción a la columna cromatográfica.

Una alternativa a las técnicas de microextracción anteriormente citadas

es la denominada DLLME que permite la rápida extracción de especies

orgánicas de muestras acuosas. Aunque existen numerosas aplicaciones de

DLLME a diferentes especies orgánicas, incluso fue usada para la

determinación de BP-3 y otras 3 benzofenonas hidroxiladas [Tarazona, 2010], las

condiciones de extracción no han sido optimizadas para filtros UV menos

polares, detectados con frecuencia en el medio acuático. Así pues, en esta Tesis

se ha desarrollado un método de preparación de muestra rápido y sensible para

la determinación de 8 filtros solares y el BzS en matrices acuosas basado en

DLLME [Negreira, 2010-A].

La segunda técnica considerada en esta parte de la memoria ha sido la

microextracción en fase sólida sobre siliconas de grado técnico, en formato

disco. El bajo coste de este material permite desecharlo después de cada ciclo de


108

extracción, evitando los problemas de contaminación cruzada entre muestras,

típicos de las técnicas de SPME y SBSE. Además de optimizar las condiciones

de extracción para un grupo amplio de filtros UV se ha comparado las eficacias

de extracción proporcionadas por siliconas, en diferentes formatos, y los

correspondientes a los Twister recubiertos con PDMS, empleados en SBSE.

La tercera aplicación presentada en esta sección de la memoria se ha

centrado en la combinación de SPME con GC-MS para la determinación de

varias benzofenonas (BP-1, BP-3, BP-8) y dos salicilatos (EHS y HMS), a niveles

de los ng L-1, en muestras de agua. Con objeto de mejorar los LOQs

previamente obtenidos en la bibliografía [Felix, 1998], en esta Tesis se ha

desarrollado un método de SPME combinado con derivatización on-fibre,

empleando reacciones de sililación para la determinación de EHS, HMS, BP-3 y

otras 2 benzofenonas hidroxiladas (BP-1 y BP-8) [Negreira, 2009-A].

Dentro de la familia de las benzofenonas, existen compuestos que no

pueden ser determinados mediante cromatografía de gases (ej. BP-4, presente a

niveles importantes en muestras de agua residual y poco eliminada en las

estaciones depuradoras convencionales) y otros cuya derivatización y

determinación mediante cromatografía de gases presenta muchas dificultades,

ej. BP-2 y BP-6. Por ello, se ha desarrollado un método de LC-MS/MS para la

determinación de varias benzofenonas, empleando SPE como técnica de

extracción. Las condiciones de trabajo han sido optimizadas con objeto de

compatibilizar la extracción y determinación simultánea de los analitos

considerados. [Negreira, 2009-B].

Adicionalmente a la optimización de métodos de preparación de

muestra, es importante conocer la estabilidad de los filtros UV en el medio

acuático. Dado que algunos presentan grupos fenólicos y aminos en sus

estructuras parece factible que puedan dar lugar a la formación de derivados

halogenados al reaccionar con el cloro empleado en la desinfección del agua. De

hecho, Sakkas y col. [Sakkas, 2003] detectaron la formación de derivados

clorados del EHPABA en agua de piscina, aunque no aportaron datos de la

velocidad de reacción ni de la estabilidad de los productos generados. En esta


109

Tesis, se llevaron a cabo estudios de degradación química de EHPABA, EHS y

BP-3 en presencia de cloro libre y se identificaron varios productos de

transformación [Negreira, 2008]. Las muestras fueron concentradas mediante

SPE, empleando GC-MS para seguir la evolución temporal de los compuestos

de interés e identificar los productos de reacción formados.

Además del desarrollo de metodología analítica, en los trabajos que se

presentan a continuación se aporta un número relevante de datos relativos a la

presencia de diversos filtros solares en muestras de agua superficial, de piscina

y aguas residuales urbanas, discutiendo también su estabilidad en las

estaciones depuradoras de aguas residuales.


110


acuosas

Separación de fases

Muestra agua10 mL

Adición 1 mL acetona (dispersante) conteniendo

60 µL clorobenceno (extractante)

Agitación durante 1 min

Inyección 2 µL GC-MS (modo SIM)

Centrifugación3 min, 3200 rpm

Formación de la emulsión

Recogida de la gota sedimentada

Figura 15: Esquema seguido para la determinación de filtros solares en agua mediante

DLLME y GC-MS.


111

Muestra agua100 mL (10% metanol)

EXTRACCIÓN: inmersión,

14 horas a Tª ambiente

Inyección 9 µL PTV GC-MS

DESORCIÓN:30 min a Tª ambiente con

200 µL acetato de etilo

Retirada del disco de silicona y

secado con papel

Figura 16: Esquema empleado en la determinación de filtros solares en agua mediante

extracción con siliconas, en formato no comercial, y GC-MS.


112

Derivatización on-fibre20 µL MSTFA10 min a 45 ºC

Muestra agua10 mL, pH 3

Inmersión, fibra PDMS-DVB, 30 min a Tª ambiente

con agitación

Retirada de la fibra, secar con papel

Extracción de los analitos

Derivatización

Desorción de la fibra2 min a 270 ºC GC-MS/MS

Figura 17: Esquema para la determinación de filtros solares en agua mediante SPME,

derivatización y GC-MS/MS.


113

•Flujo: 10 mL/min.

•Secado cartuchos: corriente N2; 30 min.

•Ajustar pH 3.

•Acondicionamiento cartuchos.

5 mL metanol-acetato amónico (5 mM)

5 mL agua milli-Q

Inyección

LC—MS/MS (15 μL)

Oas

is H

LB

, 60

mg

Muestra agua

Elución 3 mL metanol-acetato amónico

(5 mM)

..

Evaporación a sequedad

Re-disolución con 1 mL metanol:agua milli-Q (1:1)2,5 mM acetato amónico

Figura 18: Esquema de trabajo para la determinación de filtros solares en agua mediante

SPE y LC-MS/MS.


114

• Adicción de cloro libre (0 3 mg/L)

Medida de HClO/ClO- con fotómetro

• Adicción de analitos (aprox. 10 50 ng/mL)

Muestra agua tamponada (rango pH 6,2 8,2)

Reacción a Tª ambiente(0 180 min)

Parada de la reacción con tiosulfato sódico

Acidificar a pH 4,5 y concentrar mediante SPE (ver Figura 20)

Figura 19: Metodología empleada en los estudios de halogenación de filtros solares.


115

Elución 2 mL

•Flujo: 10 mL/min.

•Secado cartuchos: corriente N2; 30 min.

•Elución con acetato de etilo o CH2Cl2.

•Ajustar pH 4,5.

•Acondicionamiento cartuchos.

3 mL acetato de etilo o CH2Cl2

3 mL metanol

3 mL agua Milli-Q

InyecciónGC-MS (2 μL)

Oas

is H

LB

, 60

mg

Muestra agua

..

PREPARACIÓN MUESTRA

DERIVATIZACIÓDERIVATIZACIÓDERIVATIZACIÓDERIVATIZACIÓDERIVATIZACIÓDERIVATIZACIÓDERIVATIZACIÓ

20 µL de MTBSTFA

(Tª ambiente, 5 min.)

DERIVATIZACIÓN (BP-3 y BP-1)

500 µLEXTRACTO

+Inyección

GC-MS (2 μL)

Figura 20: Condiciones de extracción y determinación empleadas en los estudios de

halogenación.

1.3. Publicación:

DISPERSIVE LIQUID-LIQUID

MICROEXTRACTION FOLLOWED BY GAS

CHROMATOGRAPHY-MASS SPECTROMETRY

FOR THE RAPID AND SENSITIVE DETERMINATION

OF UV FILTERS IN ENVIRONMENTAL

WATER SAMPLES

N. Negreira, I. Rodríguez, E. Rubí, R. Cela

Analytical and Bioanalytical Chemistry 398 (2010) 995

(doi:10.1007/s00216-010-4009-9)


119

Dispersive liquid-liquid microextraction followed by gas chromatography-

mass spectrometry for the rapid and sensitive determination of UV filters in

environmental water samples

N. Negreira, I. Rodríguez*, E. Rubí, R. Cela

Departamento de Química Analítica, Nutrición y Bromatología, Instituto

de Investigación y Análisis Alimentario (IIAA), Universidad de Santiago de

Compostela, Santiago de Compostela 15782, Spain.

Abstract

The performance of the dispersive liquid-liquid microextraction

(DLLME) technique for the determination of eight UV filters and a structurally

related personal care species, benzyl salicylate (BzS), in environmental water

samples is evaluated. After extraction, analytes were determined by gas

chromatography combined with mass spectrometry detection (GC-MS).

Parameters potentially affecting the performance of the sample preparation

method (sample pH, ionic strength, type and volume of dispersant and

extractant solvents) were systematically investigated using both, multi- and

univariant optimization strategies. Under final working conditions, analytes

were extracted from 10 mL water samples by addition of 1 mL of acetone

(dispersant) containing 60 L of chlorobenzene (extractant), without modifying

either the pH, or the ionic strength of the sample. Limits of quantification

(LOQs) between 2 and 14 ng L-1, inter-day variability (evaluated with relative

standard deviations, RSDs) from 9 to 14% and good linearity up to

concentrations of 10000 ng L-1 were obtained. Moreover, the efficiency of the

extraction was scarcely affected by the type of water sample. With the only

exception of 2-ethylhexyl-p-dimethylaminobenzoate (EHPABA), compounds

were found in environmental water samples at concentrations between 6 ± 1 ng

L-1 and 26 ± 2 ng mL-1.


120

Keywords: UV filters; Dispersive liquid-liquid microextraction; GC-MS;

water samples.

1. Introduction

Reduction of the ozone layer increases the amount of UV radiation

reaching the surface of the earth, and thus the concern about its harmful effects

on human health. The so-called organic UV filters are compounds designed to

absorb radiation in the UV region, protecting human skin against direct

exposure to deleterious wavelengths of sunlight [1-2]. These compounds are

incorporated in sunscreen products, as well as in shampoos, lipsticks, facial

day-creams, after-shave products, make-up formulations and even in plastics,

varnishes and clothes to enhance their light stability [1-4]. These uses have led

to the appearance of significant concentrations of several UV filters in the

aquatic environment [5-6]. Particularly, compounds belonging to

benzophenone and salicylate classes, as well as 3-(4-methylbenzylidene)

camphor (4-MBC), 2-ethylhexyl-p-methoxycinnamate (EHMC) and octocrylene

(OCR) have been systematically detected in surface, bathing and municipal

wastewater samples, at concentrations ranging from the low ng L-1 up to several

ng mL-1 [7-12]. The latter three species and 2-hydroxy-4-methoxybenzophenone

(BP-3) have been also reported in river and lake sediments [13], sludge [14-15]

and aquatic organisms [3,16]. Moreover, the results of several studies suggest

that certain UV filters, which have been found in the above refereed matrices,

behave as endocrine disrupters [17-19]. Together, the above findings indicate

the need to evaluate the fate of the UV filters in the aquatic media in order to

assess their mobility, bio-accumulation potential and possible effects on aquatic

organisms.

The determination of UV filters in water samples is normally

accomplished with solid-phase extraction (SPE) followed by gas

chromatography (GC) [3,8-9] or liquid chromatography (LC) [7,10-11], both

combined with mass spectrometry (MS). Although SPE offers considerable


121

advantages over liquid-liquid extraction (LLE), it still requires large sample

volumes (from 0.3 to 1 L), a moderate consumption (10-15 mL) of organic

solvents for analytes desorption, and further clean-up to compensate for its

limited selectivity when applied to wastewater [3,9]. Miniaturization of solid-

and liquid-phase extraction methodologies overcomes some of the above

limitations offering a similar performance to SPE. In this way, solid-phase

microextraction (SPME) [20-22], stir-bar sorptive extraction (SBSE) [23-24],

microextraction using packed sorbents (MEPS) [25] and liquid-phase

microextraction (LPME), with non-porous membranes [26] as well as in the

single drop modality [27-28], have been applied to the extraction of UV filters

from aqueous matrices. In most of the above applications an excellent

sensitivity has been achieved; however, relatively long extraction steps, fragile

extraction devices and/or dedicated equipment are required.

An alternative to the above microextraction techniques is the so-called

dispersive liquid-liquid microextraction (DLLME) [29]. This approach allows

the rapid extraction of organic species from aqueous solutions by addition of a

binary mixture of an extractant and a dispersant. The first is a high-density,

water insoluble solvent; whereas, acetone, methanol or acetonitrile are normally

used as dispersants. When this mixture comes in contact with the water sample

a cloudy stage, consisting of fine particles of the extractant dispersed into the

aqueous phase, is formed. After centrifugation, the high-density solvent settles

at the bottom of the extraction tube. Then, a fraction of the sedimented phase is

injected in the chromatographic system [29]. Following this initial study on

DLLME [29], more than one hundred applications have been published. Most of

them have been compiled in two recent reviews [30-31]. To the best of our

knowledge, DLLME has been only applied to the determination of BP-3 and

some hydroxylated by-products [32]; however, extraction conditions have not

been optimized for less polar UV filters, often found in the aquatic

environment.


122

The aim of this study is to develop a rapid and sensitive sample

preparation method for the extraction of eight UV filters, commonly included in

the formulation of sunscreen products and belonging to different chemical

classes, as well as a structurally related compound (benzyl salicylate, BzS) from

environmental water samples. GC-MS was used as the determination

technique. Parameters affecting the efficiency of the extraction process have

been evaluated systematically and the performance of the developed method

compared with previously reported approaches involving other

microextraction techniques and with SPE. An overview of the concentrations

for selected UV filters in a significant number of water samples is also provided.

2. Experimental

2.1. Standards and material

HPLC-grade methanol and acetonitrile were supplied by Merck

(Darmstadt, Germany). Trace analysis grade carbon tetrachloride (CCl4), 1,1,1-

trichloroethane (CH3CCl3), chlorobenzene (C6H5Cl) and acetone were obtained

from Aldrich (Milwaukee, WI, USA). Sodium chloride was also provided by

Aldrich. Ultrapure water was obtained from a Milli-Q system (Millipore,

Billerica, MA, USA). Standards of 2-ethylhexyl salicylate (EHS), 3,3,5-

trimethylcyclohexyl salicylate (Homosalate, HMS), 2-ethylhexyl-p-

dimethylaminobenzoate (EHPABA), BzS, BP-3, 4-MBC, EHMC and OCR were

acquired from Aldrich (Milwaukee, WI, USA) and Merck. Isoamyl-p-

methoxycinnamate (IAMC) was kindly provided by Dr. R. Rodil (University of

La Coruña, Spain). Individual solutions (ca. 1000 g mL-1) and mixtures of the

above analytes were prepared in methanol. Further dilutions were also made in

methanol and used to prepare the aqueous standards employed during

optimization of extraction conditions. Another series of standards was made in

chlorobenzene.

Glass tubes (12 mL volume) with a conical bottom and a screw cap,

furnished with a polytetrafluoroethylene (PTFE)-lined septum, were acquired

from Afora (Barcelona, Spain).


123

2.2. Samples

River, swimming-pool and wastewater samples, obtained from the inlet

and outlet streams of an urban sewage treatment plant (STP) equipped with

primary and activated sludge units, were processed throughout this study.

Samples were collected in amber glass bottles (250 mL), previously rinsed with

acetone and ultrapure water, and transported immediately to the laboratory.

Then, they were passed through glass fibre filters followed by cellulose acetate

ones (0.45 m pore size) and stored in the dark, at 4º C, for a maximum of 48

hours before being concentrated. In the case of swimming-pool water samples,

200 mg of sodium thiosulphate were added immediately after collection in

order to remove free chlorine. During sampling and sample preparation

operations gloves were used to minimize contamination problems arising from

the usage of UV filters and BzS in hand creams, bath gels and other personal

care products.

2.3. Sample preparation

Optimization of DLLME conditions was performed with aqueous

solutions of target species in ultrapure water. Unless otherwise is stated, the

concentration of UV filters in these aqueous standards was 5 ng mL-1. Aliquots

of 10 mL were poured into conical bottom glass tubes, previously wrapped

with aluminium foil to prevent photo-isomerization of UV filters during

extraction. The extraction mixture was added using a micropipette. Both, uni-

and multivariant optimization strategies, based on the use of experimental

factorial designs, were considered in order to assess the effect of different

experimental variables on the performance of the extraction process. In the

second case, the Statgraphics Centurion software (Manugistics, Rockville, MD,

USA) was used for data analysis.

Under optimized conditions, the binary extraction mixture consisted of 1

mL of acetone (dispersant) containing 60 L of chlorobenzene (extractant). After

addition to the water sample (or aqueous standard), the conical tube was

closed, shaken manually for 1 min and centrifuged at 3200 rpm (ca. 900 g) for 3


124

min. During centrifugation, the dispersed droplets of chlorobenzene settled at

the bottom of the vessel, building-up a drop (45 L volume) which was further

recovered and transferred to a conical insert with a 25 µL micro-syringe fitted

with a bevel-tip needle. A conventional autosampler for liquid samples was

used to inject a fraction (2 L) of this extract in the GC-MS system.

2.4. GC-MS equipment

Analytes were determined with a GC-MS system consisting of an Agilent

(Wilmington, DE, USA) 7890A gas chromatograph connected to a quadrupole

mass spectrometer (Agilent MS 5975C), operated in the SIM mode. Separations

were carried out in an HP-5ms type capillary column (30 m x 0.25 mm i.d., df:

0.25 m) supplied by Agilent. Helium (99.999 %) was used as the carrier gas at a

constant flow of 1.2 mL min-1. The GC oven was programmed as follows: 125 ºC

(held for 1 min), increased at 9 ºC min-1 to 280 ºC (held for 10 min).

Temperatures of the electron impact ionization source and the quadrupole mass

analyzer were set at 230 ºC and 150 ºC, respectively. Standards and sample

extracts in chlorobenzene (2 L) were injected in the splitless mode (splitless

time 1 min), with the injector port at 280 ºC. Retention times and m/z ratios of

selected ions for each compound are summarized in Table 1.

Table 1. Retention times and monitored ions for target species.

Compound Retention time (min) aSelected ions (m/z)

EHS

HMS

BzS

IAMC

BP-3

4-MBC

EHPABA

EHMC

OCR

9.77

10.46,10.71

10.54

b12.28

12.28

b12.57

14.57

b15.01

18.11

120,138

120,138

91

178,161

227,151

254,239

165,277

178,161

360, 232, 249

aUnderlined ions were used for quantification purposes.

bRetention times corresponding to (E) forms.


125

2.5. Method characterization and samples quantification

The performance of the optimized methodology was characterized in

terms of extraction efficiencies (%) and enrichment factors (EFs). EFs were

defined as the ratio between the concentration of each species in the

chlorobenzene extract (Cs), which was determined against a calibration curve

obtained for standards in chlorobenzene, and that added to the water sample

(C0). Efficiencies (%) were calculated as the ratio between the mass of each

species in the settled phase (Cs x Vs) and that added to the sample (C0 x Va),

multiplied by 100 [29-31]. Being Vs and Va the volumes of the settled

chlorobenzene extract and the water sample, respectively.

Potential changes in the performance of the DLLME process among

ultrapure and environmental water samples (matrix effects) were assessed

using relative recoveries (%). They were calculated as the difference between

responses measured for spiked and non-spiked aliquots of different

environmental water samples divided by those obtained for ultrapure water

standards with the same addition level and multiplied by 100 [26]. Obtained

values stayed around 100%; therefore, the levels of UV filters in real-life

samples were quantified by comparison with the responses obtained for

aqueous standards in ultrapure water, containing increased amounts of target

analytes (up to 10000 ng L-1) and subjected to the optimized DLLME process.

Procedural blanks (ultrapure water) were tested with each series of

samples in order to assess the significance of contamination problems.

Responses measured for real samples were corrected with those obtained for

procedural blanks.

3. Results and discussion

3.1. Optimization of DLLME conditions

3.1.1. Effects of extractant, sample pH, salt addition and extraction time

The type of extractant is one of the most important parameters during

optimization of DLLME methods. In a first series of assays (n=3 replicates),

mixtures of 0.9 mL of acetone with 0.1 mL of 3 different extractants (CCl4,


126

CH3CCl3 and C6H5Cl) were added to 10 mL aliquots of aqueous standards

prepared in ultrapure water. The lowest responses for all compounds

corresponded to CCl4 (data not shown) which was therefore rejected as

extractant.

The effects of sample pH, salt (NaCl) addition, extraction time (defined

as the period during which the sample is shaken after addition of the binary

extraction mixture and before centrifugation) and type of extractant on the

performance of the method were simultaneously investigated using a two

levels 24-1 type experimental fractional factorial design, with four replicates of

the central point. Low and high levels for each variable are shown in Table 2. In

all experiments, a binary mixture of acetone (0.9 mL) with 0.1 mL of the

corresponding extractant was used. The experimental conditions explored

within the domain of this design exerted a limited effect on the volume of the

sedimented phase (85 ± 4 L); thus, the peak areas for each compound were

directly considered as variables response.


127

Tab

le 2

. Exp

erim

enta

l dom

ain

and

stan

dard

ized

mai

n ef

fect

s of

fact

ors

cons

ider

ed in

the

24-1

exp

erim

enta

l fra

ctio

nal f

acto

rial

des

ign.

Fact

or

Lev

el

Stan

dar

diz

ed m

ain

effe

ct v

alue

s

L

ow

Hig

h E

HS

HM

S B

zS

IAM

C

BP-

3 4-

MB

C

EH

PA

BA

E

HM

C

OC

R

NaC

l (%

) 0

5 -0

.58

-0.8

3 -0

.53

-1.1

0.

36

-0.0

10

-0.6

0 -0

.64

-1.1

Sam

ple

pH

2 6

0.17

0.

54

0.29

0.

64

1.1

0.78

0.

001

0.14

-0

.07

Ext

ract

ant

CH

3CC

l 3

C6H

5Cl

7.0

a 8.

5 a

8.9

a 15

a

17 a

11

a

10 a

6.

6 a

5.3

a

Ext

ract

ion

tim

e (m

in)

1 10

-0

.32

-0.2

0 0.

054

-0.0

67

0.18

0.

54

-0.2

9 -0

.09

-0.1

0

a St

atis

tica

lly s

igni

fica

nt fa

ctor

s at

the

95%

con

fid

ence

leve

l.


128

Standardized main effects summarized in Table 2 indicate the same trend

for all species. The type of extractant was the only variable with a significant

influence (95% confidence level) on the performance of the extraction, with the

highest responses corresponding to C6H5Cl. On the other hand, the addition of

a 5% of NaCl to water samples and the extraction time played negligible effects

on the obtained responses; thus, they were maintained at their lower levels (0%

of NaCl and 1 min of shaking, respectively), to simplify sample handling and

also to speed up the extraction process. Finally, between pH 2 and 6 the

efficiency of the extraction also remained unaffected. For this latter factor,

considering that (1) environmental water samples can present pH values above

6 units and that (2) salicylates and BP-3 show a slightly acidic behaviour (their

pKa are comprised between 7.5 and 8 units), an additional series of extractions

was carried out with aqueous standards adjusted to pHs 6 and 9. Again, no

significant differences were observed in the responses of target analytes (Fig. 1);

therefore, the decision was to process the real water samples without any pH

adjustment. Data shown in Fig. 1 demonstrated that the ionized forms of EHS,

HMS and BP-3 are also effectively extracted by chlorobenzene. Likely, the

capability of this solvent to establish - interactions with the aromatic region

of target compounds significantly contributes to enhance the efficiency of the

extraction process. Two-factor interactions, obtained from the experimental

factorial study, also showed very low standardized values, far below the

statistically significant threshold (data not shown).


129

0.0E+00

1.0E+05

2.0E+05

3.0E+05

4.0E+05

5.0E+05

6.0E+05

7.0E+05

8.0E+05

9.0E+05

EHS HMS BzS IAMC BP‐3 4‐MBC EHPABA EHMC OCR

Compound

Peak area

pH 6 pH 9

Fig. 1. Responses obtained for ultrapure water samples (spiked concentration 5 ng mL-1)

adjusted at two different pHs, n=3 replicates.

3.1.2. Selection of dispersant

Fig. 2 shows the responses obtained for spiked aliquots of ultrapure

water using the above optimized conditions and considering methanol or

acetonitrile as alternative to acetone. Overall, the lowest responses

corresponded to methanol, the most polar of the 3 tested dispersants, whereas

no differences were noticed between acetone and acetonitrile. Taking into

account its lower cost and larger commercial availability, acetone was

maintained as dispersant.


130

0.0E+00

2.0E+05

4.0E+05

6.0E+05

8.0E+05

1.0E+06

EHS HMS BzS IAMC BP-3 4-MBC EHPABA EHMC OCR

Compound

Pe

ak

are

a

Methanol Acetone Acetonitrile

Fig. 2. Comparison of responses (peak areas) obtained with 0.1 mL of chlorobenzene as

extractant and 0.9 mL of three different dispersants: methanol, acetone and acetonitrile,

n= 3 replicates.

3.1.3. Volumes of extractant and dispersant

The optimal volumes of dispersant (acetone) and extractant

(chlorobenzene) were evaluated with a multilevel experimental factorial design.

The investigated levels were 0.5, 1, 1.5 and 2 mL for acetone and 0.06 and 0.1

mL of chlorobenzene. Each experiment was performed in duplicate. Analysis of

variance (ANOVA) was used in order to determine the contribution of both

factors, and their first order interaction, to responses (peak areas) measured for

target species. P-values for extractant and dispersant volumes remained under

0.05 for all the analytes, indicating a significant influence, at the 95% confidence

level, on their responses. Furthermore, the interaction acetone-chlorobenzene

was also statistically significant for four (IAMC, BP-3, EHPABA and OCR) of

the nine considered species. Numerical results of ANOVA are provided as

supplementary data, Table S1. Fig. 3 shows the interaction plots corresponding

to the average responses of selected compounds. As observed, the volume of

extractant was the factor affecting in a higher extension to the obtained peak

areas, with the most favourable situation corresponding to the lower volume of


131

C6H5Cl. Considering 0.1 mL of C6H5Cl, the analytes peak areas decreased

steadily with the volume of acetone; however, for 0.06 mL of C6H5Cl the

maximum responses were obtained using 1 mL of acetone, Fig. 3.

Consequently, 1 mL of acetone and 60 L of chlorobenzene were fixed as

optimal values of both variables. Under these conditions, the volume of the

settled phase (45 L) was large enough to be transferred to a conical insert and

loaded in the autosampler of the GC-MS instrument.

Table S1. F-ratios and P-values resulting from ANOVA of data obtained in the

multilevel experimental factorial design of extractant and dispersant volumes.

Compound Main effects Two-factor interaction

Acetone volume Chlorobenzene volume Acetone-chlorobenzene

F-Ratio P-Value F-Ratio P-Value F-Ratio P-Value

EHS

HMS

BzS

IAMC

BP-3

4-MBC

EHPABA

EHMC

OCR

9.5

9.7

10.4

13.2

18.1

10.20

10.8

11.07

23.7

0.0052

0.0049

0.0039

0.0018

0.0006

0.0041

0.0034

0.0032

0.0002

299

381

272

427

304

279

370

318

466

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

3.2

3.4

4.0

5.1

4.4

4.0

4.4

2.9

6.8

0.086

0.072

0.052

0.029

0.041

0.051

0.042

0.10

0.014


132

Acetone (mL)

47

67

87

107

127

Pea

k ar

ea (

x 10

4 )

0.5 1 1.5 2

EHS BP-3

18

28

38

48

58

0.5 1 1.5 2

Pea

k ar

ea (

x 10

4 )

Acetone (mL)

4-MBC

15

18

21

24

27

30

33

0.5 1 1.5 2

Acetone (mL)

Pea

k ar

ea (

x 10

4 )

EHPABA

65

85

105

125

145

0.5 1 1.5 2

Acetone (mL)

Pea

k ar

ea (

x 10

4 )

OCR

37

47

57

67

77

87

97

0.5 1 1.5 2

Acetone (mL)

Pea

k ar

ea (

x 10

4 )

EHMC

62

82

102

122

142

0.5 1 1.5 2

Acetone (mL)

Pea

k ar

ea (

x 10

4)

60 l of C6H5Cl

100 l of C6H5Cl

Fig. 3. Interaction plots obtained from the multilevel factorial experimental design for

selected compounds.

3.2. Performance of the method

Table 3 shows the extraction efficiencies and the EFs obtained for

ultrapure water standards under optimized conditions. Efficiencies varied from

74 to 95% and the averaged EFs stayed between 170 and 200 times. Likely, a

further reduction in the volume of extractant would provide even higher EFs.

Consequently, lower LOQs may be achieved. Nevertheless, this possibility was


133

not investigated since extracts (settled drops) with a significant lower volume

have to be manually injected.

Table 3. Extraction efficiencies (%) and average enrichment factors (EFs) provided by

the DLLME method for ultrapure water standards, n=4 replicates.

Compound Extraction efficiencies (%) ± SD

Average EFs a1000 ng L-1 a200 ng L-1

EHS

HMS

BzS

IAMC

BP-3

4-MBC

EHPABA

EHMC

OCR

90 ± 4

84 ± 4

79 ± 7

84 ± 4

81 ± 7

83 ± 4

92 ± 5

83 ± 5

85 ± 6

90 ± 4

85 ± 4

74 ± 2

95 ± 5

82 ± 5

81 ± 2

83 ± 6

84 ± 9

79 ± 6

200 ± 9

187 ± 9

170 ± 10

199 ± 10

181 ± 13

182 ± 7

194 ± 12

186 ± 16

182 ± 13

aConcentration level

The precision of the method was evaluated with aqueous standards of

different concentrations and extracted within the same day (repeatability) or on

different days (reproducibility). In the first case, relative standard deviations

(RSDs) from 2 to 11% were observed for triplicate extractions of aqueous

solutions fortified at four different concentrations: 50, 100, 1000 and 5000 ng L-1,

Table 4. Reproducibility was assessed with an aqueous standard (500 ng L-1)

processed in triplicate during 5 consecutive days. In this case, RSDs between 9

and 14% were attained. The linearity of the overall method was investigated

with standards prepared at eight different concentrations from 10 to 10000 ng L-

1. Fig. 4 shows the GC-MS chromatogram corresponding to a 50 ng L-1 aqueous

calibration solution. Within the above range, the plot of peak areas versus the

concentration of each analyte fitted a linear model with R2 values higher than

0.998, Table 4.


134

Table 4. Repeatability (n=3 extractions in the same day), reproducibility (n=15

extractions in 5 different days), linearity and limits of quantification (LOQs) of the

method.

Compound

Repeatability,

(RSDs, %)

Reproducibility

(RSDs, %)

Linearity, R2

a (10- 10000,

8 levels)

LOQs

(ng L-1) a50 a100 a1000 a5000 a500

EHS

HMS

BzS

IAMC

BP-3

4-MBC

EHPABA

EHMC

OCR

5

3

11

10

11

9

9

7

3

2

4

2

3

4

2

5

3

4

2

4

2

3

4

2

5

3

4

3

3

2

2

3

2

3

3

4

11

12

10

10

9

12

10

10

14

0.9995

0.9996

0.9997

0.9996

0.9993

0.9993

0.9994

0.9991

0.9982

2

9

3

4

7

3

3

14

10

aConcentration level (ng L-1)

EHS

HMS

(E) 4-MBC

EHPABABzS

OCR

(E) EHMC

BP-3, (E) IAMC

HMS

10.00 11.00 12.00 13.00 14.00 15.00 16.00 17.000

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

2400

Time (min)

Abundance (counts)

Fig. 4. Total ionic current (TIC) chromatogram corresponding to a 50 ng L-1 aqueous

standard.


135

Procedural blanks, performed with samples of ultrapure water, showed

the absence of significant contamination problems for most species. However,

traces of EHMC and OCR were systematically detected in blanks extracts, even

when different ultrapure water samples were tested. The exact source of this

contamination could not be identified; however, our finding is concordant with

the previous observations of Rodil et al. using SBSE as concentration technique

[23]. For those analytes not affected by contamination processes, the LOQs were

estimated as the concentration providing a peak with a signal to noise ratio

(S/N) 10 times higher than the baseline noise. In the case of EHMC and OCR,

LOQs were calculated as 10 times the standard deviations of blanks signals

(n=5 replicates) divided by the slope of their calibration curves. Achieved LOQs

remained below 14 ng L-1 for all species, Table 4. Globally, these values are

similar to those reported for SBSE followed by GC-MS determination (LOQs

from 2 to 26 ng L-1) [23-24], and liquid-phase microextraction (LPME) through

non-porous membranes in combination with LC-MS/MS detection (LOQs from

3 to 45 ng L-1) [25]. The combination of MEPS with GC-MS offers slightly higher

limits of detection (from 35 to 87 ng L-1) [26] and the same comment is also valid

for some SPME based methods [20]. SPE followed by GC-MS detection [9] and

LC-MS/MS [7] renders LOQs in the same range of values than those

summarized in Table 4.

Fig. 5 shows the relative recoveries of the optimized DLLME method for

river water, swimming-pool water (Fig. 5A) and wastewater samples (Fig. 5B)

fortified at different concentration levels. In the case of swimming-pool and

river water relative recoveries between 87% and 109% were achieved (Fig. 5A).

Similar results were attained for treated wastewater; whereas, values from 80%

to 117% were obtained for raw urban wastewater (Fig. 5B). In all cases, the

relative standard deviations remained around or below 10%. In view of these

data, it is evident that the efficiency of the sample preparation method is

scarcely affected by the characteristics of the matrix; therefore, external

calibration, against fortified aliquots of ultrapure water, was selected as

quantification technique.


136

0

20

40

60

80

100

120


Compound

Rel

ativ

e re

cove

ry (

%)

Treated wastewater, 500 ng L-1 Treated wastewater, 1000 ng L-1

Raw wastewater, 2000 ng L-1

0

20

40

60

80

100

120


Compound

Rel

ativ

e re

cove

ry (

%)

Swimming pool, 250 ng L-1 River water, 100 ng L-1

River water, 500 ng L-1 River water, 1000 ng L-1

A

B

Fig. 5. Relative recoveries (%), with their standard deviations, obtained for river,

swimming-pool water (A) and wastewater samples (B), spiked at different concentration

levels, n= 3 replicates.


137

3.3. Application to environmental water samples

The proposed method was applied to the analysis of 21 samples

corresponding to wastewater (codes 1-6), rivers without (codes 7-12) and with

bathing areas (codes 13-17) and public swimming-pools (codes 18-21), Table 5.

Most samples were collected in the summer of 2009 and all were processed

within 48 hours, after received. Except EHPABA, which remained below the

LOQ of the method in all samples, the rest of species were quantified in some of

the processed samples, with the highest detection frequency and the maximum

concentrations corresponding to OCR, followed by 4-MBC and EHMC. As

shown in Fig. 6, 4-MBC and EHMC were found in environmental samples as a

mixture of isomers (E and Z forms). The sum of their peak areas was compared

with the calibration curve obtained for the E form, which is the commercial

isomer included in the formulation of personal care products.


138

Tab

le 5

. Con

cent

rati

ons

(ng

L-1 )

, wit

h th

eir

stan

dard

dev

iati

ons,

mea

sure

d in

env

iron

men

tal w

ater

sam

ples

, n=

3

repl

icat

es.

Cod

e T

ype

Sam

plin

g d

ate

Con

cent

rati

on (n

g L

-1) ±

SD

EH

S H

MS

BzS

IA

MC

B

P-3

4-

MB

C

EH

MC

O

CR

1

R.W

. 4/

05/0

9 6

± 1

N.D

. 28

± 1

N

.D.

17 ±

2

7 ±

1 N

.D.

66 ±

6

2 T

. W

4/05

/09

N.D

. N

.D.

N.D

. N

.D.

N.D

. N

.D.

N.D

. 15

± 1

3

R.W

. 9/

07/0

9 18

± 3

11

± 2

18

9 ±

21

N.D

. 11

1 ±

8 10

3 ±

9 51

± 6

15

8 ±

9 4

T. W

9/

07/0

9 N

.D.

N.D

. 8

± 1

N.D

. 10

± 1

66

±3

N.D

. 40

± 5

5

R.W

. 27

/7/0

9 32

± 1

20

± 1

13

7 ±

2 N

.D.

227

± 10

15

3 ±

3 12

4 ±

2 44

0 ±

34

6 T

. W

27/7

/09

N.D

. N

.D.

N.D

. N

.D.

58 ±

3

94 ±

5

N.D

. 59

± 4

0 7

Riv

er

9/07

/09

N.D

. N

.D.

6 ±

1 N

.D.

N.D

. N

.D.

N.D

. N

.D.

8 R

iver

9/

07/0

9 N

.D.

N.D

. 7

± 1

N.D

. 8

± 1

7 ±

1 N

.D.

18 ±

1

9 R

iver

9/

07/0

9 N

.D.

N.D

. 79

± 2

N

.D.

42 ±

3

7 ±

1 N

.D.

46 ±

4

10

Riv

er

9/07

/09

60 ±

5

N.D

. 31

± 4

N

.D.

N.D

. N

.D.

N.D

. N

.D.

11

Riv

er

26/0

7/09

12

± 1

N

.D.

12 ±

1

N.D

. 12

± 2

13

4 ±

4 81

3 ±

7 80

2 ±

33

12

Riv

er

26/0

7/09

N

.D.

N.D

. N

.D.

N.D

. N

.D.

N.D

. N

.D.

79 ±

19

13

Riv

er

11/0

7/09

N

.D.

N.D

. N

.D.

N.D

. N

.D.

N.D

. 10

4 ±

16

209

± 43

14

R

iver

11

/07/

09

N.D

. N

.D.

N.D

. N

.D.

N.D

. N

.D.

20 ±

1

48 ±

6

15

Riv

er

11/0

7/09

N

.D.

N.D

. N

.D.

N.D

. N

.D.

N.D

. 19

± 2

36

± 3

16

R

iver

25

/07/

09

31 ±

5

18 ±

3

N.D

. N

.D.

N.D

. 73

± 1

2 12

7± 3

42

56 ±

311

17

R

iver

25

/07/

09

62 ±

8

124

± 6

N.D

. N

.D.

15 ±

2

1132

± 1

43

362

± 41

33

90 ±

459

18

S.

P.

11/0

7/09

19

± 1

N

.D.

N.D

. 65

5 ±

48

N.D

. 10

70 ±

93

1274

± 1

40

1421

± 2

61

19

S.P.

25

/07/

09

17 ±

1

13 ±

1

14 ±

1

537

± 8

N.D

. 12

50 ±

16

1462

± 5

3 27

52 ±

167

20

S.

P.

25/0

7/09

17

8 ±

7 45

0 ±

20

N.D

. N

.D.

2326

± 1

18

4035

± 9

2 20

7 ±

7 25

967

± 16

22

21

S.P.

25

/07/

09

12 ±

1

N.D

. N

.D.

33 ±

2

N.D

. 43

± 3

10

7 ±

12

2997

± 3

26

Det

ecti

on fr

eque

ncy

52%

29

%

48%

14

%

43%

67

%

62%

90

%

R

.W. r

aw w

aste

wat

er

T.W

. tre

ated

was

tew

ater

S.

P. s

wim

min

g-po

ol w

ater

N

.D. b

elow

det

ecti

on li

mit

s


139

Code 17Code 11Procedural blank

9.60 9.80 10.00 10.20 10.40 10.60 10.80 11.00

406080

100120140160180200220240260280

Time (min)

Abundance m/z 120

EHS HMS

Time (min)11.50 11.70 11.90 12.10 12.30

405060708090

100110120130140150160170180

Abundancem/z 227

BP-3

13.40 13.80 14.20 14.60 15.00 15.400

500100015002000250030003500400045005000550060006500

Abundance

Time (min)

(Z) EHMC

(E) EHMC

m/z 178

11.80 12.20 12.60 13.00 13.400

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

Abundance

Time (min)

(Z) 4-MBC

(E) 4-MBC

m/z 254

18.6017.60 17.80 18.00 18.20 18.400

400

800

1200

1600

2000

2400

2800

3200

3600

4000

4400

Abundance

Time (min)

m/z 360

OCR

Fig. 6. Selected ion chromatograms for a procedural blank and samples code 11 and 17

(Table 5).

Samples 1 to 6 were collected in the same STP, from a non-costal city of

100,000 inhabitants, equipped with primary and activated sludge treatments.

Although data summarized in Table 5 correspond to grab samples, obtained


140

without considering the residence time of the STP, all species appear to be

removed to a considerable extent in the plant, with 4-MBC displaying the

lowest degradation efficiency. This trend is consistent with the results reported

for time-average samples from several STPs in Switzerland [3].

Analytes were also found in rivers without bathing or recreational areas

(codes 7-12). In some cases, these rivers received treated municipal wastewaters

(codes 8-9); however, the highest concentrations of 4-MBC, EHMC and OCR

were measured in a small stream (code 11) flowing through a highly

industrialized area and a 400,000 inhabitants city in the Northwest of Spain.

Some of the river water samples, collected in bathing areas (codes 13-17),

and particularly all swimming-pool water samples (codes 18-21), presented

very high concentrations of several UV filters (4-MBC, EHMC and OCR),

whereas the salicylates remained at lower levels. In fact, the concentration of

OCR in sample code 20 surpassed the linear response range of the method.

Thus, this sample was processed twice. First directly to quantify those species

presented at low levels, and then after a 5-fold dilution to establish the

concentration of OCR. It is worth noting that IAMC was detected only in

swimming-pool water but not in sewage or in river water. This finding suggest

a limited usage of this UV filter in sunscreen products in comparison with

EHMC, the other cinnamate type UV filter involved in this study.

4. Conclusions

DLLME constitutes an advantageous, rapid and inexpensive alternative

for the sensitive extraction of selected UV filters from environmental water

samples. The proposed method requires a small volume of sample, extraction is

completed in a few minutes (using a four positions centrifuge sample

preparation remains around 2-3 min per sample), dedicated instrumentation

and sorbents are not required and the extractant is compatible with GC-based

determination techniques. Moreover, the extraction efficiency is barely affected

by the type of water sample, allowing the comparison against spiked aliquots of

ultrapure water (aqueous standards) as quantification technique. The volume


141

and the type of extractant and dispersant solvents were the variables with the

highest influence on the efficiency of the extraction, whereas the ionic strength

and the pH of the sample did not affect the extraction process. Data obtained

for different environmental water samples revealed the presence of eight of the

nine investigated species in the aquatic environment, with the highest

occurrence frequency corresponding to OCR, followed by 4-MBC and EHMC.

Acknowledgements

Financial support from the Spanish Government and E.U. FEDER funds

(project CTQ2009-08377) is acknowledged. N.N. is grateful for an FPU grant

from the Spanish Ministry of Education and Science.


142

References

1. Salvador A, Chisvert A (2005) Anal Chim Acta 537:1-14

2. Giokas DL, Salvador A, Chisvert A (2007) Trends Anal Chem 26:360-374

3. Balmer ME, Buser HR, Müller MD, Poiger T (2005) Environ Sci Technol 39:953-962

4. Salvador A, Chisvert A (2005) Anal Chim Acta 537:15-24

5. Díaz-Cruz MS, Llorca M, Barceló D (2008) Trends Anal Chem 27:873-887

6. Richardson SD (2008) Anal Chem 80:4373-4402

7. Rodil R, Quintana JB, López-Mahía P, Muniategui-Lorenzo S, Prada-Rodríguez D (2008)

Anal Chem 80:1307-1315

8. Cuderman P, Health E (2007) Anal Bioanal Chem 387:1343-1350

9. Poiger T, Buser HR, Balmer ME, Bergqvist PA, Müller MD (2004) Chemosphere 55:951-

963

10. Trenholm RA, Vanderford BJ, Drewes JE, Snyder SA (2008) J Chromatogr A 1190:253-

262

11. Zenker A, Schmutz H, Fent K (2008) J Chromatogr A 1202:64-74

12. Giokas DL, Sakkas VA, Albanis TA (2004) J Chromatogr A 1026:289-293

13. Rodil R, Moeder M (2008) Anal Chim Acta 612:152-159

14. Plagellat C, Kupper T, Furrer R, Alencastro LF, Grandjean D, Tarradellas J (2006)

Chemosphere 62:915-925

15. Nieto A, Borrull F, Marcé RM, Pocurull E (2009) J Chromatogr A 1216:5619-5625

16. Buser HR, Balmer ME, Schmid P, Kohler M (2006) Environ Sci Technol 40:1427-1431

17. Klammer H, Schlecht C, Wuttke W, Schmutzler C, Gotthardt I, Köhrle J, Jarry H (2007)

Toxicology 238:192-199

18. Díaz-Cruz MS, Barceló D (2009) Trends Anal Chem 28:708-717

19. Schlumpf M, Cotton B, Conscience M, Haller V, Steinmann B, Lichtensteiger W (2001)

Environ Health Perspect 109:239-244

20. Lambropoulou DA, Giokas DL, Sakkas VA, Albanis TA, Karayannis MI (2002) J

Chromatogr A 967:243-253

21. Felix T, Hall BJ, Brodbelt JS (1998) Anal Chim Acta 371:195-203

22. Negreira N, Rodríquez I, Ramil M, Rubí E, Cela R (2009) Anal Chim Acta 638:36-44

23. Rodil R, Moeder M (2008) J Chromatogr A 1179:81-88

24. Kawaguchi M, Ito R, Honda H, Endo N, Okanouchi N, Saito K, Seto Y, Nakazawa H

(2008) J Chromatogr A 1200:260-263

25. Rodil R, Schrader S, Moeder M (2009) J Chromatogr A 1216:4887-4894

26. Moeder M, Schrader S, Winkler U, Rodil R (2010) J Chromatogr A 1217:2925-2932

27. Okanouchi N, Honda H, Ito R, Kawaguchi M, Saito K, Nakazawa H (2008) Anal Sci

24:627-630


143

28. Vidal L, Chisvert A, Canals A, Salvador A (2010) Talanta 81:549-555

29. Rezaee M, Assadi Y, Hosseini M-R M, Aghaee E, Ahmadi F, Berijani S (2006) J


30. Rezaee M, Yamini Y, Faraji M (2010) J Chromatogr A 1217:2342-2357

31. Zang XH, Wu QH, Zhang MY, Xi GH, Wang Z (2009) Chin J Anal Chem 37:161-168

32. Tarazona I, Chisvert A, León Z, Salvador A (2010) J Chromatogr A, in press;

doi:10.1016/j.chroma.2010.05.047

1.4. Publicación:

SILICONE DISCS AS DISPOSABLE

ENRICHMENT PROBES FOR GAS

CHROMATOGRAPHY-MASS SPECTROMETRY

DETERMINATION OF UV FILTERS

IN WATER SAMPLES


Analytical and Bioanalytical Chemistry, submitted


147

Silicone discs as disposable enrichment probes for gas chromatography-mass

spectrometry determination of UV filters in water samples



de Investigación y Análisis Alimentario, Universidad de Santiago de


Abstract

This work describes an effective, low solvent consumption and

affordable sample preparation approach for the determination of eight UV

filters in surface and wastewater samples. It involves sorptive extraction of

target analytes in a disposable, technical grade silicone disc (5 mm diameter x

0.6 mm thickness) followed by organic solvent desorption, large volume

injection (LVI) and gas chromatography-mass spectrometry determination (GC-

MS). Parameters affecting the performance of sampling and desorption steps

are systematically investigated and the observed trends are related with the

polarity of UV filters. Final working conditions involved overnight extraction of

100 mL samples, containing a 10% of methanol, followed by analytes

desorption with 0.2 mL of ethyl acetate. The method provides linear responses

between the limits of quantification (from 0.003 to 0.040 ng mL-1) and 10 ng mL-

1, an intra-day precision below 13%, and low matrix effects for surface,

swimming pool and treated sewage water samples. Except in the case of

benzophenone-3 (BP-3), extraction efficiencies are in reasonable agreement with

values predicted from octanol-water partition coefficients of UV filters, sample

and sorbent volumes. Globally, no differences are noticed between the

extraction yields provided by the bulk, technical grade silicone sorbent and

those corresponding to polydimethylsiloxane covered stirring bars. Several UV

filters were found in surface and sewage water samples, with the maximum

concentrations corresponding to octocrylene (OC).


