7/31/2019 Berrier SAB 2011

1/8

Sensors andActuatorsB 160 (2011) 181188

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical

journa l homepage: www.elsevier .com/locate /snb

Enhancing the gas sensitivity ofsurface plasmon resonance with a nanoporoussilica matrix

Audrey Berrier a,, Peter Offermansb, Ruud Coolsa, Bram van Megena, Wout Knobenb,Gabriele Vecchia, Jaime Gmez Rivasa,c, Mercedes Crego-Calamab, Sywert H. Brongersmab

a Centre for Nanophotonics, FOMInstitute AMOLF c/o Philips Research Laboratories, High Tech Campus 4, 5656 AEEindhoven, The Netherlandsb Holst Centre/imec-nl,High Tech Campus 31, 5656 AEEindhoven, The Netherlandsc COBRAResearch Institute, EindhovenUniversity ofTechnology, P.O. Box 513,5600MB Eindhoven, The Netherlands

a r t i c l e i n f o

Article history:

Received 11 October 2010

Received in revised form 20June 2011

Accepted 14July 2011

Available online 27 July 2011

Keywords:

Nitrogen dioxide (NO2)

Gas sensing

Surface plasmon resonance (SPR)

Porphyrin

Porous matrix

Nanoporous silica

a b s t r a c t

Thedevelopmentofsensing schemes for the detectionofhealth-threateninggases is an attractive subject

for research towards novel integrated autonomous sensor systems. We report here on a novel way of

sensing NO2 by surface plasmon resonance (SPR) using a gas-sensitive layer composed of5,10,15,20-

Tetrakis(4-hydroxyphenyl)-21H,23H-porphine (2H-OHTPP) embedded in a nanoporous silicamatrix on

topofa gold thin film. Thesensingmechanismis basedonthemodification ofthe SPR condition due to gas

induced changes in theoptical properties ofthe sensing layer.We demonstrate that theuse ofnanoporous

silica as embeddingmatrix enhances the detection sensitivity compared to a polymer matrix with low

porosity. The second important finding ofthis work is that the active layer thickness plays a significant

role in the enhancement ofthe sensing response. The improvement is explained by the optimization of

the overlap between the field ofthe surface plasmon polariton and the active dielectric layer.

2011 Elsevier B.V. All rights reserved.

1. Introduction

Thedetectionofgreenhouse gases (CO2 , NO2, SO2, CH4) for out-

door monitoring and the control of health-threatening gases for

indoor monitoring and building health assessment sets steadily

growingdemands fora simple, cost-effective andsensitivemethod

for the detection of such gases in the ppb range. Many of the

existing sensors are bulky or power hungry due to the need of

a high operating temperature or large optical input power. For

outdoor monitoring, remote sensing ofgreenhouse gases is pos-

sible with tunable mid infrared laser sources which require mW

up to W of optical power [1]. For autonomous indoor monitor-

ing systems, optical methods using light emitting diodes (LED)

or halogen lamps with W of optical power are, in this respect,

more appropriate. Moreover, sensors relying on changes of the

optical properties ofan active sensing material operate at room

temperature and remain immune to electrical interferenceeffects.

In this work, we demonstrate surface plasmon resonance (SPR)

baseddetectionofNO2 with a gas sensitive layerconsisting ofpor-

phyrins [2] embedded in a nanoporous silica layer. We show that

the use ofporous silica as embeddingmatrix enhances the detec-

Corresponding author. Tel.: +31 402740158.

E-mail address: a.berrier@amolf.nl (A. Berrier).

tion sensitivity compared to a polymer matrix with low porosity.

We focus our attention towards ways ofenhancing the interac-

tion between the gas molecules, the porphyrins and the surface

plasmon polariton field. It is not the purpose ofthis work to fully

characterize a sensor, but rather to provide a method to improve

the gas sensing mechanism based on surface plasmon resonances

(SPRs).

The sensingmechanismatworkhere is based onSPRs,originat-

ing from the interaction ofsurface plasmon polaritons (SPPs) with

their surrounding environment. SPPs are electromagnetic modes

propagatingat theinterfacebetweenadielectric andametal,which

is oftengoldowingto its goodchemicalstability. TheSPPmodes are

extremely sensitive to changes in the permittivity oftheir dielec-

tric environment [3]. SPR sensors have been investigated formany

years in fields as varied as biotechnology [4], food monitoring [5]

or health-care[6]. SPR sensing ofNO2 has beendemonstratedwith

a limit ofdetection of1ppm [7,8]. Very recently a detection limit

of 70ppb was achieved using a silver layer in combination with

lock-in detection [9]. However, most ofthe proposed systems use

very thin layersofa few nanometers for the gas capture anddo not

fully exploit the decay ofthe intensity ofthe SPP from the surface,

which is typically a few hundreds ofnanometers. The concept of

a porous layer in combination with the SPR method has already

been reported in the literature [1012]. Here, we study a SPR-

basedsystemwithppb-leveldetectionwithouttheneed forlock-in

0925-4005/$ see frontmatter 2011 Elsevier B.V. All rights reserved.

doi:10.1016/j.snb.2011.07.030

http://dx.doi.org/10.1016/j.snb.2011.07.030http://dx.doi.org/10.1016/j.snb.2011.07.030http://www.sciencedirect.com/science/journal/09254005http://www.elsevier.com/locate/snbmailto:a.berrier@amolf.nlhttp://dx.doi.org/10.1016/j.snb.2011.07.030http://dx.doi.org/10.1016/j.snb.2011.07.030mailto:a.berrier@amolf.nlhttp://www.elsevier.com/locate/snbhttp://www.sciencedirect.com/science/journal/09254005http://dx.doi.org/10.1016/j.snb.2011.07.030