148

Keywords: sample preparation, sorptive extraction, silicone sorbents, UV

filters, water analysis

1. Introduction

Organic UV filters belong to the group of personal care chemicals

considered as emerging environmental pollutants [1]. Direct release in bathing

areas and diffuse discharges through domestic wastewater represent the main

inputs of these species in the aquatic media [2-3]. The available information

related with removal efficiencies of UV filters in sewage treatment plants (STPs)

[4], occurrence in surface water samples, [5-6], accumulation in solid matrices

and biota [7-11] and eco-toxicological effects [12] indicate the need of long term

monitoring studies in order (1) to detect seasonal variations in surface and

sewage water, (2) to fully understand their removal and/or accumulation

routes and (3) to assess the impact of UV filters in the environment. The

feasibility of such long term monitoring studies, involving the analysis of many

samples, depends on the availability of cost affordable analytical

methodologies.

Trace level determination of UV filters in water samples is normally

accomplished by gas (GC) or liquid chromatography (LC) followed by mass

spectrometry (MS), after an effective sample preparation step. Sample

preparation contributes significantly not only to the performance of the overall

analytical method but also to the cost, complexity and automation possibilities

of the analysis. The above comments justify the plethora of extraction and

microextraction approaches proposed for the concentration of UV filters in

water samples. Among solid-phase microextraction modalities, stir bar sorptive

extraction (SBSE) combines some interesting features, such as high extraction

yields for medium and low polarity compounds (as it is the case of many UV

filters), simple set up, unattended operation and suitability of the


149

polydimethylsiloxane (PDMS) coated bars (Twisters) for thermal and organic

solvent desorption [13-14]. So far, several research groups have developed

successful applications of SBSE, in combination with GC-MS and LC-MS/MS,

for the determination of UV filters in water samples [15-18]. Despite the above

commented features, two relevant limitations of the SBSE technique are (1) the

relatively high cost of Twisters and (2) the risk of cross-contamination when

they are re-used. As an alternative to the commercial version of SBSE, in 2004,

Popp and co-workers [19-20] reported the use of disposable silicone sorbents

(DSS) as high capacity probes for the extraction and concentration of organic

compounds from water samples. After they early works, bulk silicone sorbents

have been used for direct and headspace (HS) sorptive extraction of organic

pollutants and natural compounds from water and food samples [21-24], as

well as for time-average passive sampling [25].

Technical grade silicone sorbents are inexpensive; thus, they are

normally considered as single use devices, avoiding carry-over problems

related with the incomplete desorption of target analytes. Moreover, the

amount and the format (tube, rods, sheets) of the sorbent can be adjusted for

each particular application. On the other hand, they may contain other

polymers, in addition to pure PDMS, as well as variable percentage of

additives, which can modify the efficiency of the extraction and/or interfere

with the determination of target analytes [26-27].

In this study, an effective, low cost sample preparation method for the

sensitive determination of eight UV filters in water samples is proposed. It

involves sorptive extraction of target analytes with a DSS followed by solvent

desorption, large volume injection (LVI) and GC-MS determination. Extraction

conditions were optimized using a small silicone disc (5 mm diameter x 0.6 mm

thickness), which can be easily desorbed in an autosampler vessel (1.5 mL

volume) with a few microlitres of a suitable organic solvent. Thereafter, the

efficiency of the extraction process was compared with (1) the theoretical values


150

predicted from octanol-water partition coefficients (Kow) of UV-filters, and (2)

with the extraction yields provided by another silicone sorbent, in a rod format,

and with those achieved with the commercial Twisters. Finally, the applicability

of the method was demonstrated with river and sewage water samples.

2. Experimental

2.1. Solvents, standards and sorbents

HPLC-grade methanol, ethyl acetate and dichloromethane for trace

analysis were supplied by Merck (Darmstadt, Germany). UV filters standards

were purchased from Aldrich (Milwaukee, WI, USA) and Merck. Table 1

summarizes the analytes considered in this study and some properties of

relevance to optimize the sorptive extraction process. Individual solutions of

each species (ca. 1000 g mL-1) were prepared in methanol. Further dilutions

and mixtures of the UV filters were also dissolved in methanol and used to

fortify the water samples employed during optimization of extraction

conditions. Another set of standards (from 1 to 1000 ng mL-1) was made in ethyl

acetate and used to determine the amount of each analyte in the organic extract

obtained from silicone sorbents.

Technical grade silicone sorbents were acquired from Goodfellow (Bad

Nauheim, Germany) in two different formats: cord with a diameter of 1 mm

and sheets with a 0.6 mm thickness. Accordingly to the supplier, the

composition of the sorbent corresponds to phenyl-vinyl-methyl polysiloxane;

however, the relative percentage of the above substituents in the polysiloxane

skeleton is not provided. Rods (15 mm length x 1 mm diameter) were cut using

a sharp blade; discs (5 mm diameter x 0.6 mm thickness) were obtained by

pressing the silicone sheet with a sharp hollow punch (internal diameter 5 mm).

Discs and rods contained the same volume of silicone (ca. 12 L) and their cost

stayed below 0.1 Euro per unit. Twisters covered with 24 L of PDMS were from

Gerstel (Mühlheim, Germany). Twisters and silicone sorbents were soaked twice

with a methanol: acetone (1:1) solution, for 15 min, and then conditioned

overnight at 200 ºC, before being used for first time. Silicone discs and rods


151

were discarded after each extraction-desorption cycle. Twisters were

additionally soaked with methanol:acetone (1:1) after each use.

Oasis HLB (60 mg) solid-phase extraction (SPE) cartridges were acquired

from Waters (Milford, MA, USA).

Table 1. Abbreviated names, octanol-water partition coefficients (Kow), pKa values, GC-

MS retention times and quantification ions of target analytes.

Analyte Abbreviation aLog

Kow apKa

Retention

time (min)

Quantification

Ions (m/z)

2-Ethylhexyl salicylate EHS 5.77 8.13 15.41 120+138

Homosalate HMS 5.82 8.10 15.93,16.14 120+138

Isoamyl-p-methoxycinnamate IAMC 4.06 - 17.35b 161+178

2-Hydroxy-4-

methoxybenzophenone BP-3 3.64 7.56 17.36 151+227

3-(4-Methylbenzylidene)

camphor 4-MBC 4.95

- 17.57b 254

2-Ethylhexyl-p-

dimethylaminobenzoate EHPABA 6.15 2.39 19.10 161+178

2-Ethylhexyl-p-

methoxycinnamate EHMC 5.66 - 19.42b 165+277

Octocrylene OC 7.53 - 21.80 232+249+360

aValues compiled from SciFinder Scholar Database

bRetention time values for the E isomers

2.2. Samples and sample preparation

Ultrapure (Milli-Q), river, swimming pool and sewage water were used

in this study. Except ultrapure water, the rest of samples were passed through

0.45 m pore size filters before extraction. Unless otherwise stated, the

enrichment step was performed at room temperature (20 ± 2 ºC), with the

sorbent dipped into the water sample. Extractions were carried out in vessels

with two different nominal volumes (20 and 100 mL), furnished with a PTFE-

coated silicone septum and crimped with an aluminium cap. Silicone discs and

rods were pierced in a stainless steel pin, passed through the septum of the

vessel. During extraction, water samples were stirred with a PTFE covered


152

magnetic bar. Extractions with Twisters were carried out under same

experimental conditions, using the PDMS covered bar as stirring device. After

finishing the sampling step, Twisters and silicone rubber sorbents were rinsed

with ultrapure water, dried using a soft tissue and desorbed with a small

volume of organic solvent.

Under final working conditions, extractions were carried out with a

silicone disc (5 mm x 0.6 mm) exposed overnight (14 hours) to 100 mL samples,

which contained a 10% of methanol. Thereafter, the disc was introduced in a

GC vial (1.5 mL volume) and desorbed with 0.2 mL of ethyl acetate. An aliquot

(9 L) of this extract was injected in the GC-MS system.

2.3. GC-MS determination

Analytes were determined by GC-MS, using a Varian (Walnut Creek,

CA, USA) 450 GC instrument connected to an ion-trap Varian 240 mass

spectrometer (MS) and equipped with an electron impact (EI) ionization source,

in the external configuration mode. Separations were carried out in an Agilent

(Wilmington, DE, USA) HP-5ms type capillary column (30 m x 0.25 mm i.d., df:

0.25 µm), operated at a constant helium flow of 1.2 mL min-1. The GC oven was

programmed as follows: 70 ºC (held for 4 min), first rate at 12 ºC min-1 to 280 ºC

(held for 5 min). The injector was furnished with Siltek fritted liner. The

injection volume was 9 L. The temperature of the injector was maintained at 60

ºC, for 1 min and then increased to 280 ºC with a rate of 200 ºC min-1. During the

first 0.2 min, the solenoid valve was maintained in the split position (split flow

20 ml min-1) to remove the excess of solvent and then, it was switched to the

splitless mode until 4 min. Thereafter, it was turned again to the split mode

and, a flow rate of 80 mL min-1 used to sweep the liner. Transfer line, ion source

and trap temperatures were set at 280, 200 and 150 ºC, respectively. The helium

damping gas flow in the mass analyzer was set at 2.5 mL min-1.

The mass spectrometer was operated in the electron impact ionization

mode (70 eV), with a filament emission current of 50 A. MS spectra were

acquired in the m/z range between 80 and 400 a.m.u. The electromultiplier


153

voltage was set at 1800 V. The m/z ratios summarized in Table 1 were used to

monitor the extracted ion chromatograms for target analytes.

2.4. Extraction efficiencies and sample quantification

The concentrations of UV filters in the ethyl acetate extracts from Twisters

and silicone sorbents were established by comparison with the responses (peak

areas) measured for calibration standards in the same solvent. The extraction

efficiency (EE) of the sample preparation method (extraction plus desorption)

was defined as the ratio between the mass of each analyte in the ethyl acetate

extract and that added to the water sample, multiplied by 100. Theoretical

extraction efficiencies (TE) were estimated with the following equation:

100

1

1x

K

TE

OW

, being the ratio between sample and sorbent volumes,

and Kow the octanol-water partition coefficients of UV filters summarized in

Table 1, [14,28].

The levels of UV filters in river and treated sewage water samples were

determined by comparison with aqueous standards, prepared in ultrapure

water, submitted to whole sample preparation procedure. Raw sewage water

samples were quantified with the standard addition methodology.


3.1. Sample preparation conditions

Sample preparation (extraction and desorption) parameters were

optimized with fortified (10 ng mL-1) aliquots of ultrapure water. Unless

different conditions are specified, the extraction step was performed overnight

(12-14 hours), with the silicone disc exposed directly to the spiked samples

placed on a multi-position magnetic stirrer.


154

3.1.1. Desorption parameters

Dichloromethane and ethyl acetate were considered to recover the

analytes previously concentrated in the silicone discs. Both solvents have been

recommended for the elution of UV filters from SPE cartridges [5,29] and, they

are compatible with GC-based determinations. In the initial experiments, 0.5

mL of solvent and 30 min of desorption were employed. Under these

conditions, ethyl acetate provided between 15% and 20% higher desorption

efficiencies than dichloromethane; thus, the former solvent was selected. Fig. 1

shows the normalized responses obtained in the consecutive desorptions of

silicone discs (5 mm diameter x 0.6 mm thickness) with 0.2 mL aliquots of ethyl

acetate. Around 95% of the extracted amount corresponded to the first fraction,

Fig. 1. Taking into account that discs were employed as single use devices, 0.2

mL was adopted as the working value for this factor. In a further series of

assays, desorption times of 10 and 20 min were also investigated; however, the

obtained responses were slightly lower than those corresponding to 30 min,

which was maintained as the optimum desorption time.

85%

90%

95%

100%

EHS HMS IAMC BP-3 4-MBC EHPABA EHMC OC

No

rma

lize

d re

spo

nse

Fraction 1 Fraction 2 Fraction 3

Fig. 1. Normalized responses in the consecutive desorptions of silicone discs with 0.2

mL of ethyl acetate. Desorption time 30 min. Average values for triplicate experiments.


155

3.1.2. Optimization of extraction conditions

3.1.2.1. Sampling mode and temperature

Sorptive extractions can be performed in direct and headspace (HS)

modes. The major advantage of the latter is the higher selectivity of the process,

avoiding the co-extraction of non-volatile organic interferences. In this study,

we have compared the extraction yields obtained for the direct mode, at room

temperature, versus those corresponding to HS at room temperature, 55 ºC and

90 ºC. In all cases, the sampling time was 14 hours and the sample volume 100

mL. HS extractions at room temperature rendered negligible yields for all

species. Responses obtained under the rest of conditions are compared in Fig. 2.

Globally, the most favourable situation corresponded to direct sampling at

room temperature; nevertheless, some analytes (both salicylates, 4-MBC,

EHPABA and EHMC) were also extracted in a considerable extent in the HS

mode, Fig. 2. For the above UV filters, the combination of HS sampling with

high temperatures offers an interesting balance between efficiency and

selectivity. Obviously, in this study, direct sampling and room temperature

were the adopted conditions.

0%

20%

40%

60%

80%

100%

120%

140%

160%

180%


Nor

mal

ized

res

pons

e

Direct, room temperature HS, 55 ºC HS, 90 ºC

Fig. 2. Effect of the sampling mode and the temperature in the performance of the

sorptive extraction step. Normalized responses to those obtained in the direct mode, at

room temperature. Sample volume 100 mL, extraction time 14 hours, n=3 replicates.


156

3.1.2.2. Effects of salt and methanol

The influence of these variables on the yield of the sorptive extraction

was investigated using a sequential approach, considering three levels for the

concentration of sodium chloride (0, 5 and 15%) and four for the percentage of

methanol (0, 2, 5 and 10%). The obtained data followed two different trends,

which can be related with the octanol-water (Kow) partition coefficients of UV

filters. Fig. 3 depicts both trends using IAMC and OC as model analytes. For

medium polarity species (IAMC, 4-MBC and BP-3, log Kow from 3.6 to 5 units),

the extraction yield decreased steady with the percentage of methanol added to

the sample, whereas it improved slightly with the addition of sodium chloride.

This behaviour is the result of the negative (sodium chloride) and positive

(methanol) effects of the above factors in their water solubility. On the other

hand, for the more lipophilic compounds (EHS, HMS, EHPABA, EHMC and

OC, log Kow above 5.6 units), the extraction efficiency improved with the

percentage of methanol, achieving a maximum between 5% and 10% of organic

modifier, and underwent a dramatic diminution with the concentration of

sodium chloride, Fig. 3. Salt addition reduced the solubility of the compounds,

increasing sorption losses on the walls of glass vessels; moreover, it increased

the viscosity of the sample, which limits the migration rate of the less polar

analytes from the sample to the interface with the silicone disc. Obviously, the

addition of methanol prevented competitive adsorption losses. On the basis of

the above comments, sodium chloride was not added to water samples,

whereas the percentage of methanol was adjusted to 10%. These conditions

matched with those recommended for the sorptive extraction of UV filters using

PDMS covered stir bars [17].


157

0%

20%

40%

60%

80%

100%

120%

IAMC OC

Ext

ract

ion

effic

ienc

y

Without methanol 3% Methanol

5% Methanol 10% Methanol

A

0%

20%

40%

60%

80%

100%

120%

IAMC OC

Ext

ract

ion

effic

ienc

y

Without salt 5% salt 15% salt

B

Fig. 3. Influence of methanol (A) and sodium chloride (B) on the extraction efficiencies

of IAMC and OC. Sample volume 100 mL, extraction time 14 hours, n=3 replicates.

3.1.2.3. Stirring and sample pH

The extraction efficiency was enhanced significantly for stirred versus

non-stirred samples; however, for a sampling step of 14 hours, no differences

were appreciated using stirring rates of 300, 600 and 1000 rpm. The effect of

sample pH was investigated at three levels (3, 6 and 9). Despite BP-3 and

salicylate type UV filters are slightly acidic compounds, within the above range,

the pH showed a negligible effect in the extraction yield for all compounds,

data not shown. Further extractions were performed without adjusting the pH


158

of water samples, unless it falls outside of the above range, and using a stirring

speed of 600 rpm.

3.1.2.4. Time and sample volume

In solid-phase microextraction processes, for a given amount of sorbent,

an increase in the volume of the sample leads to longer equilibrium times and

lower extraction efficiencies [30]. Fig. 4 depicts the time course of the sorptive

extraction for HMS and OC (all compounds showed similar profiles) in 20 and

100 mL samples. The Y axis in the figure represents the total amount (mass, in

ng) of each analyte in the ethyl acetate extract (0.2 mL) from the silicone disc.

For 20 mL samples, equilibrium was achieved after 5 hours of extraction,

whereas more than 12 hours were required for 100 mL ones. It is also evident

that, for sampling steps longer than 5 hours, the extracted amount of all

compounds was higher for 100 mL samples than for 20 mL ones. In further

experiments, 100 mL samples and 14 hours extractions (overnight) were

employed.


159

HMS

0

250

500

750

1000

0 5 10 15 20 25

Time (h)

Ext

ract

ed m

ass

(ng)

100 mL 20 mL

OC

0

250

500

750

1000

0 5 10 15 20 25

Time (h)

Ext

ract

ed m

ass

(ng)

100 mL 20 mL

Fig. 4. Time course of sorptive extraction for HMS and OC considering two different

sample volumes, n=2 duplicates.

3.2. Performance of the method

The performance of the proposed method, sorptive extraction with

disposable silicone discs followed by ethyl acetate desorption and GC-MS

determination, was first characterized in terms of linearity, intra- and inter-day

precision and limits of quantification (LOQs), Table 2. Linearity was assessed

with samples fortified at seven concentration levels, prepared in duplicate,

comprised between 0.01 and 10 ng mL-1. The plots of peak area versus

concentration were adjusted to a linear model obtaining determination


160

coefficients (R2) values between 0.992 and 0.999 (Table 2); moreover, the

ANOVA lack-of-fit test demonstrated the suitability of the linear model to

describe the dependence between responses (peak areas) and concentration, at a

95% confidence level (data not given). Intra- and inter-day precision stayed

below 10% and 13%, respectively. These relative standard deviations (RSDs) are

similar to those reported for PDMS covered stir bars used in combination with

thermal desorption and GC-MS determination [17].

Procedural blanks, corresponding to ultrapure water samples (Fig. 5),

showed some traces of OC and EHMC, as it has been usually reported in

previous works [4,17,31]. The achieved LOQs, defined as the concentration of

each UV filter providing a signal 10 higher than the baseline noise, or the

standard deviation of 5 consecutive blanks (case of EHMC and OC), stayed at

the low ng L-1 level, except in case of BP-3 (LOQ 0.040 ng mL-1), which was the

compound showing the lowest extraction efficiency. Overall, LOQs

summarized in Table 2 are similar to those reported using Twisters in

combination with thermal desorption and GC-MS determination for 20 mL

samples [17] and other effective microextraction techniques, such as dispersive

liquid-liquid microextraction [32]; moreover, they remained below those

reported for solid-phase extraction and GC-MS determination [5].


161

Tab

le 2

. Lin

eari

ty (

0.01

0-10

ng

mL-

1 ), r

epea

tabi

lity,

rep

rodu

cibi

lity

and

limit

s of

qua

ntifi

cati

on (

LOQ

s) o

f the

pro

pose

d m

etho

d. D

epic

ted

data

corr

espo

nd to

100

mL

sam

ples

usi

ng a

sili

cone

dis

c (5

mm

dia

met

er x

0.6

mm

thic

knes

s) a

s so

rben

t.

Ana

lyte

L

inea

rity

(0.0

1-10

ng

mL

-1, n

=7

leve

ls)a

R

epea

tabi

lity,

RSD

(%)

(n=5

rep

licat

es, s

ame

day

)

Rep

rodu

cibi

lity,

RSD

(%)

(n=

12 r

eplic

ates

, 4 d

ays)

L

OQ

s (n

g m

L-1 )

Sl

ope

(SD

) In

terc

ept (

SD)

R2

0.2

ng m

L-1

2

ng m

L-1

0.

5 ng

mL

-1

EH

S

HM

S

IAM

C

BP-

3

4-M

BC

EH

PA

BA

EH

MC

OC

8.9E

4 (3

.4E

3)

5.9E

4 (2

.4E

3)

6.7E

4 (2

.3E

3)

1.3E

4 (2

.1E

2)

1.1E

4 (2

.7E

2)

1.7E

5 (6

.2E

3)

1.3E

5 (4

.1E

3)

8.4E

4 (2

.2E

3)

542

(124

3)

-106

0 (9

220)

-128

0 (8

112)

-117

0 (7

80)

323

(106

3)

3800

(179

0)

1135

0 (1

4800

)

1410

0 (7

800)

0.99

3

0.99

2

0.99

4

0.99

9

0.99

7

0.99

4

0.99

5

0.99

7

5.6

6.2

8.4

9.9

2.0

1.0

1.0

3.8

4.7

6.0

2.8

4.4

1.4

5.5

6.2

1.0

8.4

11.4

11.4

8.5

12.5

7.6

12.2

8.5

0.00

3

0.00

8

0.00

6

0.04

0.01

0.00

4

0.01

5

0.01

1

a The

line

ar r

espo

nse

rang

e of

BP

-3 w

as 0

.04-

10 n

g m

L-1

.


162

15 16 17 18 19 minutes

0

5

10

15

20

(x 103) Counts

m/z 120+138+161+178

EHS

HMS

IAMCEHMC

16.75 17.00 17.25 17.50 17.75 minutes0.00

0.25

0.50

0.75

1.00

1.25

(x 103) Counts

m/z 151+227+254

BP‐3

4‐MBC

19.0 19.5 20.0 20.5 21.0 21.5 22.0 minutes0

1

2

3

4

5

6

7

8

(x 103) Counts

m/z 165+277+360+232+249

EHPABA OC

Fig. 5. Extracted ion chromatograms for a procedural blank (dotted line) and a low level

(0.07 ng mL-1) spiked ultrapure water sample (solid line).


163

The potential influence of the matrix on the efficiency of the sorptive

extraction was evaluated comparing the responses obtained for different water

samples, fortified at the same concentration level. River, swimming-pool and

sewage water were first filtered and then divided in two fractions. Thereafter,

one of them was fortified with UV filters at the same level as the aliquots of

ultrapure water. The difference between responses measured for spiked and

non-spiked fractions of real water samples were normalized to those obtained

for ultrapure water. Relative recoveries over 75% were obtained for river,

swimming-pool and treated wastewater, Table 3. On the other hand, the

efficiency of the extraction underwent a reduction up to 50% for the most

lipophilic species (OC) in raw sewage water. Thus, external calibration against

aqueous standards, dissolved in ultrapure water, can be used to measure the

levels of UV filters in surface, bathing and treated sewage water; however, the

standard addition methodology is required for untreated wastewater.

Table 3. Influence of the sample matrix on the efficiency of the sorptive extraction using

silicone discs. Normalized responses to those measured for ultrapure water fortified at

the same concentration level (2 ng mL-1), n=3 replicates.

Analyte Normalized response (%) SD

River Swimming pool Treated wastewater Raw wastewater

EHS

HMS

IAMC

BP-3

4-MBC

EHPABA

EHMC

OC

104 3

96 3

92 6

98 7

90 7

94 1

98 5

104 5

93 6

87 2

87 5

76 6

82 12

89 2

88 2

89 2

93 11

92 9

86 4

75 4

86 7

91 1

91 11

92 1

62 1

57 5

82 9

108 7

91 8

66 5

60 3

49 3


164

3.3. Silicone discs versus rods and Twisters

Table 4 compiles the experimental extraction efficiencies (EE) obtained

with silicone discs, rods and Twisters for 20 and 100 mL water samples, spiked

at different concentration levels. Theoretical extraction efficiencies (TE),

calculated as described in section 2.4, are also given. In order to prevent

changes in the extraction efficiencies caused by incomplete desorptions, the

volume of ethyl acetate was increased up to 0.5 mL for rods and Twisters, which

were not completely covered, in the 1.5 mL vessels, with 0.2 mL of solvent.

Overall, an excellent agreement was observed among EE values provided by

silicone discs and rods containing 12 L of silicone for 20 and 100 mL samples.

The same comment is also valid for the experimental efficiencies achieved using

two discs (total silicone volume 24 L) and a single Twister (24 L of pure

PDMS), Table 4. For the most lipophilic UV filters (EHS, HMS, EHMC,

EHPABA and OC), EE values were similar to TE ones; moreover, the above UV

filters were recovered in an extension higher than 90% in all the tested

conditions. In the case of IAMC and 4-MBC (log Kow 4.06 and 4.95,

respectively), the EE were in agreement with TE for 20 mL samples and slightly

lower for the 100 mL ones. Finally, the EE for BP-3 always remained 2-3 times

lower than the TE value. Solid-phase extraction of the water sample [33], after

finishing the sorptive extraction step, confirmed that BP-3 remained un-

extracted in the water phase, without undergoing significant degradation

processes.


165

Table 4. Experimental extraction efficiencies (EE) and theoretical values (TE) for spiked

samples of ultrapure water (pH 6, 10% methanol) using different sorbents. Sampling

time 14 hours, n=3 replicates.

Analyte Sample volume 20 mLa

EE (%) SD TE (%)

Disc (12 L) Rod (12 L) Disc (2 x 12 L) Twister (24 L) (12 L) (24 L)

EHS

HMS

IAMC

BP-3

4-MBC

EHPABA

EHMC

OC

105 3

101 6

79 3

18 5

88 4

95 3

99 1

104 2

104 2

109 6

86 10

20 6

87 5

99 2

105 7

106 2

95 4

91 2

87 8

28 2

92 2

94 2

93 2

94 5

104 8

104 7

98 4

36 3

102 1

108 10

105 3

102 7

100

100

87

72

98

100

100

100

100

100

93

84

99

100

100

100

aAddition level 1 ng mL-1.

Table 4 cont. Experimental extraction efficiencies (EE) and theoretical values (TE) for

spiked samples of ultrapure water (pH 6, 10% methanol) using different sorbents.

Sampling time 14 hours, n=3 replicates.

Analyte Sample volume 100 mLb

EE (%) SD TE (%)

Disc (12 L) Rod (12 L) Disc (2 x 12 L) Twister (24 L) (12 L) (24 L)

EHS

HMS

IAMC

BP-3

4-MBC

EHPABA

EHMC

OC

92 8

101 5

47 6

8 2

54 5

93 2

94 6

104 9

95 8

96 6

44 64

7 2

50 3

91 2

85 6

91 7

96 4

102 1

54 2

19 3

76 3

96 7

98 1

101 1

106 4

106 1

65 8

26 5

92 1

102 1

99 2

94 5

99

99

58

34

91

99

98

100

99

99

73

51

96

100

99

100

bAddition level 0.2 ng mL-1.


166

A potential explanation of the low extraction yields observed for BP-3 is

that, the methanolic content of the samples (10%) resulted in a significant

reduction of the partition coefficients for UV filters between the silicone (or

PDMS) sorbent and the matrix. Despite such reduction, the most lipophilic

species were still quantitatively transferred to silicone sorbents; however, in the

case of BP-3 the efficiency of the extraction was significantly reduced. Another

feasible hypothesis is that, Kow values are not valid to predict the distribution of

certain compounds between water samples and polysiloxane sorbents. In this

sense, Popp and co-workers [34] have also found important disagreements

between predicted and observed extraction yields for other phenolic species,

using technical grade silicone sorbents. Serodio and co-workers [35] reported

similar deviations, for medium polarity species, using PDMS covered stir bars.

Fig. 6 shows the total ion current (TIC) chromatograms corresponding to

desorption of a Twister and two silicone discs. In both cases, the corresponding

ethyl acetate extracts were adjusted to a final volume of 0.2 mL, and an aliquot

(9 L) was injected in the GC-MS system. As appreciated, any of both materials

released relevant levels of interfering species.

16 17 18 19 20 21 22 min.0

10

20

30

40

(x 103 Counts)

Silicone disc

Twister

Fig. 6. Total ion chromatograms (TIC) corresponding to ethyl acetate extracts (final

volume 0.2 mL) from a Twister and two silicone discs.


167

3.3. Application to water samples

The developed method was applied to a total of twelve grab samples of

river and sewage water collected in three different dates. Concentrations for

compounds found above their LOQs are provided in Table 5. Samples code 1 to

3 correspond to a river which receives the discharges from an urban STP,

equipped with primary and biological treatments, in the Northwest of Spain.

The highest concentrations in the above river water samples were found at the

end of summer and the beginning of autumn, as a consequence of the reduction

in the flow rate of the river. Samples 4 to 6 were obtained from bathing areas in

small streams, located in the same geographic area. The pairs of samples 7-8, 9-

10 and 11-12 were simultaneously collected in the inlet and outlet streams of the

urban STP. EHS and OC were found in the six samples at concentrations above

their LOQs. For the latter compound, the measured levels were slightly higher

than those previously reported by our group for samples taken in the same

plant in 2009 [32], and those found in other STP in Spain [18,31]. On the other

hand, they are similar to those reported for STPs in Switzerland [4]. Besides

EHS and OC, 4-MBC and EHMC were also found in a significant percentage of

the processed sewage water samples, achieving concentrations above 100 ng L-1,

Table 5. Finally, IAMC and EHPABA were never detected.


168

Table 5. UV filter concentrations measured in river and wastewater samples. IAMC

and EHPABA were not detected in any sample, n= 3 replicates.

Code Type Sampling

date

Concentration ( ng L-1) ± SD

EHS HMS BP-3 4-MBC EHMC OC

1 River July 2010 - - - - - 27 ± 3

2 River September 2010 4.0 ± 0.3 8 ± 2 - 46 ± 1 39 ± 4 243 ± 18

3 River October 2010 15 ± 1 15 ± 2 - 48 ± 6 - 251 ± 30

4 River September 2010 - - - - 27 ± 1 19 ± 2

5 River September 2010 - - - - 24 ± 1 27 ± 11

6 River September 2010 - - 53 ± 6 - - -

7 R.W. July 2010 105 ± 11 40 ± 7 - 17 ± 5 757 ± 60 831 ± 73

8 T.W. July 2010 4.5 ± 0.1 - - 223 ± 5 - 106 ± 11

9 R.W. September 2010 36 ± 1 25 ± 1 236 ± 5 131 ± 16 129 ± 3 437 ± 25

10 T.W. September 2010 4.8 ± 0.1 15 ± 1 - 140 ± 7 31 ± 1 111 ± 1

11 R.W. October 2010 22 ± 2 14 ± 1 - 61 ± 5 27 ± 4 209 ± 26

12 T.W. October 2010 5.3 ± 0.1 - - - - 65 ± 1

R.W. raw wastewater T.W. treated wastewater - below detection limits

4. Conclusions

Sorptive extraction using technical grade, disposable silicone sorbents

followed by solvent desorption and GC-MS determination, with large volume

injection, provides suitable figures of merit for the determination of UV filters

in surface and sewage water samples. Optimum extraction conditions are

similar to those previously reported in the literature for Twisters covered with

PDMS; moreover, very close experimental extraction efficiencies were

calculated for both materials. Practical advantages of the methodology

described in this study are (1) the low cost of silicon discs (ca. 0.1 Euro per unit),


169

(2) the low volume of the sorbent (12 L), which can be easily accommodate at

the bottom of a GC vial and desorbed using just 0.2 mL of organic solvent, and

(3) the possibility to re-analyze the extract, without repeating the whole sample

preparation process, as it occurred in the case of thermal desorption methods.

The slow kinetics of the sorptive extraction for 100 mL samples can be

compensated by processing simultaneously, in an unattended mode, a large

number of specimens, using a multi-position stirrer.

Acknowledgements

This study has been supported by the Spanish Government and E.U.

funds (projects CTQ2009-08377 and DE2009-0020). N.N. thanks a FPU pre-

doctoral contract to the Spanish Ministry of Education.


170

References

1. Richardson SD (2010) Anal Chem 82:4742-4774

2. Giokas DL, Salvador A, Chisvert A (2007) Trends Anal Chem 26:360-374

3. Díaz-Cruz MS, Llorca M, Barceló D (2008) Trends Anal Chem 27:873-887

4. Balmer ME, Buser HR, Müller MD, Poiger T (2005) Environ Sci Technol 39:953-962

5. Cuderman P, Health E (2007) Anal Bioanal Chem 387:1343-1350

6. Poiger T, Buser HR, Balmer ME, Bergqvist PA, Müller MD (2004) Chemosphere 55:951-

963

7. Plagellat C, Kupper T, Furrer R, de Alencastro LF, Grandjean D, Tarradellas J (2006)

Chemosphere 62:915-925

8. Buser HR, Balmer ME, Schmid P, Kohler M (2006) Environ Sci Technol 40:1427-1431

9. Zenker A, Schmutz H, Fent K (2008) 1202:64-74

10. Nieto A, Borrull F, Marcé RM, Pocurull E (2009) J Chromatogr A 1216:5619-5625

11. Fent K, Zenker A, Rapp M (2010) Environ Pollut 158:1817-1824

12. Díaz-Cruz MS, Barceló D (2009) Trends Anal Chem 28:708-717

13. Baltussen E, Sandra P, David F, Cramers C (1999) J Microcolumn Sep 11:737-747

14. Prieto A, Basauri O, Rodil R, Usobiaga A, Férnandez LA, Etxebarria N, Zuloaga O

(2010) J Chromatogr A 1217:2642-2666

15. Kawaguchi M, Ito R, Endo N, Sakui N, Okanouchi N, Saito K, Sato N, Shiozaki T,

Nakazawa H (2006) Anal Chim Acta 557:272-277

16. Kawaguchi M, Ito R, Honda H, Endo N, Okanouchi N, Saito K, Seto Y, Nakazawa H

(2008) J Chromatogr A 1200:260-263

17. Rodil R, Moeder M (2008) J Chromatogr A 1179:81-88

18. Pedrouzo M, Borrull F, Marcé RM, Pocurull E (2010) Anal Bioanal Chem 397:2833-2839

19. Popp P, Bauer C, Paschke A, Montero L (2004) Anal Chim Acta 504:307-312

20. Montero L, Popp P, Paschke A, Pawsliszyn J (2004) J Chromatogr A 1025:17-26

21. Montes R, Rodríguez I, Rubí E, Cela R (2007) J Chromatogr A 1143:41-47

22. Van Pinxteren MS, Montero L, Jäsch S, Paschke H, Popp P (2009) Anal Bioanal Chem

393:767-775

23. Burger BV, Marz B, Roux M, Burger WJG (2006) J Chromatogr A 1121:259-267

24. Carpinteiro I, Abuín B, Rodríguez I, Ramil M, Cela R (2010) J Chromatogr A 1217:7208-

7214

25. Paschke A, Schwab K, Brümmer J, Schüürmann G., Paschke H, Popp P (2006) J


26. Schellin M, Popp P (2007) J Chromatogr A 1152:175-183

27. Van Pinxteren MS, Paschke A, Popp P (2010) J Chromatogr A 1217:2589-2598

28. Bicchi C, Cordero C, Rubiolo P, Sandra P (2003) J Sep Sci 26:1650-1656


171

29. Giokas DL, Sakkas VA, Albanis TA (2004) J Chromatogr A 1026:289-293

30. Pawliszyn J (1997) Solid phase microextraction, theory and practice, Wiley-VCH

31. Rodil R, Quintana JB, López-Mahía P, Muniategui-Lorenzo S, Prada-Rodríguez D (2008)

Anal Chem 80:1307-1315

32. Negreira N, Rodríguez I, Rubí E, Cela R (2010) Anal Bioanal Chem 398:995-1004

33. Negreira N, Canosa P, Rodríguez I, Ramil M, Rubí E, Cela R (2008) J Chromatogr A

1178:206-214

34. Schellin M, Popp P (2007) J Chromatogr A 1152:175-183

35. Serodio P, Nogueira JMF (2004) Anal Chim Acta 517:21-32

1.5. Publicación:

SENSITIVE DETERMINATION OF

SALICYLATE AND BENZOPHENONE

TYPE UV FILTERS IN WATER SAMPLES

USING A SOLID-PHASE MICROEXTRACTION,

DERIVATIZATION AND GAS CHROMATOGRAPHY

TANDEM MASS SPECTROMETRY

N. Negreira, I. Rodríguez, M. Ramil, E. Rubí, R. Cela

Analytica Chimica Acta 638 (2009) 36

(doi:10.1016/j.aca.2009.02.015)


175

Sensitive determination of salicylate and benzophenone type UV filters in

water samples using solid-phase microextraction, derivatization and gas

chromatography tandem mass spectrometry

N. Negreira, I. Rodríguez*, M. Ramil, E. Rubí, R. Cela


de Investigación y Análisis Alimentario, Universidad de Santiago de


Abstract

A sensitive procedure for the determination of three UV filters:

ethylhexyl salicylate (EHS), 3,3,5-trimethylcyclohexyl salicylate (Homosalate,

HMS), 2-hydroxy-4-methoxybenzophenone (BP-3) and two related

hydroxylated benzophenones (2,4-dihydroxybenzophenone, BP-1 and 2,2´-

dihydroxy-4-methoxybenzophenone, BP-8) in water samples is presented.

Analytes were first concentrated on the coating of a solid-phase microextraction

(SPME) fibre, on-fibre silylated and then determined using gas chromatography

combined with tandem mass spectrometry (GC-MS/MS). Factors affecting the

performance of extraction and derivatization steps are thoroughly evaluated

and their effects on the yield of the sample preparation discussed. Under final

working conditions, a PDMS-DVB coated SPME fibre was exposed directly to

10 mL of water, adjusted at pH 3, for 30 min. After that, the fibre was placed in

the headspace (HS) of a 1.5 mL vial containing 20 L of N-methyl-N-

(trimethylsilyl)-trifluoroacetamide (MSTFA). On-fibre silylation of hydroxyl

groups contained in the structure of target compounds was performed at 45 ºC

for 10 min. The whole sample preparation process was completed in 40 min,

providing limits of quantification from 0.5 to 10 ng L-1 and acceptable precision

(RSDs under 13%) for samples spiked at different concentrations. All

compounds could be accurately determined in river and treated wastewater

(relative recoveries from 89 to 115%) using standards in ultrapure water,


176

whereas standard addition is recommended to quantify their levels in untreated

wastewater. Analysis of wastewater revealed the systematic presence of BP-3

and BP-1 in raw samples with maximum concentrations close to 500 and 250 ng

L-1, respectively.

Keywords: UV filters; water samples; solid-phase microextraction;

derivatization; gas chromatography.

1. Introduction

Concern about skin damage produced by UV radiation has fostered the

consumption of sunscreen products in developed countries. In the European

Union (EU), 26 organic compounds have been approved to be used as UV filters

in sunscreen products, with maximum individual concentrations up to 10%

[1,2]. Additionally, same compounds, as well as other species with similar

structures, are also included in the formulation of many other personal care

products (PCPs) such as hair shampoos, lipsticks and even in packaging

materials to enhance their light stability [3-6].

Direct release of sunscreens from the skin in bathing areas and indirect

inputs through domestic wastewater are responsible for local and diffuse

discharges of UV filters in the aquatic environment [1,7]. Among them, 2-

hydroxy-4-methoxybenzophenone (BP-3) is one of the most often detected,

reaching concentrations up to 100 ng L-1 in surface water from bathing areas

[8,9]. It is also present in raw [10] and treated domestic wastewater [11], and

even in fatty tissues from river and lake fish [10]. Moreover, when applied on

the skin, BP-3 is partially absorbed by the human body and excreted as more

polar metabolites, such as 2,4-dihydroxybenzophenone (BP-1) and 2,2´-

dihydroxy-4-methoxybenzophenone (BP-8) [12,13]. BP-1 and BP-8 are also used

as UV absorbers to protect goods against UV radiation [3]. In vivo and in vitro

studies have proved that BP-3 and BP-1 present estrogenic activity [14,15];

moreover, they are prone to evolve into halogenated by-products when mixed


177

with chlorinated water [16]. On the other hand, BP-8 presents activity as

bacterial mutagen [1,17].

In addition to BP-3, two salicylates: 2-ethylhexyl salicylate (EHS) and

3,3,5-trimethylcyclohexyl salicylate (Homosalate, HMS) are also often employed

as UV filters in sunscreens [5]. Although they are more lipophilic than the

above benzophenones, the presence of a phenolic group in their structures

might provide them certain mobility in the aquatic environment. In fact, HMS

and EHS have been recently detected in surface water samples [8,18].

Solid-phase extraction (SPE) is the preferred technique for the

concentration of UV filters in water samples [1]. Although it offers considerable

advantages in comparison with liquid-liquid extraction (LLE), SPE still requires

large sample intakes (from 0.3 to 1 L), a considerable volume (10-15 mL) of

organic solvents for analytes desorption and a further clean-up to compensate

for its limited selectivity when applied to wastewater [1,8,9,11]. Theoretically,

microextraction techniques, based on equilibrium processes, and particularly

solid-phase microextraction (SPME), should allow overcoming some of the

above drawbacks. Although SPME, followed by gas chromatography with mass

spectrometry detection (GC-MS), has been previously evaluated for the

concentration of several UV filters in aqueous matrices (including BP-3 and

some related benzophenones) the achieved limits of quantification (LOQs), in

the ng mL-1 range, were unsuitable for waste and surface water analysis [12,19].

The well-known trend of phenolic species to produce tailing GC peaks, added

to a non-exhaustive optimization of extraction conditions, can explain the poor

sensitivity of the resulting methods. The former limitation can be overcome

integrating a derivatization step in the SPME process. In this sense, SPME

combined with on-fibre derivatization has been described as a successful

approach for the determination of emerging pollutants, with carboxylic and

phenolic groups, in water samples at the low ng L-1 level [20,21].

The goal of this work is to evaluate the suitability of SPME, followed by

on-fibre silylation, for the determination of HMS, EHS, BP-3 and two other

hydroxylated benzophenones (BP-1 and BP-8) in waste and surface water


178

samples, using GC with tandem mass spectrometry (MS/MS) as determination

technique. Although, the use of silylation reactions is quite common to enhance

the detectability of salicylates and hydroxylated benzophenones in GC methods

[8,16,22,23], to the best of our knowledge, the combination of these reactions

with SPME has not yet been proposed for the determination of phenolic UV

absorbers in water samples.

2. Experimental

2.1. Reagents, SPME equipment and samples

HPLC-grade methanol, acetone and ethyl acetate (trace analysis grade),

and glacial acetic acid were obtained from Merck (Darmstadt, Germany).

Standards of BP-3, BP-1, BP-8, EHS and HMS were acquired from Aldrich

(Milwaukee, WI, USA) and Merck. Their chemical structures, pKa and octanol-

water partition coefficients (log Kow) are summarized in Table 1. Derivatization

reagents N-methyl-N-(tert-butyldimethylsilyl)-trifluoroacetamide (MTBSTFA)

and N-methyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA) were also

purchased from Aldrich. Individual solutions of each compound were prepared

in methanol, further dilutions and mixtures of them, used to fortify water

samples, were made in the same solvent. Silylated compounds in ethyl acetate,

employed during optimisation of GC-MS/MS conditions, were prepared

adding 20 L of MSTFA to a standard solution of target species in the above

solvent. The reaction was accomplished at 60 ºC for 1 hour [8].


179

Table 1. Abbreviated names, structures, log Kow and pKa values of target species.

Abbreviation Name Structure Log

Kowa pKaa

EHS 2-ethylhexyl salicylate

5.97 8.13

HMS

3,3,5-trimethylcyclohexyl

salicylate

(Homosalate)

6.16 8.09

BP-3 2-hydroxy-4-

methoxybenzophenone

3.79 7.56

BP-1 2,4-

dihydroxybenzophenone

3.17 7.53

BP-8 2,2´-dihydroxy-4-

methoxybenzophenone

3.93 6.99

aValues obtained from SciFinder Scholar Database,

http://www.cas.org/products/sfacad/

A manual SPME holder and fibres coated with different polymers:

poly(dimethylsiloxane) (PDMS, 100 m film thickness), polyacrylate (PA, 85

m film thickness), Carboxen-PDMS (CAR-PDMS, 75 m film thickness) and

PDMS-divinylbenzene (PDMS-DVB, 65 m film thickness) were obtained from

Supelco (Bellefonte, PA, USA). Before being used for first time, SPME fibres

were thermally conditioned following conditions recommended by the

supplier.

Ultrapure (Milli-Q), river and wastewater samples, obtained from the

inlet and outlet streams of an urban sewage treatment plant (STP), equipped

with primary and activated sludge units, were employed throughout this study.


180

River and wastewater samples were filtered using cellulose acetate membranes

(0.45 m pore size) and stored in the dark at 4º C until analysis.


Extractions were performed in glass vessels of different nominal

capacities (10, 22 and 110 mL), containing a magnetic stir bar (10 mm x 4 mm)

and sealed with a PTFE layered rubber septa. Under optimised conditions, a

PDMS-DVB fibre was exposed directly to aqueous standards, prepared in

ultrapure water, and real life water samples, previously adjusted at pH 3, for 30

min. Extractions were carried out in the smallest vessels containing 10 mL of

water, at room temperature (20 ± 2 ºC), with magnetic stirring (1200 rpm). After

that, the fibre was retracted into the SPME holder. Water drops attached to the

outlet surface of the metallic needle were removed with a soft paper tissue and

the fibre was exposed the HS of a vessel containing 20 L of MSTFA. On-fibre

silylation of salicylates and benzophenones was performed at 45 ºC for 10 min.

2.3. Determination

Analytes were determined by GC-MS/MS using a Varian (Walnut Creek,

CA, USA) CP 3900 gas chromatograph connected to an ion-trap mass

spectrometer (Varian Saturn 2100). Separations were carried out in a HP-5ms

capillary column (30 m x 0.25 mm I.D., df: 0.25 m) supplied by Agilent

(Wilmington, DE, USA). Helium (99.999 %) was used as carrier gas at a constant

flow of 1 mL min-1. The GC oven was programmed as follows: 70 ºC (held for 3

min), at 12 ºC min-1 to 280 ºC (held for 10 min). The GC-MS interface and the ion

trap temperatures were set at 280 ºC and 220 ºC, respectively. Silylated

standards prepared in ethyl acetate were injected in the splitless mode (splitless

time 1 min), with the injector port at 280 ºC. SPME fibres were desorbed at 270

ºC in case of PDMS-DVB, and 280 ºC for the rest of coatings, for 3 min,

maintaining the injector in the splitless mode during this period.

The mass spectrometer was operated in the electron impact ionisation

mode (70 eV). MS spectra were recorded in the range from 70 to 500 m/z units.


181

The base peak in the spectra of each compound, as trimethylsilyl derivative,

was isolated with a window of 3 m/z units and subjected to collision induced

dissociation.


3.1. Performance of GC-MS/MS determination

Optimisation of GC-MS/MS conditions was carried out using silylated

standards prepared in ethyl acetate. Initially, MTBSTFA was selected as

derivatization reagent, since it leads to derivatives with an excellent stability, it

has been previously employed for the silylation of BP-3 and BP-1 [16], and it is

commonly employed in on-fibre derivatization methodologies [20,21].

Although different experimental conditions (reaction time, temperature and

volume of MTBSTFA) were tested, the two salicylates did not react

quantitatively with MBTSTFA. On the other hand, using MSTFA, all

compounds were converted in the corresponding trimethylsilyl derivatives. As

reported previously, two chromatographic peaks were obtained for HMS [8].

The base peak in the MS spectra of the three benzophenones showed the lost of

a methyl group, appearing at 285, 343 and 373 m/z units for BP-3, BP-1 and BP-

8, respectively [22]. In the case of salicylates, an intense fragment at 195 m/z

units, corresponding to the dimethylsilyl-2-hydroxy benzoic acid moiety, was

observed. The above ions were isolated in the trap and further fragmented

using a resonant waveform. The most intense transitions in the MS/MS spectra

of the two salicylates revealed the loss of H2O and CO2, Fig. 1. BP-3 and BP-1

showed a main transition corresponding to the removal of fragments with 43

and 72 m/z units, respectively. The first corresponds to a SiCH3 moiety,

whereas the second might represent the replacement of a Si(CH3)3 group,

bonded to one of the aromatic hydroxyls in the silylated BP-1, by hydrogen. On

the other hand, BP-8 underwent a much more complex fragmentation leading

to several product ions, Fig. 1. The most intense ones, at 329, 330 and 301 m/z

units (corresponding to the loss of 44, 43 and 72 units of mass, respectively)

were used for the quantification of this compound. It is worth noting that it was


182

not possible to fragment BP-8 using the non-resonant mode. Under optimised

conditions, see Table 2, the GC-MS/MS system provided linear responses for all

compounds (HMS was quantified as sum of peaks) in the range between 5 and

1000 ng mL-1. Limits of quantification (LOQs), defined as the concentration of

each compound producing a signal ten times higher than the baseline noise in

the corresponding GC-MS/MS chromatograms, ranged from 0.2 to 1.1 ng mL-1,

Table 2. Except for BP-8, these values were 10 times lower than those achieved

using single MS detection.

75 100 125 150 175 m/z

0%

25%

50%

75%

100%

91 135

151

177

O

O

O

Si

O

O

O

Bu-n

Et

SiEHS

HMS

150 200 250 300 m/z

0%

25%

50%

75%

100%

253

271

325343

BP-1

100 150 200 250 m/z

0%

25%

50%

75%

100%

212

242

267

285

O

MeO

Ph

O

Si

BP-3

150 200 250 300 350 m/z

0%

25%

50%

75%

100%

144

181209

239

256 283

301

315

329

343357

373

BP-8

Fig. 1. MS/MS spectra for UV absorbers.


183

Table 2. Optimal MS/MS detection conditions, correlation coefficients and

instrumental LOQs for the trimethylsilyl derivatives of analytes.

Compound Parent

ion

(m/z)

Product ions

(m/z)

Excitation

amplitude

(V)

Storage

level

(m/z)

Correlation

Coefficient

(R2)a

LOQs

(ng mL-

1)b

EHS

HMS

BP-3

BP-1

BP-8

195

195

285

343

373

177,151

177,151

242

271

329,330,301

0.75

0.90

1.10

1.19

1.50

86

86

126

151

164

0.998

0.999

0.995

0.995

0.997

0.2

0.3

0.2

0.3

1.1

a Evaluated with standards at 6 different concentrations between 5 and 1000 ng mL-1.

b Defined as the concentration producing a peak with a signal to noise (S/N) ratio of 10.

3.2. Optimisation of SPME conditions

3.2.1. Preliminary experiments

In order to asses the feasibility of the on-fibre silylation reaction, as well

as to obtain a first evaluation of the extraction capabilities of different SPME

coatings, spiked aliquots of ultrapure water, adjusted at pH 3, were extracted in

the direct mode using four different SPME fibres, for 30 min. The Carbowax-

DVB fibre, whose affinity for benzophenones had been previously

demonstrated [12], was not included in this study since its commercialization

has been recently stopped. After extraction, fibres were exposed to the HS of a

vial containing 20 L of MSTFA at 60 ºC for 15 min. GC-MS was used as

detection technique. Whatever the SPME coating, neither BP-3 nor salicylates

were obsereved in the GC-MS chromatograms monitored selecting the most

intense ions in the mass spectrum of free compounds (151+227 and 120+138,

respectively); thus, it can be assumed that they are converted quantitatively into

the corresponding silyl derivatives. The same behaviour was noticed for BP-1

and BP-8; however, it must be noted that the presence of two hydroxyl groups

in the structure of both species prevents their sensitive detection using GC

based methods. Fig. 2 compares the responses, as peak areas, obtained with the

tested SPME fibres. The lowest extraction efficiency corresponded to the CAR-

PDMS one. Probably, the molecular size of the analytes is too large to allow


184

their diffusion into the porous structure of the Carboxen sorbent. The PDMS

coating provided high responses for the less polar salicylates but failed to

extract the di-hydroxylated benzophenones. The highest efficiencies were

achieved with PDMS-DVB, except in case of BP-1 (the most polar specie), which

presented a higher affinity for the PA coating. On the basis of these data,

PDMS-DVB and PA fibres were selected for further experiments. The relative

standard deviation values (RSDs) obtained with both fibres for replicate

extractions (n=4) of samples spiked at the 4 ng mL-1 level remained under 12%.

Carry-over effects were evaluated by desorbing each fibre twice. Relative

responses up to 5% were noticed for some compounds, particularly EHS and

HMS, in the second desorption; therefore, to avoid cross contamination

problems, fibres were additional desorbed for 3 min at 270 and 280 ºC (PDMS-

DVB and PA, respectively) using a nitrogen flow of 30 mL min-1 in the split

injector of a non-operative GC instrument.

0,0E+00

5,0E+05

1,0E+06

1,5E+06

2,0E+06

2,5E+06

3,0E+06

3,5E+06

EHS HMS BP-3 BP-1 BP-8

Pe

ak

are

a

PDMS-DVB PA PDMS CAR-PDMS

Fig. 2. Comparison of responses for different SPME fibres. Direct sampling for 30 min.


185

3.2.2. Extraction parameters

The influence of different parameters on the performance of the

microextraction step was evaluated with aliquots of ultrapure water, spiked

with target compounds at 4 ng mL-1. The SPME fibre was immersed in the

water sample or maintained in the headspace (HS) of the vessel, depending on

the experiment, and then exposed to the HS of a 1.5 mL GC autosampler vial

containing MSTFA. Unless otherwise is stated, 22 mL glass vessels were used in

the sampling step. On-fibre derivatization was carried out under conditions

reported in the above paragraph and GC-MS/MS was used as detection

technique.

3.2.2.1. Sample pH, stirring and fibre selection

Optimal values for these factors were simultaneously investigated using

a two-level factorial experimental design, with two central points. Extractions

were carried out in the direct sampling mode, using 20 mL samples, without

salt addition, for 30 min. Table 3 shows extraction conditions corresponding to

the 10 experiments involved in the design, as well as the responses (peak areas)

measured for target species. Fig. 3 depicts the main effects for each factor. The

length of plotted bars is proportional to the change in the response of a given

compound when the associated factor varies from the low to the high level

within the domain of the design. A positive sign indicates an increase in the

observed response, whereas a negative value shows the opposite effect. Dotted

lines correspond to the statistic significance limit, established for a 95%

confidence level. The pH of water samples showed a negative effect on the

efficiency of the extraction step for all species. On the other hand, stirring

played a positive effect and it was statistically significant for the less polar

species, which presumably present the slower diffusion kinetics from the bulk

of the sample to the interface with the SPME fibre. The effect of the SPME

coating was compound dependant. BP-3 and BP-8 showed at higher affinity for

the PDMS-DVB fibre than for the PA one, whereas, the second coating is

preferred for the most polar BP-1. In case of salicylates, the influence of the


186

factor fibre was negligible. Two-factor interactions played a minor influence on

the SPME process, figure not shown. Stirring of water samples at 1200 rpm,

using a PTFE covered magnetic bar, pH 3 and PDMS-DVB fibres were selected

for further experiments.

Table 3. Experimental conditions employed in the 23 experimental factorial design and

responses (peak areas) attained for each compound.

Exp. Fibre pH Stirring (rpm) Response


1 PA 6 600 1961102 1732152 869855 2291319 885819

2 PDMS-DVB 3 1200 2859116 2517074 2542495 2312986 5986178

3 PA 3 600 2360417 2158650 1223590 3139505 1398965

4 PA 3 1200 3026911 2968394 1150444 3991787 3300669

5 PDMS-DVB 6 1200 3074981 2965423 2855940 2705516 6943826

6 PDMS-DVB 3 600 2537813 2224669 2136804 1911934 2207861

7 PDMS-DVB 6 600 1854872 1662130 1483205 1461656 1566350

8 PDMS-DVB 4.5 900 2552540 2399962 1409212 668191 1744285

9 PA 4.5 900 2566060 2372379 1200280 2771724 2076543

10 PA 6 1200 2669330 2513666 1373519 3112942 2674027


187

-4 -2 0 2 4 6 8

EHS

HMS

BP-3

BP-1

BP-8

Standardized main effect values

pH (3-6) Stirring (600-1200 rpm) Fibre (PA-PDMS/DVB)

Fig. 3. Combined Pareto chart showing the main effects of pH, stirring and type of fibre

on the performance of the extraction step. Low and high values of each factor are given

in the legend of the figure.

3.2.2.2. Sampling mode, temperature and ionic strength

The influence of the sampling mode was assessed with 15 mL aliquots of

a spiked water sample placed in 22 mL vessels. Extractions were carried out for

30 min, at room temperature (20 ± 2 ºC) and 100 ºC, maintaining the fibre in the

HS of the vessel or dipping it into the sample. For the two salicylates, little

differences were observed among responses obtained under different explored

experimental conditions. On the other hand, benzophenones could be hardly

detected using HS extraction, even at 100 ºC. As regards direct sampling,

around 10-fold higher responses were achieved at room temperature versus 100

ºC for these three compounds. Although, increasing the temperature of the

sample speeds up mass transference kinetics from the water sample to the fibre,

it produces a diminution in the affinity of compounds for the PDMS-DVB


188

coating. Direct sampling, at room temperature was maintained as the extraction

mode.

The effect of the ionic strength on the extraction process was investigated

with samples containing different concentrations of sodium chloride between 0

and 250 mg mL-1. Fig 4 shows the pattern followed by EHS, BP-8 and BP-1.

HMS and BP-3 behaved as EHS and BP-8, respectively. For the most polar

compound, BP-1, the efficiency of the extraction rose slightly with the ionic

strength of the water sample; whereas for the most lipophilic species (EHS and

HMS) it remained constant between 0 and 50 mg mL-1 of NaCl and then started

to decrease. For BP-8 and BP-3, with intermediate log Kow constants, a

maximum was noticed for 50 mg mL-1 of NaCl. Operating in direct immersion,

the addition of salt to the water sample benefices the thermodynamics of the

SPME but slows down its kinetics due to the increase in the viscosity of the

sample. Normally, the result of this balance is an improvement in the efficiency

of the extraction for the most polar species and a diminution for those with

lower water solubility [20], which matches with results depicted in Fig 4.

Despite above results, in order to simplify the setup of the extraction process

and also to avoid the build up of solid residues in the liner of the GC, no salt

was added to the samples in further experiments.


189

0,0E+00

1,0E+06

2,0E+06

3,0E+06

4,0E+06

5,0E+06

6,0E+06

7,0E+06

8,0E+06

0 50 100 150 200 250

NaCl (mg mL-1)

Pea

k a

rea

EHS BP-1 BP-8

Fig. 4. Influence of NaCl concentration on the efficiency of the SPME using the PDMS-

DVB fibre. Values for duplicate extractions.

3.2.2.3. Sample volume and extraction time

According to the SPME theory, in a two-phase system (sample and SPME

coating) the amount of analyte incorporated in the fibre increases initially with

the volume of sample and then, after a given value, it reaches a plateau. In

practise, the size of the sample affects also to kinetics of SPME processes. Fig. 5

compares the responses obtained for aliquots of the same spiked water sample,

poured in vessels with different nominal capacities: 10, 22 and 110 mL. In all

cases, the free HS was limited to the minimum (around 0.5 mL to allow effective

stirring of the samples) and extractions were performed in the direct mode, for

30 min. As observed, the neat effect of the sample volume was negligible, thus

the smaller vessels with capacity to accommodate a stir bar and 10 mL of water

were selected to continue this study.


190

0,0E+00

2,0E+06

4,0E+06

6,0E+06

8,0E+06

1,0E+07


Pe

ak

are

a10 mL 20 mL 110 mL

Fig. 5. Effect of sample volume on responses obtained after 30 min of sampling, n= 3 replicates.

Fig. 6 depicts the time-course of the SPME process. As observed, the

kinetics of the extraction was rather slow and only EHS and HMS reached

equilibrium conditions after 2 hours of sampling. For practical reasons,

considering that extractions were carried out manually, 30 min was adopted as

working value. The use of longer times, to increase the sensitivity of the

method, is only advisable if the proposed method is automated.

0,0E+00

3,0E+06

6,0E+06

9,0E+06

1,2E+07

0 30 60 90 120 150 180

Time (min)

Pe

ak

are

a

EHS HMS BP-3 BP-1 BP-8/2

Fig. 6. Extraction profile using the PDMS-DVB coated fibre. Direct sampling at room

temperature using magnetic stirring (1200 rpm). Responses for BP-8 have been divided

by a factor of 2.


191

3.3. On-fibre derivatization

Fine optimization of on-fibre derivatization conditions (temperature,

MSTFA volume, and time) was performed using a mixed mode 31 x 22

experimental factorial design. The factor temperature was studied at three

different levels (45, 60 and 75 ºC), whereas the volume of MSTFA and the

derivatization time were varied from 20 to 60 L and from 5 to 15 min,

respectively. Derivatization conditions and experimental responses obtained in

this design are summarized in Table 4. Fig. 7 shows the main effects plot for

each compound. The slope and the length of depicted lines correspond to the

variation in the peak area for each UV filter, when the associated factor changes

from the low to the high level within the domain of the design. Temperature

affected negatively to the performance of the derivatization for all compounds,

being statistically significant (95% confidence level) for EHS, HMS and BP-3.

Thus, it was fixed at 45 ºC. Time showed positive or negative effects depending

on the considered species; therefore, an intermediate value of 10 min was

chosen. Finally, the volume of MSTFA was the factor with a lower influence on

the responses of the analytes, 20 L was maintained as the working value for

this variable. Further attempts to perform the on-fibre derivatization reaction at

room temperature showed a decrease in the peak areas of silylated salicylates.

Under these conditions (room temperature, 10 min and 20 L of MSTFA), the

use of GC-MS detection demonstrated the existence of trace signals at retention

times of non-silylated EHS and HMS.


192

Table 4. Working conditions and responses (peak areas) obtained during optimization of

the on-fibre derivatization step using a 31 x 22 experimental factorial design.

Exp. Temperature

(ºC)

Time

(min)

MSTFA

vol. (µL)

Response


1 45 15 20 2002034 1719899 2549875 1680316 3670781

2 75 15 60 1722501 1525167 1812293 1667950 3473920

3 75 15 20 2061090 1864535 1950972 1472269 3496397

4 75 5 60 2162177 2022074 2045982 1517034 3932099

5 60 5 20 1773382 1597682 1811395 1344895 3488291

6 60 5 60 3378060 3213493 1763688 1391516 3345540

7 45 15 60 2390730 2139211 2906665 1832605 5201832

8 45 5 20 3207076 3073160 2204482 1634691 4087105

9 45 5 60 3500059 3346986 1978389 1442321 3677789

10 60 15 20 1956833 1838614 2477505 1972256 4802980

11 75 5 20 1597714 1425535 2063501 1678704 2738220

12 60 15 60 1769710 1587229 1969200 1400062 3659315


193

EHS

Temp (ºC)

45 75Time (min)

5 15MSTFA (L)

20 6018

20

22

24

26

28

(x 105)

Peak

are

a

HMS

Temp (ºC)

45 75

Time (min)

5 15

MSTFA (L)

20 6017

19

21

23

25

27

(x 105)

Peak

are

a

BP-1

Temp (ºC)

45 75

Time (min)

5 15MSTFA (L)

20 6014

15

16

(x 105)

Peak

are

a

BP-8

Temp (ºC)45 75

Time (min)5 15

MSTFA (L)

20 6034

36

38

40

42(x 105)

Peak

are

a

BP-3

Temp (ºC)

45 75

Time (min)

5 15

MSTFA (L)

20 6018

20

22

24

26

(x 105)

Peak

are

a

Fig. 7. Main effect plots showing the influence of temperature, volume of MSTFA and

time on the efficiency of the on-fibre silylation reaction.

3.4. Analytical figures of merit

The linear response range of the developed method was evaluated using

standard solutions, prepared in ultrapure water, at seven different

concentration levels between 10 and 4000 ng L-1. Table 5 shows the correlation

coefficients (R2) corresponding to the representation of peak area versus

concentration. Values higher than 0.993 were obtained for all compounds.

Repeatability was assessed with samples spiked at three different

concentrations (40, 400 and 4000 ng L-1) within the calibration range. RSDs for

quadruplicate extractions remained below 13% for all species. The achieved


194

limits of quantification (LOQs) ranged from 0.5 to 10 ng L-1, Table 5. In case of

benzophenones, they were controlled by the sensitivity of the GC-MS/MS

system and the efficiency of the developed sample preparation process. For the

two salicylates, the lowest quantifiable concentrations (5 ng L-1) were

determined by the signals of these species in procedural blanks. Standard

deviations of peak areas for both compounds in procedural blanks (n=5

replicates) were multiplied by 10 and divided by the slope of their calibration

curves. Achieved LOQs are similar to those reported for stir bar sorptive

extraction (SBSE), followed by thermal desorption of the PDMS coated stir bar

and GC-MS determination [18,24], and nearly three orders of magnitude lower

than LOQs previously obtained for some of these compounds using SPME

without considering the derivatization step [12,19].

Table 5. Linearity, repeatability and limits of quantification (LOQs) of the proposed

method.

Compound Correlation

coefficient (R2)

Repeatability (RSD, %), n=4 replicates LOQ

(ng L-1) Added concentration (ng L-1)

(10-4000 ng L-1) 40 400 4000

EHS

HMS

BP-3

BP-1

BP-8

0.998

0.998

0.998

0.993

0.995

2

7

7

12

12

6

1

6

10

8

7

4

8

13

9

5a

5a

0.5b

10b

2b

a Calculated as 10 times the standard deviation of their responses in procedural blanks (n=5)

divided by the slope of calibration curves.

b Defined as the concentration which produces a signal (peak area) 10 times higher than the

baseline noise.

The effect of the type of matrix on the performance of the sample

preparation method was studied with ultrapure, river, treated and raw

wastewater. After filtration, each sample was divided into two aliquots, one

was processed directly and the other was fortified with the considered

compounds (200 ng L-1). Differences between responses measured for spiked


195

and non-spiked aliquots of each sample were compared with those

corresponding to ultrapure water with the same addition level. Normalized

values are summarized in Table 6. Relative recoveries around 100% were

noticed for all compounds in river and treated wastewater. Benzophenones also

showed relative recoveries over 80% for raw wastewater; however, the yield of

the extraction for salicylates underwent a significant reduction in this matrix.

EHS and HMS are the most lipophilic of the tested compounds; thus,

interaction with dissolved organic compounds, presented in untreated

wastewater, reduces their affinity for the PDMS-DVB coating in a higher

extension than for the more polar benzophenones.

Table 6. Relative recoveries, with their standard deviations (n=4 replicates), for

different water samples spiked at the 200 ng L-1 level.

Compound River water Treated wastewater Raw wastewater

EHS

HMS

BP-3

BP-1

BP-8

110 ± 8

109 ± 8

108 ± 13

99 ± 14

97 ± 16

95 ± 11

89 ± 10

115 ± 6

97 ± 10

108 ± 4

53 ± 5

48 ± 5

93 ± 7

92 ± 5

80 ± 5

3.5. Application to real water samples

Table 7 summarizes the levels of EHS, BP-3 and BP-1 measured in grab

samples of wastewater obtained, in different dates, from the inlet and outlet

streams of the same STP, as well as in a sample of river water. The other two

compounds considered in this study, HMS and BP-8, remained under the limits

of detection of the method in all samples. BP-3 and BP-1 were ubiquitous in the

influent of the STP reaching maximum concentrations of 460 and 245 ng L-1.

According to the information given in a recent review [17], this is one of the first

reports of the existence of BP-1 in the aquatic environment. Although grab

samples have a limited usefulness to evaluate the efficiency of treatment plants,

on the basis of concentrations measured for influents and effluents collected on

the same day, it appears that BP-3 and BP-1 are effectively removed in the


196

studied STP (primary and activated sludge treatments are applied). In case of

BP-3, this conclusion is in agreement with the results obtained by Balmer et al.

for several municipal STPs in Switzerland [10]; however, in their study, the

level of BP-3 in raw wastewater reached levels as high as 8 ng mL-1. EHS was

found only in one of the processed pairs of wastewater samples at much lower

concentrations than BP-3 and BP-1. Sample 9 corresponds to the river, which

receives the effluent of the STP, as well as some non-controlled discharges of

raw wastewater. Levels of BP-3 and BP-1 in this sample were similar to those

existing in the effluent of the plant; moreover, EHS was also detected, although

not quantified, in the river water. Fig. 8 shows the chromatograms

corresponding to non-spiked and spiked (150 ng L-1) fractions of this river

water sample (code 9, Table 7), as well as a procedural blank for ultrapure

water.

Table 7. Levels of EHS, BP-3, and BP-1 in non-spiked water samples, n=3 replicates.

Code Type Sampling date Conc. (ng L-1) with their standard deviations

EHS BP-3 BP-1

1

2

3

4

5

6

7

8

9

Influent

Efluent

Influent

Efluent

Influent

Efluent

Influent

Efluent

River water

30/11/07

30/11/07

29/2/08

29/2/08

9/6/08

9/6/08

8/9/08

8/9/08

8/9/08

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

28 (1)

7.5 (0.3)

n.q.

224 (3)

n.d.

216 (27)

13 (4)

271 (22)

n.d.

462 (74)

44 (8)

52 (5)

199 (14)

n.d.

161 (11)

n.d.

131 (13)

n.d.

245 (20)

41 (2)

37 (6)

n.d.: under detection limits n.q.: under quantification limits


197

16.50 16.75 17.00 17.25 17.50 minutes

0

50

100

150

200

kCounts

BP-3 m/z 242

15.75 16.00 16.25 16.50minutes

0

25

50

75

100

125

kCounts

EHS

HMS

m/z 177

17.3 17.4 17.5 17.6 17.7 17.8 17.9minutes

0

10

20

30

40

kCounts

BP-1 m/z 271

18.0 18.1 18.2 18.3 18.4 18.5 minutes

0

5

10

15

20

kCounts

BP-8 m/z 301+329+330

ABC

Fig. 8. Overlay of GC-MS/MS chromatograms. A, procedural blank. B, Un-spiked river

water (code 9, Table 7). C, river water spiked at 150 ng L-1.

4. Conclusions

The combination of SPME with GC-MS/MS is an interesting approach

for the sensitive determination of salicylates and hydroxylated benzophenones

in water samples. The proposed method uses only 10 mL of sample, requires a

moderate sample preparation time and provides acceptable precision (RSDs

under 13%) even for samples spiked at the low ng L-1 level; moreover, it

improves considerably the LOQs of previous methods using SPME as


198

concentration technique. Significant matrix effects were only observed for two

of the considered species in raw wastewater. As far as we could trace, this work

constitutes the first application of GC-MS/MS, using electron impact ionization,

to the determination of silylated salicylates and benzophenones. Results

obtained for a limited number of municipal raw wastewater samples revealed

the ubiquity of BP-3 and BP-1 in this matrix. Thus, urban wastewater

contributes significantly to the discharge of both species in the aquatic

environment. On the other hand, both compounds (BP-3 and BP-1) seemed to

be effectively removed in the activated sludge STP.

Acknowledgments

This study has been supported by Spanish Government, Xunta de

Galicia, and E.U. FEDER funds (projects DGICT CTQ2006-03334 and

PGIDIT06PXIB237039PR). N. N and M. R. thank the Spanish Ministry of Science

and Innovation and Xunta de Galicia for a FPU grant and a Parga Pondal

contract, respectively.


199

References

[1] D.L. Giokas, A. Salvador, A. Chisvert, Trends Anal. Chem. 26 (2007) 360.

[2] A. Salvador, A. Chisvert, Anal. Chim. Acta 537 (2005) 1.

[3] A. Zenker, H. Schmutz, K. Fent, J. Chromatogr. A 1202 (2008) 64.

[4] R. Rodil, J.B. Quintana, P. López-Mahía, S. Muniategui-Lorenzo, D. Prada-Rodríguez, Anal.

Chem. 80 (2008) 1301.

[5] K.W. Ro, J.B. Choi, M.H. Lee, J.W. Kim, J. Chromatogr. A 688 (1994) 375.

[6] A. Salvador, I. Gadea, A. Chisvert, M.C. Pascual-Martí, Chromatographia 54 (2001) 795.

[7] S. D. Richardson, Anal. Chem. 80 (2008) 4373.

[8] P. Cuderman, E. Heath, Anal. Bioanal. Chem. 387 (2007) 1343.

[9] T. Poiger, H.R. Buser, M.E. Balmer, P.A. Bergqvist, M.D. Müller, Chemosphere 55 (2004) 951.

[10] M.E. Balmer, H.R. Buser, M.D. Müller, T. Poiger, Environ. Sci. Technol. 39 (2005) 953.

[11] W. Li, Y. Ma, C. Guo, W. Hu, K. Liu, Y. Wang, T. Zhu, Water Res. 41 (2007) 3506.

[12] T. Felix, B.J. Hall, J.S. Brodbelt, Anal. Chim. Acta 371 (1998) 195.

[13] H. Gonzalez, C.E. Jacobson, A.M. Wennberg, O. Larkö, A. Farbrot, Anal. Chem. Insights 3

(2008) 1.

[14] M. Schlumpf, B. Cotton, M. Conscience, V. Haller, B. Steinmann, W. Lichtensteiger,

Environ. Health Perspect. 109 (2001) 239.

[15] M. Heneweer, M. Muusse, M. van den Berg, J.T. Sandersson, Toxicol. Appl. Pharmacol. 208

(2005) 170.

[16] N. Negreira, P. Canosa, I. Rodríguez, M. Ramil, E. Rubí, R. Cela, J. Chromatogr. A 1178

(2008) 206.

[17] M. S. Díaz-Cruz, M. Llorca, D. Barceló, Trends Anal. Chem. 27 (2008) 873.

[18] R. Rodil, M. Moeder, J. Chromatogr. A 1179 (2008) 81.

[19] D.A. Lambropoulou, D.L. Giokas, V.A. Sakkas, T.A. Albanis, M.I. Karayannis, J.

Chromatogr. A 967 (2002) 243.

[20] P. Canosa, I. Rodríguez, E. Rubí, M.H. Bollaín, R. Cela, J. Chromatogr. A 1124 (2006) 3.

[21] I. Rodríguez, J. Carpinteiro, J.B. Quintana, A. M. Carro, R. A. Lorenzo, R. Cela, J.

Chromatogr. A 1024 (2004) 1.

[22] H.K. Jeon, Y. Chung, J.C. Ryu, J. Chromatogr. A 1131 (2006) 192.

[23] R. Rodil, M. Moeder, Anal. Chim. Acta 612 (2008) 152.

[24] M. Kawaguchi, R. Ito, H. Honda, N. Endo, N. Okanouchi, K. Saito, Y. Seto, H. Nakazawa, J.

Chromatogr. A 1200 (2008) 260.

1.6. Publicación:

SOLID-PHASE EXTRACTION

FOLLOWED BY LIQUID CHROMATOGRAPHY


FOR THE DETERMINATION OF

HIDROXILATED BENZOPHENONE UV ABSORBERS

IN ENVIRONMENTAL WATER SAMPLES

N. Negreira, I. Rodríguez, M. Ramil, E. Rubí, R. Cela


(doi:10.1016/j.aca.2009.09.033)


203

Solid-phase extraction followed by liquid chromatography tandem mass

spectrometry for the determination of hydroxylated benzophenone UV

absorbers in environmental water samples

N. Negreira, I. Rodríguez, M. Ramil*, E. Rubí, R. Cela

Departamento de Química Analítica, Nutrición y Bromatología, Instituto de

Investigación y Análisis Alimentario, Universidad de Santiago de Compostela,

Santiago de Compostela 15782, Spain.

Abstract

A procedure for the determination of six derivatives of 2-

hydroxybenzophenone, used as UV absorbers, in water samples is presented.

Compounds were first concentrated using a solid-phase extraction (SPE)

cartridge and then selectively determined by liquid chromatography tandem

mass spectrometry (LC-MS/MS) using electrospray ionization (ESI). The effect

of different parameters on the performance of concentration and determination

steps is discussed. The highly polar and acidic 2-hydroxy-4-

methoxybenzophenone 5-sulfonic acid (BP-4) required the use of ammonium

acetate as modifier during desorption of SPE cartridges and also to improve the

performance of its separation in the LC column. Under optimised conditions,

the proposed method provided limits of quantification from less than 1 to 32 ng

L-1, depending on the compound and the type of water sample. Recoveries from

the SPE step (83-105%) remained unaffected by the nature of the matrix;

however, the efficiency of electospray ionization was compound and sample

dependant. Real sample analysis reflected the presence of three of the six

investigated species (BP-4; 2-hydroxy-4-methoxybenzophenone, BP-3, and 2,4-

dihydroxybenzophenone, BP-1) in the aquatic environment, particularly in raw

wastewater samples. In this latter matrix, BP-4 was the compound presenting

the highest concentrations; moreover, it was poorly removed in sewage

treatment plants and consequently it also appeared in river water.


204

Keywords: Hydroxylated benzophenones, UV absorbers, liquid

chromatography tandem mass spectrometry, water analysis.

1. Introduction

Derivatives of 2-hydroxybenzophenone are extensively employed as UV

absorbers. In the European Union (EU), 2-hydroxy-4-methoxybenzophenone-5-

sulphonic acid (BP-4) and 2-hydroxy-4-methoxybenzophenone (BP-3) have

been approved to be used as UV filters in sunscreens at maximum individual

concentrations of 5 and 10%, respectively [1]. Other countries, e.g. Japan, also

allow the incorporation of 2,4-dihydroxybenzophenone (BP-1), 2,2´,4,4´-

tetrahydroxybenzophenone (BP-2) and 2,2´-dihydroxy-4,4´-

methoxybenzophenone (BP-6) in sunscreens [2]. In addition, the above

compounds, as well as other related benzophenones such as 2,2´-dihydroxy-4-

methoxybenzophenone (BP-8), are included as photostabilizers in many other

personal care products (e.g. hair dyes and shampoos), varnishes, clothes and

food container plastics [3,4]. BP-1 is also the main metabolite of BP-3, identified

in human urine after topical application of sunscreen products containing the

latter compound [5,6].

The increasing usage of hydroxylated benzophenones, combined with

their moderate to high water solubility, has led to the appearance of some of

these compounds in the aquatic environment. BP-3 has been often detected in

different water samples, from recreational areas (e.g. swimming pool and

bathing waters) to surface water and municipal wastewater [7-10] and even in

fish tissues, which suggests that it is bio-accumulated in the food chain [11,12].

More recently, BP-4 [13,14] and BP-1 [15] have been also found in water

samples and the first species seems to be able to pass through sewage treatment

plants (STPs) without being significantly removed [13,14,16]. On the other

hand, the levels and fate of BP-2, BP-6 and BP-8 in the aquatic environment

remain mostly unknown. As regards toxicological effects, in-vivo and in-vitro

studies have demonstrated that BP-1, BP-2, BP-3 and BP-4 exert estrogenic and

anti-androgenic actions and they might affect the reproduction of fish [17-20].


205

On the other hand, BP-8 is considered as a genotoxic compound [6]. Globally,

the above data raises the concerns about medium and long-term environmental

effects of hydroxylated benzophenones.

Until recently, gas chromatography in combination with mass

spectrometry (GC-MS) was the most common technique for the determination

of benzophenone-type UV absorbers in water and other environmental matrices

[21]. When combined with effective sample preparation techniques, such as

liquid-liquid extraction (LLE) [4,9], solid-phase extraction (SPE) [8,11], solid-

phase microextraction (SPME) [15] or stir-bar sorptive extraction (SBSE) [22,23],

it provides limits of detection in the low ng L-1 range, particularly, if acetic

anhydride [23] or a silylation reagent [4,15,24] are used to derivatize the

phenolic groups contained in the structure of target benzophenones.

Derivatization is particularly important for sensitive detection of those

compounds with two or more phenolic moieties, such as BP-1, BP-2, BP-6 and

BP-8. Unfortunately, BP-4, which contains a sulfonic group attached the

aromatic ring, in addition to the phenolic one, is not amenable to GC analysis.

The above limitations can be overcome using liquid chromatography (LC)-

based methods. BP-3 and BP-4 have been often included in multi-residue LC-

MS/MS methods focused on the determination of personal care products [16]

and multi-class UV filters [13,25] in water samples; however, the performance

of LC-MS/MS for the determination of other hydroxyl benzophenones in water

has received less attention. Although MS in combination with LC has been

recently applied to a broader group of benzophenones [14], this work was

devoted to biological samples and it did not deal either with the concentration

of water samples, or with the use of tandem MS.

Thus, the aim of this research is to evaluate the efficiency of LC-MS/MS,

using electrospray ionization (ESI), for the determination of 6 hydroxylated

benzophenones: BP-1 to BP-4, BP-6 and BP-8 in water samples and to assess

their levels and fate in this matrix, with special attention focused on

wastewater. Samples were concentrated using a reversed-phase SPE sorbent in


206

order to improve the sensitivity of the method. The influence of different

parameters on the performance of sample preparation and determination steps

is thoroughly evaluated. Data related to the levels of target compounds in

different environmental water samples are also provided.

2. Experimental

2.1. Standards, solvents and material

Standards of BP-1, BP-2, BP-3, BP-4, BP-6 and BP-8 were purchased from

Aldrich (Milwaukee, WI, USA) and Riedel de Haën (Seelze, Germany). The

chemical structures of these compounds and some properties of relevance to

optimize extraction (SPE) and LC separation steps are given in Table 1. Formic

and hydrochloric acid, ammonium acetate and HPLC-grade methanol were

acquired from Merck (Darmstadt, Germany). Ultrapure water was obtained

from a Milli-Q system (Millipore, Billerica, MA, USA). Stock solutions of each

compound were prepared in methanol. Further dilutions and mixtures of them

were made in the same solvent. Standards in methanol were stored in the dark

at 4 ºC for a maximum of two months. Calibration standards in a

methanol:water (1:1), containing a 5 mM concentration of ammonium acetate,

were prepared when needed, at different concentrations in the range from 5 to

1000 ng mL-1.

SPE cartridges, containing 60 mg of the OASIS HLB sorbent, were

provided by Waters (Milford, MA, USA).


207

Table 1. Abbreviated names, CAS number, structures, pKa and octanol-water partition

coefficients of selected compounds.

Abbreviated

name

CAS

number

Molecular

weight (a.m.u.) Structure apKa aLog Kow

BP-1 131-56-6 214.2

7.53 3.17

BP-2 131-55-5 246.2

6.98 3.16

BP-3 131-57-7 228.2

7.56 3.79

BP-4 4065-45-6 308.3

-0.70 0.89

BP-6 131-54-4 274.3

6.80 4.10

BP-8 131-53-3 244.2

6.99 3.93

a Values obtained from SciFinder Scholar Database, http://www.cas.org/products/sfacad/

2.2. Samples and sample preparation

Ultrapure (Milli-Q), river and wastewater samples, obtained from the

inlet and outlet streams of an urban STP, equipped with primary and activated

sludge units, were employed throughout this study. River and wastewater

samples were stored in the dark, at 4º C, for a maximum of 72 hours until being

concentrated by SPE. Under optimised sample preparation conditions, from 200

to 500 mL of acidified water (adjusted to pH 2 using HCl 0.1 M) were filtered,

using glass fibre and cellulose acetate (0.45 m pore size) filters, and passed

through an OASIS HLB cartridge (c.a. 10 mL min-1), previously conditioned


208

with methanol and ultrapure water (3 mL each) at the same pH as water

samples. After that, the beaker containing the sample and connections with the

SPE cartridge were rinsed with 20 mL of ultrapure water. Finally, cartridges

were dried using a stream of nitrogen and analytes eluted with 3 mL of

methanol, containing a 5 mM concentration of ammonium acetate. This extract

was evaporated to dryness and reconstituted with 1 mL of a methanol:water

(1:1) solution, 5 mM in ammonium acetate.

Performance of the SPE process was assessed by spiking the acidified

samples after the filtration step. Unless otherwise stated, 10 ng mL-1 was used

as the addition level during optimisation of SPE conditions.

2.3. Equipment

Analytes were determined using a Varian (Walnut Creek, CA, USA) LC-

MS/MS system. The LC instrument comprised two isocratic, high-pressure

mixing pumps (Varian ProStar 210), an autosampler and a thermostated

compartment for the column (Varian ProStar 410). The mass spectrometer was

a U-shaped triple quadrupole (Varian MS 1200L) furnished with an

electrospray ionization (ESI) interface. The LC-MS/MS instrument was entirely

controlled by the Varian MS Workstation Version 6.9 software.

Compounds were separated using a Kromasil C18 column (100 mm x 2.1

mm; 5 m) acquired from Sugelabor (Madrid, SPAIN) and connected to a C18

(4 mm x 2 mm) guard cartridge supplied by Phenomenex (Torrance, CA, USA).

Ultrapure water (A) and methanol (B), containing different amounts of formic

acid or ammonium acetate as modifiers, were employed as mobile phases.

Under final conditions, a 5 mM concentration of the latter modifier was added

to both solvents and compounds were separated using the following gradient:

0-2 min, 15% B; 6 min, 40% B; 10 min, 70% B; 18 min, 75% B; 21-24 min, 100% B;

27-36 min, 15% B. The mobile phase flow was set at 0.2 mL min-1 and the

temperature of the column fixed at 35 ºC. Injection volume for standards and

sample extracts was 15 L.


209

Nitrogen (99.999%), used as nebulising (50 PSI) and drying gas (300 ºC,

21 PSI) in the ESI source, was provided by a high purity generator (Domnick

Hunter, Durham, UK). The temperature of the ESI housing was maintained at

50 ºC and the voltage of the ESI needle fixed at 5000 V. Argon (99.999%) was

employed as collision gas (1.5 mTorr) in the mass spectrometer. Benzophenones

were recorded in the multiple reaction monitoring (MRM) mode, using two

transitions per compound. The most intense one was used to quantify the

response of each species in standards and SPE extracts from real water samples.

Under final working conditions, the six benzophenones were grouped in four

segments according to their elution order. Table 2 summarizes retention times,

most intense MS/MS transitions, ionization modes (positive or negative),

capillary voltages and collision energies for target species.

Table 2. Retention times and optimized ESI-MS/MS conditions. CV (capillary voltage,

V); CE (collision energy, eV).

Compound Ret. time

(min)

ESI Segment MRM1

(quantification)

CV/CE MRM2

(confirmation)

CV/CE

BP-4

BP-2

BP-1

BP-8

BP-6

BP-3

10.5

11.8

14.1

15.6

18.2

19.5

-

-

-

-

-

+

1

2

3

3

4

4

307>210

245>135

213>91

243>93

273>123

229>151

80/34

40/14

60/29

30/17

40/20

52/17

307>227

245>109

213>135

243>123

273>108

229>105

80/26

40/22

60/19

30/16

40/31

52/18


210


3.1. Optimization of LC-MS/MS parameters

3.1.1. MS/MS transitions

MS/MS fragmentation conditions were optimized by infusion of

individual standards (10 g mL-1 in methanol:water, 1:1) at a constant flow of 50

L min-1, operating the electrospray source in the positive (ESI+) and negative

(ESI-) modes. Except for BP-4, which could be ionized only in negative mode,

the rest of compounds yielded the corresponding protonated ([M+H]+)and

deprotonated ([M-H]-) parent ions in ESI+ and ESI-, respectively. Using MS/MS

detection, [M+H]+ ions underwent the cleavage of the bond between the

carbonyl group and one aromatic ring, with the positive charge remaining in

the fragment attached to the carbonyl moiety. In case of BP-3, the most intense

products appeared at 151 and 105 m/z units, Fig. 1A. The same bond was also

broken during fragmentation of [M-H]- precursors. For those compounds

containing just a hydroxyl substituent over each phenolic ring, e.g. BP-3, BP-6

and BP-8, the carbonyl moiety did not remain attached to MS/MS phenolate

product ions, Fig. 1B. Fig. 1C shows the MS/MS fragmentation pattern of those

benzophenones with two hydroxyl moieties in the same aromatic ring (BP-1

and BP-2), using BP-1 as model compound. In addition to the formation of

dihydroxylated phenolic ions, appearing at a m/z of 109 units, the carbonyl

moiety may also remain attached to the negative charged product ion (m/z

135), being further removed together with one atom of oxygen as CO2

(transition 135 > 91 m/z). Finally, the most intense ions in the MS/MS spectrum

of BP-4 maintained the benzophenone structure and corresponded to the loss of

sulfonic (307>227) and sulfonic plus hydroxyl moieties (307>210), Fig. 1D.

Other transitions, e.g. 307>80 and 307>211, previously reported in the literature

[13,16], were also observed; however, their intensities were lower than those

corresponding to the former ones.


211

m/z 229

m/z 105m/z 151

[M+H]+ A

D

C

[M-H]-

m/z 307 m/z 227

m/z 210

O O

MeO

m/z 91

m/z 213

m/z 135

m/z 109

[M-H]-

B

- 44

(CO2)

[M-H]-m/z 243

m/z 93 m/z 123

Fig. 1. Proposed MS/MS fragmentation mechanisms. A, BP-3 (ESI+). B, BP-8 (ESI-).

C, BP-1 (ESI-). D, BP-4 (ESI-).


212

Flow injection analysis (FIA) was used to investigate the effect of the

ionization mode on the intensity of MS/MS responses for each compound. ESI-

was preferred for all analytes, except for BP-3, which gave an increased

response (three orders of magnitude) in positive mode.

3.1.2. Mobile phase modifiers

Addition of formic acid to the mobile phase (concentrations up to 0.1%

were evaluated) produced a reduction in the responses measured for BP-1, BP-

2, BP-6 and BP-8. These compounds present pKa values from 6.8 to 7.5 units

(Table 1), thus formic acid shifts their acid-base equilibrium towards the neutral

forms, leading to a diminution of their responses in ESI-. On the other hand,

signals recorded for BP-3 in ESI+, and the strong acidic BP-4 remained

unchanged in presence of formic acid. Ammonium acetate, at increased levels

from 0 to 10 mM, produced also a slight reduction in the MRM responses of all

compounds. However, when included in the LC mobile phase, it improved the

chromatographic behavior of BP-4. Without this salt, using just methanol and

water as mobile phases, the retention of this highly polar compound (log Kow

0.89) was low and two peaks with same MS/MS spectra were observed. Adding

a 5 mM of ammonium acetate to the mobile phase increased the retention time

of BP-4 around 5 min, leading to a single and symmetrical peak for this

compound.

3.1.3. Instrumental performance

Using the gradient given in the experimental section, selected

compounds were separated in about 20 min, Fig. 2. In order to improve the

achieved limits of quantification (LOQs), they were grouped into four different

segments according to their retention times. In the latter one, negative and

positive ionization modes were combined to record BP-6 and BP-3 under

optimal conditions. The total dwell time per segment was maintained at 1.2 s.


213

Other MS/MS detection conditions, as well as retention times of selected

benzophenones are compiled in Table 2.

10.0 12.5 15.0 17.5 20.0Min.

0

20

40

kCounts

0.0

0.5

1.0

1.5

MCounts

0.50

1.00

1.50MCounts

0

100

200

kCounts

0

100

200

300

kCounts

0

20

40

kCounts

BP-4307 > 210

BP-2245 > 135

BP-1213 > 91

BP-8243 > 93

BP-6273 > 123

BP-3229 > 151

0.0

Fig. 2. LC-MS/MS chromatogram for a standard of benzophenones (25 ng mL-1 per

compound).

Relevant data related to the performance of the optimized LC-MS/MS

method are summarized in Table 3. The dependence between peak areas and

analytes concentration was investigated with standards (injection volume 15


214

L) at 7 different concentrations in the range from 5 to 1000 ng mL-1. BP-3, BP-4,

BP-6 and BP-8 gave a linear response in the above range, with correlation

coefficients (R2) between 0.998 and 0.999; whereas, BP-1 and BP-2 fitted better a

quadratic plot with R2 values of 0.999 and 0.997, respectively. Instrumental

limits of quantification, defined for a signal to noise ratio of 10 (S/N=10) varied

from 0.2 for BP-4 up to 4 ng mL-1 in case of BP-3 and BP-8, Table 3. The

repeatability in the responses of the system was evaluated with standards at

two different concentrations: 25 and 125 ng mL-1. Relative standard deviations

(RSDs, %) for 5 injections made in the same day ranged from 1 to 6%, Table 3.

For injections performed on consecutive days RSDs between 3 and 14% were

obtained.

Table 3. Linearity, instrumental limits of quantification (LOQs), repeatability and

reproducibility of the LC-MS/MS system.

Compound Correlation

coefficient (R2)

LOQs

(ng mL-1)

aRepeatability (RSD, %) bReproducibility (RSD, %)

125 ng mL-1 25 ng mL-1 125 ng mL-1

BP-4

BP-2

BP-1

BP-8

BP-6

BP-3

0.998

c0.997

c0.999

0.999

0.999

0.998

0.2

0.4

3

4

3

4

6.1

4.0

3.2

4.0

5.0

3.5

3.7

3.4

2.4

1.1

3.5

2.0

3.4

3.1

4.5

8.3

13.5

2.6

aN=5 injections in the same day

bN=12 injections in 3 consecutive days

cQuadratic model

3.2. Solid-phase extraction conditions

The optimisation of SPE conditions was performed with aliquots of

ultrapure water spiked at 10 ng mL-1 and passed through 2 cartridges

connected in series. After that, sample containers and connections with the

sorbents were rinsed with ultrapure water, ca. 20 mL. Cartridges were then


215

disassembled, dried and eluted separately with different volumes (from 1 to 10

mL) of methanol or methanol containing a 5 mM concentration of ammonium

acetate. Extracts in methanol were diluted (1:1) with ultrapure water 5 mM in

ammonium acetate before injection in the LC-MS/MS system. The pH of the

sample showed a strong effect on the breakthrough volume of BP-4. When

samples were acidified at pH 2, up to 1000 mL of water could be concentrated

without noticing the presence of any compound in the extract from the second

cartridge. At neutral pH values (between 6 and 8 units), BP-4 showed a

breakthrough volume around 200 mL, whereas the rest of species were still

quantitatively retained in the first SPE cartridge. In the elution step,

hydroxylated benzophenones could be recovered with 3 mL of methanol;

however, BP-4 presented a lower affinity for this solvent, requiring much larger

volumes of eluent. This problem was overcome using methanol containing

ammonium acetate (5 mM). All compounds were recovered with 3 mL of this

mixture, a volume significantly lower than those reported in previous works

dealing with SPE of BP-4 and BP-3 [13, 25]. It is probable that the ammonium

ions neutralize the negative charge of the sulfonic group, increasing the

solubility of BP-4 in methanol.

In a further series of assays, performed with samples spiked at a lower

concentration level (ca. 1 ng mL-1), extracts from SPE cartridges were

concentrated to dryness and reconstituted with 1 mL of the same solution used

to prepare calibration standards. Losses of the analytes were not detected when

evaporation was carried out at room temperature with a gentle stream of

nitrogen. Table 4 shows the absolute recoveries, estimated against external

calibration, for 500 and 1000 mL aliquots of spiked ultrapure water (1 ng mL-1)

using above conditions. Average recoveries from 87 to 103%, with acceptable

precision, were attained for all compounds. These values are similar to the

recoveries obtained for BP-3 and other UV filters in previous works [7,11] using

different SPE sorbents.


216

Table 4. Extraction yields of the solid-phase extraction process for ultrapure water

samples. Added concentration 1 ng mL-1, n=3 replicates.

Compound Extraction yield (%) ± SD

a500 mL a1000 mL

BP-4

BP-2

BP-1

BP-8

BP-6

BP-3

87 ± 1

96 ± 1

96 ± 6

90 ± 8

87 ± 9

103 ± 6

92 ± 1

87 ± 6

99 ± 3

91 ± 4

89 ± 3

97 ± 7

aVolume of sample

3.3. Matrix effects and figures of merit

The efficiency of the ESI interface is prone to changes depending on the

complexity of the sample. Matrix effects for the proposed method were assessed

comparing the differences between responses obtained for spiked and non-

spiked extracts from different water samples (extracts were fortified after SPE at

the 500 ng mL-1 level) with those measured for a standard of the same

concentration, prepared in methanol: water (1:1), both 5 mM in ammonium

acetate. Fig. 3 shows the results obtained for river (500 mL), treated (300 mL)

and raw (200 mL) wastewater samples. In case of river and treated wastewater,

BP-2 was the only compound undergoing strong signal suppression (up to 80%

in the latter matrix); whereas, responses measured for the rest of hydroxylated

benzophenones represented between 75 and 113% of those corresponding to the

reference standard. The response of BP-4 experienced a moderate enhancement

(around 30%) for the above matrices, similar to previous data published by

Rodil and co-workers [13]. Probably, the elution of polar compounds at short

retention times impaired the ionization of BP-2 in the negative mode. Although

BP-4 elutes even earlier, the high acidity of the sulfonic group, led to a more

robust ionization ESI-. As expected, for raw wastewater the signal suppression

for BP-2 was even more acute. The reduction in the efficiency of the ionization

for the rest of species ranged from 25 to 50% in this matrix; however, no


217

problems were observed for BP-4, Fig. 3. Signal suppression effects for BP-3 and

BP-4 in raw wastewater samples were less significant than those reported for

the same sample intake in a recent work [25]. Differences between (1) employed

SPE conditions, (2) selected MS/MS transitions for BP-4 and (3) configurations

of the ESI interface in LC-MS/MS systems from different suppliers might

contribute to these findings. On the basis of data depicted in Fig. 3, and

considering the lack of isotopic labeled analogous of target compounds, the

standard addition method is recommended to assess the levels of hydroxylated

benzophenones in raw wastewater. For the rest of matrices, this quantification

approach is only mandatory for the accurate determination of BP-2. Anyhow,

matrix effects have to be systematically checked since they might change among

different samples, even of the same type (e.g. surface of treated wastewater),

depending on the levels of dissolved salts and organic compounds.

0%

20%

40%

60%

80%

100%

120%

140%

BP-4 BP-2 BP-1 BP-8 BP-6 BP-3

No

rma

lize

d r

esp

on

se

River Treated wastewater Raw wastewater

Fig. 3. Assessment of matrix effects for river (500 mL), treated (300 mL) and raw

wastewater samples (200 mL). Normalized responses to a standard of same

concentration, n=4 replicates.

Table 5 shows the recoveries of the method for river, treated and raw

wastewater after correcting the responses measured for the final extract from

the SPE cartridge with above referred matrix effects. Each sample was divided

into several fractions, some of them were processed directly and the others


218

fortified with different concentrations of target compounds in the range from

0.15 to 2.0 ng mL-1, depending on the matrix being investigated. Corrected

recoveries ranged from 83 to 105%, indicating that the efficiency of the SPE

process was not comprised by the type of water sample. Despite some

researchers have reported cross contamination problems for UV filters [26],

procedural blanks, corresponding to the concentration of ultrapure water

samples, did not reveal noticeable responses for target species. The lower log

Kow values of hydroxylated benzophenones (from 0.9 to 4.1), in comparison

with other UV filters such as camphors and cinnamates (e.g. octocrylene log

Kow 7.5), may explain the absence of carryover problems due to analytes

adsorption on glass material and connectors of the SPE sample preparation

station. Anyhow, the use of gloves during manipulation of water samples and

SPE extracts is mandatory in order to prevent contamination from personal care

products used by operators. Taking these results into account, the limits of

quantification (LOQs) of the proposed method, defined for a signal to noise

(S/N) of 10, are controlled by the sensitivity of the LC-MS/MS system, the

concentration factor provided by the sample preparation method: 500, 300 and

200-fold for river, treated and raw wastewater, respectively and signal

suppression effects depicted in Fig. 3. Estimated values ranged from less than 1

ng L-1 up to 32 ng L-1, depending on the compound and the type of water

sample. In case of BP-4 and BP-3, LOQs achieved in this work are significantly

lower than those reported recently by Kasprzyk-Hordern and co-workers in

wastewater samples: 10 and 80 ng L-1 for BP-4 and BP-3, respectively [16]. The

detection of BP-3 in the negative ionization mode probably contributed to the

limited sensitivity of their method for this compound [16]. LOQs reported on

Table 5 are also similar to those achieved combining SPME with GC-MS/MS

detection after on-fiber derivatization of target species [15]; however, this latter

procedure cannot be applied to BP-4.


219

T

able

5. C

orre

cted

rec

over

ies

(n=

4 r

eplic

ates

) and

est

imat

ed L

OQ

s fo

r th

e pr

opos

ed m

etho

d.

R

iver

(500

mL

) T

reat

ed w

aste

wat

er (3

00 m

L)

Raw

was

tew

ater

(200

mL

)

Com

poun

d

Rec

over

y (%

) S

D

LO

Q (n

g L

-1)

Rec

over

y (%

) S

D

LO

Q (n

g L

-1)

Rec

over

y (%

) S

D

LO

Q (n

g L

-1)

a 0.

15 n

g m

L-1

a 0

.4 n

g m

L-1

a 2.0

ng

mL

-1

a 2

.0 n

g m

L-1

BP

-4

BP

-2

BP

-1

BP

-8

BP

-6

BP

-3

100

± 6

101

± 5

105

± 3

104

± 5

103

± 3

98 ±

6

90 ±

5

91 ±

5

96 ±

3

95 ±

3

93 ±

4

84 ±

4

0.4

b 2 6 8 6 8

104

± 2

97 ±

4

93 ±

2

91 ±

2

91 ±

2

96 ±

2

0.7

b 14

10

13

10

13

83 ±

6

96 ±

4

101

± 2

98 ±

3

90 ±

4

87 ±

5

1

b 20

b 30

b 32

b 25

b 32

a Ad

ded

con

cent

rati

on

b Rep

orte

d v

alu

es h

ave

been

cor

rect

ed w

ith

mat

rix

effe

cts

dep

icte

d in

Fig

. 3


220

3.4. Real samples analysis

The developed method was applied to the analysis of several river and

wastewater samples taken during the period from June to September of 2008.

This season corresponds to the highest consumption of sunscreen products,

urban wastewater is not diluted by rain and rivers show their lowest flow rates.

Thus, these samples are representative of a worst-case scenario for the

investigated geographical area. The resultant concentrations for BP-1, BP-3 and

BP-4 are shown in Table 6, the rest of species remained below the detection

limits (LODs) of the method. External calibration and standard addition over

sample extracts, before dryness evaporation and re-constitution to 1 mL, were

used to quantify their levels in river and wastewater, respectively. All

wastewater samples (codes 1-8, Table 6) corresponded to the same STP,

equipped with primary and secondary treatments, which receives municipal

wastewater from a 125000 inhabitants, non-coastal city. Grab samples, without

considering the residence time of the STP, were obtained from the inlet and

outlet of the plant. BP-1, BP-3 and BP-4 were found in the four influents with

concentrations increasing in the above order from 30 ng L-1 for BP-1 up to 1600

ng L-1 for BP-4. This trend agrees with levels found in STPs from Wales [16].

Fig. 4 shows the LC-MS/MS chromatograms for spiked and non-spiked

aliquots of a raw wastewater (code 1, Table 6), as well as a procedural blank of

ultrapure water. Dissolved concentrations of BP-1 and BP-3 were considerably

reduced in the effluent of the plant; however, its efficiency was poor for BP-4,

Table 6. Although this conclusion needs to be confirmed with integrated

samples, the resistance of BP-4 to degradation agrees with (1) its high polarity,

(2) recent results published during last year for other STPs [13, 16], and (3) the

reports of this species in surface water obtained with passive sampling [14]. BP-

4 was also present in the four considered river samples (codes 9-12, Table 6).

Samples 9 and 10 were collected in the same river, five Km downstream from

the STP, in two different months. In addition to BP-4, they showed levels of BP-

1 (only sample 9) and BP-3 similar to those measured in the effluent of the STP

(codes 6 and 8). Codes 11 and 12 corresponded to small rivers flowing through


221

urban areas, which do not receive direct discharges of STPs and without

recreational bathing areas. In spite of this, BP-4 was over the quantification

limits of the method in both samples.

Table 6. Concentrations (ng L-1) of BP-4, BP-1 and BP-3 in environmental water

samples, n= 3 replicate extractions.

Code Type Sampling data Concentration, ng L-1,

(standard deviation)

BP-4 BP-1 BP-3

1

2

3

4

5

6

7

8

9

10

11

12

Raw wastewater

Treated wastewater

Raw wastewater

Treated wastewater

Raw wastewater

Treated wastewater

Raw wastewater

Treated wastewater

River

River

River

River

June 2008

June 2008

July 2008

July 2008

August 2008

August 2008

September 2008

September 2008

August 2008

September 2008

July 2008

September 2008

1293 (74)

821 (84)

1237 (10)

1028 (61)

1354 (147)

890 (97)

1596 (36)

765 (22)

416 (45)

283 (16)

20 (5)

147 (3)

40 (3)

< LOD

148 (7)

13 (2)

120 (4)

11 (1)

31 (2)

< LOD

24 (1)

< LOD

< LOD

< LOD

184 (8)

< LOD

317 (24)

83 (12)

429 (23)

77 (4)

369 (12)

84 (3)

54 (3)

87 (8)

< LOD

< LOD


222

9.5 10.0 10.5 11.0 min.

0

100

200

300

400

500

600

(x103 Counts)

BP-4

16 17 18 19 20.

0.0

0.5

1.0

1.5

2.0

(x106 Counts)

BP-6

13.0 14.0 15.0 min,

0.0

0.5

1.0

1.5

2.0

2.5

(x106 Counts)

BP-1

11.0 12.0 Min.0

100

200

300

400

500

600

700

(x103 Counts)

BP-2

16 17 18 19 20 min.

0

100

200

300

400

500

600

700

(x103 Counts)

BP-3

14.0 15.0 16.0 min.0.00

0.25

0.50

0.75

1.00

1.25

(x106 Counts)

BP-8

CBA

Fig. 4. Overlay of LC-MS/MS chromatograms. A, procedural blank for 500 mL of

ultrapure water. B, raw wastewater (code 1, Table 6). C, same sample spiked at 2 ng

mL-1.


223

4. Conclusions

LC-MS/MS combined with an SPE pre-concentration step allows the

sensitive determination of six hydroxylated benzophenones in water samples

avoiding the need of analytes derivatization, as required in GC-based methods.

The differences in polarity and acidity of BP-4 when compared with the rest of

analytes required a careful tuning of SPE and LC conditions, using an ion pair

reagent to reduce the elution volume for this compound in the SPE step and to

improve its chromatographic behavior. Conversely to the rest of species, BP-3 is

more efficiently ionized in the positive than in the negative ESI mode, thus both

ionization modes must be combined in the same chromatographic run. Under

optimized conditions, the proposed method provided enough sensitivity for the

determination of the six considered compounds in environmental water

samples. With the exception of BP-2, the rest of compounds were affected by

strong matrix suppression effects just in raw wastewater, but not in river and

treated sewage samples. Analysis of real samples confirmed the ubiquitous

presence of BP-4 in the aquatic environment, as well as its recalcitrant behavior

in STPs. This behavior added to its high polarity turns BP-4 in a mobile

compound in the aquatic media, with the potential risk to reach potable water

sources. BP-3 and BP-1 were also detected in waste and river samples;

however, they seemed to be removed to a considerable extent during

conventional wastewater treatments.

Acknowledgments

Financial support from the Spanish Government, Xunta de Galicia and

FEDER funds (projects DGICT CTQ2006-03334 and PGIDIT06PXIB237039PR) is

acknowledged. N. N. and M. R. thank the Spanish Ministry of Sciene and the

Xunta de Galicia for a FPU grant and an I. Parga Pondal contract, respectively.


224

References


[2] N.A. Shaath, The Encyclopedia of Ultraviolet Filters, Allured, Illinois, 2007.

[3] S.D. Richardson, Anal. Chem. 80 (2008) 4373.

[4] H.K. Jeon, Y. Chung, J. C. Ryu, J. Chromatogr. A 1131 (2006).

[5] H. Gonzalez, C.E. Jacobson, A.M. Wennberg, O. Larkö, A. Farbrot, Anal. Chem. Insights 3

(2008) 1.

[6] M. S. Díaz-Cruz, M. Llorca, D. Barceló, Trends Anal. Chem. 27 (2008) 873.


[8] W. Li, Y. Ma, C. Guo, Y. Hu, K. Liu, Y. Wang, T. Zhu, Water Res. 41 (2007) 3506.

[9] D. L. Giokas, V. A. Sakkas, T. A. Albanis, D.A. Lampropoulou, J. Chromatogr. A 1077 (2005)

19.

[10] D. L. Giokas, A. Salvador, A. Chisvert, Trends Anal. Chem. 26 (2005) 360.

[11] M. E. Balmer, H.R. Buser, M.D. Müller, T. Poiger, Environ. Sci. Technol. 39 (2005) 953.

[12] M.A. Mottaleb, S. Usenko, J.G. O´Donnell, A.J. Ramirez, B.W. Brooks, C.K. Chambliss, J.

Chromatogr. A 1216 (2009) 815.


Chem. 80 (2008) 1307.


[15] N. Negreira, I. Rodríguez, M. Ramil, E. Rubi, R. Cela, Anal. Chim. Acta 638 (2009) 36.

[16] B. Kasprzyk-Hordern, R.M. Dinsdale, A.J. Guwy, Anal. Bional. Chem. 391 (2008) 1293.

[17] P.Y. Kunz, H.F. Galicia, K. Fent, Toxicol. Sci. 90 (2006) 349.

[18] K. Fent, P.Y. Kunz, E. Gomez, Chimia 62 (2008) 368.

[19] C.J. Weisbrod, P.Y. Kunz, A.K. Zenker, K. Fent, Toxicol. Appl. Pharmacol. 225 (2007) 255.

[20] M. Heneweer, M. Muusse, M. van den Berg, J.T. Sanderson, Toxicol. Appl. Pharmacol. 208

(2005) 170.

[21] T. Felix, B.J. Hall, J.S. Brodbelt, Anal. Chim. Acta 371 (1998) 195.


[23] M. Kawaguchi, R. Ito, H. Honda, N. Endo, N. Okanouchi, K. Saito, Y. Seto, H. Nakazawa, J.

Chromatogr. A 1200 (2008) 260.

[24] K.W. Ro, J.B. Choi, M.H. Lee, J.W. Kim, J. Chromatogr. A 688 (1994) 375.

[25] R. Rodil, S. Schrader, M. Moeder, Rapid Commun. Mass Spectrom. 23 (2009) 580.

[26] M. Meinerling, M. Daniels, Anal. Bioanal. Chem. 386 (2007) 1465.

1.7. Publicación:

STUDY OF SOME UV FILTERS STABILITY

IN CHLORINATED WATER AND IDENTIFICATION

OF HALOGENATED BY-PRODUCTS BY

GAS CHROMATOGRAPHY-MASS SPECTROMETRY

N. Negreira, P. Canosa, I. Rodríguez, M. Ramil, E. Rubí, R. Cela

Journal of Chromatography A 1178 (2008) 206

(doi:10.1016/j.chroma.2007.11.057)


227

Study of some UV filters stability in chlorinated water and identification of

halogenated by-products by gas chromatography- mass spectrometry

N. Negreira, P. Canosa, I. Rodríguez*, M. Ramil, E. Rubí, R. Cela




Abstract

This work studies the stability of three UV filters: 2-ethylhexyl salicylate

(ES), 2-ethylhexyl 4-(dimethylamino) benzoate (EHPABA) and 2-hydroxy-4-

methoxybenzophenone (BP-3), in water samples containing low concentrations

of free chlorine. Moreover, 2,4-dihydroxybenzophenone (2,4-DHBP), a

metabolite of BP-3, was also included in some of the performed assays.

Experiments were carried out considering free chlorine and analytes

concentrations at the g mL-1 and ng mL-1 level, respectively. Gas

chromatography with mass spectrometry was used to follow the time course of

target compounds and to identify their halogenated by-products. Concentration

of water samples with solid-phase extraction cartridges and derivatization

(silylation) of some species were also employed to improve their detectability.

Under the experimental conditions explored in this work, ES showed an

acceptable stability whereas the rest of species reacted with free chlorine at

significant rates following pseudo-first-order kinetics. Their half-lives ranged

from 0.4 to 25 min depending on the UV filter, chlorine concentration, water pH

and presence of bromide traces. For EHPABA a relatively simple degradation

pathway was established. It consisted of aromatic substitution of one atom of

hydrogen per chlorine or bromide. The same reaction pattern was observed for

BP-3 leading, in this case, to mono- and di-halogenated by-products. In

addition, several halogenated forms of 3-methoxyphenol were identified as BP-

3 cleavage by-products.


228

Keywords: UV filters, halogenated by-products, chlorine, bromide, water

samples, gas chromatography-mass spectrometry

1. Introduction

Awareness regarding harmful effects of solar radiation on the skin has

increased the production and consumption of the so-called UV filters. These

compounds are classified as either inorganic or organic species. Most of the

organic UV filters are relatively lipophilic compounds, which contain aromatic

rings and conjugated carbon-carbon double bonds in their structures. They

absorb radiation at certain wavelengths in the range of 280-400 nm. UV filters

are considered as cosmetics which are incorporated in the formulation of

several personal care products, such as lipsticks, hair dyes, beauty creams,

shampoos and in particular sunscreen lotions [1-3]. In the latter application,

several organic filters are normally combined in the same product to provide

UV protection in a wide range of wavelengths. Their individual concentrations

may represent up to 10% of the product weight, according to the EU legislation

[4-6].

The presence of organic UV filters in the aquatic media, added to the

controversy about the potential endocrine disrupting activity of some of these

compounds [7-9], has increased concern about their possible environmental

effects. In the case of water samples, measured concentrations ranged from the

low ng L-1 in surface water [10-14] up to the ng mL-1 level in swimming pool,

bathing and wastewater [5,14-16]. Direct release from skin in swimming pools

and sunbathing areas, industrial wastewater and indirect discharges (e.g.

during showering, clothes washing and urinary excretion) through domestic

wastewater, represent the main input routes of these compounds in water

bodies [2,5-6,17]. UV filters have been detected in sludge from sewage

treatment plants with average concentrations over the 1 g g-1 level for the most

lipophilic ones, such as 3-(4-Methylbenzylidene) camphor (4-MBC), octocrylene


229

(OC) and octyl triazone (OT) [18]. 2-Ethylhexyl methoxycinnamate (EHMC), 4-

MBC, OC and 2-hydroxy-4-methoxybenzophenone (BP-3) have been also found

in fish tissues from Swiss rivers and lakes receiving discharges of waste and

bathing water [14, 19]. In summary, the available information suggests that

some UV filters behave as persistent and bioaccumulative pollutants in the

aquatic environment.

Additionally to monitoring studies, a realistic estimation of potential

environmental risks associated with the presence of UV filters in the aquatic

media requires also to evaluate their degradation rates and the formation of by-

products. Up to now, very few articles have dealt with this topic. Most of them

have been focussed on the study of aqueous photolysis reactions [20-22]. These

works estimated the half-lives and the rate constants of several UV filters under

different experimental conditions [20-21]. In addition, Sakkas et al. [20] have

reported the formation of several chlorinated forms of 2-ethylhexyl 4-

(dimethylamino) benzoate (EHPABA) after a prolonged (60 hours) exposure of

chlorinated swimming-pool water samples, previously spiked with this

compound, to solar irradiation. This work constitutes the first evidence of the

reactivity between EHPABA and chlorine in water samples; however, it does

not provide data neither about the rate of EHPABA halogenation reactions nor

about the stability of the generated by-products.

Formation of halogenated disinfection by-products in chlorinated water

samples is a relevant process for many organic substances. This kind of

reactions is particularly favourable for species with phenolic and/or amino

groups in their structures [23-25]. In this sense, the reactivity of several

pharmaceuticals and personal care products in chlorinated water samples has

been studied recently [23,26-28]. For some compounds, kinetically favourable

reactions leading to the formation of relatively stable and toxic by-products

have been reported [23,29]. As far as we know, such information is not available

for organic UV filters.


230

In this work, the reactivity of three commonly used UV filters containing

phenolic or amino groups, BP-3, EHPABA and 2-ethylhexyl salicylate (ES), in

chlorinated water samples was assessed. Aims of the study were: (1) to evaluate

the stability of parent species in presence of free chlorine (sodium hypochlorite

plus hypochlorous acid) at neutral pH values, (2) to obtain their half-lives under

different experimental conditions and (3) to identify the corresponding

halogenated by-products, as well as to determine their stability under oxidative

conditions. Experiments were carried out using aliquots of ultrapure water

spiked with free chlorine concentrations at the low g mL-1 level, similar to

those found in tap and swimming pool water, and standard solutions of each

UV filter in the ng mL-1 range. In further assays, commercial personal care

products, containing UV filters, were mixed directly with chlorinated tap water

to confirm the formation of those by-products observed in model experiments.

Solid-phase extraction and gas chromatography-mass spectrometry (GC-MS)

were employed for the concentration of water samples, the determination of

remaining UV filters levels and the identification of generated by-products.

2. Experimental

2.1. Reagents, standards and material

Methanol, dichloromethane and ethyl acetate, trace analysis grade, were

acquired from Scharlab (Barcelona, Spain). Standards of BP-3, ES, EHPABA and

2,4-dihydroxybenzophenone (2,4-DHBP), Fig. 1, as well as the silylation reagent

N-methyl-N-(tert-butyldimethylsilyl)trifluoroacetamide (MTBSTFA) were

purchased from Aldrich (Milwaukee, WI, USA). Sodium thiosulphate,

potassium bromide and potassium dihydrogen phosphate were obtained from

Merck (Darmstadt, Germany). Individual solutions of each compound were

prepared in methanol. Two series of calibration standards were made in ethyl

acetate (BP-3 and 2,4-DHBP) and dichloromethane (ES and EHPABA). BP-3, 2,4-

DHBP, and their by-products, were converted into the corresponding tert-

butyldimethylsilyl derivatives to improve their detectability by GC-MS.


231

Derivatization was carried out adding 20 L of MTBSTFA to 500 L aliquots of

calibration standards or sample extracts in ethyl acetate [27,29]. The mixture

was shaken manually for 5 minutes and then injected in the GC-MS system.

MeO

C

OOH

BP-3

Bu-n

Et

Me2N

O CH2 CHC

O

EHPABA

C

O

Bu-n

Et

O CH2 CH

OH ES

HO

C

OOH

2,4-DHBP

Fig. 1. Structures of the considered UV filters and 2,4-DHBP.

Sodium hypochlorite with a nominal free chlorine content of 4% (w/v)

was purchased from Aldrich. This solution was stored at 4 ºC and its exact

concentration determined weekly by iodometric titration as reported elsewhere

[29]. Oasis HLB (60 mg) SPE cartridges were acquired from Waters (Milford,

MA, USA).

2.2. Chlorination experiments

Stability of selected UV filters and formation of halogenated by-products

was evaluated at room temperature (18 2 ºC) considering initial

concentrations of parent species and free chlorine in the ranges of 10-50 ng mL-1

and 0.1-3 g mL-1, respectively. Reactions were carried out in amber glass

vessels containing between 100 and 250 mL of ultrapure or tap water. Ultrapure

water samples were spiked with a titrated sodium hypochlorite solution to get

the required initial concentration of free chlorine. In the case of tap water

samples, their initial free chlorine content was determined using the N,N-

diethyl-p-phenylenediamine method with photometric detection [30]. After

that, samples were adjusted at different pHs, between 6 and 8 units (with the

intent to cover the range of values expected in real water samples, e.g. tap,


232

swimming-pool and wastewater), and spiked with a solution of the considered

UV filter in methanol. In some experiments, personal care products, containing

UV filters, replaced standard solutions. Then, vessels were closed, shaken

manually for 2 minutes and allowed to stand. After an established reaction

time, the excess of chlorine was quenched with 10 mg of sodium thiosulphate

and samples processed as described below.

2.3. Analytical procedure

Water samples were concentrated using Oasis HLB 60 mg cartridges,

previously conditioned with methanol and ultrapure water (3 mL each). In

order to maximise the yield of the concentration step, slightly different

conditions were used depending on the considered compound. Samples spiked

with BP-3 or 2,4-DHBP were adjusted at pH 3, passed through the SPE cartridge

and analytes eluted with 2 mL of ethyl acetate. An aliquot of this extract was

derivatized as indicated in previous sections. For ES and EHPABA, samples

were adjusted to pH 4.5 and methanol (20% of water volume) was added to

prevent sorption losses on glass vessels and/or connections with the SPE

cartridge. After the enrichment step, analytes were recovered with 2 mL of

dichloromethane.

Parent UV filters and their by-products were determined by GC-MS

using a Varian CP 3900 gas chromatograph (Walnut Creek, CA, USA)

connected to an ion-trap mass spectrometer (Varian Saturn 2100T). Normally,

separations were carried out in a HP-5ms type capillary column (30 m x 0.25

mm I.D., df: 0.25 m) purchased from Agilent. Moreover, a HP-35 column, from

the same supplier and with same dimensions as the HP-5 one, was also used.

For both columns, the temperature of the GC oven was programmed as follows:

70 ºC (1 min), rate at 12 ºC min-1 to 280 ºC (10 min). Helium (99.999 %) was used

as carrier gas at a constant flow of 1 mL min-1. The injector was maintained at

280 ºC and injections (1-2 l) were made in the splitless mode with a purge time

of 1 min. Transfer line and ion trap temperatures were set at 280 and 220 ºC,


233

respectively. Electron Impact (EI) mass spectra (MS) were recorded in the range

of 100-650 m/z units.


3.1. Performance of the analytical procedure

Table 1 summarizes the most relevant data related to the performance of

the analytical procedure. Reported retention times were obtained with the HP-

5ms column, and, in the case of BP-3 and 2,4-DHBP correspond to their

silylated derivatives. The linearity in the response of the GC-MS system was

evaluated with standards at seven concentration levels between 2 and 2000 ng

mL-1. Correlation coefficients of the resulting graphs varied from 0.997 to 0.999,

and the instrumental quantification limits of the GC-MS system, defined for a

signal to noise ratio of 10 (S/N = 10), remained under 5 ng mL-1 (data not

shown). If the silylation step was not considered, no signal was observed for

2,4-DHPB, whereas a wide, tailing peak was obtained for BP-3. Yields of the

SPE process (n= 4 replicates) were calculated by external calibration. Under

conditions reported in the previous section, recoveries over 80 % and limits of

quantification (LOQs) between 8 and 25 ng L-1 were achieved for 500 mL

volume samples. Although, the performance of the analytical procedure for the

potential by-products of UV filters was not evaluated, it was assumed that they

behave in a similar way to parent species (particularly those presenting very

close structures), as regards the silylation reaction and affinity for the SPE

sorbent.

Table 1. Figures of merit of the analytical procedure for parent UV filters.

Analyte Retention time

(min)

Quantification

Ions (m/z)

Linearity, R2

(range, ng mL-1)

Recovery (%)

SD

LOQs

(ng L-1)

ES

EHPABA

BP-3a

2,4-DHBPa

13.09

16.91

17.34

21.65

138+120

277+165

285

385

0.999 (5-2000)

0.997 (5-2000)

0.999 (2-2000)

0.999 (2-2000)

84 7

82 6

98 8

89 8

24

25

8

8

a Data for their tert-butyldimethylsilyl derivatives


234

3.2. Stability of UV filters at different pHs and free chlorine

concentrations

Previously to investigate the formation of halogenated by-products from

selected UV filters, the stability of BP-3, EHPABA and ES in presence of

increasing chlorine levels (up to 3 g mL-1) was assessed. Experiments were

carried with ultrapure water (100 mL aliquots) buffered at three different pHs,

covering the range of values expected in different water samples, and spiked at

50 ng mL-1 with only one of the considered compounds. After a fixed time (from

2 to 30 min), the excess of chlorine was removed and samples were

concentrated to determine the remaining amount of parent compound.

Obtained results are depicted in Fig. 2. Normalised values in the Y-axis

correspond to the ratios between the responses for each analyte in the SPE

extracts from chlorinated and non-chlorinated aliquots of ultrapure water,

multiplied by 100. Each point represents the average value for duplicate

experiments. As observed, the stability of the UV filters increased in the

following order: BP-3 < EHPABA < ES. For the latter, at the 3 investigated pHs,

more than 60% of the added amount was found in the sample even when a

relatively high chlorine concentration (3 g mL-1) and a long reaction time (30

min) were considered, Fig. 2. In a real life situation, with several organic species

competing for the available chlorine, the extension of ES halogenation reactions

was estimated as negligible; consequently, it was not longer considered in this

study. On the other hand, EHPABA and BP-3 showed a lower and pH

dependant stability. EHPABA was more stable at pH 8.2 than at pHs 6.2 and

7.2, Fig. 2. This behaviour suggests that the kinetics of the reaction between the

UV filter and sodium hypochlorite (pKa 7.5) is less favourable than with

hypochlorous acid. For BP-3 (pKa 7.6), at the two considered reaction times: 10

and 2 min, its stability decreased with the increase in the pH of the water

samples, Fig. 2. Therefore, within the explored range of pH, the ratio between

non-protonated and protonated forms of the UV filter seems to be the key factor

controlling its stability.


235

BP-3

0%

20%

40%

60%

80%

100%

120%

0 0.5 1 1.5 2 2.5 3 3.5

Free chlorine (g mL-1)

Nor

mal

ised

resp

onse

pH 7.2

pH 6.2

pH 8.2

2 min

0%

20%40%

60%

80%100%

120%

0.0 0.2 0.4 0.6Free chlorine (g mL-1)

No

rmal

ised

resp

onse10 min

EHPABA, 10 min

0%

20%

40%

60%

80%

100%

120%

0 0.5 1 1.5 2 2.5 3 3.5


No

rmal

ised

resp

ons

e

pH 7.2

pH 6.2

pH 8.2

ES, 30 min

0%

20%

40%

60%

80%

100%

120%

0 0.5 1 1.5 2 2.5 3 3.5


Nor

mal

ised

resp

onse

pH 7.2

pH 6.2

pH 8.2

Fig. 2. Effect of free chlorine concentration and water pH on the stability of BP-3,

EHPABA and ES.


236

3.3. Reaction kinetics and half-lives

Time course of BP-3 and EHPABA concentrations were followed using

250 mL aliquots of ultrapure and chlorinated tap water. Samples were buffered

at pH 7.2 (unless otherwise stated) and spiked with one of the above

compounds at the 10 ng mL-1 level (from 36 to 44 nM). The percentage of

methanol in the samples was maintained under 0.05% to avoid any possible

effect on the reaction rates. In addition to above compounds, a series of

experiments was carried out with 2,4-DHBP, which is the major metabolite of

BP-3. Initial concentrations of free chlorine added to ultrapure water varied

from 0.3 to 1 g mL-1 (4.4 to 14.5 M); moreover, in some assays, bromide (as

potassium salt) was also added to the samples to investigate its effect on the

stability of parent species. Tap water was collected when needed and the

content of free chlorine measured as indicated in the experimental section. After

a given time, the excess of chlorine was removed and the remaining amount of

each compound determined. At least 5 data were obtained between zero (the

quenching reagent was added to chlorinated water samples previously to the

UV filter) and 2-3 times the half-life of the considered compound. In excess of

chlorine, the removal of BP-3, EHPABA and 2,4-DHBP followed pseudo-first-

order kinetics, Fig. 3. For the latter compound, the reaction was so fast that it

could be studied only in samples with a relatively low concentration of

chlorine: 0.1 g mL-1, Fig. 3. Under these conditions, a half-life of 0.4 min was

measured. Taking into account that 2,4-DHBP only differs from BP-3 in the

substitution of the methoxy moiety by a hydroxyl one (Fig. 1), the effect of the

latter group on the reactivity of the compound is evident. This finding is in

agreement with differences between the stabilities of phenol and 1,3-

dihydroxibenzene in chlorinated water [25].


237

BP-3Initial chlorine 0.3 g mL-1

y = -0.302x + 11.504

R2 = 0.997

8

9

10

11

12

0 1 2 3 4 5 6 7

Reaction time (min)

Ln (P

eak

area

)

EHPABAInitial chlorine 0.6 g mL-1

y = -0.026x + 12.474

R2 = 0.995

10

10.5

11

11.5

12

12.5

13

0 5 10 15 20 25 30 35 40 45

Reaction time (min)

Ln (P

eak

area

)

2,4-DHBPInitial chlorine 0.1 g mL-1

y = -1.742x + 13.251

R2 = 0.991

23456789

1011121314

0 0.5 1 1.5 2 2.5 3 3.5

Ln (P

eak

area

)

Fig. 3. Natural logarithm plots of analytes responses versus reaction time. Data for

ultrapure water buffered at pH 7.2.


238

Half-lives measured for BP-3 and EHPABA, under different

experimental conditions, are given in Table 2. In agreement with the

information depicted in Fig. 2, BP-3 was less stable at pH 8.2 than at pH 7.2.

Addition of bromide to ultrapure water, even at the low ng mL-1 level, reduced

the stability of both UV filters, with the most significant influence on EHPABA.

This effect can be explained due to the formation of bromine, which shows a

strong tendency to react with aromatic compounds. It is also remarkable that

the half-life of BP-3 in chlorinated tap water was similar to that measured in

ultrapure water, spiked with a similar level of free chlorine. On the other hand,

significant differences were noticed for EHPABA.

Table 2. Half-lives (t1/2) measured for BP-3 and EHPABA under different

experimental conditions.

Analyte Matrix pH Free chlorine

(g mL-1)

Bromide

(ng mL-1)

t1/2 (min)

BP-3

Ultrapure

water

7.2 0.30a 0 2.7

7.2 0.60a 0 1.2

8.2 0.30a 0 1.8

8.2 0.60a 0 0.8

7.2 0.30a 1 2.8

7.2 0.30a 10 0.8

Tap water 7.2 0.76b Not determined 1.0

EHPABA

Ultrapure

water

7.2 0.6a 0 26.7

7.2 1.0a 0 18.7

7.2 0.60a 1 21.3

7.2 0.60a 10 4.6

Tap water 7.2 0.63b Not determined 12.1

a Added concentration

b Measured concentration


239

3.4. Halogenated by-products

Retention times and most intense ions in the MS spectra of major BP-3

and EHPABA by-products are shown in Table 3. They were the result of

hydrogen replacement per chlorine and/or bromine in the aromatic rings of

both UV filters. Positions where those replacements occurred were not

confirmed experimentally; however, considering the structures of parent

species (Fig. 1) and the activation effects of hydroxyl and amino groups towards

electrophilic substitution reactions, the most probable ones are the carbons in

ortho- to the amino moiety (EHPABA), and those in ortho- and para- to the

hydroxyl group (BP-3). For EHPABA only mono-halogenated species were

detected, whilst for BP-3 mono and di-substituted by-products were identified.

Although EHPABA and its chlorinated by-product could not be separated with

the HP-5ms column, using a HP-35 one, it was confirmed that signals in their

mass spectra appeared at different m/z ratios (Fig. 4); therefore, their relative

amounts can be determined even when they co-eluted in the same peak.

Table 3. Retention times (HP-5ms column), identities and most intense ions, with their

relative abundances, for the halogenated substitution by-products of BP-3 and

EHPABA.

UV filter By-product Retention time (min) m/z ions

(relative abundances)

BP-3

Cl-BP-3 (1) 18.55 319 (100), 321 (33), 276 (12)

Cl-BP-3 (2) 18.71 319 (100), 321 (33), 276 (12)

DCl-BP-3 19.03 353 (100), 355 (66), 310 (17)

Br-BP-3 (1) 19.25 363 (100), 365 (100), 320 (11), 323 (4)

Br-BP-3 (2) 19.41 363 (100), 365 (100), 320 (11), 323 (4)

DBr-BP-3 20.70 443 (100), 441 (50), 445 (50), 428 (13)

Br-Cl-BP-3 (1) 19.75 399 (100), 397 (82), 401 (53), 384 (20)

Br-Cl-BP-3 (2) 19.80 399 (100), 397 (82), 401 (53), 384 (20)

EHPABA Cl-EHPABA 16.91 198 (100), 311 (75), 313 (25)

Br-EHPABA 17.50 355 (100), 357 (100), 243 (94)

(1) and (2) refer to isomeric by-products


240

Ultrapure water

Ultrapure water plus 0.6 g mL-1

of chlorine, 10 min

198

150 200 250 300 350m/z

0%

25%

50%

75%

100%

118

154

182

311

Cl-EHPABA

150 200 250 300 m/z

0%

25%

50%

75%

100%

120

148

165

277

EHPABA

21.25 21.50 21.75 22.00 22.25 22.50 22.75 minutes

0

100

200

300

400

(x103 Counts)

Fig. 4. Total ionic current (TIC) chromatograms (HP-35 column) and MS spectra for

EHPABA and its chlorinated by-product.

Table 4 summarizes which substitution by-products were observed

under different experimental conditions. Formation of brominated species

when BP-3 and EHPABA standards were mixed with chlorinated tap water

confirmed the presence of bromide in this matrix and explained the differences

between the half-lives of EHPABA in ultrapure (26.7 min) and tap water (12.1

min) containing a similar concentration of free chlorine, see Table 2. Last

column on Table 4 presents the most concerning results: with the only

exception of DBr-BP-3, the rest of EHPABA and BP-3 substitution by-products,

identified in model experiments, were also noticed after mixing tap water with

two personal care products including the parent UV filters in their formulation.


241

Table 4. Summary of BP-3 and EHPABA substitution by-products detected under

different experimental conditions.

Ultrapure

water plus

UV filter

standard

Ultrapure

water plus

UV filter

standard

Ultrapure

water plus

UV filter

standard

Tap water

plus

UV filter

standard

Tap water

plus

suncare

product

BP-3

Free

chlorine

(g mL-1)

0.3a 0.3a 0.3a 0.76b 0.55b

Bromide

(ng mL-1) 0 1a 10a

not

determined

not

determined

Cl-BP-3 (1) X X X X X

Cl-BP-3 (2) X X X X X

DCl-BP-3 X X X X X

Br-BP-3 (1) -- X X X X

Br-BP-3 (2) -- X X X X

DBr-BP-3 -- X X -- --

Br-Cl-BP-3

(1) -- X X X X

Br-Cl-BP-3

(2) -- X X X X

EHPABA

Free

chlorine

(g mL-1) 0.6a 0.6a 0.6a 0.63b 0.89b

Bromide

(ng mL-1) 0 1a 10a

not

determined

not

determined

Cl-

EHPABA X X X X X

Br-

EHPABA -- X X X X

aadded concentration

bmeasured concentration

Additionally to the above mentioned species, in the case of BP-3 another

group of by-products was detected. They were tentatively identified as


242

halogenated forms of 3-methoxyphenol, which could be generated from

cleavage of the carbonyl bond between the two aromatic rings in the molecule

of BP-3 (Fig. 1), followed by halogenation of the methoxyphenol fragment.

Moreover, mono- and di-halogenated substitution by-products of BP-3 might

also break down rendering different halogenated methoxyphenols. Fig. 5 shows

the MS spectra, as tert-butyldimethylsilyl derivatives, for two of these cleavage

by-products. Retention times, most intense ions, and the number of chlorine

and/or bromine atoms for all the BP-3 cleavage by-products are shown in Table

5. As observed, only di- and tri-halogenated species were detected.

Chromatographic responses of the first represented less than 2% of the BP-3

peak area in the reference experiment, performed in absence of chlorine;

whereas, for tri-halogenated phenols, they reached up to 10%. Moreover, the

trichloro-methoxyphenol was the only cleavage by-product detected when

mixing tap water with a BP-3 containing suncare lotion.

150 200 250 300 350 400m/z

0%

25%

50%

75%

100%

213

294

373

A B

150 200 250 300 350m/z

0%

25%

50%

75%

100%

248

268

285

m/z 283+285

MeO

O

Cl

Cl

Cl

Si

m/z 248+250

MeO

OCl

Si

m/z 373+375

Br

Br

m/z 292+294

Fig. 5. MS spectra and tentative structures for two cleavage halogenated by-products of

BP-3.


243

Table 5. Halogenated cleavage by-products of BP-3.

By-product structure Atoms of

chlorine

Atoms of

bromine

Retention time

(min)

Most intense ions

(m/z)

Methoxy-phenol

2 0 13.63 249, 251

3 0 14.47 283, 285

0 2 15.20 337, 339, 341

0 3 16.79 417, 419

1 2 16.04 373, 375

3.5. Stability of the halogenated by-products

Stability of major EHPABA and BP-3 by-products was followed

considering reaction times up to 180 and 120 min, respectively and chlorine

concentrations up to 1 g mL-1. Some of the obtained data are plotted in Fig. 6.

Normalised responses represented in the Y-axis correspond to the ratio between

the peak area of each by-product, at a given reaction time, and that measured

for the parent UV filter in the reference experiment at zero time; therefore, they

serve as a rough estimation for the yield of depicted transformations. EHPABA

by-products showed an excellent stability under explored conditions, Fig. 6A.

The presence of just 10 ng mL-1 of bromide, led to a significant diminution in

the production of Cl-EHPABA and shifted the reaction towards the brominated

specie. Di-halogenated forms of EHPABA were never detected, thus, insertion

of one atom of chlorine or bromide in the structure of the parent specie seems to

deactivate it to further electrophilic substitution reactions. This finding differs

from the data published by Sakkas et al. for chlorinated swimming-pool water

[20]. The use of longer reaction times added to the effect of solar radiation

(exposure time 60 hours) might be responsible for the formation of di-

halogenated EHPABA by-products reported in the above work [20].


244

0

10

20

30

40

50

60

0 50 100 150 200

Reaction time (min)

No

rmal

ised

res

po

nse

(%

)Cl-EHPABA, 0 ng mL-1 Br-

Cl-EHPABA, 10 ng mL-1 Br-

Br-EHPABA, 10 ng mL-1 Br-

0

10

20

30

40

50

0 20 40 60 80 100 120

Reaction time (min)

No

rmal

ized

res

po

nse

(%

)

DCl-BP-3Cl-BP-3 Br-Cl-BP-3 DBr-BP-3 x 5Br-BP-3

A

B

C

0

1

2

3

4

5

6

7

8

9

0 20 40 60 80 100 120 140

Reaction time (min)

No

rmal

ised

res

po

nse

(%

)

Ultrapure water (1 g mL-1 chlorine) plus BP-3 standard

Tap water (0.55 g mL-1 chlorine) plus sun care product

Fig. 6. Time course of some UV filters halogenated by-products. A, EHPABA by-

products for 0.6 g mL-1 of free chlorine and different levels of bromide. B, Mono- and

di-halogenated BP-3 species for 0.3 g mL-1 of chlorine and 1 ng mL-1 of bromide. C,

Trichloro-3-methoxyphenol in chlorinated ultrapure and tap water.


245

Fig. 6B presents the time course of BP-3 substitution by-products for 0.3

g mL-1 of chlorine and 1 ng mL-1 of bromide. The maximum responses for

mono-halogenated species were observed for short reaction times, whereas the

di-halogenated ones reached a plateau after 20 min and then remained stable.

Additional experiments, using higher concentrations of free chlorine, showed

that the latter are also slowly degraded. Particularly, for 1 g mL-1 of free

chlorine in absence of bromide, the normalised response of DCl-BP-3 reached a

maximum (around 50%) after 20 min and then decreased to 20% in 2 hours,

figure not shown.

Trichloro-methoxyphenol, the most abundant of the BP-3 cleavage by-

products, showed a considerable stability in ultrapure water containing 1 g

mL-1 of free chlorine and also when a suncare lotion was mixed with tap water

(0.55 g mL-1 of free chlorine). In both cases, its normalised response rose until

40-60 min and then decreased very slowly, Fig. 6C.

On the basis of the identified by-products and the temporal stabilities of

the most abundant ones, a tentative reaction pathway for BP-3 has been

proposed, Fig. 7. On one hand, the parent species is converted into mono and

di-halogenated substitution by-products. Moreover, BP-3 and/or its

substitution by-products may undergo a cleavage process rendering

methoxyphenols, which can be further halogenated by the excess chlorine

and/or bromine.


246

MeO

C

OH

X

MeO

C

OOH

X

X

BP-3

O

MeO

C

OH O

MeO

C

OH

X

O

Substitution by-products

X= Cl and/or Br

MeO

OH

X

X

X

Cleavage by-products

MeO

OH

X

X

Fig. 7. Proposed degradation pathway for BP-3.

4. Conclusions

Conversely to ES, EHPABA and BP-3 exhibited a limited stability at

neutral pHs in water samples containing low levels of free chlorine. Differences

among reactivities of ES, BP-3 and 2,4-DHBP highlight the effect of different

organic groups on the activation or deactivation of the phenolic ring towards

electrophilic substitution reactions. Whilst EHPABA rendered only mono-

halogenated by-products, with the same structure as the parent UV filter, BP-3

followed a more complex reaction pathway leading to mono- and di-

halogenated substitution compounds and cleavage halogenated

methoxyphenols. Traces of bromide, at similar levels to those existing in

aquifers and sources of tap water from coastal areas, shortened considerably the

half-life of EHPABA; moreover, they shifted the degradation of BP-3 and

EHPABA towards the formation of brominated by-products, which, in general,

are considered more concerning than the chlorinated ones.


247

By-products of EHPABA, di-halogenated BP-3 forms and the tri-

halogenated methoxyphenols were produced in a significant extension, showed

a considerable stability to further oxidation reactions and their formation was

even observed under quasi real-life conditions; therefore, it is expected that

they could be generated in chlorinated bath waters (e.g. swimming-pools) and

during showering, after dermal application of suncare products. Further studies

should investigate the presence of these by-products in waste and swimming-

pool water, address their potential toxicity due to dermal exposition and

investigate possible environmental effects.

Acknowledgements



PGIDIT06PXIB237039PR). N. N and P. C. thank the Spanish Ministry of

Education and Science for their FPU grants.


248

References

1. A. Salvador, A. Chisvert, Anal. Chim. Acta 537 (2005) 1.

2. S.D. Richardson, Anal. Chem. 79 (2007) 4295.

3. D.L. Giokas, V.A. Sakkas, T.A. Albanis, D.A. Lampropoulou, J. Chromatogr. A 1077

(2005) 19.

4. A. Salvador, I. Gadea, A. Chisvert, M.C. Pascual-Martí, Chromatographia 54 (2001) 795.

5. W. Li, Y. Ma, C. Guo, W. Hu, K. Liu, Y. Wang, T. Zhu, Water Res. 41 (2007) 3506.

6. D.L. Giokas, A. Salvador, A. Chisvert, Trends Anal. Chem. 26 (2007) 360.

7. M. Schlumpf, B. Cotton, M. Conscience, V. Haller, B. Steinmann, W. Lichtensteiger,


8. M. Heneweer, M. Muusse, M. Van den Berg, J. T. Sanderson, Toxicol. Appl. Pharmacol.

208 (2005) 170.

9. K. Maerkel, W. Lichtensteiger, S. Durrer, M. Conscience, M. Schlumpf, Environ. Toxicol.

Pharmacol. 19 (2005) 761.

10. T. Poiger, H.R. Buser, M.E. Balmer, P.A. Bergqvist, M.D. Müller, Chemosphere 55 (2004)

951.

11. H.K. Jeon, Y. Chung, J.C. Ryu, J. Chromatogr. A 1131 (2006) 192.

12. P. Cuderman, E. Heath, Anal. Bioanal. Chem. 387 (2007) 1343.

13. D.L. Giokas, V.A. Sakkas, T.A. Albanis, J. Chromatogr. A 1026 (2004) 289.

14. M.E. Balmer, H.R. Buser, M.D. Müller, T. Poiger, Environ. Sci. Technol. 39 (2005) 953.

15. D.A. Lambropoulu, D.L. Giokas, V.A. Sakkas, T.A. Albanis, M.I. Karayannis, J.

Chromatogr. A 967 (2002) 243.

16. T. Kupper, C. Plagellat, R.C. Brändli, L.F. de Alencastro, D. Grandjean, J. Tarradellas,

Water Res. 40 (2006) 2603.

17. L. Vidal, A. Chisvert, A. Canals, A. Salvador, J. Chromatogr. A (2007),

doi:10.1016/j.chroma.2007.07.077.

18. C. Plagellat, T. Kupper, R. Furrer, L. F. de Alencastro, D. Grandjean, J. Tarradellas,

Chemosphere 62 (2006) 915.

19. H.R. Buser, M.E. Balmer, P. Schmid, M. Kohler, Environ. Sci. Technol. 40 (2006) 1427.

20. V.A. Sakkas, D.L. Giokas, D.A. Lambropoulou, T.A. Albanis, J. Chromatogr. A 1016

(2003) 211.

21. D.L. Giokas, A.G. Vlessidis, Talanta 71 (2007) 288.

22. J.J. Inbaraj, P. Bilski, C.F. Chignell, Photochem. Photobiol. 75 (2002) 107.

23. M. Bedner, W.A. MacCrehan, Environ. Sci. Technol. 40 (2006) 516.

24. J.Y. Hu, T. Aizawa, S. Ookubo, Environ. Sci. Technol. 36 (2002) 1980.

25. H. Gallard, U. von Gunten, Environ. Sci. Technol. 36 (2002) 884.

26. G.R. Boyd, S. Zhang, D.A. Grimm, Water Res. 39 (2005) 668.


249

27. P. Canosa, I. Rodríguez, E. Rubí, N. Negreira, R. Cela, Anal. Chim. Acta 575 (2006) 106.

28. K.L. Rule, V.R. Ebbett, P.J. Vikesland, Environ. Sci. Technol. 39 (2005) 3176.

29. P. Canosa, S. Morales, I. Rodríguez, E. Rubí, R. Cela, M. Gómez, Anal. Bioanal. Chem.

383 (2005) 1119.

30. L.S. Clesceri, A.E. Greenberg, A.D. Eaton (Eds.), Standard Methods for the Examination

of Water and Wastewater, American Water Works Association, Baltimore, MD, 20th ed.,

1998, pp. 461-465.

Metodología desarrollada-Muestras sólidas

253

2. MUESTRAS SÓLIDAS

2.1. Introducción

Algunos filtros solares presentan carácter fuertemente lipofílico y

elevada estabilidad en condiciones ambientales. Por tanto, son propensos a

acumularse en matrices sólidas ambientales, tales como lodos y sedimentos.

Desde un punto de vista analítico, la preparación de muestra para la

determinación de contaminantes emergentes en lodos es un proceso complejo,

que requiere múltiples etapas de clean-up y un elevado consumo de disolventes

orgánicos. Además, a diferencia de lo que ocurre con las matrices acuosas, el

número de aplicaciones desarrolladas es limitado, así como el número de

analitos incluidos en las mismas. En esta Tesis, se propone un protocolo semi-

automatizado combinando PLE con una primera etapa de limpieza on-line

usando carbón grafitizado para la eliminación de pigmentos, seguido de una

etapa adicional de purificación en modo off-line, utilizando PSA, para la

determinación de 8 filtros solares en lodos de depuradoras mediante GC-MS. El

método desarrollado ha sido validado y aplicado al análisis de lodos de

diferentes orígenes, confirmando la acumulación de determinados analitos en

esta matriz [Negreira, in press].

Además de su utilización en protectores solares, los filtros UV se

emplean también en otros productos de cuidado personal y materiales

ampliamente usados en el hogar, tales como pinturas y barnices, plásticos,

tapicerías y alfombras. Estos usos permiten inferir su presencia en atmósferas

interiores, lo que provocaría una exposición continua e inadvertida a estos

compuestos a través de la inhalación e incluso, la ingestión. Dada su limitada

volatilidad, parece lógico que los filtros UV se encuentren asociados a polvo y

material particulado, aunque su existencia en estas matrices no había sido

descrita previamente en la bibliografía. En esta Tesis, se desarrolló un método


254

rápido, sencillo y de bajo coste para investigar la presencia de filtros solares en

muestras de polvo procedentes de viviendas particulares, edificios públicos y

vehículos. Las etapas de extracción y purificación se integraron en un solo paso

empleando MSPD como técnica de preparación de muestra y GC-MS/MS en la

etapa de determinación [Negreira, 2009-C]. La aplicación del método

desarrollado puso de manifiesto la existencia de niveles significativos de

algunos filtros solares en atmósferas interiores, asociados a partículas de polvo.


255


sólidas

Extracción

Mezclar con 2 g tierra de diatormeas

Muestra lodo 0,5 g

Relleno de la celda de extracción:

PLE:Hexano: Diclorometano (80:20), 1 ciclo, 5 min, 75 ºC, 1500 psi,

100% flush, 2 min purga

Concentración a 1 mL

Limpieza

Clean-up en cartucho de PSA-sílica (0,5 g)

Fracción 2ª :5 mL de hexano: éter (1:1)

Fracción 1ª : 1 mL hexano

Añadir 2 mL isooctano

Inyección 2 µLGC-MS (modo SIM)

Concentrar a 1 mL

1ºFiltros (1 de celulosa + 1 de fibra de vidrio)2ºTierra de diatomeas (1,5 g)3ºCarbón (0,5 g)4ºTierra de diatomeas (0,5 g)5ºMezcla de lodo + tierra de diatomeas6ºFiltro de celulosa

Figura 21: Esquema seguido para la determinación de filtros solares en lodo mediante

PLE y GC-MS.


256

Frita de polietileno

Dispersión con1,25 g C18

Evaporación a 1 mL

Polvo0,5 g

Secar con 0,5 g sulfato sódico anhidro

2 g de Sílica (Co-adsorbente)

Transferir la mezcla al cartucho de MSPD

4 mL acetonitrilo

..Frita de polietileno

Elución

Inyección 2 µL GC-MS/MS

Figura 22: Esquema empleado para la determinación de filtros solares en polvo mediante

MSPD y GC-MS/MS.

2.3. Publicación:

OPTIMIZATION OF

PRESSURIZED LIQUID EXTRACTION

AND PURIFICATION CONDITIONS FOR

GAS CHROMATOGRAPHY-MASS SPECTROMETRY

DETERMINATION OF UV FILTERS IN SLUDGE


Journal of Chromatography A, in press

(doi:10.1016/j.chroma.2010.11.028)

Journal of Chromatography A, in press (doi:10.1016/j.chroma.2010.11.028)

259

Optimization of pressurized liquid extraction and purification conditions for

gas chromatography-mass spectrometry determination of UV filters in sludge



Investigación y Análisis Alimentario (IIAA), Universidad de Santiago de


Abstract

This work presents an effective sample preparation method for the

determination of eight UV filter compounds, belonging to different chemical

classes, in freeze-dried sludge samples. Pressurized liquid extraction (PLE) and

gas chromatography-mass spectrometry (GC-MS) were selected as extraction

and determination techniques, respectively. Normal-phase, reversed-phase and

anionic exchange materials were tested as clean-up sorbents to reduce the

complexity of raw PLE extracts. Under final working conditions, graphitized

carbon (0.5 g) was used as in-cell purification sorbent for the retention of co-

extracted pigments. Thereafter, a solid-phase extraction cartridge, containing

0.5 g of primary secondary amine (PSA) bonded silica, was employed for off-

line removal of interferences overlapping the chromatographic peaks of some

UV filters. Extractions were performed with a n-hexane:dichloromethane (80:20,

v:v) solution at 75 ºC, using a single extraction cycle of 5 min at 1500 psi. Flush

volume and purge time were set at 100% and 2 min, respectively. Considering

0.5 g of sample and 1 mL as the final volume of the purified extract, the

developed method provided recoveries between 73% and 112%, with limits of

quantification (LOQs) from 17 to 61 ng g-1. Total solvent consumption remained

around 30 mL per sample. The analysis of non-spiked samples confirmed the

sorption of significant amounts of several UV filters in sludge with average

concentrations above 0.6 g g-1 for 3-(4-methylbenzylidene) camphor (4-MBC),

2-ethylhexyl-p-methoxycinnamate (EHMC) and octocrylene (OC).


260

Keywords: UV filters; sludge; purification; pressurized liquid extraction;

GC-MS.

1. Introduction

Organic UV filters are compounds designed to absorb the ultraviolet

wavelengths of solar radiation preventing photo-aging and other harmful

effects in human health. The concentration of UV filters in sunscreen lotions

may represent up to 10% of the product weight; moreover, they are included, at

lower levels, in the formulation of many other personal care products [1-2]. The

above uses contribute to the direct input of UV filters in bathing waters and

their indirect release in the aquatic environment through domestic sewage

water [3-7]. The activity of some UV filters as endocrine disrupters [8-10],

added to their ubiquity in sewage and surface water, has awaken the concern

about their potential medium-term environmental effects.

Gas and liquid chromatography-mass spectrometry techniques,

combined with effective sample concentration approaches [5,11-13], have been

applied to obtain an overview of UV filters occurrence in different water

samples, including wastewater from sewage treatment plants (STPs). However,

understanding the behaviour of UV filters in STPs requires not only measuring

their concentrations in the water phase, but also determining the fraction which

remains attached to sludge particles [7]. This latter information is necessary to

distinguish between biodegradation and sorption processes, and to assess the

risk of introducing the UV filters in the terrestrial environment through the

application of sludge as fertilizer in agriculture.

From the analytical point of view, sludge is an extremely complex matrix

which requires well-tuned sample preparation approaches providing a balance

among efficiency, selectivity, extraction time and cost. These constraints explain

the limited number of studies dealing with the analysis of UV filters in sludge


261

versus the plethora of publications focussed on water samples. The first method

for sludge was proposed by Plagellat and co-workers [14]. It involved three

consecutive liquid-liquid extractions of fresh sludge samples (60 g), followed by

dryness evaporation of the combined extract and column purification with

activated silica. The optimized protocol provided an excellent performance for

selected analytes; however, it was time and solvent consuming (more than 200

mL of organic solvent per sample), as well as difficult to automate.

Pressurized liquid extraction (PLE) is a popular sample preparation

technique for solid matrices showing limited solvent consumption, excellent

extraction yields and possibility to integrate extraction and purification steps.

The applications of PLE to the extraction of personal care compounds from

sludge have been compiled in a recent review [15]. PLE, combined with in-cell

clean-up using activated silica, has been reported as a straight forward

alternative for gas chromatography-mass spectrometry (GC-MS) determination

of UV filters in low carbon content sediment samples [16]; however, the above

strategy provided too complex extracts in the case of sludge [17]. Although, the

selectivity of the extraction could be improved by enclosing the sludge sample

in a non-porous polyethylene membrane bag, within the cell, the efficiency of

the extraction underwent a dramatic reduction, with recoveries around or

below 50% for most UV filters [17]. In addition to the above procedures,

specifically designed for UV filters, Nieto and co-workers [18] have developed a

PLE method for the extraction of several personal care products, including three

UV filters (benzophenone-3, BP-3; octocrylene, OC; and 2-ethylhexyl-p-

dimethylaminobenzoate, EHPABA), from sludge samples. Analytes were

recovered with methanol followed by methanol:water mixtures and

simultaneously purified in a layer of alumina packed inside the extraction cell.

Considering a sample intake of 1 g and 25 mL as the volume of the final extract,

recoveries over 79% and low signal suppression effects (below 15%) were

observed in the further LC-(ESI)-MS/MS determination. Unfortunately, the

performance of this method has not been assessed for other UV filters of


262

environmental relevance, such as salicylates, methoxycinnamates and

camphors; moreover, the employed aqueous extraction mixture is not

compatible with GC-based determinations.

In this study, we optimize an alternative sample preparation method for

the determination of eight UV filters, belonging to different chemical classes, in

freeze-dried sludge samples. PLE was selected as extraction technique due to its

high automation capabilities. Purification conditions were optimized in order

(1) to reduce the content of interferences in the final extract and (2) to maintain

the consumption of organic solvents and the complexity of the method at

acceptable levels. GC-MS was considered as determination technique on the

basis of (1) its worldwide availability in environmental monitoring laboratories

and (2) the poor detection limits reported for salicylate type UV filters using

LC-(ESI)-MS systems [19]. Finally, the applicability of the method was

demonstrated with sludge samples from urban STPs.

2. Experimental

2.1. Solvents, standards and sorbents

N-hexane, isooctane, acetone, dichloromethane and ethyl ether (trace

analysis grade) and HPLC-grade methanol were supplied by Merck

(Darmstadt, Germany). The list of UV filters included in this study is compiled

on Table 1. Standards of target analytes were acquired from Aldrich

(Milwaukee, WI, USA) and Merck, except isoamyl-p-methoxycinnamate

(IAMC), which was kindly provided by Dr. R. Rodil (University of Santiago de

Compostela, Spain). Individual solutions of each species (ca. 1000 g mL-1) were

prepared in methanol. Further dilutions and mixtures of them were dissolved

in acetone (when used to prepare the spiked sludge samples employed during

optimization and validation of sample preparation conditions) and in isooctane

(case of calibration standards).


263

Table 1. Abbreviations, retention times, selected ions and instrumental limits of

quantification (LOQs) of the GC-MS system of target analytes.

Analyte Abbreviation Retention

time (min)

Quantification

ion (m/z)

Qualification

ion (m/z)

LOQs

(ng mL-1)

(S/N 10)

2-Ethylhexyl salicylate EHS 9.65 120 138 2

Homosalate HMS 10.26,10.40 120 138 4

Isoamyl-p-

methoxycinnamate IAMC 11.63a 178 161 2

2-Hydroxy-4-

methoxybenzophenone BP-3 11.63 227 151 6

3-(4-

Methylbenzylidene)

camphor

4-MBC 11.87a 254 239 3

2-Ethylhexyl-p-

dimethylaminobenzoate EHPABA 13.41 277 165 1

2-Ethylhexyl-p-

methoxycinnamate EHMC 13.73a 178 161 1

Octocrylene OC 16.15 360 232, 249 2 aRetention time values for the E isomers

Alumina, Florisil and silica solid-phase extraction (SPE) cartridges (0.5 g)

were acquired from Waters (Milford, MA, USA). Cartridges containing 0.5 g of

silica bonded to ethylenediamine-N-propyl groups (PSA sorbent) and 0.25 g of

graphitized carbon were purchased from Supelco (Bellefonte, PA, USA). Both

sorbents, in the bulk format, were also obtained from Supelco. Diatomaceous

earth was provided by Aldrich.

2.2. Samples

Optimization of sample preparation (extraction and purification)

conditions was performed with a freeze-dried pooled matrix of primary and

biological sludge, fortified with 5 g g-1 of each UV filter. The total carbon (TC)

content of the pooled matrix was 33%. The spiking procedure consisted of the

addition of a measured volume of a standard in acetone to an accurately


264

weighed fraction of sludge. Approximately, 1 mL of standard was used per g of

freeze-dried sludge. The resulting slurry was protected from light,

homogenized periodically and kept in a hood until complete elimination of the

acetone. The recoveries of the method were evaluated with individual samples

of primary and biological sludge fortified at different concentrations. All spiked

samples were aged for a minimum of two weeks before extraction.


Extractions were performed with a pressurized liquid extractor, ASE 200

Dionex (Sunnyvale, CA, USA), furnished with 11 mL stainless-steel cells. A

cellulose filter, followed by a glass fibre one, was placed on the bottom of each

cell. Under final working conditions, cells were filled (bottom to top) with 1 g of

diatomaceous earth, 0.5 g of graphitized carbon, 0.5 g of diatomaceous earth

and 0.5 g of sludge, previously homogenized with 2 g of diatomaceous earth.

Analytes were extracted with n-hexane:dichloromethane (80:20), at 75 ºC,

considering a single static extraction cycle of 5 min with the cell pressurized at

1500 psi. The flush volume was 100% and the purge time 2 min.

PLE extracts were evaporated, ca. 1 mL, and additionally purified with a

PSA cartridge (0.5 g) previously conditioned with n-hexane:ether (1:1) and n-

hexane (5 mL each). After loading the concentrated extract, the sorbent was

rinsed with n-hexane (1 mL). Analytes were further recovered with 5 mL of n-

hexane:ether (1:1). Thereafter, 1 mL of isooctane was added as a keeper to the

purified extract, which was evaporated and adjusted to a final volume of 1 mL

with the same solvent.

2.3. GC-MS equipment

UV filters were determined with a GC-MS system consisting of an

Agilent (Wilmington, DE, USA) 7890A gas chromatograph connected to a

quadrupole type mass spectrometer (Agilent MS 5975C), furnished with an

electron-impact (EI) ionization source. Separations were carried out in a HP-

5ms type capillary column (30 m x 0.25 mm i.d., df: 0.25 m) supplied by


265

Agilent. Helium (99.999 %) was used as carrier gas at a constant flow of 1.2 mL

min-1. The GC oven was programmed as follows: 110 ºC (held for 1 min),

increased at 12 ºC min-1 to 280 ºC (held for 10 min). Ionization source, mass

analyzer and transfer line temperatures were set at 230, 150 and 290 ºC,

respectively. Standards and sample extracts were injected in the splitless mode,

maintaining the injection port at 280 ºC. The splitless time and the split flow

were set at 1 min and 20 mL min-1, respectively. The mass spectrometer was

operated in the SCAN mode (m/z range from 45 to 400) to assess the efficiency

of the purification process, and in the SIM mode for quantification purposes.

Retention times and ions monitored for each compound are summarized in

Table 1. Analytes were grouped in three chromatographic segments. The dwell

time per ion was 100 ms in the first and third segment and 50 ms for the second

one.

2.4. Recoveries and procedural blanks

Signals (peak areas) measured for sample extracts were compared with

those obtained for calibration standards in isooctane, covering the range of

concentrations between 5 and 5000 ng mL-1. Within this range, the GC-MS

system provided linear response plots with determination coefficients (R2)

higher than 0.996 for all compounds. The instrumental limits of quantification

(LOQs), defined as the concentration of each compound producing a response

ten times higher than the baseline noise in the SIM acquisition mode, ranged

from 1 to 6 ng mL-1, Table 1. Recoveries were calculated as the difference

between concentrations obtained for spiked and non-spiked fractions of the

same sludge sample divided by the added amount and multiplied by 100.

Procedural blanks represent the whole sample preparation process (extraction

plus purification) performed without sludge.


266


3.1. Preliminary experiments

Previous applications of PLE to the extraction of UV filters from sludge

employed rather different conditions as regards the extraction solvent and the

temperature of the cell [17-18]. Likely, the selected in-cell clean up strategies,

based on the use of permeable non-porous membranes [17] or a normal-phase

sorbent [18], conditioned the optimum extraction parameters.

In this study, in order to prevent the influence of clean-up conditions on

the yield of PLE, a first series of extractions was carried out considering just

diatomaceous earth as inert dispersant of sludge (0.5 g of sludge plus 2 g of

diatomaceous earth) and filling material in the extraction cell. Flush volume,

pressure and extraction time were set at 100%, 1500 psi and 5 min, respectively.

Samples were first extracted with n-hexane (50 ºC, 1 cycle) followed by

dichloromethane (60 ºC, 3 cycles). Extracts were collected in separated vessels,

adjusted to 25 mL, filtered (0.45 m) and injected in the GC-MS system. N-

hexane, at low temperature, has been proposed for the removal of low polar

interferences, previously to analytes extraction from complex samples [20-21].

On the other hand, Chu and co-workers [22] described the use of

dichloromethane, under above instrumental conditions, for the PLE extraction

of medium-polar personal care compounds from sludge.

Analysis of n-hexane and dichloromethane extracts revealed that BP-3

and OC were distributed between both fractions, whereas 95% of the responses

measured for the rest of UV filters corresponded to the n-hexane fraction, data

not shown. Despite the high dilution of sample extracts (25 mL), their GC-MS

chromatograms showed a considerable complexity. Visually, the n-hexane

extract was colourless whereas the dichloromethane fraction presented a

yellowish appearance and turbidity. This preliminary data indicate the

suitability of n-hexane:dichloromethane mixtures for the extraction of UV filters

from sludge at relatively low temperatures. However, there is no possibility to

improve the selectivity of the process using n-hexane as pre-extraction solvent.


267

3.2. Clean-up conditions

Several SPE sorbents were tested in order to reduce the complexity of

PLE extracts from sludge samples. In all cases, extractions were carried out at 70

ºC using a n-hexane:dichlorometane (8:2) mixture with the cells pressurized at

1500 psi. Two static cycles of 2 min each and a flush volume of 100% were

employed. Extracts were concentrated to 1 mL and loaded on top of the

considered SPE cartridge, previously conditioned as described in the

experimental section. Thereafter, 1 mL of n-hexane was passed through the

sorbent and discarded. Subsequently, analytes were eluted using 5 mL of a n-

hexane: ether (1:1, v:v) mixture. The extract was mixed with 1 mL of isooctane

and evaporated to a final volume of 1 mL. No differences were noticed between

the turbidity and the colour of raw PLE extracts versus those purified with

alumina and silica cartridges. Florisil and PSA cartridges rendered transparent,

although yellowish, extracts and graphitized carbon transparent, colourless

ones. The efficiency of the above clean-up sorbents was evaluated operating the

GC-MS system in the SCAN mode. PSA was the only sorbent able to remove

two broad chromatographic bands (tentatively identified as fatty acids)

overlapping the peaks of HMS, BP-3, IAMC and 4-MBC, and to reduce

significantly the baseline level of the GC-MS chromatograms, Fig. 1. Likely,

fatty acids and other interferences with anionic moieties remain strongly

retained in the PSA cartridge, as it has been early described for the purification

of extracts from vegetal samples [23]. Except graphitized carbon, the rest of

sorbents failed to remove pigments contained in the raw extract. Although

pigments exerted a little effect in the complexity of the GC-MS chromatograms

(Fig. 1), they might impair the efficiency of the GC column due to irreversible

contamination of the stationary phase. Thus, PSA and graphitized carbon were

selected as clean-up sorbents.


268

8.5 9.5 10.5 11.5 12.5 13.5 14.5 15.5 16.5 17.5 18.5 19.50

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Time (min)

Abundance (x 107)

A

BCD

No clean-up PSA clean-up Carbon clean-up

Fig. 1. GC-MS chromatograms (SCAN mode) and pictures corresponding to PLE

extracts from sludge purified with different SPE cartridges: A, without clean-up; B,

Florisil; C, graphitized carbon; D, PSA. Extraction conditions: n-

hexane:dichloromethane (80:20), 70 ºC, 2 cycles of 2 min,100 % flush volume, 1500 psi.

In a second series of extractions, the feasibility of integrating extraction

and clean-up steps, placing a layer of the above sorbents (from 0.5 to 2 g) inside

the PLE cell, was investigated. Using the above described extraction

parameters, graphitized carbon (0.5 g) allowed an efficient removal of

pigments; however, the purification efficiency of PSA underwent a dramatic

reduction. Probably, the ability of this sorbent to retain fatty acids interferences

is reduced due to the temperature of the PLE cell (70 ºC versus room

temperature in the off-line modality), as well as the differences in the volume

and the composition of the organic mixture flowing through the layer of PSA,

packed inside the cell, versus those used in the SPE mode [24]. Graphitized

carbon (0.5 g) was introduced in the PLE cell for on-line removal of pigments;

thereafter, the extract was submitted to an additional off-line clean-up with a

SPE cartridge containing 0.5 g of PSA.


269

3.3. PLE parameters

3.3.1. Time, dichloromethane percentage and temperature

The influence of the above factors on the efficiency of the extraction step

was simultaneously investigated using a Box-Behnken experimental factorial

design with each variable considered at three levels, Table 2. The flush volume

was 100%, the pressure 1500 psi and two extraction cycles were applied. The

purified extracts were injected in the GC-MS system, operated in the SIM

acquisition mode, and peak areas were used as the response variable in order to

calculate the main effects associated with each experimental factor, their

quadratic terms and the two-factor interactions. Table 3 summarizes the

numerical values of the standardized main effects and their quadratic terms.

The absolute value of a main effect is proportional to the influence of the

associated factor on the efficiency of the PLE extraction. A positive sign

indicates an improvement in the yield of the process when the factor varies

from the low to the high level, within the domain of the design, and a negative

one the opposite trend.

Table 2. Experimental domain of the Box-Behnken design.

Factor Code Level

Low Medium High

Time (min)

CH2Cl2 (%)

Temperature (ºC)

A

B

C

2

5

40

6

22.5

65

10

40

90


270

Table 3. Standardized main effects and quadratic terms provided by the Box-Behnken

design.

Compound Main effects Quadratic terms

A B C AA BB CC

EHS

HMS

IAMC

BP-3

4-MBC

EHPABA

EHMC

OC

-0.42

-2.7a

-1.0

-0.37

-2.8a

-1.7

-2.2

-4.0a

-2.7a

-3.2a

-1.5

7.7a

-3.5a

-4.1a

-1.7

9.4a

0.03

-0.52

0.82

6.3a

-1.7

-0.60

0.96

-0.75

-2.0

-2.8a

-1.4

-0.96

-3.6a

-3.3a

-3.0a

-2.5

-3.7a

-3.6a

-3.3a

-7.6a

-4.3a

-2.5

-3.8a

-14.4a

-0.94

-1.0

-0.84

-1.1

-0.98

-0.78

-1.5

0.56

a Statistical significant factors and quadratic terms

Data summarized in Table 3 show that the percentage of

dichloromethane (code B) played a positive and statistical significant effect

(95% confidence level) in the extraction of BP-3 and OC, whereas the opposite

trend was observed for the rest of analytes. The temperature of the cell (code C)

affected positively to the extraction of BP-3 and the extraction time (code A)

showed a negative influence on the yield of the extraction, being statistically

significant for three (HMS, 4-MBC and OC) of the investigated species.

Quadratic terms associated with the extraction time (AA) and the percentage of

dichloromethane (BB) also presented statistically significant effects for many

compounds (Table 3). These data suggest a non-linear variation in the efficiency

of the extraction within the domain of the design. The main effect plots for

selected compounds confirmed that maximum yields were achieved at

intermediate extraction times and dichloromethane percentages, Fig. 2. Finally,

two-factor interactions remained below the statistical significance threshold,

data not shown.


271

2.0CH2Cl2 (%)

40.0 90.0

HMS

11

12

13

14

15

16

(x104)

Time (min)

10.0 5.0Temperature (ºC)

40.0

Pea

k ar

eaBP-3

14

19

24

29

34

39

44

(x104)

Pea

k ar

ea

2.0 40.0 90.010.0 5.0 40.0CH2Cl2 (%)Time (min) Temperature (ºC)

4-MBC

18

19

20

21

22

(x104)

Pea

k ar

ea


OC

10

12

14

16

18

(x104)

Pea

k ar

ea


Fig. 2. Main effect graphs provided by the experimental factorial design for selected

compounds.

The best compromise conditions, which maximized the efficiency of the

extraction for all analytes, were calculated with a global desirability (D)

function. D is defined as the geometric mean of the normalized (between 0 for

the minimum and 1 for the maximum) individual responses (di) predicted by

the Box-Behnken design for each UV-filter. The maximum value of D (0.89) was

obtained at 75 ºC, using a n-hexane:dichloromethane (80:20, v:v) mixture and

considering an extraction time of 5 min, Fig. 3.


272

Extraction time, 5 min

5 10 15 20 25 30 35 40CH2Cl2 (%)

4050

6070

8090

0

0.2

0.4

0.6

0.8

1D

esir

abili

ty (

D)

Extraction temperature, 75 ºC

2 4 6 8 10Time (min)

1020

3040

0

0.2

0.4

0.6

0.8

1

Des

irab

ility

(D

)

Fig. 3. Plots of the global desirability function.

3.3.2. Extraction cycles, flush volume and purge time

The potential influence of these parameters on the efficiency of the

extraction was evaluated with an univariant approach. No differences were

observed using 1, 2 or 3 extraction cycles of 5 min. Thus, a single cycle was

considered to speed up the extraction step. Fig. 4 shows the responses (peak

areas) for three different flush percentages, referred to the volume of the PLE

cell (11 mL). Similar responses were measured for flush values of 100% and

140%, whereas a slight reduction was appreciated for several analytes using a

percentage of 60%. This factor was set at 100%. Operating under above

conditions (1 cycle and 100% flush), the volume of the extract collected from the


273

PLE cell remained around 20 mL. Purge times higher than 2 min were also

studied without significant changes in the extraction efficiency; thus, 2 min was

maintained as working value for this variable.

Pe

ak

are

a (

x 1

05 )

0

5

10

15

20

25


60% flush 100% flush 140% flush

Fig. 4. Effect of flush percentage on the responses obtained for a spiked (5 g g-1) sludge

sample, n=3 replicates.

3.4. Recoveries and quantification limits

The recoveries of the method were evaluated using two freeze-dried

samples of primary and biological sludge spiked at two different concentration

levels (300 and 1000 ng g-1). Non-spiked fractions of each matrix and procedural

blanks were also processed, Fig. 5. Found recoveries ranged from 73% to 112%,

with relative standard deviation values below 12%, Table 4. The above data are

similar to those reported by Plagellat et al. [14] for 4-MBC, EHMC and OC

using liquid-liquid extraction of wet sludge samples and Nieto et al. [18] for BP-

3, EHPABA and OC considering methanol:water mixtures for PLE of several

personal care compounds from freeze-dried sludge.


274

9.40 9.80 10.20 10.600

2000

4000

6000

8000

10000

12000

Time (min)

Abundancem/z 120

EHS

HMS

11.60 12.00 12.40 12.80 13.20 13.600

2000

4000

6000

8000

10000

12000

14000

Time (min)

Abundancem/z 178

IAMC(E) EHMC

(Z) EHMC

11.25 11.35 11.45 11.55 11.650

500

1000

1500

2000

2500

3000

3500

4000

Time (min)

Abundancem/z 227

BP-3

11.00 11.20 11.40 11.60 11.80 12.000

500

1000

1500

2000

2500

3000

3500

4000

4500

Time (min)

Abundancem/z 254

(E) 4-MBC

(Z) 4-MBC

15.80 16.00 16.20 16.40 16.600

2000400060008000

10000120001400016000180002000022000

Time (min)

Abundancem/z 360

OC

12.40 12.80 13.20 13.60 14.000

200400600800

1000120014001600180020002200

Time (min)

Abundancem/z 277

EHPABA

A

B

C

Fig. 5. Selected ion monitoring chromatograms corresponding to a procedural blank

(A), a non-spiked sample of biological sludge (B), and same sample fortified with 300 ng

g-1 of each analyte (C).


275

Table 4. Recoveries of the method for spiked samples (n=3 replicates) and estimated

limits of quantification (LOQs) of the method.

Analyte Primary sludge (TC 30%) Biological sludge TC (35%) LOQs (ng g-1)

a300 ng g-1 a1000 ng g-1 a300 ng g-1 a1000 ng g-1

EHS

HMS

IAMC

BP-3

4-MBC

EHPABA

EHMC

OC

101 7

96 6

107 6

89 11

86 5

93 7

90 5

85 5

95 7

78 5

80 4

106 6

79 4

88 6

73 5

84 12

102 8

103 1

90 4

112 4

91 7

83 7

90 9

112 5

103 3

100 7

98 6

100 4

107 3

104 3

88 7

98 8

17

34

34

61

26

22

24

33


The reproducibility of the method was investigated with a sample of

biological sludge fortified at 500 ng mL-1. The relative standard deviations

(RSDs, %) for nine extractions in three consecutive days varied between 6 and

13%.

As shown in Fig. 5, analytes were not detected in the procedural blanks;

therefore, the LOQs of the method (defined for a S/N of 10) were estimated

from chromatographic peaks of UV filters in non-spiked samples, or in the low

level spiked fraction for those species not detected in sludge (IAMC and

EHPABA). The achieved LOQs varied between 17 ng g-1 for EHS and 61 ng g-1

for BP-3, Table 4. They are similar to the LOQs (from 7 to 67 ng g-1) reported for

same compounds in sediment samples with TC below 0.2%, using GC-MS as

detection technique and a less elaborated clean-up procedure [16]. Plagellat and

co-workers [14] achieved LOQs between 9 and 18 ng g-1 for 4-MBC, EHMC and

OC considering a three times larger sample intake (60 g of fresh sludge at 3%)

and using also GC-MS as determination technique.


276

3.5. Application to real samples

The proposed method was applied to freeze-dried sludge from different

urban sewage plants. EHPABA and IAMC were not detected in any of the

processed samples. The concentrations measured for the rest of species are

compiled in Table 5. 4-MBC and EHMC appear in sludge as mixtures of E and

Z forms. The sum of peak areas for both isomers was compared with calibration

curves obtained for the commercial available E forms. Samples code 1 and 2, on

Table 5, corresponded to wet sludge from a plant receiving the wastewater

from a 100000 inhabitants city, located in the Northwest of Spain. Both samples

were obtained in March of 2010 and lyophilized in our laboratory. The rest of

specimens (codes 3 to 9) were from STPs in the same geographical area,

although their exact locations are not revealed due to a confidentiality

agreement. They were collected between February and May 2010, in-situ

lyophilized and further submitted to the laboratory for analysis. Samples code 5

to 9 (Table 5) are mixtures of primary and biological sludge.

Table 5. Summary of concentrations (ng g-1) measured in sludge samples, n=3

replicates. IAMC and EHPABA were not detected in any sample.

Code Type Concentration (ng g-1) SD

EHS HMS BP-3 4-MBC EHMC OC

1

2

3

4

5

6

7

8

9

Primary

Biological

Primary

Biological

Mixture

Mixture

Mixture

Mixture

Mixture

n.d.

270 14

n.d.

133 26

298 5

n.d.

200 36

188 9

268 11

n.d.

207 31

n.d.

110 10

401 35

n.d.

240 8

256 18

180 ± 27

n.d.

n.d.

n.d.

93 11

n.d.

n.d.

n.d.

n.d.

295 14

1543 26

1439 49

106 5

97 8

223 9

120 3

372 10

351 41

1579 51

3287 98

856 98

213 3

104 5

160 7

192 15

125 5

100 10

2776 137

2242 16

3263 176

1039 50

377 30

1766 72

1038 63

1934 222

523 58

2240 45

Meana 226 232 194 648 868 1602

aAverage value of quantified concentrations


277

4-MBC, EHMC and OC were ubiquitous pollutants in sludge, with

average concentrations increasing the following order: 4-MBC< EHMC < OC,

Table 5. The mean levels of 4-MBC and OC were lower than those reported for

sewage sludge samples collected in Switzerland; however, a higher value was

obtained for EHMC [14]. EHS and HMS were quantified in six of nine samples

with maximum values below 400 ng g-1, and BP-3 showed a lower detection

frequency, Table 5. Globally, the occurrence frequency and the relative

concentrations of UV filters in sludge followed the same pattern as in water

samples taken in the same geographic area [13]. Moreover, they are in

agreement with the high sorption coefficients reported for 4-MBC, EHMC and

OC in sludge [6].

4. Conclusions

PLE extraction, combined with the use of graphitized carbon for in-cell

retention of pigments and additional clean-up with a PSA cartridge, constitutes

a suitable approach in terms of extraction efficiency and selectivity for the GC-

MS determination of a broad group of UV filters in sludge samples. As far as

we could trace, this study reports the first application of both materials for the

clean-up of PLE extracts from sludge samples, achieving an improved

selectivity in comparison with the commonly used normal-phase sorbents. The

analysis of sludge samples confirmed the significant accumulation of three UV

filters (4-MBC, EHMC and OC) in this matrix, with average concentrations

higher than 600 ng g-1. This information must be considered in order to (1)

properly calculate their removal rates during wastewater treatments and (2) to

evaluate the risk of re-introducing the above species in the terrestrial

environment through the disposal of sludge as fertilizer in agriculture fields.

Acknowledgements

Financial support from the Spanish Government and E.U. FEDER funds

(project CTQ2009-08377) is acknowledged. N.N. is grateful for a FPU grant from


278

the Spanish Ministry of Education and Science. We are also in debt with Dr.

Carballa and Labaqua for supplying, or providing access, the sludge samples.


279

References

[1] A. Salvador, A. Chisvert, Analysis of Cosmetic Products, Elsevier, Italy, 2007.



[4] K. Fent, A. Zenker, M. Rapp, Environ. Poll. 158 (2010) 1817.


[6] T. Kupper, C. Plagellat, R.C. Brändli, L.F. de Alencastro, D. Grandjean, J. Tarradellas, Water

Res. 40 (2006) 2603.

[7] T. Poiger, H.R. Buser, M.E. Balmer, P.A. Bergqvist, M.D. Müller, Chemosphere 55 (2004) 951.

[8] H. Klammer, C. Schlecht, W. Wuttke, C. Schmutzler, I. Gotthardt, J. Köhrle, H. Jarry,

Toxicology 238 (2007) 192.

[9] M. Schlumpf, B. Cotton, M. Conscience, V. Haller, B. Steinmann, W. Lichtensteiger, Environ.

Health Perspect. 109 (2001) 239.

[10] M. S. Díaz-Cruz, D. Barceló, Trends Anal. Chem. 28 (2009) 708.


Chem. 80 (2008) 1307.


[13] N. Negreira, I. Rodriguez, E. Rubí, R. Cela, Anal. Bioanal. Chem. 398 (2010) 995.

[14] C. Plagellat, T. Kupper, R. Furrer, L.F. de Alencastro, D. Grandjean, J. Tarradellas,


[15] A. Nieto, F. Borrull, E. Pocurull, R.M. Marcé, Trends Anal. Chem. 29 (2010) 752.


[17] R. Rodil, S. Schrader, M. Moeder, J. Chromatogr. A 1216 (2009) 8851.

[18] A. Nieto, F. Borrull, R.M. Marcé, E. Pocurull, J. Chromatogr. A 1216 (2009) 5619.

[19] R. Rodil, S. Schrader, M. Moeder, Rapid Commun. Mass Spectrom. 23 (2009) 580.

[20] P. Canosa, D. Pérez-Palacios, A. Garrido-López, M.T. Tena, I. Rodríguez, E. Rubí, R. Cela, J.

Chromatogr. A 1161 (2007) 105-112.

[21] P. Haglund, E. Spinnel, LCGC North America 28 (2010) 544.

[22] S. Chu, C.D. Metcalfe, J. Chromatogr. A 1164 (2007) 212.

[23] B. Gilbert-López, J.F. García-Reyes, A. Molina-Díaz, Talanta 79 (2009) 109.

[24] O. Shimelis, Y. Yang, K. Stenerson, T. Kaneko, M. Ye, J. Chromatogr. A 1165 (2007) 18.

2.4. Publicación:

DETERMINATION OF

SELECTED UV FILTERS IN INDOOR DUST

USING MATRIZ SOLID-PHASE DISPERSION

AND GAS CHROMATOGRAPHY




(doi:10.1016/j.chroma.2009.06.020)


283

Determination of selected UV filters in indoor dust using matrix solid-phase

dispersion and gas chromatography tandem mass spectrometry





Abstract

A simple and inexpensive sample preparation procedure, based on the

matrix solid-phase dispersion (MSPD) technique, for the determination of six

UV filters: 2-ethylhexyl salicylate (EHS), 3,3,5-trimethylcyclohexyl salicylate

(Homosalate, HMS), 3-(4-methylbenzylidene) camphor (4-MBC), isoamyl-p-

methoxycinnamate (IAMC), 2-ethylhexyl-p-methoxycinnamate (EHMC) and

octocrylene (OCR) in dust from indoor environments is presented and the

influence of several operational parameters on the performance of the

extraction discussed. Under final working conditions, sieved samples (0.5 g)

were mixed with the same amount of anhydrous sodium sulfate and dispersed

with 2 g of C18 in a mortar with a pestle. This blend was transferred to

polypropylene solid-phase extraction cartridge containing 2 g of activated silica,

as clean-up co-sorbent. The cartridge was first rinsed with 5 mL of n-hexane

and then analytes were recovered with 4 mL of acetonitrile. This extract was

adjusted to 1 mL, filtered and compounds were determined by gas

chromatography combined with tandem mass spectrometry (GC-MS/MS).

Recoveries for samples spiked at two different concentrations ranged from 77 to

99% and the limits of quantification (LOQs) of the method remained between 10

and 40 ng g-1. Analysis of settled dust from different indoor areas, including

private flats, public buildings and vehicle cabinets, showed the ubiquitous of

EHMC and OCR in this matrix, with maximum concentrations of 15 and 41 g

g-1, respectively. Both UV filters were also quantified in dust reference material


284

SRM 2585 for first time. EHS, 4-MBC and IAMC were detected in some of the

analyzed samples, although at lower concentrations than EHMC and OCR.

Keywords: UV filters; dust; indoor atmospheres; matrix solid-phase

dispersion; gas chromatography tandem mass spectrometry.

1. Introduction

UV filters are compounds designed to mitigate the deleterious effects of

sunlight. Most of them are organic substances, characterized by single or

multiple aromatic structures, often with attached hydrophobic groups [1]. One

of their most known uses is in sunscreens, which are applied directly on the

skin as protection against UV radiation. The number and maximum allowable

concentrations of UV filters in these products have been legislated in many

countries. As example, in the European Union 26 organic compounds have been

approved to be incorporated in sunscreens at individual concentrations up to

10%, for most of them [2,3]. Moreover, they are also included in the formulation

of other personal care products (cosmetics, hair dyes, shampoos) and used in

the protection of goods, plastics, varnishes and clothes [4-7]; however, no data

could be traced related to the type of UV absorbers and concentrations

incorporated in these materials.

As many other daily usage compounds, UV filters are continuously

discharged in the aquatic environment. Washing off from the skin, during

bathing or swimming, and indirect releases from towels and clothes contribute

significantly to their presence in surface and wastewater samples [8]. Their

further behavior in the aquatic media depends on a number of factors such as

(1) their stability during wastewater treatments [4,9-10], (2) physicochemical

properties, particularly their polarity, and (3) potential transformation through

photochemical and/or oxidation reactions [11,12]. Medium and highly polar

UV filters (e.g. 2-hydroxy-4-methoxybenzophenone, BP-3, and 2-hydroxy-4-

methoxybenzophenone-5-sulphonic acid, BP-4) have been often detected in


285

surface and wastewater [6,7,13]; moreover, BP-4 is not effectively removed

during conventional sewage treatments [6]. More lipophilic species, such as 3-

(4-methylbenzylidene) camphor (4-MBC), 2-ethylhexyl-p-methoxycinnamate

(EHMC) and octocrylene (OCR) have been found in river and lake sediments

[14], sludge [15] and even in biota [4,16]. In fact, concentrations up to 2 g g-1

have been reported for 4-MBC and OCR in fish [16] and as high as 15 g g-1 for

OCR in sludge [15], suggesting that certain UV filters can be bio-accumulated

and concentrated on solid matrices. This information added to reports about the

estrogenic activity of some species, such as 4-MBC and EHMC [17-21], have

risen the concern about their effects on wildlife and humans.

Conversely to above studies dealing with environmental samples, as far

as we could trace, no data are available in relation to the levels of UV filters in

indoor areas. Breath and oral intake of suspended particulate matter and settled

dust are considered as sources of continuous exposure to organic and inorganic

chemicals used in building materials and daily activities [22,23]. Most

compounds present an excellent stability in indoor and confined areas, since

removal through photochemical reactions as well as airborne dispersion have a

minor importance. Some of these species might contribute to the increasing

incidence of allergies, asthma and other respiratory diseases in developed

countries, where citizens spend more and more time in indoor areas [24]. In this

sense, recent studies have shown that certain personal care products reach

higher concentrations in dust than in other heavily polluted environmental

matrices, such as sludge [25,26]. The same trend has been observed for other

compounds added to building materials and furniture, e.g. organophosphorous

[27] and polybrominated flame retardants [28].

The goal of this work was to develop a procedure for the determination

of a group of selected UV filters: EHS, HMS, isoamyl p-methoxycinnamate

(IAMC), 4-MBC, EHMC, and OCR, in dust from different indoor environments.

Matrix solid-phase dispersion (MSPD) was considered as sample preparation


286

technique on the basis of its low cost, use of mild conditions and feasibility to

integrate extraction and clean-up in the same step, which can be achieved with

a suitable combination of elution solvents, dispersant and clean-up co-sorbents

[29,30]. Although MSPD was initially centered on biota samples [31], recent

works have proven its applicability to the extraction of organic compounds

from dust [25,32] and other solid matrices, such as freeze-dried sludge [33,34].

After extraction, analytes were determined by gas chromatography in

combination with tandem mass spectrometry (GC-MS/MS). The second goal of

this work was to provide a first overview of the levels for some UV filters in

dust samples collected from different confined environments, including a

reference material of indoor dust.

2. Experimental


HPLC-grade methanol, acetone, n-hexane, dichloromethane, ethyl

acetate and acetonitrile (trace analysis grade) were supplied by Merck

(Darmstadt, Germany). Standards of EHS, HMS, 4-MBC, EHMC and OCR were

acquired from Aldrich (Milwaukee, WI, USA) and Merck. IAMC was kindly

provided by Dr. R. Rodil (University of La Coruña, Spain). Chemical structures

and octanol-water partition coefficients (log Kow) of above compounds are

depicted in Fig. 1. As observed, all of them show moderate to high lipophilic

character with the subsequent risk of being adsorbed on dust particles.

Individual solutions of each compound were prepared in methanol. Further

dilutions and mixtures of them were made in acetone and acetonitrile.


287

EHS

log Kow 5.77

HMS

log Kow 5.82

IAMC

log Kow 4.06

4-MBC

log Kow 4.95

EHMC

log Kow 5.66

OCR

log Kow 7.53

Fig. 1. Chemical structures and octanol-water partition coefficients (Log Kow) of selected

UV filters.

Anhydrous sodium sulphate, Florisil (60-100 mesh), C18 (70-230 mesh),

silica (230-400 mesh) and alumina (150 mesh) sorbents, used in the MSPD

extraction process, were acquired from Aldrich and Merck. Normal-phase

materials were activated at 130 ºC, for 24 h, and then allowed to cool down in a

desiccator before being used in the extraction process. C18 was used as

received. Polypropylene solid-phase extraction cartridges (15 mL capacity) and

20 µm polyethylene frits were purchased from International Sorbent

Technology (Mid Glamorgan, UK). Syringe filters (Millex GV, 13 mm, 0.22 µm)

were obtained from Millipore (Billerica, MA, USA).


288

2.2. Samples

Dust from private houses, public buildings and vehicles cabins were

collected using domestic vacuum cleaners equipped with paper filter bags.

After opening, the content of bags was sieved and the fraction with a particle

size below 60 µm was retained for this study. Sieved samples were stored at 4

ºC, in amber glass vessels and their total carbon (TC) content characterized.

Obtained TC values ranged from 13.5 to 33.6%. Reference material SRM 2585,

organic compounds in house dust, was acquired from NIST (Gaithersburg, MD,

USA). MSPD extraction conditions were optimised using a pooled dust matrix

with a TC of 20.8%. Fractions of this sample were fortified with different UV

filters. More details about added compounds and their concentrations are given

in the results and discussion section. During method validation, individual dust

samples were also spiked with all compounds at different levels. In all cases,

the spiking procedure consisted of mixing the sieved samples with a standard

solution of UV filters in acetone. Approximately, 1 mL of standard was added

per g of dust in order to obtain an homogeneous slurry, which was left at room

temperature until complete evaporation of the solvent. After that, spiked

samples were aged, at 4 ºC, for at least 2 weeks before extraction.


Dust (0.5 g) was mixed with anhydrous sodium sulphate, 0.5 g, and

dispersed with C18 in a glass mortar, with a pestle, until a visually

homogeneous blend was obtained. Then, it was transferred to a cartridge

containing a polyethylene frit and a given amount of co-sorbent. A second frit

was placed over the dispersed sample before slight compression. Cartridges

were eluted by gravity. Under optimised conditions, 2 g of C18 and the same

mass of silica were used as dispersant and co-sorbent, respectively. MSPD

cartridges were first rinsed with 5 mL of n-hexane to remove organic

compounds with a lower polarity than analytes. Then, analytes were recovered

using just 4 mL of acetonitrile.


289

2.4. Determination

Analytes were determined by GC-MS/MS using a Varian (Walnut Creek,

CA, USA) CP 3900 gas chromatograph connected to an ion-trap mass

spectrometer (Varian Saturn 2100). Separations were carried out in a HP-5ms

capillary column (30 m x 0.25 mm i.d., df: 0.25 m) supplied by Agilent

(Wilmington, DE, USA). Helium (99.999 %) was used as carrier gas at a constant

flow of 1.2 mL min-1. The GC oven was programmed as follows: 70 ºC (held for

1 min), at 12 ºC min-1 to 280 ºC (held for 10 min). GC-MS interface and ion-trap

temperatures were set at 280 ºC and 220 ºC, respectively. Standards and sample

extracts in acetonitrile (1-2 L) were injected in the splitless mode (splitless time

1 min), with the injector port at 280 ºC. The mass spectrometer was operated in

the electron impact ionisation mode (70 eV). MS spectra were recorded in the

range from 70 to 400 m/z units. The base peak in the spectra of each compound

was isolated with a window of 3 m/z units and subjected to collision induced

dissociation. GC-MS and GC-MS/MS were considered as detection techniques

during the development of this work. Concentrations of target species in non-

spiked samples and evaluation of method performance were done by GC-

MS/MS, considering external calibration, against standards prepared in

acetonitrile, as quantification technique.


3.1. GC-MS/MS conditions

Fig. 2 shows the MS/MS spectra obtained for considered compounds

and the structures proposed for their most intense product ions. In the case of

salicylates (EHS and HMS), the parent ion (138 m/z), corresponding to

replacement of the aliphatic chain bonded to the ester moiety by hydrogen,

underwent a removal of water leading to a single product at 120 m/z units. The

precursor ion in the MS spectra of both cinnamates (178 m/z units), which

reflected also the substitution of the branched alkyl chain attached to the ester

moiety by hydrogen, rendered two main transitions corresponding to the

removal of the hydroxyl and carbonyl groups (178 > 161 and 161 > 133 m/z


290

units, respectively). The main transition from the [M-H]+ OCR ion, appearing at

360 m/z units, reflected the loss of organic chains (butyl and ethyl groups)

bonded to the tertiary carbon, Fig. 2. Finally, the MS/MS spectra of 4-MBC

showed multiple transitions, the most intense one, indicating the loss of a

methyl group (254 > 239 m/z units), was used for quantitative purposes. Table

1 summarizes optimal MS/MS detection parameters, as well as correlation

coefficients (R2) values and instrumental limits of quantification (LOQs),

defined for a signal to noise ratio (S/N) of 10, corresponding the injection of

standards prepared in acetonitrile. In comparison to the single MS mode,

MS/MS detection provided around twice lower LOQs values, except for 4-

MBC. Retention times given in Table 1 for 4-MBC and both cinnamates

correspond to their E-isomers, which are the forms incorporated in sunscreens

and other personal care products. It must be kept on mind that Z-isomers,

detected in some environmental samples due to photochemical isomerisation

[14,15], appear at lower retention times; thus, the multiple reaction monitoring

(MRM) mode was used to record the MS/MS transitions of 4-MBC, IAMC and

EHMC in the same segment.


291

225 250 275 300 325 350 m/z

0%

25%

50%

75%

100%

232246258

264

276

288304

318

330

342

360

OCR

m/z 276

100 125 150 175 200 225 m/z

0%

25%

50%

75%

100%

103

121

133

149

161

178

EHMC, IAMC

m/z 133

MeO

(E)

O

m/z 161

125 150 175 200 225 250 m/z

0%

25%

50%

75%

100%

132

149

162

170

183

197

211

226

239

254

4-MBC

m/z 239

100 150 200 250 300 350 m/z

0%

25%

50%

75%

100%120

138

HMS, EHS

m/z 120

Fig. 2. MS/MS spectra and structures proposed for most intense product ions.


292

Tabl

e 1.

Ret

enti

on ti

mes

, MS/

MS

oper

atio

nal c

ondi

tion

s an

d pe

rfor

man

ce o

f GC

-MS/

MS

and

GC

-MS

for

stan

dard

s of

UV

filte

rs p

repa

red

in a

ceto

nitr

ile. L

OQ

s va

lues

def

ined

for

a S/

N r

atio

of 1

0.

Com

poun

d

Ret

. tim

e

(min

)

Pare

nt io

n

(m/z

)

Prod

uct i

ons

(m/z

)

Exc

itati

on

ampl

itud

e (V

) St

orag

e le

vel (

m/z

) C

orre

lati

on

Coe

ffic

ient

(R2 )

b

LO

Qs

(ng

mL

-1)b

Cor

rela

tion

Coe

ffic

ient

(R2 )

c

LOQ

s

(ng

mL

-1)c

EH

S

HM

S

IAM

C

4-M

BC

EH

MC

OC

R

12.7

13.3

,13.

5

14.7

a

14.9

a

16.8

a

19.2

138

138

178

254

178

360

120

120

161+

133

239

161+

133

276

0.45

0.45

0.91

0.82

0.91

1.29

53

53

68

97

68

137

0.99

9

0.99

8

0.99

9

0.99

7

0.99

9

0.99

9

5 6 5 20

5 6

0.99

6

0.99

7

0.99

6

0.99

6

0.99

3

0.99

9

11

10

15

17

15

12

a Ret

enti

on ti

mes

cor

resp

ondi

ng to

E-i

som

ers.

b Dat

a fo

r G

C-M

S/M

S.

c Dat

a fo

r G

C-M

S.

b,c V

alue

s ob

tain

ed fo

r st

and

ard

s at

six

leve

ls b

etw

een

LO

Qs

and

3

g m

L-1 .


293

3.2. Optimisation of MSPD parameters

Preliminary extraction assays were performed with fractions of a pooled

dust sample (TC 20.8%) spiked with all compounds at 3 g g-1, except IAMC,

which was not considered in the earliest steps of this research. On the basis of a

previous application of MSPD to the extraction of bactericides from dust [25],

samples (0.5 g) were mixed with 0.5 g of sodium sulphate and dispersed using

1.25 g of C18 in a mortar with a pestle. The resulting blend was transferred to a

polypropylene cartridge containing 2 g of different normal-phase materials,

which worked as clean up co-sorbents retaining species with a higher polarity

than target UV filters through adsorption processes. N-hexane,

dichloromethane, ethyl acetate and acetonitrile were tested as extraction

solvents. In all cases, 5 mL of solvent were recovered from the MSPD cartridge

and analyzed by GC using single MS detection. None of the compounds could

be eluted with n-hexane. 4-MBC, EHMC and OCR were noticed in

dichloromethane extracts; however, this solvent failed to recovered EHS and

HMS, figure not shown. Probably, the presence of a phenolic group in the

structure of these two species (Fig. 1) resulted in a stronger interaction with the

dust matrix and/or with the normal-phase co-sorbent than the rest of analytes.

Responses (peak areas) obtained using ethyl acetate and acetonitrile are plotted

in Fig. 3. Whatever the type of co-sorbent, for EHMC, OCR and, in a lesser

extension, 4-MBC acetonitrile provided higher responses than ethyl acetate. For

EHS and HMS, the co-sorbent played a more important effect on the extraction

yield than the elution solvent. Both salicylates were strongly retained by

alumina, particularly when combined with acetonitrile as eluent. On the basis

of these results, Florisil, silica, ethyl acetate and acetonitrile were selected as co-

sorbents and elution solvents, respectively, for further experiments. Extraction

of non-spiked fractions of the pooled dust sample, using above conditions,

demonstrated the existence of noticeable concentrations of OCR and EHMC in

this matrix, figure not shown.


294

0,E+00

4,E+05

8,E+05

1,E+06

2,E+06

2,E+06

EHS HMS 4-MBC EHMC OCR

Pe

ak

are

aAlumina ACN Florisil-ACN Silica-ACN Alumina-AcOEt Florisil-AcOEt Silica-AcOEt-

Fig. 3. Comparison of responses for different combinations of elution solvents and co-

sorbents in the MSPD process, n=3 replicates. Data for 0.5 g of spiked dust dispersed

with 1.25 g of C18.

In a second series of extractions, the effects of co-sorbent, elution solvent

and mass of dispersant (C18), on the efficiency of the MSPD process, were

simultaneously assessed using a 23 type experimental factorial design, Table 2.

The mass of sample was 0.5 g and the volume of extraction solvent 10 mL in all

experiments; moreover, MSPD cartridges were rinsed with 5 mL of n-hexane to

remove less polar interferences, previously to the extraction of target species.

Obviously, n-hexane extracts were discarded. The study was accomplished

with a different fraction of the same pooled matrix used in the preliminary

extraction experiments. In this case, it was fortified with EHS, HMS, 4-MBC and

IAMC (3 g g-1, each). Extracts were filtered and processed directly using GC-

MS as detection technique. Peak areas obtained for above species, as well as

EHMC and OCR (already present in the sample) were considered as variable

response in the experimental factorial design. For EHMC the sum of responses

for E and Z isomers was taken. On the other hand, for IAMC and 4-MBC just

one peak corresponding to the isomer added to the sample (E-form) was


295

noticed in the GC-MS chromatograms. This fact suggests the absence of

isomerisation reactions during aging of the spiked sample and through the

extraction process. Standardized main effects associated to each factor are

reported in Table 2. Their absolute values are proportional to the variation in

the efficiency of the extraction when the considered factor changes from the low

to the high level, within the domain of the design. A positive sign indicates an

increase in the yield of the process and, a negative one the opposite behaviour.

Data on Table 2 suggest to different trends. On one hand, for 4-MBC, OCR,

IAMC and EHMC, the extraction solvent was the only factor playing a

significant influence (95% confidence level) on the yield of the extraction, with

the highest responses corresponding to acetonitrile. On the other hand, EHS

and HMS followed a different pattern. In this case, the most relevant factor was

the type of co-sorbent placed at the bottom of the extraction cartridge. It seemed

that both species were strongly retained on Florisil, thus lower responses were

observed for this material than for silica. A similar trend has been reported by

Rodil and co-workers [14] in the extraction of hydroxylated UV filters from

sediments using pressurized solvents. Moreover, the extraction of EHS and

HMS was favoured significantly by the use of large masses of C-18 as

dispersant. Regarding the elution solvent, HMS was more efficiently recovered

with acetonitrile, whereas EHS presented a higher affinity for ethyl acetate,

although without achieving the statistically significance boundary. In general,

two factor interactions exerted a minor influence on the yield of the extraction

process, figure not shown. As regards the effect of extraction parameters on the

complexity of the total ionic current (TIC) GC-MS chromatograms, little

changes were noticed among different conditions explored in the experimental

design. The most intense chromatographic peaks corresponded to phthalates,

although they appeared at different retention times than target UV filters. On

the basis of above comments, acetonitrile and silica were selected as extraction

solvent and co-sorbent in the MSPD process, respectively; whereas, the mass of

dispersant was fixed at 2 g.


296

Table 2. Domain and standardized main effects of factors considered in the experimental

factorial design.

Factor Level Standardized value

Low High EHS HMS 4-MBC IAMC EHMC OCR

Solvent

Co-sorbent

C-18 mass (g)

Ethyl acetate

Florisil

0.5

Acetonitrile

Silica

2.0

-5.8

18a

9.1

19a

88a

42a

26a

0.06

3.8

15a

1.2

3.2

17a

0.96

1.4

12a

0.11

0.15

a Statistically significant factors at the 95% confidence level

The following parameter to optimise in the extraction process was the

volume of acetonitrile. The study was carried out by collecting consecutive

fractions of 2 mL from the MSPD cartridge. Normalised responses for duplicate

experiments are shown in Fig. 4. As observed, compounds could be eluted

using just 4 mL of acetonitrile. Thus, the whole sample preparation method

required a total of 9 mL of two different organic solvents, a volume significantly

smaller than that reported for the extraction of same compounds from other

solid matrices with lower carbon contents, such as sediments [14]. Evaporation

of acetonitrile extracts to a final volume of 1 mL, under mild conditions (room

temperature using a gentle stream of nitrogen), did not lead to noticeable losses

of any compound, thus this step was incorporated in the sample preparation

scheme.


297

0%

20%

40%

60%

80%

100%

120%

EHS HMS IAMC 4-MBC EHMC OCR

Compound

No

rmal

ised

res

po

nse

Fraction 1 Fraction 2 Fraction 3 Fraction 4

Fig. 4. Normalised signals for consecutive acetonitrile fractions (2 mL) eluted from the

MSPD cartridge.

3.3. Performance evaluation

Recoveries of the optimized method were estimated with dust spiked at

two different concentration levels: 0.3 and 3 g g-1, per compound. This study

was carried out with a sample, obtained from a public building (TC 13.5%),

containing relatively low levels of target species. After sieving, it was divided in

three fractions, one was used as blank and the others spiked at the above

referred levels and aged for 2 weeks. Each fraction was processed (n=4

replicates) and the corresponding extracts concentrated to 1 mL. Table 3 shows

the recoveries obtained using GC-MS and GC-MS/MS as detection techniques.

The first led to values over 100% for some analytes in the sample spiked at the

lower level, probably due the co-elution of other compounds which contributed

to peak areas of UV filters recorded in chromatograms monitored for the most

intense m/z ions in their MS spectra (see 3rd column on Table 1), this drawback

was overcome using MS/MS detection, which was chosen as quantification

technique for the analysis of non-spiked samples. Globally, the proposed


298

method provided recoveries between 77 and 99%, with associated standard

deviations below 10, for all species.

Table 3. Summary of recoveries obtained for a spiked dust sample, n=4 replicates, and

LOQs of the method defined for a S/N ratio of 10.

Compound

Percentage of recovery (%) ± SD LOQs (ng/g)b

GC-MS GC-MS/MS

300 ng g-1a 3000 ng g-1a 300 ng g-1a 3000 ng g-1a

EHS

HMS

IAMC

4-MBC

EHMC

OCR

89 ± 3

91 ± 4

115 ± 6

73 ± 10

120 ± 5

109 ± 2

82 ± 5

81 ± 5

90 ± 3

97 ± 6

100 ± 5

96 ± 5

92 ± 2

87 ± 4

85 ± 4

99 ± 9

93 ± 4

79 ± 1

81 ± 6

83 ± 2

80 ± 4

77 ± 7

89 ± 5

77 ± 5

10

12

10

40

10

12


b Referred to GC-MS/MS detection

Considering a sample intake of 0.5 g and adjusting the volume of the

final extract to 1 mL, the limits of quantification for the proposed method,

defined as the concentration of analyte which produced a signal to noise ratio of

10, ranged from 10 to 40 ng g-1, Table 3. Analysis of procedural blanks

demonstrated the absence of contamination problems (Fig. 5), thus above

values were mainly controlled by the instrumental LOQs of the GC-MS/MS.


299

Procedural blankUn-spiked sampleSpiked at 300 ng g-1

13.0 13.5 min.

0.0

2.5

5.0

7.5

10.0

12.5

15.0

m/z 120

EHS

HMS

kCounts

19.0 19.5 20.0 min.

0.0

2.5

5.0

7.5

m/z 276

OCR

kCounts

14.5 15.0 15.5 min.

0.0

2.5

5.0

7.5

10.0m/z 239

4-MBC

kCounts

14 15 16 17 min.

0

25

50

75

100

125

kCounts

m/z 161+133

E-IAMC

E-EHMC

Z-EHMC

Fig. 5. Overlay of GC-MS/MS chromatograms for a procedural blank, a non-spiked

dust sample (code 1, Table 4) and the same matrix fortified at 300 ng g-1.

3.4. Real samples analysis

The proposed method was applied to the analysis of several dust

samples collected in confined environments, from two different geographic

areas: Galicia (Northwest Spain) and La Rioja (North Spain). Obtained

concentrations are given in Table 4. Codes 1 to 8 correspond to dust from

private flats and public building, samples 9-10 were obtained from vehicles


300

cabins and code 11 corresponds to SRM 2585. HMS remained below the LOQs

of the method in all samples, whereas the rest of species were noticed at

different levels from 35 ng g-1 up to 41 g g-1, depending on the compound and

the sample. In some cases, extracts needed to be diluted up to 10 times to fit

within the linear response range of the optimised method (up to 3 g mL-1

referred to the acetonitrile extract). Data reported on Table 4 for IAMC, 4-MBC

and EHMC corresponded to the sum of E and Z isomers. For the first two

species, the ratio between peak areas of Z and E forms remained under 0.05;

however, EHMC showed a higher percentage of isomerisation with (Z/E) ratios

from 0.4 to 1, depending on the sample. In general, the highest UV filters

concentrations corresponded to EHMC and OCR, which were found in

practically all the processed samples. The average concentration of OCR in dust

(11.4 g g-1) was twice higher than that reported by Plagellat and co-workers in

freeze-dried sludge [15]. As regards EHMC, from the best of our knowledge,

these are the highest levels (up to 15 g g-1) ever found in any environmental or

biota sample. Concentrations of both UV filters, OCR and EHMC, in reference

material SRM 2585 (code 11, Table 4) are in the same range than the certified

values of BDEs 99 and 209 in this sample [28]. As this reference material was

prepared from dust samples obtained in North America, a different geographic

region than that for the rest of samples considered in this study, the presence of

high levels of EHMC and OCR in dust seems to be a world wide reality. EHS,

IAMC and 4-MBC were also found in some of the processed samples; however,

their occurrence frequency and found concentrations were lower than those

corresponding to EHMC and OCR.


301

Table 4. Concentrations of UV filters in non-spiked dust samples, n=3 replicates.

Code TC (%) Concentrations (ng g-1) with their standard deviations

EHS IAMC 4-MBC EHMC OCR

1

2

3

4

5

6

7

8

9

10

11ª

13.5

21.7

33.6

22.4

18.6

17.9

20.2

26.5

15.7

15.1

23.3

35 (1)

650 (20)

1440 (50)

2747 (104)

n.d.

144 (14)

114 (17)

n.d.

114 (13)

n.d.

n.d.

n.d.

1290 (60)

n.d.

n.d.

n.d.

n.d.

58 (1)

92 (1)

n.d.

n.d.

n.d.

n.d.

700 (50)

1990 (240)

n.d.

n.d.

n.d.

n.d.

116 (52)

n.d.

n.d.

n.d.

560 (40)

15000 (500)

2400 (30)

1680 (34)

1460 (20)

2050 (80)

1080 (30)

6220 (330)

6570 (30)

177(12)

6460 (150)

700 (40)

7482 (12)

19300 (760)

34400 (3000)

458 (8)

1400 (300)

n.d.

41000 (1100)

7700 (300)

450 (50)

880 (30)

Meanb 750 480 935 3970 11400

a SRM 2585 b Average of values over the LOQs of the method n.d. below detection limits

Likely, different sources are responsible for the presence of UV filters in

indoor dust. On one hand, personal care products contribute directly (sprays)

and indirectly (accidental spillage, volatilisation, skin cells) to the levels of UV

filters in confined areas; moreover, furniture, upholstery, paints and polymeric

materials used in vehicles cabins, flats and public buildings, as well as clothes

lint, might content some of the investigated compounds as UV absorbers.


302

4. Conclusions

MSPD followed by GC-MS/MS constitutes a fast, straightforward

approach for the determination of six UV filters in dust samples. The proposed

sample preparation method is particularly attractive since it does not require

the acquisition of dedicated instrumentation and it consumes low volumes of

organic solvents, providing recoveries over to 77% for all compounds and a

ready to inject extract. The performance of the extraction was mainly controlled

by the elution solvent and the type of co-sorbent placed at the bottom of the

MSPD cartridge. Results obtained for dust samples from different confined

environments demonstrated the presence of several UV filters in this matrix.

Particularly, mean values in the g g-1 range were detected for two of the

investigated species: EHMC and OCR. To the best of our knowledge, this is the

first report of both compounds in indoor areas. Further studies are necessary in

order to establish whether they proceed just from personal care products or

they might diffuse out from clothes, upholstery and building materials used in

indoor areas.

Acknowledgments



PGIDIT06PXIB237039PR). N. N thanks a FPU grant to the Spanish Ministry of

Science and Innovation. We also acknowledge Dr. P. Canosa for the supply of

most dust samples.


303

References





[5] A. Salvador, A. Chisvert., Anal. Chim. Acta 537 (2005) 15.


Chem. 80 (2008) 1307.


[8] S. D. Richardson, Anal. Chem. 80 (2008) 4373.

[9] W. Li, Y. Ma, C. Guo, W. Hu, K. Liu, Y. Wang, T. Zhu, Water Res. 41 (2007) 3506.

[10] T. Kupper, C. Plagellat, R.C. Brändi, L.F. Alencastro, D. Grandjean, J. Tarradellas, Water

Res. 40 (2006) 763.

[11] N. Negreira, P. Canosa, I. Rodríguez, M. Ramil, E. Rubí, R. Cela, J. Chromatogr. A 1178

(2008) 206.

[12] V.A. Sakkas, D.L. Giokas, D.A. Lambropoulou, T.A. Albanis, J. Chromatogr. A 1016 (2003)

211.



[15] C. Plagellat, T. Kupper, R. Furrer, L. F. Alencastro, D. Grandjean, J. Tarradellas,


[16] H.R. Buser, M.E. Balmer, P. Schmid, M. Kohler, Environ. Sci. Technol. 40 (2006) 1427.

[17] H. Klammer, C. Schlecht, W. Wuttke, C. Schmutzler, I. Gotthardt, J. Köhrle, H. Jarry,

Toxicology 238 (2007) 192.

[18] M. Heneweer, M. Muusse, M. Van den Berg, J. T. Sanderson, Toxicol. Appl. Pharmacol. 208

(2005) 170.

[19] M. Schlumpf, B. Cotton, M. Conscience, V. Haller, B. Steinmann, W. Lichtensteiger,


[20] K. Maerkel, W. Lichtensteiger, S. Durrer, M. Conscience, M. Schlumpf, Environ. Toxicol.

Pharmacol. 19 (2005) 761.

[21] D. Seidlová-Wuttke, J. Christoffel, G. Rimoldi, H. Jarry, W. Wuttke, Toxicol. Appl.

Pharmacol. 214 (2006) 1.

[22] S. Harrad, C. Ibarra, M. Diamond, L. Melymuk, M. Robson, J. Douwes, L. Roosens, A.C.

Dirtu, A. Covaci, Environ. Intern. 34 (2008) 232.

[23] H.A. Jones-Otazo, J.P. Clarke, M.L. Diamond, J.A. Archbold, G. Ferguson, T. Harner, M.G.

Richardson, J.J. Ryan, B. Wilford, Environ. Sci. Technol. 39 (2005) 5121.

[24] K. Bröms, K. Svärdsudd, C. Sundelin, D. Norbäck, Indoor Air 16 (2006) 227-235.


304

[25] P. Canosa, I. Rodríguez, E. Rubí, R. Cela, Anal. Chem. 79 (2007) 1675.

[26] S. Morales, P. Canosa, I. Rodríguez, E. Rubí, R. Cela, J. Chromatogr. A 1082 (2005) 128.

[27] G. Ingerowski, A. Friedle, J. Thumulla, Indoor Air 11 (2001) 145.

[28] H.M. Stapleton, T. Harner, M. Shoeib, J.M. Keller, M.M. Schantz, S.D. Leigh, S.A. Wise,

Anal. Bional. Chem. 384 (2006) 791.

[29] M. García-López, P. Canosa, I. Rodríguez, Anal. Bioanal. Chem. 391 (2008) 963.

[30] E.M.Kristenson, L. Ramos, U.A.Th. Brinkman, Trends Anal. Chem. 25 (2006) 96.

[31] S.A. Barker, J. Chromatogr. A 885 (2000)115.

[32] M. García, I. Rodríguez, R. Cela, Anal. Chim. Acta 590 (2007) 17.

[33] M.T. Pena, M.C. Casais, M.C. Mejuto, R. Cela, Anal. Chim. Acta 626 (2008) 155.

[34] C. Sánchez-Brunete, E. Miguel, J.L. Tadeo, Talanta 74 (2008) 1211.

Metodología desarrollada-Fotoiniciadores en alimentos

307

B. Fotoiniciadores

1. FOTOINICIADORES EN ALIMENTOS

1.1. Introducción

El uso de fotoiniciadores en la cara externa de los envases de alimentos

puede provocar la contaminación de éstos. De hecho, diversos fotoiniciadores

han sido detectados en leche infantil, yogur, té y zumos. Sus efectos en la salud

humana son desconocidos; sin embargo, son compuestos indeseables cuya

presencia en alimentos debe ser controlada.

Los métodos de preparación de muestra desarrollados hasta el momento

requieren elevados tiempos de análisis, consumo de grandes volúmenes de

disolvente y exhaustivos procesos de concentración, con los posibles riesgos de

pérdida de analitos. Por ello, es necesario el desarrollo de nuevas metodologías

que minimicen la manipulación de la muestra, el empleo de disolventes

orgánicos, el coste y el tiempo de análisis. En esta Tesis, se estudió la

posibilidad de emplear la SPME como técnica de extracción y concentración, en

una única etapa, para la determinación de varios fotoiniciadores en muestras de

leche distribuidas en envases del tipo brick [Negreira, 2010-B]. La elección de la

matriz se ha realizado en base a las alertas sanitarias y a los datos relativos a la

presencia de varios compuestos orgánicos, usados como fotoiniciadores en la

polimerización de tintas, en muestras de leche. También se ha tenido en cuenta

la importancia y el elevado consumo de este alimento, en especial, por parte de

la población infantil. Además de la optimización de las condiciones de

extracción, se ha evaluado la influencia del contenido graso de las muestras en

la eficacia del proceso de concentración.


308

1.2. Esquema del método desarrollado para fotoiniciadores en

alimentos

Muestra leche 1,5 mL

SPME:Inmersión, fibra PDMS-DVB, 100ºC, 40 min con agitación

Retirada de la fibra, secar con papel

Adicción de agua milli-Q (8,5 mL)

+

Desorción de la fibra 2 min a 270 ºC

GC-MS

Figura 23: Esquema de trabajo seguido para la determinación de fotoiniciadores en leche

mediante SPME y GC-MS.

1.3. Publicación:

SOLID-PHASE MICROEXTRACTION FOLLOWED BY

GAS-CHROMATOGRAPHY MASS SPECTROMETRY

FOR THE DETERMINATION OF

INK PHOTO-INITIATORS IN PACKED MILK


Talanta 82 (2010) 296

(doi:10.1016/j.talanta.2010.04.037)

Talanta 82 (2010) 296

311

Solid-phase microextraction followed by gas chromatography mass

spectrometry for the determination of ink photo-initiators in packed milk





Abstract

A novel, single step method for the determination of seven ink photo-

initiators in carton packed milk samples is described. Solid-phase

microextraction (SPME) and gas chromatography (GC), combined with mass

spectrometry (MS), were used as sample preparation and determination

techniques, respectively. Parameters affecting the performance of the

microextraction process were thoroughly evaluated using uni- and multivariate

optimization strategies, based on the use of experimental factorial designs. The

coating of the SPME fibre, together with the sampling mode and the

temperature were the factors playing a major influence on the efficiency of the

extraction. Under final conditions, 1.5 mL of milk and 8.5 mL of ultrapure water

were poured in a glass vessel, which was closed and immersed in a water

boiling bath. A poly(dimethylsiloxane)-divinylbenzene (PDMS-DVB) coated

fibre was exposed directly to the diluted sample for 40 min. After that, the fibre

was desorbed in the injector of the GC-MS system for 3 min. The optimized

method provided limits of quantification (LOQs) between 0.2 and 1 g L-1 and a

good linearity in the range between 1 and 250 g L-1. The inter-day precision

remained below 15% for all compounds in spiked whole milk. The efficiency of

the extraction changed for whole, semi-skimmed and skimmed milk; however,

no differences were noticed among the relative recoveries achieved for milk

samples, from different brands, with the same fat content.


312

Keywords: ink photo-initiators, packed food, milk, solid-phase

microextraction, gas chromatography, mass spectrometry.

1. Introduction

Photo-initiators are low molecular weight compounds added to some

printing inks. Under ultraviolet (UV) irradiation these substances are

decomposed generating reactive free radicals, which activate the

polymerization of ink components allowing a fast drying of the printing film.

Generally, photo-initiators present one or two aromatic rings in their molecules.

Some of them, case of benzophenone and amino benzoate derivatives, show

similar chemical structures to those compounds used as UV-filters in

sunscreens and personal care products [1].

The use of ink photo-initiators in the external face of multilayered

packaging cartons can lead to their occurrence in food. In 2005, the European

Food Safety Authority (EFSA) reported the presence of 2-

isopropylthioxanthone (ITX) in several liquid foods, particularly packed milk,

and solid infant formula [2]. Further studies confirmed the presence of ITX not

only in milk [3-5], but also in yoghurts [6], fruit juices [7] and even wine [8], at

concentrations up to several hundreds of g L-1. In addition, other photo-

initiators such as 2-ethylhexyl-4-dimethylaminobenzoate (EHPABA) [4,8] and

benzophenone [8] have been also found in carton packed foods and beverages.

Migration through multilayer materials and/or contamination of the inner face

during storage of rolled bobbins of printing packages may lead to the presence

of ink photo-initiators in food [4,9,10]. Potential long-term effects of photo-

initiators exposure on human health remain unknown; however, they are

considered as undesirable compounds, whose presence in packed foodstuff has

to be controlled. Milk is a particularly concerning matrix, since it is considered a

basic, worldwide consumed nourishment.

Talanta 82 (2010) 296

313

Gas chromatography (GC) and liquid chromatography (LC) based

techniques are often applied to the determination of ink photo-initiators in

packed food. In most cases, they are combined with single or tandem mass

spectrometry (MS) detection [4,8,11]. Moreover, the complexity of foodstuff

matrices makes necessary a previous step to isolate target compounds from the

rest of constituents. Liquid-liquid extraction (LLE), using acetonitrile as

extractant, is one of the most popular approaches for the extraction of photo-

initiators (particularly ITX) from liquid and powdered milk, as well as from

other fatty samples [4-6, 9]. Acetonitrile is compatible with the use of LC, in the

reversed-phase mode, as separation technique and provides extracts with a low

level of lipids. N-hexane extraction followed by a further clean-up using a silica

cartridge has also been reported [8]. Solid-phase extraction (SPE) of diluted

milk samples [3], or of the primary LLE extract from the same matrix [7,12],

allows a reduction in the consumption or organic solvents and a further

improvement in the selectivity of the sample preparation process, respectively.

Solid-phase microextraction (SPME) is a valuable, solvent-free alternative

for the extraction and concentration of organic compounds from different

matrices. SPME integrates extraction and concentration in the same step;

therefore, it competes with multistep strategies in terms of cost and, many

times, SPME provides a higher selectivity since it is based on equilibrium

processes rather than in exhaustive extractions. Liquid foodstuffs, such as milk,

constitute complex matrices, limiting the yield of SPME extractions in

comparison with water samples. In spite of this, several authors have

demonstrated the suitability of SPME for the determination of g per L levels of

different organic compounds in milk [13-16]. When sample pre-treatment

and/or headspace sampling are not possible, dilution of the matrix with water

is a straightforward solution to preserve the integrity of the SPME coating and

to limit the co-extraction of interfering compounds which might damage the GC

column [15-16].


314

The aim of this study is to develop a single step sample preparation

method, based on the SPME technique, for the determination of a group of

seven ink photo-initiators in packed milk samples. To the best of our

knowledge this is the first application of SPME to the determination of this

family of compounds in milk. Parameters affecting the performance of the

extraction were systematically evaluated using univariant and also

experimental factorial designs studies. After extraction, fibres were thermally

desorbed and compounds determined by GC-MS, in the selected ion

monitoring (SIM) mode.

2. Experimental

2.1. Solvents, standards and SPME equipment

Methanol, acetonitrile (HPLC-grade) and ethyl acetate (trace analysis

grade) were obtained from Merck (Darmstadt, Germany). Sodium chloride was

provided by Aldrich (Milwaukee, WI, USA). Ultrapure water was obtained

from a Milli-Q system (Millipore, Billerica, MA, USA). Standards of

benzophenone (BP), 1-hydroxycyclohexyl-phenylketone (CPK), ethyl-4-

dimethylaminobenzoate (EDMAB), 4-methylbenzophenone (4-MBP), 2,2-

dimethoxy-2-phenylacetophenone (2,2-DMPA), EHPABA and ITX were

acquired from Aldrich. Their chemical structures and some properties of

relevance to predict their behaviour during extraction are summarized in Table

1. In general, target compounds show medium to low polarities and those with

ionisable moieties (CPK, EDMAB and EHPABA) remain in the neutral form at

the pH of milk (6.6-6.8 units). Individual solutions of each species were

prepared in methanol, further dilutions and mixtures of them were also made

in methanol, when used to fortify milk samples, and in ethyl acetate when

considered to optimize GC-MS determination conditions.

Talanta 82 (2010) 296

315

Table 1. Abbreviated names, CAS numbers, structures, log Kow and vapour pressure

(Pv) values of target species.

AAbbbbrreevviiaattiioonn NNaammee CCAASS

nnuummbbeerr SSttrruuccttuurree

aaLLoogg

KKooww

aaPPvv

((mmTToorrrr))

BP Benzophenone 119-61-

9

3.18 0.82

CPK 1-Hydroxycyclohexyl-

phenylketone

947-19-

3

2.34 0.037

EDMAB Ethyl-4-

dimethylaminobenzoate

10287-

53-3

3.14 1.43

4-MBP 4-Methylbenzophenone 134-84-

9

3.64 0.19

2,2-DMPA 2,2-dimethoxy-2-

phenylacetophenone

24650-

42-8

4.76 0.011

EHPABA

2-Ethylhexyl-4-

dimethylaminobenzoate

21245-

02-3

6.15 0.0046

ITX 2-

Isopropylthioxanthone

5495-

84-1

5.33 0.0014

a Values obtained from SciFinder Scholar Database, http://www.cas.org/products/sfacad/

A manual SPME holder and fibres coated with different polymers:

poly(dimethylsiloxane) (PDMS, 100 m film thickness), polyacrylate (PA, 85

m film thickness), Carboxen-PDMS (CAR-PDMS, 75 m film thickness),

PDMS-divinylbenzene (PDMS-DVB, 65 m film thickness) and DVB-CAR-

PDMS (50/30 m film thickness) were obtained from Supelco (Bellefonte, PA,

USA). Before being used for first time, SPME fibres were thermally conditioned

following conditions recommended by the supplier.


316

2.2. Samples and SPME procedure

The whole (3.6% fat), half-skimmed (1.55% fat) and skimmed (0.30% fat)

milk samples were bought in local supermarkets. All samples were

commercialized in multilayered Tetra Pak or Combibloc type carton packages.

Optimization of SPME conditions was carried out with spiked aliquots (100 g

L-1) of whole milk. The percentage of methanol in this matrix was maintained at

1%. In further experiments, samples with different fat contents were spiked at

increased levels in the range between 1 and 250 g L-1. Spiked samples were

thoroughly homogenized and stored overnight at 4 ºC to simulate the

interactions between analytes and matrix occurring in polluted samples.

SPME experiments were carried out in 10 and 22 mL glass vials

furnished with a PTFE-faced septum and an aluminium crimp cap. A given

volume of milk (from 1.5 to 3 mL) and the corresponding amount of ultrapure

water were poured in the vessels, which contained a magnetic stir bar (10 mm x

4 mm). After being closed, vessels were stirred for 5 min and then stabilized at

the selected temperature for the same period. Then, a SPME fibre was exposed

to the headspace (HS) of the vial or dipped directly into the liquid matrix for a

pre-established period. In some experiments, sodium chloride (NaCl) was also

added to the SPME vessel in order to assess the effect of the ionic strength on

the yield of the extraction. Under optimized conditions, extractions were carried

out in 10 mL (nominal volume) vessels containing 1.5 mL of milk and 8.5 mL of

ultrapure water, without addition of NaCl. A PDMS-DVB fibre was exposed

directly to the stirred sample (700 rpm), previously thermostated at 100 ºC, for

40 min. After this time, the fibre was retracted into the SPME holder. Drops of

sample attached to the outlet surface of the metallic needle were removed with

a soft paper tissue and the fibre was desorbed for 3 min at 270 ºC in the injector

of the GC-MS system.

Talanta 82 (2010) 296

317

2.3. Carton packages

Additionally to the optimization of SPME conditions for milk samples,

ink photo-initiators were also investigated in a limited number of carton

packages. Samples were extracted using the method previously developed and

validated by Sanches-Silva and co-workers [17]. In brief, packages were opened,

the internal side was rinsed with ultrapure water and then, they were cut in

small pieces (around 2 mm x 2 mm). One gram of the above matrix was

accurately weighed and extracted with 10 mL of acetonitrile, at 70 ºC for 24 h.

After filtration, the supernatant solution was concentrated to 2 mL, using a

gentle stream of nitrogen, and injected directly in the GC-MS system.

2.4. Determination

Analytes were determined using a GC-MS system consisting of an

Agilent (Wilmington, DE, USA) 7890A gas chromatograph connected to a

quadrupole type mass spectrometer (Agilent MS 5975C), furnished with an

electron-impact (EI) ionization source. The mass analyzer was operated in the

selected ion monitoring (SIM) mode. Separations were carried out in a HP-5ms

type capillary column (30 m x 0.25 mm i.d., df: 0.25 m) supplied by Agilent.

Helium (99.999 %) was used as carrier gas at a constant flow of 1.0 mL min-1.

The GC oven was programmed as follows: 70 ºC (held for 3 min), at 10 ºC min-1

to 280 ºC (held for 10 min). Ionisation source, mass analyzer and transfer line

temperatures were set at 230, 150 and 290 ºC, respectively. Standards prepared

in ethyl acetate were injected in the splitless mode (splitless time 3 min), with

the injector port at 280 ºC. SPME fibres were desorbed at 270 ºC, case of PDMS-

DVB, PDMS and DVB-CAR-PDMS, or 290 ºC for PA and CAR-PDMS. A

desorption step of 3 min, maintaining the injector in the splitless mode during

this time, was used in all cases. Retention times and m/z ratios of ions used to

monitor the signal of each compound are summarized in Table 2.


318

Table 2. Performance of the GC-MS quadrupole system for direct injection (2 L) of

photo-initiators standards prepared in ethyl acetate.

Compound Ret.time

(min)

aSelected

ions (m/z)

Linearity,

R2

(10-2000

g L-1)

bRepeatability

(RSDs, %),

50 g L-1

cReproducibility

(RSD, %),

50 g L-1

LOQs

(g L-1)

BP

CPK

EDMAB

4-MBP

2,2-DMPA

EHPABA

ITX

14.98

15.67

15.86

16.41

17.71

21.24

22.29

105,77,182

99,81

148,193,164

119,196

151,105

165,277

239,254

0.9995

0.9995

0.9993

0.9993

0.9999

0.9999

0.9997

1.5

0.6

2.6

5.3

2.0

3.2

2.8

3.4

5.4

2.8

3.4

1.6

4.8

3.6

1.6

2.3

0.7

1.9

0.7

1.1

3.5

a Underlined ions were used for quantification purposes. b data for n=3 consecutive replicates. c data for n=9 replicate injections in 3 different days.

An ion-trap type Varian (Walnut Creek, CA, USA) 240 mass

spectrometer (MS), furnished with an EI source and connected to 450 model GC

instrument, from the same supplier, was also used to confirm the presence of

photo-initiators in some milk cartons. The system was also equipped with a

Factor Four (Varian) BP-5 type capillary column (30 m x 0.25 mm i.d., df: 0.25

m). Injector, transfer line temperatures and rest of chromatographic conditions

were the same as those reported in the above paragraph for the quadrupole GC-

MS instrument; however, the splitless time was reduced to 1 min. Source and

trap temperatures were set at 200 and 150 ºC, respectively. MS spectra were

recorded in the range from 50 to 400 m/z.


3.1. GC-MS determination conditions

Table 2 summarizes some relevant features of the GC-MS (quadrupole)

system for the determination of ink photo-initiators. Under conditions reported

in the experimental section, all compounds were baseline separated showing

retention times comprised between 15 and 23 min. The plots of peak area,

Talanta 82 (2010) 296

319

corresponding to quantification ions (see Table 2), versus concentration fitted a

linear model with determination coefficients (R2) higher than 0.999, within the

interval between 10 and 2000 g L-1. Limits of quantification (LOQs), defined as

the concentration of each species producing a signal 10 times higher than the

baseline noise, remained between 0.7 and 3.5 ng mL-1. Relative standard

deviations (RSDs) corresponding to injections performed within the same day

and in consecutives days stayed below 6%.

3.2. Optimization of SPME parameters

3.2.1. Preliminary experiments

In the initial steps of this study, microextraction experiments were

carried out in 10 mL volume vessels, which contained 2 mL of spiked (100 g L-

1) whole milk plus 8 mL of ultrapure water. Vessels were equilibrated at 100 ºC

and fibres were exposed directed to the diluted samples for 25 min. After that,

they were desorbed using conditions provided in the experimental section. Fig.

1 depicts the responses (peak areas) obtained for triplicate assays using three

SPME coatings. With the only exception of EHPABA, the most hydrophobic of

the considered species, the PDMS fibre provided much lower responses than

PDMS-DVB and PA ones. The latter two coatings showed similar extraction

efficiencies for EHPABA and ITX, whereas PDMS-DVB was preferred for the

rest of photo-initiators. CAR-PDMS and DVB-CAR-PDMS fibres showed a very

low affinity for ITX, as well as a poor repeatability for the rest of species (data

not shown). Thus, PDMS-DVB and PA were selected for additional

experiments. Carry-over effects were evaluated by desorbing each fibre twice at

270 ºC (PDMS-DVB) and 290 ºC (PA). Relative responses in the second

desorption remained below 0.2% of those observed in the first one. In order to

eliminate any risk of cross contamination, they were additionally desorbed (3

min), at the above temperatures, in the split injector of a non operative GC

instrument under a nitrogen stream of 30 mL min-1.


320

0

1

2

3

BP CPK EDMAB 4-MBP 2,2-DMPA EHPABA ITX

Pea

k ar

ea

PDMS-DVB PDMS PA

(x 105)

Fig. 1. Responses obtained with different fibres for spiked whole milk samples, n=3

replicates. Direct sampling at 100 ºC for 25 min.

3.2.2. Multifactor optimization of SPME conditions

Efficiency of SPME methods is affected by a considerable number of

factors, which are sometimes correlated. A strategy based on the use of

experimental factorial designs was adopted to identify those parameters

playing a major effect on the performance of the SPME process, and to achieve

optimal conditions with a minimum effort and cost.

Initially, a two levels 25-1 type fractional factorial design was used to

assess the effects of temperature, sampling mode, ionic strength, dilution factor

and fibre coating on the efficiency of the extraction. Low and high values for

each of these parameters are given Table 3. Previous assays showed poor

efficiencies operating at room temperature; therefore, the domain of this factor

was established between 55 and 100 ºC. The dilution ratio was chosen according

to the information reported in a previously work, dealing with the application

of SPME to the determination of pesticides in cow milk [16]. The volume of

milk used in each experiment was 1.5 or 3 mL, depending on the dilution ratio,

and the total volume in the SPME vessel was made up to 15 mL with ultrapure

water in all cases. Extractions were performed in 22 mL vials to allow working

Talanta 82 (2010) 296

321

in direct and HS modes, depending on the conditions defined by the

experimental design. Peaks areas obtained for each compound in the 16

extractions involved in the above design were used as variable responses.

Standardized values for main effects corresponding to each factor were

calculated with the Statgraphics Centurion XV software (Manugistics,

Rockville, MD, USA). Their graphical representation is summarized in the main

effect plots grouped in Fig. 2. The length of depicted lines is proportional to the

variation in the response of the investigated species when a given factor

changes from the low (-) to the high (+) level, in the domain of the design. A

positive slope indicates an improvement in the efficiency of the extraction and a

negative one the opposite effect.

Table 3. Experimental domain of the 25-1 fractional design.

Factor Level

Low (-) High (+)

Fibre

Temperature (ºC)

Sampling mode

NaCl (%)

Dilution factor

PDMS-DVB

55

Direct

0

1:5

PA

100

HS

15

1:10


322

Temperature Salt

EDMAB

31

41

51

61

71

81(x 103)

Fibre Mode Dilution

-

+ -

+-

+

- + -+

Pea

k ar

ea

Temperature Salt

BP

93

113

133

153

173

(x 103)

Fibre Mode Dilution

-

+ -

+

- + - + -+P

eak

area

Temperature Salt

2,2-DMPA

43

63

83

103

123(x 103)

Fibre Mode Dilution

-

+ -

+-

+

- + -+P

eak

area

Temperature Salt

EHPABA

02

4

68

1012

(x 103)

Fibre Mode Dilution

- +

-

+-

+

-

+

-+

Pea

k ar

ea

Temperature Salt

ITX

0

3

6

9

12

15(x 103)

Fibre Mode Dilution

- +

-

+ -

+

-

+

-

+Pea

k ar

ea

Temperature Salt

4-MBP

46

66

86

106

126

(x 103)

Fibre Mode Dilution

-

+ -

+

-+

-

+

-

+Pea

k ar

ea

Temperature Salt

CPK

28

38

48

58

68

(x 103)

Fibre Mode Dilution

-

+-

+-

+-

+-

+Pea

k ar

ea

Fig. 2. Main effect plots corresponding to the 25-1 fractional design.

Again, similar responses were attained for EHPABA and ITX with both

fibres, whereas for the rest of species the PDMS-DVB coating provided higher

yields. Direct exposure of the fibre to the diluted samples was preferred to HS

Talanta 82 (2010) 296

323

sampling and higher extraction efficiencies were attained at 100 ºC versus 55 ºC.

The effect of the ionic strength was compound dependent and the dilution

factor affected negatively to the obtained responses. Considering these features,

direct sampling using a PDMS-DVB fibre was fixed for further experiments.

Extractions were carried out in 10 mL vessels containing 1.5 mL of milk plus 8.5

mL of water. This dilution factor (ca. 6.6 times) constitutes a reasonable

compromise between extraction efficiency and stability of the SPME coating,

avoiding the existence of free HS in the vessel.

Optimal values corresponding to temperature, sodium chloride and

extraction time were evaluated with more detail using a 31 x 22 type factorial

design, with three replicates of the central point. The domain of this second

design, the standardized values of main effects associated to each factor and

some relevant two-factor interactions are compiled in Table 4. The absolute

values of the main effects are proportional to the variation in the efficiency of

the extraction when the considered factor changes from the low to the high

level. A positive sign points out to an increase in the yield of the process and a

negative one indicates the opposite behaviour. The temperature was the most

important of the considered factors, with a positive and statistically significant

effect (95% confidence level) in the efficiency of the extraction for all

compounds. In the case of ITX and EHPABA, the quadratic term associated to

this factor (AA) was also significant, Table 4. The corresponding main effect

plot suggested an exponential increase in the efficiency of the SPME with the

temperature of the sample for both species, figure not shown. The sampling

time also affected positively to the extraction process, although it overpass the

level of statistically significance only for EHPABA and ITX. Finally, sodium

chloride exerted a negative effect on the extraction and it was statistically

significant for 4-MBP, EHPABA and ITX. The exception to the above pattern

corresponded to CPK, which is the most polar of the considered compounds.

Consequently, the thermodynamic of CPK extraction increases significantly

with the ionic strength of the solution, whereas the kinetics of its migration


324

from the bulk of the solution to the surface of the SPME coating is reduced in a

minor extension than for the more hydrophobic compounds [18]. The

interaction temperature-salt (AB) was also statistically significant for CPK,

EHPABA and ITX. The corresponding interaction plots (figure not given)

demonstrated that, for EHPABA and ITX, the negative effect of salt was more

relevant at 100 than at 60 ºC. Taking these comments into account, further

extractions were performed at 100 ºC, without addition of sodium chloride to

the SPME vessel.

Talanta 82 (2010) 296

325

Tab

le 4

. Exp

erim

enta

l dom

ain,

sta

ndar

dize

d m

ain

effe

cts

and

rele

vant

inte

ract

ions

of f

acto

rs in

volv

ed in

the

31

x 22

exp

erim

enta

l

desi

gn. Fa

ctor

C

ode

Lev

el

Stan

dar

diz

ed v

alu

e

Low

M

ediu

m

Hig

h B

P

CP

K

ED

MA

B

4-M

BP

2,2-

DM

PA

E

HP

AB

A

ITX

Tem

per

atu

re (º

C)

A

60

80

100

6.3*

4.

6*

6.7*

8.

2*

6.6*

18

* 21

*

NaC

l (%

) B

0

- 10

-2

.2

0.26

-2

.3

-2.6

* -2

.3

-10*

-9

.0*

Tim

e (m

in)

C

20

- 40

1.

5 1.

0 1.

8 1.

5 1.

7 3.

4*

4.5*

A

A

-

- -0

.32

-0.5

5 0.

20

1.24

1.

23

7.86

* 8.

81*

A

B

-

- 1.

52

3.11

* 1.

58

0.69

0.

77

-8.0

2*

-5.9

6*

*Sig

nifi

cant

fact

ors

at th

e 95

% c

onfi

den

ce le

vel.


326

A detailed study of the extraction kinetics showed that EHPABA and ITX

required more than 2 hours of direct sampling at 100 ºC to achieve equilibrium

conditions, Fig. 3. For the rest of species, the efficiency of the extraction

normally reached a maximum between 30 and 45 min and then it decreased

slightly, Fig. 3. This trend might be the result of competitive adsorption

processes on the surface of the PDMS-DVB coating. An exposure time of 40 min

was adopted.

0

1

2

3

4

5

0 20 40 60 80 100 120

Pe

ak

are

a

Time (min)

BP

CPK

EDMAB

(x 106)

0

1

2

3

4

5

6

7

0 20 40 60 80 100 120

Pe

ak

are

a

Time (min)

4-MBP

2,2-DMPA

(x 106)

0

1

2

3

0 20 40 60 80 100 120

Pe

ak

are

a

Time (min)

EHPABA

ITX

(x 106)

Fig. 3. Kinetics of the SPME for whole milk. Direct sampling at 100 ºC using a PDMS-

DVB fibre.

Talanta 82 (2010) 296

327

3.2.3. Stirring and methanol addition

Agitation is expected to increase the kinetics of the extraction, improving

the transport of the compounds between the liquid sample and the interface

with the SPME coating. On the other hand, PTFE covered stirrers are a potential

source of cross contamination problems. Experiment data (Fig. 4) demonstrated

that stirring improved significantly the yield of the extraction for all

compounds except CPK. As previously commented, this is the most polar of the

species involved in this study; therefore, it is expected to be that showing the

higher diffusion rates towards the interface between the solution and the SPME

fibre and thus the less affected by stirring. This type of dependence between the

efficiency of stirring and the polarity of analytes has been previously reported

in the literature [19]. In order to avoid the risk of cross contamination problems,

stir bars were wrapped with PTFE tape, which was removed after each

extraction.

0%

20%

40%

60%

80%

100%

120%


Rel

ativ

e re

spo

nses

With stirring Without stirring

Fig. 4. Effect of stirring (700 rpm) on the relative efficiency of the SPME. Sampling

time 40 min, n= 3 replicates.

Addition of an organic solvent to the SPME vessel contributes to reduce

competitive adsorptions of hydrophobic species on the walls of glass vessels,

improving the efficiency of microextraction for these compounds. On the other

hand, the yield of the process decreases for hydrophilic species, which turn out


328

more soluble in the sample. During optimization of SPME conditions, fortified

aliquots of milk containing a 1% of methanol were employed. After dilution (1.5

mL to 10 mL), the percentage of methanol in the sample was 0.15%. A series of

extractions was performed using samples containing also 1% and 3% of

methanol. Except in the case of EHPABA and ITX, the efficiency of the

extraction was negatively affected by the addition of methanol, figure not

shown. To limit the contribution of this parameter to the variability of the

extraction, the percentage of methanol in the sampling vessel was always

maintained below 0.5%.

3.2.4. Fat content

In general, the performance of microextraction techniques is affected by

the characteristics of the sample. As a general rule, the higher is the complexity

of the matrix, the lower the yield of the extraction. The complexity of milk

samples is mainly related with their lipidic content. Fig. 5 depicts the responses

obtained for samples of whole, half-skimmed and skimmed cow milk spiked

with target photo-initiators at 50 g L-1. Their declared fat percentages were

3.6%, 1.55% and 0.3%, respectively. The yield of the extraction was inversely

proportional to the fat content in the sample.

0

1

2

3


Pea

k ar

ea

full-fat half-skimmed skimmed(x 105)

Fig. 5. Comparison of peak areas for spiked (50 g L-1) milk samples with different fat

contents, n= 3 replicates.

Talanta 82 (2010) 296

329

3.3. Analytical figures of merit

The developed method was characterized in terms of linearity,

repeatability and reproducibility using aliquots of whole milk fortified at

different concentrations in the range between 1 and 250 g L-1. Plots of peak

areas versus added concentrations fitted a linear model with determination

coefficients higher than 0.993, Table 5. RSDs of extractions carried out in the

same day (repeatability study) for samples spiked at three different

concentrations (between 5 and 50 g L-1), remained below 7%, only slightly

higher than the repeatability of the GC-MS system for direct injection of

standards in ethyl acetate, see Table 2. RSDs for extractions performed during 3

consecutive days varied between 8 and 15%, Table 5. LOQs of the proposed

method, defined as the concentration of each specie providing a

chromatographic peak with a signal to noise ratio (S/N) of 10, were estimated

from the lowest addition level in the linearity study. Obtained values varied

between 0.2 g L-1 for BP and 1 g L-1 for EDMAB, Table 5. In the case of ITX,

the photo-initiator more often investigated in packed milk, a LOQ of 0.4 g L-1

was achieved. Globally, these values are similar to those attained using LLE,

with additional SPE purification, followed by GC-MS [8] or LC-MS/MS [12];

they are also equivalent to LOQs reported for ITX using SPE with GC-MS/MS

determination [3], and LLE followed by LC-MS/MS [6]. Gallart-Ayala and co-

workers [11] achieved a 100-fold reduction in the LOQ of ITX at the expense of

using a last generation LC-MS/MS instrument providing accurate mass

measurements [11]. Taking into account the trend depicted in Fig. 5, LOQs

reported in Table 5 can be further reduced for semi-skimmed and skimmed

milk. The only exception was BP. For this compound the lower attainable LOQ

was limited by the presence of this specie in procedural blanks corresponding

to the extraction of ultrapure water.


330

Table 5. Linearity, repeatability (n= 3 replicates), reproducibility (n= 9 replicates in 3

different days) and LOQs of the proposed method for full-fat milk samples.

Compound Linearity (R2)

1-250 g L-1 a

Repeatability (RSDs, %) Reproducibility

(RSDs, %),

16 g L-1 a

LOQs

g L-1 5 g L-1 a 25 g L-1 a 50 g L-1 a

BP

CPK

EDMAB

4-MBP

2,2-DMPA

EHPABA

ITX

0.9962

0.9966

0.9964

0.9969

0.9977

0.9931

0.9969

7.0

4.7

0.6

5.0

6.5

1.0

5.4

5.5

5.3

4.8

3.5

1.9

4.1

4.7

5.1

1.3

4.4

3.3

4.2

3.4

2.4

8.2

8.3

9.5

11.9

15.1

9.6

9.4

0.2

0.6

1

0.4

0.3

0.8

0.4


Accuracy is a major issue during the validation of any analytical

procedure. Microextraction methodologies are prone to variations in their

efficiency depending on the characteristics of the matrix as it has been already

proved for whole, semi-skimmed and skimmed milk samples, Fig. 5. A further

series of assays was carried out to investigate whether the yield of the SPME

method varies also among samples, with the same fat content, from different

brands, or not. Aliquots corresponding to specimens of whole, semi-skimmed

and skimmed milk, from three different suppliers, were fortified at

concentrations in the range from 10 to 20 g L-1. Non-spiked aliquots of each

specimen were also processed, Fig. 6. One of the specimens was used as

reference and the responses measured for the other two normalized to the first.

Significant differences were not observed among milk samples with the same

fat content, Table 6. Thus, levels of ink photo-initiators in unknown samples can

be quantified by external calibration, using matrix-matched standards. The only

requirement is that both, unknown samples and calibration standards, present

the same fat content.

Talanta 82 (2010) 296

331

21.0 21.1 21.2 21.3 21.460

100

140

180

220

260

300

340

380

Time (min)

Abu

nd

ance

EHPABAm/z 277

14.6 14.8 15.0 15.20

1

2

3

4

5

Time (min)

Abu

nd

ance

(x 104)

BPm/z 105

(x 103)

22.0 22.2 22.40

1

2

3

4

5

Time (min)

Abu

nd

ance

ITXm/z 254

17.5 17.6 17.7 17.8

1

2

3

4

(x 104)

Time (min)

Abu

nd

ance

0

2,2-DMPAm/z 151

15.5

(x 103)

15.8 15.90

123456789

10111213

Time (min)

Abu

nd

ance

EDMABm/z 148

16.015.7 16.30 16.40 16.50

1

2

3

4

(x 104)

Time (min)

Abu

nd

ance

0

4-MBPm/z 119

16.60

15.6 15.7 15.815.4

2

4

6

8

10

12

14

16

18

20

(x 103)

Abu

nd

ance

Time (min)

CPKm/z 99

Fig. 6. GC-MS chromatograms for half-skimmed milk. Dotted line, un-spiked sample.

Solid line, same matrix fortified at 10 g L-1.


332

Table 6. Relative recoveries provided by the SPME method for milk samples with

different fat contents, n= 3 replicates.

Compound

Relative recoveries (%) ± SD

a Skimmed a Semi-skimmed b Whole

Brand 1 Brand 2 Brand 1 Brand 2 Brand1 Brand 2

BP

CPK

EDMAB

4-MBP

2,2-DMPA

EHPABA

ITX

101±1

94±1

100±4

100±3

101±3

108±3

101±6

101±12

95±11

94±8

104±2

98±4

100±0.2

106±7

102±4

97±1

106±3

97±2

100±2

106±9

100±8

98±7

94±3

94±7

93±7

98±4

101±6

97±3

103±2

101±1

101±5

101±5

105±3

89±5

98±2

94±9

98±3

99±6

100±13

98±14

92±4

96±1

a Added concentration 10 g L-1.

b Added concentration 20 g L-1.

3.4. Real samples

The presence of ink photo-initiators was investigated in a total of 20

samples corresponding to whole, semi-skimmed and skimmed milk. Around

60% of them were distributed in Tetra Pak cartons and the rest in Combibloc ones.

All species remained under the LOQs of the method, although BP was detected,

at concentrations below the LOQ of this species, in some of them. In order to

establish if these results indicate the phase-out of ink photo-initiators in milk

cartons, or simply the lack of significant migration problems, six packages

(Tetra Pak and Combibloc types) were extracted under conditions reported in the

experimental section. A peak at the retention time of BP was found in the

chromatograms provided by the quadrupole GC-MS instrument for the six

samples, whereas the rest of photo-initiators were not noticed. The identity of

this compound was confirmed with the ion-trap GC-MS system, Fig. 7. A

detailed quantification of BP levels in milk packages was beyond the scope of

this research. A rough estimation points out to concentrations below 1 g g-1 in

the six processed samples.

Talanta 82 (2010) 296

333

12.75 13.00 13.25 13.50 13.75 minutes

0.0

0.5

1.0

1.5

2.0

2.5

3.0

kCounts

m/z 105

50 100 150 200 250 300 350 m/z

0%

25%

50%

75%

100%

51

77105

182

50 100 150 200 250 300 350 m/z

0%

25%

50%

75%

100%

51

77

105

182

A

B

BPNIST spectra

Fig. 7. GC-MS (ion-trap) chromatograms and spectra showing the signal of BP in the

extract from a Tetra Pak package. A, procedural blank. B, carton extract.

4. Conclusions

The suitability of SPME for the extraction of seven ink photo-initiators in

packed milk has been demonstrated for first time. Its major advantages over

previously published methods are integration of extraction and concentration in

the same step and null consumption of organic solvents. Although the yield of

the microextraction is conditioned by the fat content in the sample, no

differences were noticed among milk specimens from different brands, with the

same nominal fat content. When combined with GC-MS detection, the

developed procedure provides adequate figures of merit to screen the potential

contamination of milk with photo-initiators used during the printing of


334

packaging materials. Although this possibility has not been explored in this

study, it is expected that the proposed method can be also adapted to other

liquid foodstuffs, such as fruit juices and wine. Target compounds could not be

quantified in any of the processed milk samples; however, BP was found in

carton packages.

Acknowledgments

This study has been supported by the Spanish Government and E.U.

FEDER funds (project CTQ2009-08377). N.N. thanks a FPU contract to the

Spanish Ministry of Science and Innovation.

Talanta 82 (2010) 296

335

References


[2] R. Anton, S. Barlow, D. Boskou, L. Castle, R. Crebelli, W. Dekant, K.H. Engel, S. Forsythe, W.

Grunow, M. Heinonen, J.C. Larsen, C. Leclercq, W. Mennes, M.R. Milana, I. Pratt, I. Rietjens,

K. Svensson, P. Tobback, F. Toldrá, EFSA J 293 (2005) 1.

[3] G. Allegrone, I. Tamaro, S. Spinardi, G. Grosa, J. Chromatogr. A 1214 (2008) 128.

[4] A. Gil-Vergara, C. Blasco, Y. Picó, Anal. Bioanal. Chem. 389 (2007) 605.

[5] R. Bagnati, G. Bianchi, E. Marangon, E. Zuccato, R. Fanelli, E. Davoli, Rapid Commun. Mass

Spectrom. 21 (2007) 1998.

[6] C. Benetti, R. Angeletti, G. Binato, A. Biancardi, G. Biancotto, Anal. Chim. Acta 617 (2008)

132.

[7] C. Sun, S.H. Chan, D. Lu, H.M.W. Lee, B.C. Bloodworth, J. Chromatogr. A 1143 (2007) 162.

[8] G. Sagratini, G. Caprioli, G. Cristialli, D. Giardiná, M. Ricciutelli, R. Volpini, Y. Zuo, S.

Vittori, J. Chromatogr. A 1194 (2008) 213.

[9] A. Sanches-Silva, S. Pastorelli, J.M. Cruz, C. Simoneau, I. Castanheira, P. Paseiro-Losada, J.

Agric. Food Chem. 56 (2008) 2722.

[10] G. Morlock, W. Schwack, Anal. Bioanal. Chem. 385 (2006) 586.

[11] H. Gallart-Ayala, E. Moyano, M.T. Galceran, J. Chromatogr. A 1208 (2008) 182.

[12] D. Shen, H. Lian, T. Ding, J. Xu, C. Shen, Anal. Bioanal. Chem. 395 (2009) 2359.

[13] M.J. González Rodríguez, F.J. Arrebola Liébanas, A. Garrido Frenich, J.L. Martínez Vidal,

F.J. Sánchez López, Anal. Bioanal. Chem. 382 (2005) 164.

[14] C. Mardones, D. Von Baer, J. Silva, M.J. Retamal, J. Chromatogr. A 1215 (2008) 1.

[15] N. Aguinaga, N. Campillo, P. Viñas, M. Hernández-Córdoba, Anal. Chim. Acta 596 (2007)

285.

[16] M. Fernandez-Alvarez, M. Llompart, J.P. Lamas, M. Lores, C. García-Jares, R. Cela, T.

Dagnac, Anal. Chim. Acta 617 (2008) 37.

[17] A. Sanches-Silva, S. Pastorelli, J.M. Cruz, C. Simoneau, P. Paseiro-Losada, J. Food Sci. 73

(2008) C92.

[18] J. Pawliszyn, Solid Phase Microextraction, Theory and Practice, Wiley-VCH, New York,

1997.

[19] P. Canosa, I. Rodríguez, E. Rubí, M.H. Bollaín, R. Cela, J. Chromatogr. A 1124 (2006) 3.

Conclusiones

339

IV. CONCLUSIONES

Tras la presentación y discusión de los resultados obtenidos es

conveniente remarcar las conclusiones más relevantes y revisar si se han

alcanzado los objetivos inicialmente planteados.

El objetivo fundamental a la hora de optimizar cada una de las

metodologías recogidas en esta Tesis ha sido el desarrollo de métodos sencillos,

rápidos y de bajo coste, que minimicen al máximo la generación de residuos y

proporcionen unos bajos límites de cuantificación, así como prestaciones

adecuadas en lo referente a su precisión y exactitud. Además, se ha trabajado

con matrices complejas, ej. polvo, agua residual, lodos y alimentos, intentando

combinar la etapa de extracción con la de clean-up. Finalmente, la aplicación de

dichas metodologías a muestras reales evidencia sus posibilidades y sus

limitaciones, además de aportar información relativa a la distribución y el

comportamiento de los analitos en las matrices consideradas.

A continuación se exponen las conclusiones más significativas de cada

uno de los estudios realizados:

1) “Dispersive liquid-liquid microextraction followed by gas

chromatography-mass spectrometry for the rapid and sensitive

determination of UV filters in environmental water samples”.

DLLME constituye una alternativa sensible, rápida y sencilla para la

concentración y extracción de 9 filtros solares (EHS, HMS, BzS, BP-3, 4-MBC,

IAMC, EHMC, EHPABA y OCR) de muestras de agua superficial y residual. El

método propuesto no requiere instrumentación dedicada, el volumen de

muestra es pequeño (10 mL), el tiempo de preparación de muestra es de tan

sólo 5 minutos y la fase extractante es compatible con el uso de cromatografía

de gases como técnica de determinación. La eficacia de extracción no se ve

afectada por la fuerza iónica, ni por el pH, ni por la naturaleza de la muestra,

Conclusiones

340

mientras que el volumen y el tipo de extractante y dispersante sí mostraron una

gran influencia. Por otro lado, las principales limitaciones del método

desarrollado son la dificultad para su automatización, así como el empleo de

disolventes halogenados en la etapa de extracción.

2) “Silicone discs as disposable enrichment probes for gas

chromatography-mass spectrometry determination of UV filters in

water samples"

El uso de absorbentes desechables de silicona resultó adecuado para la

extracción de filtros solares en muestras de agua superficial y residual,

proporcionando LOQs en la región de los bajos ng L-1, cuando se combina con

detección mediante GC-MS, empleando inyección de grandes volúmenes. Las

condiciones óptimas de extracción son similares a las publicadas en la

bibliografía para Twisters recubiertos de PDMS; además, las eficacias de

extracción son equivalentes en ambos casos. La principal limitación del método

es la lenta cinética del proceso de extracción; sin embargo, la utilización de una

placa agitadora multiposición permite procesar de manera simultánea, y con

una mínima atención por parte del operador, un número muy importante de

muestras.

La aplicación de las anteriores técnicas de microextracción a muestras de

agua superficial y residual ha puesto de manifiesto que los filtros UV más

frecuentemente detectados han sido OCR, EHMC y 4-MBC, mientras que

EHPABA y IAMC no han sido encontrados en prácticamente ninguna de las

muestras procesadas.

3) “Sensitive determination of salicylate and benzophenone type UV

filters in water samples using solid-phase microextraction,

derivatization and gas chromatography tandem mass spectrometry".

La combinación de la SPME con una etapa posterior de sililación on-fiber

hace posible la determinación de BP-1 y BP-8 mediante GC-MS, además de

Conclusiones

341

mejorar la forma de pico y los límites de cuantificación alcanzados para los

filtros UV: BP-3, EHS y HMS. Las condiciones óptimas de trabajo

correspondieron a muestreo directo, a temperatura ambiente, empleando una

fibra de PDMS-DVB y considerando MSTFA como agente sililante en la etapa

de derivatización. El método propuesto alcanzó LOQs en el nivel de los bajos

ng L-1, claramente inferiores a los descritos previamente en la bibliografía para

SPME, y una repetibilidad aceptable (RSD ≤ 14%). En el caso de muestras de

agua de río y residual tratada, la eficacia de la extracción fue similar a la

obtenida para agua ultrapura. El método, se aplicó a muestras

medioambientales encontrando concentraciones significativas de BP-3 y BP-1 en

aguas residuales.

4) “Solid-phase extraction followed by liquid chromatography

tandem mass spectrometry for the determination of hydroxylated

benzophenone UV absorbers in environmental water samples".

En este trabajo, se optimizaron las condiciones experimentales para la

determinación simultánea de 2 filtros solares (BP-3 y BP-4) y 4 benzofenonas

relacionadas (BP-1, BP-2, BP-6 y BP-8) mediante LC-MS/MS, empleando la SPE

como técnica de concentración. Las diferencias de polaridad y acidez entre la

BP-4 y el resto de benzofenonas han requerido una optimización exhaustiva

tanto de las condiciones de extracción en SPE, usando un adsorbente en fase

reversa (Oasis HLB, 60 mg), como de separación en la columna de LC. Además,

ha sido preciso combinar de modo simultáneo la ionización mediante ESI en

modo positivo y negativo, dado que, a diferencia de lo que ocurre con el resto

de compuestos, la BP-3 genera el ión precursor [M+H]+ con una eficacia muy

superior al [M-H]-.

Los LOQs alcanzados en este trabajo (desde 0,4 ng L-1 para agua de río

hasta 32 ng L-1 para agua residual sin tratar) permiten la determinación de las

seis benzofenonas consideradas en muestras de agua superficial y residual. La

BP-4 se cuantificó en todas las muestras analizadas, incluso en agua residual

Conclusiones

342

tratada, lo que indica que no es eliminada en las estaciones depuradoras. Este

hecho, añadido a su alta polaridad y elevada solubilidad en agua, indica que la

BP-4 podría alcanzar incluso los suministros de agua potable. Por su parte, la

BP-3 y la BP-1 también fueron detectadas en agua residual sin tratar y en río; sin

embargo, los datos obtenidos apuntan a una eliminación considerable de ambos

compuestos en las estaciones depuradoras de aguas residuales urbanas.

En futuros trabajos es necesario mejorar la selectividad de la etapa de

extracción, con objeto de reducir la influencia de la matriz en la eficacia de la

ionización mediante electrospray. Aprovechando el carácter ácido de estos

analitos parece factible que adsorbentes en modo mixto (ej. Oasis MAX)

permitan llevar a cabo un fraccionamiento de las benzofenonas y otros

compuestos con carácter neutro y básico.

5) "Study of some UV filters stability in chlorinated water and

identification of halogenated by-products by gas chromatography-

mass spectrometry".

En primer lugar, se desarrolló un método para la determinación de BP-3,

BP-1, EHPABA y EHS en muestras de agua usando GC-MS como técnica de

determinación, SPE en la etapa de concentración y derivatización de los

compuestos más polares.

BP-3, BP-1 y EHPABA presentan una elevada reactividad en muestras de

agua que contienen bajas concentraciones de cloro libre, similares a las

existentes en agua de grifo o de piscina. Por el contrario, el EHS presenta una

mayor estabilidad. En exceso de cloro libre, tanto BP-1 como BP-3 y EHPABA se

degradan siguiendo cinéticas de orden 1. La vida media de los compuestos es

función de variables tales como la concentración de cloro libre, el pH del agua y

la presencia de trazas de bromuro. Como productos de estas reacciones, se han

identificado compuestos halogenados (clorados y bromados) resultado de

reacciones de sustitución de hidrógenos por halógenos y, en el caso de la BP-3,

también productos halogenados donde no se conserva la estructura química de

Conclusiones

343

la benzofenona. Algunos de los productos de transformación son resistentes a

reacciones posteriores de oxidación, por lo que se podrían encontrar en matrices

medioambientales, ej. agua y sedimentos.

6) "Optimization of pressurized liquid extraction and purification

conditions for gas chromatography-mass spectrometry

determination of UV filters in sludge".

La extracción mediante PLE combinada con la utilización de carbón

grafitizado en la celda de PLE para la retención de pigmentos y una limpieza

adicional con un cartucho de PSA, constituye una mejora en términos de

eficacia de extracción y selectividad para la determinación mediante GC-MS de

un amplio número de filtros solares (EHS, HMS, BP-3, 4-MBC, IAMC, EHMC,

EHPABA y OCR) en muestras de lodo. Este estudio representa la primera

aplicación de ambos materiales para la purificación de extractos de PLE

correspondientes a muestras de lodo. Los factores con una mayor influencia

sobre el rendimiento global del proceso de preparación de muestra fueron los

disolventes empleados en la etapa de extracción y en la elución posterior de los

analitos del cartucho de PSA.

La aplicación del método a muestras de lodo confirmó la acumulación

significativa de tres filtros solares (4-MBC, EHMC y OCR) en esta matriz, con

concentraciones promedio superiores a 600 ng g-1.

7) "Determination of selected UV filters in indoor dust using matrix

solid-phase dispersion and gas chromatography tandem mass

spectrometry".

La combinación de MSPD con GC-MS/MS ha permitido determinar, por

primera vez, los niveles de seis filtros UV (EHS, HMS, 4-MBC, IAMC, EHMC y

OCR) en muestras de polvo. El método desarrollado integra las etapas de

extracción y clean-up en un único paso, con un consumo moderado de

Conclusiones

344

disolventes orgánicos y sin necesidad de utilizar instrumentación costosa en la

etapa de preparación de muestra. Las recuperaciones obtenidas fueron

superiores al 77% para todos los compuestos, y los LOQs se mantuvieron en la

región de los bajos ng g-1. Su aplicación a muestras de polvo, procedentes de

viviendas particulares y edificios públicos, puso de manifiesto la presencia de

filtros UV en esta matriz, a niveles de concentración un orden de magnitud

superiores a los existentes en lodos, alcanzando valores máximos de 15 y 41 μg

g-1 para EHMC y OCR, respectivamente.

8) "Solid-phase microextraction followed by gas chromatography

mass spectrometry for the determination of ink photo-initiators in

packed milk".

Con este estudio se demostró la adecuación de la SPME para la

extracción de 7 fotoiniciadores, con estructuras similares en muchos casos a

filtros UV derivados de la benzofenona y el ácido p-amino benzoico, en leche

envasada y su determinación mediante GC-MS. Sus ventajas respecto a trabajos

publicados anteriormente son el nulo consumo de disolventes orgánicos y la

integración de extracción y concentración en la misma etapa. Aunque, la

eficacia de la microextracción se vió afectada por el contenido graso de la

muestra, no se apreciaron diferencias entre distintas marcas de leche con el

mismo contenido graso nominal. El método proporcionó LOQs en el rango de

valores recogidos en la bibliografía, empleando metodologías de preparación de

muestra más complejas; sin embargo, no se detectaron niveles significativos de

estos fotoiniciadores en ninguna de las muestras de leche analizadas.

Bibliografía

347

V. BIBLIOGRAFÍA

A

Allegrone G., Tamaro I., Spinardi S., Grosa G., Journal of Chromatography A 1214 (2008) 128.

B

Bagnati R., Bianchi G., Marangon W., Zuccato W., Fanelli R., Davoli E., Rapid Communications in Mass Spectrometry 21 (2007) 1998. Balmer M. E., Buser H. R., Müller M. D., Poiger T., Environmental Science Technology 39 (2005) 953. Baltussen E., Pat Sandra, David F., Cramers C., Journal of Microcolumn Separations 11 (1999) 737. Barker S. A., Long A. R., Short C.R., Journal of Chromatography A 475 (1989) 353. Belardi R. P., Pawliszyn J. B., Water Pollution Research Journal of Canada. 24 (1989) 179. Benetti C., Angeletti R., Binato G., Biancardi A., Biancotto G., Analytica Chimica Acta 617 (2008) 132. Berijani S., Assadi Y., Anbia M., Mirani Hosseini M. R., Aghaee E., Journal of Chromatography A 1123 (2006) 1. Bicchi C., Cordero C., Liberto E., Rubiolo P., Sgorbini B., Sandra P., Journal of Chromatography A 1148 (2007) 137. Blanco E., Casais M. C., Mejuto M. C., Cela R., Analytical Chemistry 78 (2006) 2772. Bradley G., Coatings Inks 2006. Buser H. R., Balmer M. E., Schmid P., Kohler M., Environmental Science Technology 40 (2006) 1427.

C

Cámara C., Fernández P., Martín-Esteban A., Pérez-Conde C., Vidal. M., Toma y tratamiento de muestras. Ed. Síntesis (2002). Camel V., Analyst 126 (2001) 1182. Canosa P., Rodríguez I., Rubí E., Cela R., Analytical Chemistry 79 (2007) 1675. Canosa P, Rodríguez I., Rubí E., Ramil M., Cela R., Journal of Chromatography A 1188 (2008) 132. Carro A. M., Lorenzo R. A., Fernández F., Rodil R., Cela R., Journal of Chromatography A 1071 (2005) 93.

Bibliografía

348

Cela R., Lorenzo R. A., Casais Mª del C., Técnicas de separación en Química Analítica. Ed. Síntesis (2002). Chisvert A., Pascual-Martí M. C., Salvador A., Journal of Chromatography A 921 (2001-A) 207. Chisvert A., Salvador A., Pascual-Marti M. C., March J. G., Fresenius’ Journal of Analytical Chemistry 369 (2001-B) 684. Chisvert A., Salvador A., Journal of Chromatography A 977 (2002-A) 277. Chisvert A., Vidal M. T., Salvador A., Analytica Chimica Acta 464 (2002-B) 295. Concha-Graña E., Turnes-Carou M. I., Muniategui-Lorenzo S., López-Mahía P., Fernández-Fernández E., Prada-Rodríguez D., Journal of Chromatography A 1047 (2004) 147. Crescenzi C., Bayoudh S., Cormack PAG., Klein T., Ensing K., Analytical Chemistry 73 (2001) 2171. Cuderman P., Heath E., Analytical and Bioanalytical Chemistry 387 (2007) 1343. Cunha S. C., Fernandes J. O., Oliveira M. B. P. P., Food Additives and Contaminants 24 (2007) 156.

D

Debuys H., Levy S., Murray J. C., Madery D., Pinnell S. R., Clinics in Dermatology 4 (2000) 583.

Díaz-Cruz M.S., Barceló D., Trends in Analytical Chemistry 28 (2009) 708.

Díaz-Cruz M.S., Llorca M., Barceló D., Trends in Analytical Chemistry 27 (2008) 873.

F

Farahani H., Norouzi P., Dinarvand R., Ganjali M. R., Journal of Chromatography A 1172 (2007) 105. Fattahi N., Samadi S., Assadi Y., Hosseini M.R., Journal of Chromatography A 1169 (2007) 63. Felix T., Hall B. J., Brodbelt J. S., Analytica Chimica Acta 371 (1998) 195. Fent K., Kunz P. Y., Zenker A, Rapp M., Marine Environmental Research 69 (2010-A) 54. Fent K., Zenker A, Rapp M., Environmental Pollution 158 (2010-B) 1817. Ferrer C., Gómez M. J., García-Reyes J. F., Ferrer I., Thurman E. M., Fernández-Alba A. R., Journal of Chromatography A 1069 (2005) 183. Fidalgo-Used N., Blanco-Gómez E., Sanz-Medel A., Analytica Chimica Acta 590 (2007) 1. Fujii K., Kikuchi S., World Journal of Microbiology & Biotechnology 21 (2005) 1311.

Bibliografía

349

G

Gallard H., von Gunten U., Environmental Science Technology 36 (2002) 884. Gallart-Ayala H., Moyano E., Galceran M. T., Journal of Chromatography A 1208 (2008) 182. García-López, M., Rodríguez I., Cela R., Journal of Chromatography A 1166 (2007) 9. García M., Rodríguez I., Cela R., Analytica Chimica Acta 590 (2007) 17. García-Reyes J. F., Ferrer C., Gómez-Ramos M.J., Molina-Días A., Fernández-Alba A. R., Trends in Analytical Chemistry 26 (2007) 239. Garrido-López A., Tena M. T., Journal of Chromatography A 1099 (2005) 75. Gil-Vergara A., Blasco C., Picó Y., Analytical and Bionalytical Chemistry 389 (2007) 605. Giokas D. L., Sakkas V. A., Albanis T. A., Journal of Chromatography A 1026 (2004) 289. Gómez-Ariza J. L., Bujalance M., Giráldez I., Velasco A., Morales E., Journal of Chromatography A 946 (2002) 209.

H

Hany J., Nagel R., Deutsche Lebensmittel-Rundschau 91 (1995) 341. Haunschmidt M., Klampfl C.W., Buchberger W., Hertsens R., Analytical and Bioanalytical Chemistry 39 (2010) 269. Heneweer M., Muusse M., Van den Berg M., Sanderson T., Toxicology and Applied Pharmacology 208 (2005) 170. Herrera-Herrera A. V., Asensio-Ramos M., Hernández-Borges J., Rodríguez-Delgado M. A., Trends in Analytical Chemistry 29 (2010) 728. Horne C., O’Keeffe M., Desbrow C., Howella A., Analyst 123 (1998) 2517. Hu Y-Y., Zheng P., Zhang Z-X., He Y-Z., Journal of Agricultural and Food Chemistry 54 (2006) 4126. Huber M. M., Korhonen S., Ternes T. A., von Gunten U., Water Research 39 (2005) 3607.

I

Ito R., Kawaguchi M., Koganel Y., Honda H., Okanouchi N., Sakui N., Saito K., Nakazawa H., Analytical Sciences 25 (2009) 1033.

J

Jeon H-K., Cheng Y., Ryu J-C., Journal of Chromatography A 1131 (2006) 192.

Bibliografía

350

Kasprzyk-Hordern B., R. M. Dinsdale, Analytical and Bioanalytical Chemistry 391 (2008) 1293. Kawaguchi M., Ito R., Endo N., Sakui N., Okanouchi N., Saito K., Sato N., Shiozaki T., Nakazawa H., Analytica Chimica Acta 557 (2006) 272. Kawaguchi M., Ito R., Honda H., Endo N., Okanouchi N., Saito K., Seto Y, Nakazawa H., Journal of Chromatography A 1200 (2008-A) 260. Kawaguchi M., Ito R., Honda H., Endo N., Okanouchi N., Saito K., Seto Y, Nakazawa H., Analytical Sciences 24 (2008-B) 1509. Kawaguchi M., Ito R., Honda H., Koganei Y., Okanouchi N., Saito K., Seto Y, Nakazawa H., Journal of Chromatography B 877 (2009) 298. Klampfl C. W., Leitner T., Journal Separation Science 26 (2003) 1259. Kloskowski A., Chrzanowski W., Pilarczyk M., Namiesnik J., Crit. Rev. Analytical Chemistry 37 (2007) 15. Kozani R. R., Assadi Y., Shemirani R., Milani Hosseini M. R., Jamali M. R., Talanta 72 (2007) 387. Kubala-Drinic H., Bazulic D., Sapunar P., Grubelic M., Stuuhne G., Journal of Agricultural and Food Chemistry 51 (2003) 871. Kunz P. Y., Fent K., Aquatic Toxicology 79 (2006-A) 305. Kunz P. Y., Galicia H. F., Fent K., Toxicological Sciences 90 (2006-B) 349. Kupper T., Plagellat C., Brändi R. C., Alencastro L. F., Grandjean D., Tarradellas J., Water Research 40 (2006) 2603. Kwon J-W., Journal of AOAC International 92 (2009) 359.

L

Lambropoulou D. A., Giokas D. L., Sakkas V. A., Albanis T. A., Karayannis M. I., Journal of Chromatography A 967 (2002) 243. Le Boulaire S., Bauduret J. C., Andre F., Journal of Food and Chemistry 45 (1997) 2134. León Z., Chisvert A., Tarazona I., Salvador A., Analytical and Bioanalytical Chemistry 398 (2010) 831. Li W., Ma Y., Guo C., Hu W., Liu K., Wang Y., Zhu T., Water Research 41 (2007) 3506. Liu H., Liu L., Xiong Y., Yang X., Luan T., Journal of Chromatography A 1217 (2010) 6747. Liu X., Li J., Zhao Z., Zhang W., Lin K., Huang C., Wang X., Journal of Chromatography A 1216 (2009) 2220.

K

Bibliografía

351

M

Maerkel K., Lichtensteiger W., Durrer S., Conscience M., Schlumpf M., Environmental Toxicology and Pharmacology 19 (2005) 761. Martínez A., Ramil M., Montes R., Hernanz D., Rubí E., Rodríguez I., Cela R., Journal of Chromatography A 1072 (2005) 83. Meinerling M., Daniels M., Analytical and Bioanalytical Chemistry 386 (2006) 1465. Moeder M., Shrader S., Winkler U., Rodil U., Journal of Chromatography A 1217 (2010) 2925. Mohamed A-R, Journal of Chromatography A 1217 (2010) 2569. Montero L., Popp P., Paschke M., Pawliszyn J., Journal of Chromatography A 1025 (2004) 17. Montes R., Rodríguez I., Rubí E., Cela R., Journal of Chromatography A 1216 (2009) 205. Morlock G., Schwack W., Analytical and Bioanalytical Chemistry 385 (2006) 586. Mottaleb M. A., Usenko S., O’Donnell J. G., Ramirez A. J., Brooks B. W., Chambliss C. K., Journal of Chromatography A 1216 (2009) 815.

N

Negreira N., Canosa P., Rodríguez I., Ramil M., Rubí E., Cela R., Journal of Chromatography A 1178 (2008) 206. Negreira N., Rodríguez I., Ramil M., Rubí E., Cela R., Analytica Chimica Acta 638 (2009-A) 36. Negreira N., Rodríguez I., Ramil M., Rubí E., Cela R., Analytica Chimica Acta 654 (2009-B) 162. Negreira N., Rodríguez I., Rubí E., Cela R., Journal of Chromatography A 1216 (2009-C) 5895. Negreira N., Rodríguez I., Rubí E., Cela R., Analytical and Bionalytical Chemistry 398 (2010-A) 995. Negreira N., Rodríguez I., Rubí E., Cela R., Talanta 82 (2010-B) 296. Negreira N., Rodríguez I., Rubí E., Cela R., Journal of Chromatography A, 10.1016/j.chroma.2010.11.028 accepted. Nieto A., Borrull F., Marcé R. M., Pocurull E., Journal of Chromatography A 1216 (2009) 5619.

O

Okanouchi N., Honda H., Ito R., Kawaguchi M., Saito K., Nakazawa H., Analytical Sciences 24 (2008) 627. Oliveira H. M., Segundo M. A., Lima J. L. F. C., Miró M., Cerdà V., Journal of Chromatography A 1217 (2010) 3575.

Bibliografía

352

P

Paschke A., Schwab K., Brümmer J., Schüürmann G., Paschke H., Popp P., Journal of Chromatography A 1124 (2006) 187. Pawliszyn J., Sampling and sample preparation for field and laboratory: fundamentals and new directions in sample preparation, Elsevier, 2002. Pedrouzo M., Borrull F., Marcé R. M., Pocurull E., Journal of Chromatography A 1216 (2009) 6994. Pedrouzo M., Borrull F., Marcé R. M., Pocurull E., Analytical and Bionalytical Chemistry 397 (2010) 2833. Pena M. T., Casais M. C., Mejuto M. C., Cela R., Journal of Chromatography A 1165 (2007) 32. Pensado L., Casais M. C., Mejuto M. C., Cela R., Journal of Chromatography A 1077 (2005) 103. Plagellat C., Kupper T., Furrer R., Alencastro L. F., Grandjean D., Tarradellas J., Chemosphere 62 (2006) 915. Poiger T., Buser H., Balmer M. E., Bergqvist P., Müller M. D., Chemosphere 55 (2004) 951. Popp P., Bauer C., Paschke M., Montero L., Analytica Chimica Acta 504 (2004) 307. Potard G., Laugel C., Baillet A., Schaefer H., Marty J-P., International Journal of Pharmaceutics 189 (1999) 249.

R

Rahnama K. R., Assadi Y., Shemirani F., Milani Hosseini M. R., Jamali M.R., Chromatographia 66 (2007) 81. Regueiro J., Llompart M., García-Jares C., García-Monteagudo J. C., Cela R., Journal of Chromatography A 1190 (2008) 27. Rezaee M., Assadi Y., Milani Hosseini M. R., Aghaee E., Ahmadi F., Berijani S., Journal of Chromatography A 1116 (2006) 1. Rezaee M., Yamini Y., Faraji M., Journal of Chromatography A 1217 (2010) 2342. Rezaei F., Bidari A., Birjandi A. P., Milani Hosseini M. R., Assadi Y., Journal Hazardous Materials 158 2-3 (2008) 621. Richardson S. D., Trends in Analytical Chemistry 22 (2003) 666. Richardson S. D., Ternes T. A., Analytical Chemistry 77 (2005) 3807. Richardson S. D., Analytical Chemistry 78 (2006) 4021. Richardson S.D., Analytical Chemistry 80 (2008) 4373. Richter B. E., Jones B. A., Ezzell J. L., Porter N., Avdalovic N., Pohl C., Analytical Chemistry 68 (1996) 1033. Ricking M., Schwarzbauer J., Franke S., Water Research 37 (2003) 2607.

Bibliografía

353

Rodil R., Quintana J. B., López-Mahía P., Muniategui-Lorenzo S., Prada-Rodríguez D., Analytical Chemistry A 80 (2008-A) 1307. Rodil R., Moeder M., Journal of Chromatography A 1179 (2008-B) 81. Rodil R., Moeder M., Analytica Chimica Acta 612 (2008-C) 152.

Rodil R., Moeder M., Altenburger R., Schmitt-Jansen M., Analytical and Bioanalytical Chemistry

395 (2009-A) 1513.

Rodil R., Quintana J. B., López-Mahía P., Muniategui-Lorenzo S., Prada-Rodríguez D., Journal of Chromatography A 1216 (2009-B) 2958.

Rodil R., Schrader S., Moeder M., Journal of Chromatography A 1216 (2009-C) 4887.

Rodil R., Schrader S., Moeder M., Journal of Chromatography A 1216 (2009-D) 8851.

Rodil R., Schrader S., Moeder M., Rapid Communications in Mass Spectrometry 23 (2009-E) 580.

Rouessac F., Rouessac A., Análisis Químico: Método y Técnicas Instrumentales Modernas. Ed. McGraw Hill (2003). Ruiz M. J., Camean A. M., Moreno I. M., Picó Y., Journal of Chromatography A 1073 (2005) 257.

S

Sagratini G., Caprioli G., Cristalli G., Giardinà D., Ricciutelli M, Volpini R., Zuo Y., Vittori S., Journal of Chromatography A 1194 (2008) 213. Sakkas V. A., Giokas D. L., Lambropoulou D. A., Albanis T. A., Journal of Chromatography A 1016 (2003) 211. Salvador A., Chisvert A., Analytica Chimica Acta 537 (2005-A) 1. Salvador A., Chisvert A., Analytica Chimica Acta 537 (2005-B) 15. Salvador A., Chisvert A., Jaime M-A., Journal Separation Science 28 (2005-C) 2319. Salvador A., Ossa M. D., Chisvert A., Internacional Journal of Cosmetic Science 25 (2003) 97. Sanches-Silva A., Pastorelli S., Cruz J. M., Simoneau C., Castanheira I., Paseiro-Losada P., Journal of Dairy Science 91 (2008-A) 900. Sanches-Silva A., Pastorelli S., Cruz J. M., Simoneau C., Paseiro-Losada P., Food Chemistry 73 (2008-B) 92. Sarveiya V., Risk S., Benson H. A. E., Journal of Chromatography B, 803 (2004) 225. Schakel D. J., Kalsbeek D., Boer, K., Journal of Chromatography A 1049 (2004) 127.

Bibliografía

354

Schlumpf M., Durrer S., Faass O., Ehnes C., Fuetsch M., Gaille C., Henseler M., Hofkamp L., Maerkel K., Reolon S., Timms B., Tresguerres J. A. F., Lichtensteiger W., International Journal of Andrology 31 (2008-A) 144. Schlumpf M., Jarry H., Wuttke W., Ma R., Lichtensteiger W., Toxicology 199 (2004-A) 109. Schlumpf M., Kypke K., Vöktd C. C., Birchlerd M., Durrer S., Faass O., Ehnes C., Fuetsch M., Gaille c., Henseler M., Hofkamp L., Maerkel K., Reolon S., Zenker A., Timms B., Tresguerres J. A. F., Lichtensteiger W., Chimia 62 (2008-B) 345. Schlumpf M., Schmid P., Durrer S., Conscience M., Maerkel K., Henseler M., Gruetter M., Herzog I., Reolon S., Ceccatelli R., Faass O., Stutz E., Jarry H., Wuttke W., Lichtensteiger W., Toxicology 205 (2004-B) 113. Seidlová-Wuttke D., Christoffel J., Rimoldi G., Jarry H., Wuttke W., Toxicology and Applied Pharmacology 214 (2006) 1. Shaath N. A., The Encyclopedia of Ultraviolet Filters, Allured, Illinosis, 2007. Shellin M., Hauser B., Popp P., Journal of Chromatography A 1040 (2004) 251. Shen D-X., Lian H-Z., Ding T., Xu J-Z., Shen C-Y., Analytical and Bioanalytical Chemistry 395 (2009) 2359. Shen X., Jibao C., Yun G., Quingde S., Chromatograpia 64 (2006) 71. Skoog D. A., Leary J. J., Análisis Instrumental. Ed. Mc Graw Hill (1994). Smyrniotakis C. G., Archontaki H. A., Journal of Chromatography A 1031 (2004) 319. Straub J. O., Toxicology Letters 131 (2002) 29. Sun C., Chan S.H., Lu D., Lee H. M. W., Bloodworth B. C., Journal of Chromatography A 1143 (2007) 162.

T

Tarazona I., Chisvert A., León Z., Salvador A., Journal of Chromatography A 1217 (2010) 4771. Tolls J., Haller M., Shun D.T.H.M., Analytical Chemistry 71 (1999) 5242. Trenholm R. A., Vanderford B. J., Drewes J. E.,Snyder S. A., Journal of Chromatography A 1190 (2008) 253.

V

Van Pinxteren M., Paschke A., Popp P., Journal of Chromatography A 1217 (2010) 2589. Vidal L., Chisvert A., Canals A., Salvador A., Journal of Chromatography A 1174 (2007) 95. Vidal L., Chisvert A., Canals A., Salvador A., Talanta 81 (2010) 549.

Bibliografía

355

Wang L-H., Huang W-S., Tai H-M., Journal of Pharmaceutical and Biomedical Analysis 43 (2007) 1430. Wick A., Fink G., Ternes T. A., Journal of Chromatography A 1217 (2010) 2088.

Y

Ye X., Kuklenyik Z., Needham L. L., Calafat A. M., Analytical Chemistry 77 (2005) 5407.

Z

Zenker A., Schmutz H., Fent K., Journal of Chromatography A 1202 (2008) 64. Zhang K., Zuo Y., Analytica Chimica Acta 554 (2005) 190. Zhao R-S., Diao C-P., Chen Q-F., Wang X., Journal of Separation Science 32 (2009) 1069. Zhou Q. X., Bai H. H., Xie G. H., Xiao J. P., Journal of Chromatography A 1177 (2008) 43.

W

ABREVIATURAS

Y ACRÓNIMOS

Abreviaturas y acrónimos

359

VI. ABREVIATURAS Y ACRÓNIMOS

A ADN Ácido desoxirribonucleico

APCI Atmospheric pressure chemical

ionization Ionización química a presión atmosférica

API Atmospheric pressure ionization Ionización a presión atmosférica

APPI Atmospheric pressure photo-

ionization Fotoionización a presión atmosférica

ARN Ácido ribonucleico

ASE Accelerated solvent extraction Extracción acelerada con disolventes

B

BDE Brominated diphenyl ether Difenil éter polibromado

BSTFA N,O-bis-

(trimethylsilyl)trifluoroacetamide N,O-bis-(trimetilsilil)trifluoroacetamida

C

CAR-PDMS Carboxen-Polydimethylsiloxane Carboxen- Polidimetilsiloxano

CW-DVB Carbowax-Divinylbenzene Carbowax-Divinilbenceno

C18 Octadecyl modified silica phase Sílica modificada con grupos octadecilos

D

DDT Dichlorodiphenyltrichloro ethane Diclorodifeniltricloro etano

DC Direct current Corriente continua

DLLME Dispersive liquidliquid

microextraction Microextracción líquidolíquido dispersiva

E

EU European Union Unión Europea

EI Electronic impact Impacto electrónico

EPA Environmental Protection Agency Agencia de protección medioambiental

ESI Electrospray Electrospray

ESL Solid-liquid extraction Extracción sólido-líquido


360

F

FDA Food and Drug Administration Administración de alimentos y

medicamentos (USA)

G

GC Gas chromatography Cromatografía de gases

H

HCX Hydrophobic Cation Exchange Intercambio catiónico y fase reversa

HF-LPME Hollow fiber-Liquid phase

microextraction

Microextracción en fase líquida con fibra

hueca

HLB Hydrophilic-Lipophilic Balance Balance hidrófilo-lipófilo

HPLC High performance liquid

chromatography Cromatografía líquida de alta resolución

HS Headspace Espacio de cabeza

I

IL Ionic liquid Líquido iónico

IUPAC International Union of Pure and

Applied Chemistry

Unión internacional de Química pura y

aplicada

K

Kow Octanol-Water partition constant Constante de partición octanol-agua

L

LC Liquid chromatography Cromatografía líquida

LD Liquid desorption Desorción líquida

LDPE Low-density polyethylene Polietileno de baja densidad

LLE Liquidliquid extraction Extracción líquidolíquido

LLME Liquidliquid microextraction Microextracción líquidolíquido

LOD Limit of detection Límite de detección

LOQ Limit of quantification Límite de cuantificación

LPME Liquid-phase microextraction Microextracción en fase líquida

M

MAX Mixed-Mode Anion Exchange Modo mixto de intercambio aniónico y fase

reversa


361

MCX Mixed-Mode Cation Exchange Modo mixto de intercambio catónico y fase

reversa

MEPS Microextraction by packed Sorbent Microextracción con adsorbente

empaquetado

min Minutes Minutos

MR Reference material Material de referencia

MS Mass spectrometry Espectrometría de masas

MS-MS Tandem mass spectrometry Espectrometría de masas en tándem

MSPD Matrix solid-phase dispersion Dispersion de matriz en fase sólida

MSTFA N‐Methyl‐N‐(trimethylsilyl)trifuoroace

tamide N‐metil‐N‐(trimetilsilil)trifuoroacetamida

MTBE Methyl tert-butyl ether t-butilmetil éter

MTBSTFA N-Methyl-N-[tert-butyldimethyl-

silyl]trifluoroacetamide

N‐(tert‐butildimetilsilil)‐N‐metiltrifluoroacet

amida

m/z Mass/charge Masa/carga

P

PA Polyacrylate Poliacrilato

PAH Polycyclic aromatic hydrocarbon Hidrocarburo policíclico aromático

PCB Polychlorinated biphenyl Bifenilo policlorado

PDMS Polydimethylsiloxane Polidimetilsiloxano

PDMS-DVB Polydimethylsiloxane-Divinylbenzene Polidimetilsiloxano-Divinilbenceno

PE Polyethylene Polietileno

PEG Poliethylenglycole Polietilenglicol

PLE Pressurized liquid extraction Extracción con disolventes presurizados

PSA Primary-secundary amine Amina primaria y secundaria

R

R2 Determination coefficient Coeficiente de determinación

RP-HPLC Reversed phase-high pressure liquid

chromatography Cromatografía líquida en fase reversa

RSD Relative standard deviation Desviación estándar relativa

S

SBSE Stir bar sorptive extraction Extracción con barra agitadora

SDME Single drop microextraction Microextracción con gota


362

SEC Size exclusion chromatography Cromatografía de exclusión por tamaños

SIM Selected ion monitoring

S/N Signal-to-noise ratio Relación señal/ruido

SPE Solid-phase extraction Extracción en fase sólida

SPME Solid-phase microextraction Microextracción en fase sólida

T

TBDMS Tert-butyldimethylsilyl Tert-butildimetilsilil

TD Thermal desorption Desorción térmica

U

UPLC Ultra- performance liquid

chromatography

USAEME Ultrasound-assisted

emulsificationmicroextraction

Microextracciónemulsificación asistida por

ultrasonidos

UV Ultraviolet Ultravioleta

V

Vm Volumen de muestra

W

WAX “Weak” Mixed-Mode Anion Exchange Modo – mixto de intercambio aniónico débil

y fase reversa

WCX “Weak” Mixed-Mode Cation Exchange Relleno de modo – mixto de intercambio

cationico débil y fase reversa

VII. ANEXO: OTRAS PUBLICACIONES

Analytica Chimica Acta 575 (2006) 106–113

Formation of halogenated by-products of parabens in chlorinated water

P. Canosa, I. Rodrıguez ∗, E. Rubı, N. Negreira, R. CelaDepartamento de Quımica Analıtica, Nutricion y Bromatologıa, Instituto de Investigacion y Analisis Alimentario,

Universidad de Santiago de Compostela, Santiago de Compostela 15782, Spain

Received 31 March 2006; received in revised form 19 May 2006; accepted 22 May 2006Available online 27 May 2006

Abstract

Chemical transformations of four alkyl esters of p-hydroxybenzoic acid, parabens, in chlorinated water samples are investigated. Quantification ofthe parent species and identification of their reaction by-products were performed using gas chromatography in combination with mass spectrometry.Experiments were accomplished considering free chlorine and paraben concentrations at the mg L−1 and �g L−1 level, respectively. Concentrationof water samples, using solid-phase extraction, and silylation of the target species were carried out in order to improve the detectability of parentspecies and their possible transformation products, achieving quantification limits at the low ng L−1 level. Under employed experimental conditions,the decrease in the concentrations of parabens followed pseudo-first-order kinetics. Half-lives values obtained for model ultrapure water solutionswere in good agreement with those observed in tap water samples. For the first type of sample, only two by-products were detected for eachparaben. They corresponded to chlorination of the aromatic ring in one or two carbons situated in ortho-positions to the hydroxyl group. Bothspecies were also generated after the addition of parabens to chlorinated tap water. Moreover, three new transformation products were noticedfor each parent compound. They were identified as bromo- and bromochloro-parabens, formed due to the existence of traces of bromide in tapwater sources. Experiments carried out by mixing paraben-containing personal care products with tap water, containing free chlorine, confirmedthe formation of all above described halogenated by-products. In addition, the presence of the di-chlorinated forms of methyl and propyl parabenhas been detected for first time in raw sewage water samples.© 2006 Elsevier B.V. All rights reserved.

Keywords: Parabens; Chlorine; Halogenated by-products; Water samples; Gas chromatography–mass spectrometry

1. Introduction

Parabens, esters of p-hydroxybenzoic acid, are extensivelyemployed as bactericides and preservative agents in antiperspi-rants, sunscreen creams, bath gels, shampoos and toothpaste[1]. As in the case of many personal care chemicals, they arecontinuously released in the aquatic media through domesticwastewater, therefore, a growing concern has arisen in relationto their potential long term effects on wildlife. Nowadays, itis known that parabens are weak endocrine disrupters [2] and,although they are removed in a considerable extension duringconventional sewage water treatments [3], their presence hasbeen detected in river water samples at the low ng L−1 level[4]. Moreover, recent studies have suggested a possible relation-ship between breast cancer and prolonged dermal expositions toparaben-containing products [5].

∗ Corresponding author. Tel.: +34 981 563100x14387; fax: +34 981 595012.E-mail address: [email protected] (I. Rodrıguez).

The possibility that personal care chemicals and non-prescription drugs might reach potable water sources has fos-tered different studies aiming to test their stability and to studythe formation of undesirable by-products, during water chemicaldisinfection treatments [6]. This information is also relevant toimprove their removal efficiency in sewage water plants usingadvanced oxidation processes. On the other hand, less atten-tion has been paid to the reactivity of personal care productswhen mixed with tap water in our homes. Independently ofthe primary disinfection treatment applied in production plants,tap water is normally amended with free chlorine to insureits bacteriological quality throughout the distribution system.Chlorine is a potent oxidant able to react with natural organicmatter and anthropogenic chemicals rendering different halo-genated by-products. Some of them are toxic species, whichmight suppose a potential risk for human health [7,8]. Particu-larly, compounds containing phenolic groups in their structuresshow favourable chlorination kinetics [9–12]. As an example,recent studies have demonstrated that personal care productscontaining triclosan (5-chloro-2-(2,4-dichlorophenoxy) phenol)

0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.aca.2006.05.068

P. Canosa et al. / Analytica Chimica Acta 575 (2006) 106–113 107

produce several toxic and persistent by-products, such as 2,4-dichloro and 2,4,6-trichlorophenol, when they are mixed withchlorinated water [13,14].

Aims of this work are: (i) to evaluate the stability of fouralkylated parabens (methyl, ethyl, propyl and butyl paraben)in water samples containing free chlorine (sodium hypochlo-rite plus hypochlorous acid) at the low mg L−1 level, (ii) todetermine their half-lives under different experimental condi-tions and (iii) to identify their potential by-products as well as toinvestigate their further stability under oxidative conditions. Pre-liminary studies were carried out using buffered ultrapure watersamples spiked with known amounts of chlorine and parabenstandard solutions. The relevance of the obtained results wasfurther assessed by mixing chlorinated tap water with paraben-containing personal care products. Determination of parabenremaining concentrations in the above experiments and iden-tification of their by-products were performed using gas chro-matography with mass spectrometry. In order to achieve quan-tification limits at the low ng L−1 level, for the parent species,and also to identify their oxidation by-products, even if theyare formed in a minor extension, a sample preparation proce-dure was developed. It consisted of water concentration usinga solid-phase extraction (SPE) cartridge, followed by elution ofthe target species and derivatization with a silylation reagent toimprove the performance of gas chromatography separations.

2. Experimental


HPLC-grade methanol, ethyl acetate for trace analysisand sodium thiosulphate were supplied by Merck (Darm-stadt, Germany). Methyl paraben (MeP), ethyl paraben (EtP),propyl paraben (PrP) and butyl paraben (BuP), as well asthe derivatization reagent N-methyl-N-(tert-butyldimethylsilyl)-trifluoroacetamide (MTBSTFA), were purchased from Aldrich(Milwaukee, WI, USA). Standards of 3-chloromethylparaben(3-ClMeP) and 3-chloroethylparaben (3-ClEtP) were alsoobtained from Aldrich. Individual solutions of each analyte wereprepared in methanol. Further dilutions and mixtures of the fournon-halogenated parabens were made in methanol and ethylacetate. The first were employed for preparing spiked water sam-ples. Optimisation of GC–MS conditions was performed usingthe silylated derivatives of the analytes. They were obtained byaddition of 40 �L of MTBSTFA to aliquots (0.5 mL volume) ofparaben solutions in ethyl acetate [13].

Sodium hyphochlorite with a nominal free chlorine contentof 4% (w/v) was purchased from Aldrich. This solution wasstored at 4 ◦C and its exact concentration determined weekly byiodometric titration [15]. OASIS HLB (60 mg) SPE cartridgeswere acquired from Waters (Milford, MA, USA). Glass woolfilters were purchased from Millipore (Billerica, MA, USA).

2.2. Concentration of water samples

Recoveries of the sample preparation method for non-halogenated parabens were evaluated using spiked and non-

spiked samples of ultrapure, tap and raw wastewater. Wastewatersamples were taken from the main sewer of a 100,000 inhabi-tants city, filtered and stored for a maximum of 1 week at 4 ◦Cbefore analysis. Ultrapure and tap water were obtained directlyin the laboratory when needed. Recoveries of the SPE procedurewere calculated using spiked aliquots (1 L volume) of the abovedescribed water samples adjusted at pH 2.5. After the extractionstep, the SPE sorbent was dried using a nitrogen stream and thenanalytes were eluted with 2 mL of ethyl acetate. A fraction of thisextract (0.5 mL volume) was poured in a 1.5 mL GC autosam-pler vial together with 40 �L of MTBSTFA. The mixture washeated for 1 h at 70 ◦C and injected in the GC–MS system afterreturning to room temperature. Standard solutions of parabensin ethyl acetate were silylated using the same protocol. Quantifi-cation was performed using external calibration by comparisonof peak areas obtained for silylated standards and extracts fromSPE cartridges.

2.3. Chlorination experiments

The potential degradation of parabens and the formationof disinfection by-products was investigated using samples(100–250 mL) of ultrapure (Milli-Q) and tap water contain-ing a known amount of chlorine. Experiments were performedat room temperature, unless otherwise is stated, using light-protected 300 mL glass bottles. Samples were adjusted at dif-ferent pHs, in the range from 6.3 to 8.6, with 0.1 M phosphateor bicarbonate buffer solutions. After that, they were spikedwith a methanolic solution containing one or several parabensand with the corresponding volume of a sodium hypochlo-rite solution, which exact concentration was previously deter-mined by titration. Obviously, the last step was not necessaryfor experiments accomplished using chlorinated tap water. Inthis case the water free chlorine content was determined usingthe N,N-diethyl-p-phenylenediamine procedure with photomet-ric detection [16]. Bottles were closed, shaken manually for1–2 min and allowed to stand during an established reactiontime. Then, the excess of chlorine was removed by addition ofsodium thiosulphate (10 mg per 100 mL of water) and sampleswere adjusted at pH 2.5 and concentrated as described in theprevious paragraph. Initial concentrations of free chlorine andparent compounds in the above experiments ranged from 0 to5 mg L−1 and from 10 to 40 �g L−1, respectively. The percent-age of methanol in the spiked samples was maintained below0.04%.

2.4. GC–MS determination

Parabens and their transformation by-products were deter-mined by GC–MS. The system was a Varian CP 3900 GasChromatograph (Walnut Creek, CA, USA) connected to an ion-trap mass spectrometer (Varian Saturn 2100). Separations werecarried out using a HP-5 MS capillary column (30 m × 0.25 mmi.d., df: 0.25 um) supplied by Agilent (Wilmington, DE, USA).Helium (99.999%) was used as carrier gas at a constant flow of1 mL min−1. The GC oven was programmed as follows: 2 min at50 ◦C, at 10 ◦C min−1 to 270 ◦C (held for 10 min). The GC–MS

108 P. Canosa et al. / Analytica Chimica Acta 575 (2006) 106–113

interface and the ion trap temperatures were set at 270 and220 ◦C, respectively. The temperature of the injector was main-tained at 280 ◦C and injections (1 �L volume) were made in thesplitless mode (purge time 2 min) using an autosampler device.The mass spectrometer was operated in the electron impact (EI)mode (70 eV). Spectra were recorded in the range from 90 to550 m/z units.


3.1. Performance of the analytical procedure

In order to investigate the possible reactions of parabens con-sidering levels of free chlorine and parent compounds whichmimic those present in real life water samples, a concentrationstep was necessary. Samples were extracted using reversed-phase SPE cartridges (OASIS HLB, 60 mg). Extraction con-ditions were optimised to maximize the recoveries for thefour parabens considered as parent species in this study. Ethylacetate was chosen to elute the analytes from the SPE car-tridge. This solvent presents a medium polarity allowing thequantitative recovery of phenolic species from reversed-phasesorbents [13] and, moreover, it does not contain hydroxylgroups, which might consume the silylation reagent. There-fore, the derivatization reaction was accomplished directly inthe cartridge extract, without introducing an additional solventexchange step. Parabens were recovered quantitatively from theOASIS sorbent using only 2 mL of ethyl acetate. After that,0.5 mL of this extract were mixed with 40 �L of MTBSTFAto obtain the silylated derivatives. Table 1 summarizes the m/zratios employed for quantification of parabens, the linearity inthe response of the GC–MS for silylated standards contain-ing increasing concentrations, at seven different levels, from10 to 2000 ng L−1, and the recoveries of the sample prepara-tion procedure for different water samples, 1 L volume, spikedat the 0.5 �g L−1 level. In the case of tap water, previouslyto the addition of parabens, the available free chlorine wasremoved using sodium thiosulphate. Raw sewage water sam-ples contained significant levels of some parabens, particularlyMeP and PrP; therefore, recoveries were calculated by divid-ing the difference between concentrations measured for spikedand non-spiked aliquots of the same water sample by the addedone. Considering a sample intake of 1 L, quantification limitsof the developed procedure remained at the low ng L−1 level(Table 1).

Fig. 1. Influence of pH and free chlorine concentration on the stability of PrP.Reaction time 10 min.

Fig. 2. Plots of MeP concentration, logarithmic values, vs. the reaction time forwater samples buffered at pH 7.3. (A) Ultrapure water spiked with chlorine at0.4 and 1.6 mg L−1. (B) Tap water samples containing 0.46 mg L−1 of chlorineand spiked with sodium hypochlorite to achieve a chlorine concentration of1.60 mg L−1.

Table 1Quantification ions, linearity, recoveries (n = 4 replicates, 1 L volume samples) and quantification limits of the analytical procedure for paraben species

Compound Retentiontime (min)

Quantificationion (m/z)

Regressioncoefficient (R2)

%Recovery ± R.S.D. QL (S/N 10)(ng L−1)

Ultrapure water Tap water Raw wastewater

MeP 15.59 209 0.998 106.2 ± 2.3 80.4 ± 4.6 104.7 ± 4.1 10EtP 16.35 223 0.998 109.2 ± 2.2 94.3 ± 2.9 91.1 ± 5.8 10PrP 17.38 237 0.998 111.9 ± 1.0 97.9 ± 3.5 96.4 ± 4.1 6BuP 18.39 251 0.999 106.9 ± 1.3 97.3 ± 3.8 103.4 ± 2.3 3


Table 2Calculated half-lives (t1/2) for parabens at different chlorine concentrations

Sample type Chlorine concentration (mg L−1)a t1/2 (min)

MeP EtP PrP BuP

Ultrapure water 0.40b 33.8 (0.990) 26.1 (0.996) 30.9 (0.990) 27.6 (0.991)Ultrapure water 1.60b 4.6 (0.999) 5.3 (0.997) 4.2 (0.998) 5.0 (0.991)Tap water 0.46c 31.5 (0.990) 31.5 (0.990) 30.3 (0.999) 29.9 (0.997)Tap water 1.60b 4.7 (0.998) 4.7 (0.992) 4.6 (0.994) 4.4 (0.991)

Values in parentheses correspond to the correlation coefficients of graphs plotting the logarithmic value of paraben concentrations vs. the reaction time.a Initial concentration at zero time.b Water samples spiked with sodium hypochlorite.c Tap water without any extra addition of chlorine.

3.2. Behaviour of parabens in presence of chlorine

3.2.1. Effect of sample pH and chlorine concentrationThe relevance of reactions between phenolic species and free

chlorine (hypochlorous acid and hypochlorite) depends on sev-eral factors such as the concentration of chlorine, the pH of themedia and the kinetic of the process. In order to assess whetherparabens react at a significant extent with low chlorine concen-trations, such as those contained in tap and water samples usedin recreational activities, a first set of experiments was designed.Aliquots of ultrapure water (100 mL), buffered at different pHsin the range between 6 and 9, were spiked with a fixed con-centration of parabens (40–45 �g L−1) and increasing levels offree chlorine from 0 to 2.4 mg L−1. Independent series of exper-iments were carried out for each parent compound. After a reac-tion time of 10 min, the excess of chlorine was quenched usingsodium thiosulphate and samples concentrated as described inSection 2. Results obtained for PrP are shown in Fig. 1. Anoticeable diminution in the signal of the parent compoundwas observed for free chlorine concentrations over 0.4 mg L−1.

Within the investigated chlorine concentrations, the higherdegradation rates took place at pHs 7.3 and 8.0. This behaviourconfirms that, as it has already been described for other phenolicspecies, the anionic form of PrP (pKa 8.2) and the hypochlor-ous acid (pKa 7.5) react faster than the neutral paraben and thehypochlorite anion [6,9,17]. A similar trend to that presented inFig. 1 was observed for methyl, ethyl and butyl paraben, data notgiven.

3.2.2. Reaction kineticsTime course of paraben concentrations were followed using

ultrapure and tap water samples buffered at pH 7 and spiked,when necessary, with sodium hypochlorite at two different lev-els: 5.63 and 22.53 �M (equivalent to free chlorine concentra-tions of 0.4 and 1.6 mg L−1). Experiments were carried out using250 mL volume aliquots containing an initial concentration ofeach paraben between 9 and 10 �g L−1 (45–70 nM, dependingon their molecular weights). After a given time, the reaction wasstopped and the remaining concentration of the parent speciedetermined. Data at zero time were obtained by adding sodium

Table 3Retention times, proposed identities and most intense ions in MS spectra of paraben disinfection by-products

Parent compound By-products Retention time (min) m/z ratios of most intense ions with their relative abundances

MeP

3-ClMeP 17.19 243 (100), 245 (35), 163 (22)3,5-DClMeP 18.54 277 (100), 279 (71), 183 (12)3-BrMeP 18.06 287 (89), 289 (100), 119 (15)3,5-DBrMeP 20.21 363 (72), 365 (100), 367 (30)3-Br-5-ClMeP 19.38 321 (68), 323 (100)

EtP

3-ClEtP 17.88 257 (100), 259 (38), 149 (37)3,5-DClEtP 19.17 291 (100), 293 (71), 183 (34)3-BrEtP 18.69 301 (100), 303 (99), 149 (26)3,5-DBrEtP 20.79 379 (52), 381 (100), 383 (53), 313 (12)3-Br-5-ClEtP 20.01 335 (70), 337 (100)

PrP

3-ClPrP 18.82 271 (100), 273 (38), 149 (21)3,5-DClPrP 20.07 305 (100), 307 (70), 183 (15)3-BrPrP 19.60 315 (94), 317 (100), 149 (15)3,5-DBrPrP 21.60 393 (61), 395 (100), 397 (60)3-Br-5-ClPrP 20.83 349 (72), 351 (100)

BuP

3-ClBuP 19.76 285 (100), 287 (36), 149 (14)3,5-DClBuP 20.95 319 (100), 321 (74), 183 (10)3-BrBuP 20.50 329 (90), 331 (100), 149 (12)3,5-DBrBuP 22.43 407 (52), 409 (100), 411 (57)3-Br-5-ClBuP 21.69 363 (70), 365 (100)


thiosulphate to chlorinated water samples before parabens. As ahigh molar excess of chlorine was used, the reaction rates onlydepended on the instantaneous concentration of the consideredparaben. Fig. 2 shows the graphs obtained for MeP. The loga-rithmic value of its concentration was plotted versus the reactiontime for ultrapure (Fig. 2A) and drinking water (Fig. 2B). Oneof the drinking water samples was obtained directly from thetap and contained 0.46 mg L−1 of free chlorine. The other wasprepared in the laboratory by addition of sodium hypochloriteto a tap water sample containing 0.2 mg L−1 of chlorine. In allcases, experimental data fitted quite well a straight line meaningthat, in presence of an excess of chlorine, the removal of MePfollows a pseudo-first-order kinetic. Moreover, similar half-liveswere obtained for ultrapure and tap water containing similar lev-els of free chlorine. The decrease in the concentrations of theother three parabens followed also pseudo-first-order kineticsand, moreover, they presented similar half-life values to thoseobtained for MeP (Table 2). As experiments were performedin different days, without controlling exactly the temperatureof the water samples (15 ± 2 ◦C) and, in addition, several stepsare involved in the sample preparation procedure, the observeddifferences among half-lives of the four parabens are probablynon-significant.

3.2.3. Disinfection by-productsTable 3 summarizes retention times and m/z ratios for the

most intense ions in the MS spectra of paraben by-products.Five transformation species were identified for each parent com-pound. MS spectra of the two major by-products presentedthe typical isotopic pattern of molecular ions correspondingto mono- and di-chlorinated compounds. Moreover, their basepeaks were shifted 34 and 68 m/z units in comparison to thoseappearing in the spectra of the respective parent compounds,figure not shown. Thus, they correspond to substitution of oneor two atoms of hydrogen per chlorine in the aromatic ring ofparabens. It is expected that chlorination takes place in both car-bons located in ortho- to the phenolic group, since the para-position is blocked with the ester moiety [17]. Comparison ofretention times obtained for the mono-chlorinated by-productsof MeP and EtP with those corresponding to standards of 3-ClMeP and 3-ClEtP, purchased from Aldrich and submitted tothe same silylation procedure, confirmed the above assump-tion, figure not shown. It was also verified that both standardsreact with free chlorine following pseudo-first-order kineticsand rendering the same di-chlorinated by-products than methyland ethyl paraben. Their half-lives for a chlorine concentrationof 0.4 mg L−1 were 31.2 and 31.9 min for 3-ClMeP and 3-ClEtP, respectively. Conversely to the limited stability of mono-chlorinated paraben by-products, the di-chlorinated ones wererather resistant to undergo further chlorine substitution reactionsor cleavage of the aromatic ring, even in presence of relativelyhigh concentrations of chlorine, Fig. 3. Therefore, if they aregenerated in a real life situation, their presence in the aquaticenvironment appears as feasible.

Experiments with tap water revealed the presence of threeadditional by-products for each paraben. On the basis of theinformation contained on their MS spectra, they were identi-

Fig. 3. Time-course of EtP and its chlorinated by-products in ultrapure waterbuffered at pH 7.3. Initial chlorine concentrations 0.4 mg L−1 (A), 1.6 mg L−1

(B) and 5 mg L−1 (C).

fied as brominated compounds, Fig. 4. As in the case of parentspecies and chlorinated transformation products, the bromi-nated species react with MTBSTFA rendering the correspondingdimethyl-tert-butylsilyl derivatives. Base peaks in their spectracorresponded to the loss of tert-butyl moiety and therefore theyappear at [M-57] m/z units. The most probable source of thesenew by-products is the existence of bromide traces in tap watersources. In presence of free chlorine, bromide is oxidized tobromine, which reacts with aromatic compounds through elec-trophilic substitution reactions. In order to verify this hypothesis,250 mL aliquots of a tap water sample (measured chlorine con-centration 0.53 mg L−1) were spiked with a mixture of the fourconsidered parabens, at 10 �g L−1, and allowed to react for20 min. The experiment was repeated using another aliquot ofthe same water sample spiked also with 1 �g L−1 of bromide,as potassium bromide. As depicted in Fig. 5 for ethyl paraben,


Fig. 4. MS spectra and proposed structures for brominated by-products of EtP.

a diminution in the peak areas for the chlorinated species wasobserved, whereas, signals for the brominated by-products rosesignificantly. In seems that both, chlorine and bromine, competeto react with parabens through substitution reactions.

Further experiments were accomplished using ultrapurewater spiked with a fixed concentration of free chlorine(0.4 mg L−1) and increasing levels of bromide: 1, 10 and65 �g L−1. Ratios between peak areas of the mono and di-halogenated by-products showed that, for all parabens, thebromination reaction is much more favourable than the chlori-nation one (Table 4). As a consequence, brominated species areexpected to be the major paraben transformation by-products inchlorinated tap water prepared from aquifers placed in coastalregions or areas with salt deposits, which can contain up to1 mg L−1 of bromide [18–20]. Although, data regarding the toxi-

Fig. 5. Influence of bromide (addition level 1 �g L−1) on the formation of chlo-rinated and brominated derivatives of EtP. Experiments performed using tapwater (free chlorine concentration 0.53 mg L−1) buffered at pH 7.3. Reactiontime 20 min.

city of halogenated paraben derivatives were not found, it is wellaccepted that, normally, brominated by-products generated dur-ing disinfection of tap water represent a higher health risk thanthe chlorinated ones [21].

3.2.4. Paraben by-products from personal care productsand occurrence in the aquatic media

From toxicological and environmental perspectives, it is rel-evant to know if the above halogenation reactions can take placealso when paraben-containing products get in contact with chlo-rinated tap water. In this situation, the available chlorine couldbe exhausted by other chemicals included in the formulation ofpersonal care products and thus parabens might remain unaf-fected. In order to evaluate this possibility, a bath gel, purchased

Table 4Ratios between peak areas of chlorinated and brominated paraben by-productsin ultrapure water containing an initial chlorine concentration of 0.4 mg L−1 andincreasing levels of bromide

Parentspecies

Peak area ratios Initial bromideconcentration (�g L−1)

1 10 65

MeP3-ClMeP/3-BrMeP 2.19 0.13 <10−5

3,5-DClMeP/3,5-DBrMeP 9.67 0.02 2.3 × 10−4

EtP3-ClEtP/3-BrEtP 2.51 0.16 <10−5

3,5-DClEtP/3,5-DBrEtP 10.32 0.02 <10−5

PrP3-ClPrP/3-BrPrP 2.43 0.12 <10−5

3,5-DClPrP/3,5-DBrPrP 12.72 0.02 1.7 × 10−4

BuP3-ClBuP/3-BrBuP 2.71 0.15 <10−5

3,5-DClBuP/3,5-DBrBuP 10.53 0.02 <10−5

Reaction time 20 min; pH 7.3.


Fig. 6. Time-course of BuP brominated by-products after mixing a bath gel withtap water containing 0.82 mg L−1 of free chlorine.

from a local market, was first diluted 10 times, using ultrapurewater, and then added to 250 mL aliquots of ultrapure (reference)and tap water samples (containing 0.82 mg L−1 of free chlorine),buffered at pH 7.3. Two series of experiments were carried outwith samples thermostated at 15 and 38 ◦C. After a given time,from 0 to 30 min, chlorine was removed and samples processedas described previously. Reference experiments revealed that thebath gel contained only non-halogenated parabens. Their con-centrations ranged from 200 to 2240 �g g−1. On the other hand,assays with tap water confirmed the formation of all describedchlorinated and brominated by-products, for the four parabenspresented in the gel, even considering reaction times as short as2 min. The decrease in the signals of parent species was about1.5-folds faster at 38 ◦C (average half-life 5.9 min) than at 15 ◦C(average half-life 8.5 min). Moreover, as shown in Fig. 6 forBuP, the formation of the brominated derivatives was enhancedat 38 ◦C, a rather common temperature in a water bath. Anyhow,they remained as minor by-products in comparison to the chlo-rinated ones. Additional experiments using different personalcare products, e.g. mouth rinse solutions, confirmed the halo-genation of those parabens included in their composition, when

Fig. 7. Chromatogram for a non-spiked wastewater sample (code 3, Table 5).GC–MS plots for the di-chlorinated by-products of MeP (m/z 277 + 279) andPrP (m/z 305 + 307).

mixed with tap water, data not given. Standards for most of theby-products identified in this study are not commercially avail-able; therefore, the yield of the described reactions could notbe calculated. However, considering a semi-quantitative esti-mation, chlorinated parabens might justify up to 80% of thedecrease observed in the concentrations of the correspondingparent compounds.

From the best of our knowledge, halogenated parabens do nothave any commercial application; therefore, their presence in theaquatic environment will serve as a further evidence of trans-formation reactions described in this work. Fig. 7 depicts theGC–MS traces of the di-chlorinated MeP and PrP by-products,

Table 5Concentrations of MeP and PrP paraben (�g L−1) in different raw sewage watersamples

Samplecode

Concentration (�g L−1) ± S.D. Peak area ratios (%)

MeP PrP 3,5-DClMeP/MeP (%)

3,5-DClPrP/PrP (%)

1 n.q. 0.98 ± 0.09 n.d. 18.42 0.06 ± 0.01 10.84 ± 0.64 520.7 6.63 0.49 ± 0.05 12.96 ± 0.65 323.0 15.94 n.q. 0.94 ± 0.08 n.d. 32.05 1.88 ± 0.12 0.67 ± 0.06 4.2 3.4

Ratios between peak areas for their di-chlorinated by-products and parentspecies. Sample volume: 1 L, n = 3 replicates; n.q. under quantification limits;n.d. non-detected.


together with their MS spectra, for a raw wastewater sample con-centrated 500-folds (sample number 3, Table 5). Table 5 gives theconcentrations of MeP and PrP in five raw wastewater samples,as well as the ratios between responses (peak areas) obtained fortheir di-chlorinated by-products and the corresponding parentspecies. In two of the considered samples, peak areas for 3,5-DClMeP were higher than those corresponding to MeP. Otherby-products of both parabens were not detected. EtP and BuPremained around the detection limits of the method in all sam-ples and none of their halogenated derivatives was found.

4. Conclusions

Parabens react with free chlorine rendering several halo-genated by-products. Chlorine levels usually contained in tapwater are enough to produce significant amounts of their chlo-rinated by-products in a few minutes. Therefore, consideringthe extensive employment of parabens in personal care prod-ucts, daily activities such as showering and bathing constitute asource of dermal exposition to paraben chlorinated by-products.Moreover, the di-chlorinated derivatives are highly resistant toundergo further oxidation reactions and their presence has beenfound for first time in sewage water samples. If minimal amountsof bromide are presented in tap water sources, halogenationreactions are shifted to the production of brominated parabens.Further studies are necessary to: (i) evaluate the potential heathrisks and possible endocrine disrupter activity of halogenatedparaben by-products, (ii) to know their fate in the environmentand (iii) to study the kinetic of paraben oxidation reactions usingalternative disinfectants such as ozone or chlorine dioxide.

Acknowledgements

Financial support from project DGICT CTQ2005-00425 andtechnical assistance from LABAQUA with free chlorine deter-

minations in water samples is acknowledged. PC acknowledgesa FPU grant from the Spanish Ministry of Education.

References

[1] Q. Zhang, M. Lian, L. Liu, H. Cui, Anal. Chim. Acta 537 (2005) 31.[2] E. Silva, N. Rajapakse, A. Kortenkamp, Environ. Sci. Technol. 36 (2002)

1751.[3] H.B. Lee, T.E. Peart, M.L. Svoboda, J. Chromatogr. A 1094 (2005) 122.[4] T. Benijts, W. Lambert, A. De Leenheer, Anal. Chem. 76 (2004) 704.[5] P.D. Darbre, A. Aljarrah, W.R. Miller, N.G. Coldham, M.J. Sauer, G.S.

Pope, J. Appl. Toxicol. 24 (2004) 5.[6] S.A. Snyder, P. Westerhoff, Y. Yoon, D.L. Sedlak, Environ. Eng. Sci.

20 (2003) 449.[7] S.D. Richardson, T.A. Ternes, Anal. Chem. 77 (2005) 3807.[8] S.D. Richardson, Trends Anal. Chem. 22 (2003) 666.[9] H. Gallard, U. Von Gunten, Environ. Sci. Technol. 36 (2002) 884.

[10] P. Westerhoff, Y. Yoon, S. Snyder, E. Wert, Environ. Sci. Technol. 39(2005) 6649.

[11] M. Deborde, S. Rabouan, H. Gallard, B. Legube, Environ. Sci. Technol.38 (2004) 5577.

[12] A. Alum, Y. Yoon, P. Westerhoff, M. Abbaszadegan, Environ. Toxicol.19 (2004) 257.

[13] P. Canosa, S. Morales, I. Rodrıguez, E. Rubı, R. Cela, M. Gomez, Anal.Bioanal. Chem. 383 (2005) 1119.

[14] K.L. Rule, V.R. Ebbett, P.J. Vikesland, Environ. Sci. Technol. 39 (2005)3176.

[15] S. Onodera, M. Ogawa, S. Suzuki, J. Chromatogr. 392 (1987) 267.[16] L.S. Clesceri, A.E. Greenberg, A.D. Eaton (Eds.), Standard Methods for

the Examination of Water and Wastewater, 20th ed., American WaterWorks Association, Maryland, 1998, pp. 461–465.

[17] K.E. Pinkston, D.L. Sedlak, Environ. Sci. Technol. 38 (2004) 4019.[18] S.D. Richardson, A.D. Thruston, C. Rav-Acha, L. Groisman, I.

Popilevsky, O. Juraev, V. Glezer, A.B. Mckague, M.J. Plewa, E.D. Wag-ner, Environ. Sci. Technol. 37 (2003) 3782.

[19] S.D. Richardson, A.D. Thruston, T.V. Caughran, P.H. Chen, T.W. Col-lette, T.L. Floyd, K.M. Schenck, B.W. Lykins, G. Sun, G. Majetich,Environ. Sci. Technol. 33 (1999) 3378.

[20] U. von Gunten, Water Res. 37 (2003) 1469.[21] S.D. Richardson, Anal. Chem. 75 (2003) 2831.

desarrollo de metodologÍa analÍtica para la …

Documents