7/31/2019 Berrier SAB 2011

2/8

7/31/2019 Berrier SAB 2011

3/8

A. Berrier et al. / Sensors and Actuators B 160 (2011) 181188 183

Fig. 2. Schematic diagram ofthe experimental setup for gas sensing (a) diagram indicating the gas flow systemaswell as the optical flow cell and optical path. 1 indicates

the main flowpath, 2 is the purge path and 3 is the exhaust path (MFC, mass flowcontroller; sccm, standard cubic centimeter perminute; slm, standard liter per minute);

(b) schematic drawing zoomed around the sample. The electric field profile ofthe surface plasmon polariton at the interface between the gold layer and the active layer is

schematically represented by the redcurve.(For interpretation ofthe references to color in this figure legend, the reader is referred to the web version ofthe article.)

an excellent pathway for gas diffusion. The porous layer was

characterized by transmission measurements and ellipsometry.

Its thickness was controlled by spin coating speed and controlled

dilution following a calibration curve obtained by ellipsometry

and profilometer measurements. The absorbance spectrum ofthe

active layer is characterized by a peak at 420nm (Fig. 1b) corre-

sponding to the Soret band or B-band [10]. When the active layer

is exposed to NO2, the B-band red-shifts and an extra absorbancepeak appears around 680nm, the so-called Q-band. The develop-

ment ofthe latter absorbance peak induces a permittivity change

that can be sensitively detected following the evolution of the

intensity of the SPR at the corresponding wavelength. Sample

iv was prepared from an ethylcellulose solution in ethanol with

the same concentration ofporphyrins as with NPS solution. The

samplewas spin coated at 3000 rpm, and baked at 175 C for 15s.

Sample types ii, iii and ivwere obtainedby spin coating frominitial

solutions preparedwith identical porphyrin concentrations.

2.3. Gas setupandmeasurement procedure

Fig. 2a represents a schematic diagramofthe gas sensing setup

used in this work.NO2 is delivered into line 1 bya permeation tubeemitting2.3g/minwhenheatedat 40 C. N2 is used as carrier gas.

The gas flow is directed towards an optical flow cell after further

dilution in N2 to a concentration in the range 2906000ppb (line

2). Theflow is kept constant by introducing a mass flow controller

(MFC) and a vacuumpumpbehind the flow cell. Theminimumgas

concentration was limited by the emission rate ofthe permeation

tube and by the maximum flow allowed by the optical flow cell.

This limit is 290ppb in our setup. The gas concentration was cali-

brated by chemoluminescencemeasurements usingan EcoPhysics

chemoluminescence analyzer with a gas converter for catalytic

conversion ofNO2 into NO.Prior to gas exposure, nitrogengaswas

flushed over the sample in the flow cell for 20min. This procedure

removes remaining traces ofsolvent, moisture and oxygen from

the atmosphere ofthe flow cell before sensing. After gas exposure,

recovery ofthe active layer is performed within a few minutes by

gentle heating (75C) underN2 flow in order to desorb the trapped

NO2 molecules.

2.4. Optical methods

To characterize the behavior ofthe sensing layer, we have car-

ried outattenuatedtotal internal reflection(ATR)measurements intheKretschmann configuration [19] as well as specular reflectance

measurements.

The ATR method is used to detect changes in the SPR condi-

tion upon gas exposure. In this configuration (Fig. 2b), the sample

is fixed on a half-cylindrical lens acting as a prism by a droplet of

refractive indexmatchingliquid(n=1.51). A rotational stageallows

the independent setting ofthe angleofincidenceandofthe detec-

tion angle. A halogen lamp is used as the light source and a fiber

coupled spectrometer is used as a detector. Thepolarization ofthe

incident light is controlled with a polarizer.

The variation ofthe permittivity ofthe sensing layer due to the

reaction with the gas modifies the SPR condition, which is mon-

itored by a measurement ofthe spectrally and angularly resolved

ATRreflectance (SARR). TheSARR is ameasurementofthe couplingofthe incident light toSPPs.Thiscouplingtakesplacewhenthepar-

allel component ofthe wavevector ofthe incident radiation in the

prismwith respect to the interface,

k// =2

np sin , (1)

is equal to the wavevector ofthe SPP. For the case ofan interface

separating two semi-infinite media, the wavevector ofthe SPP is

given by:

kSPP 2

Re

metallayer

metal + layer

, (2)

where is the wavelength in vacuum, is the angle of inci-

dence, np is the refractive index ofthe prism, metal is the complex

7/31/2019 Berrier SAB 2011

4/8

7/31/2019 Berrier SAB 2011

5/8

A. Berrier et al. / Sensors and Actuators B 160 (2011) 181188 185

Fig. 3. ATR reflectanceofa samplewith active layer thickness t=65nm as a function ofthe wavelength andofthe angle ofincidence ofthe light with respect to the normal

to the layer. Thedip in reflectance(blue area) indicates the coupling to the surface plasmon at the gold-porphyrin layer interface. (a) before NO2 exposure; (b) after exposure

to 4.6ppm NO2; (c) after exposure to 350ppb NO2; (d) reflectance at 680nm before and after exposure to NO2 with a concentration of290ppb. (For interpretation ofthe

references to color in this figure legend, the reader is referred to the web version ofthe article.)

lowing) is defined in Fig. 5c. The magnitude ofthe splitting ismeasured at the anglewhere both reflectance dips reach down to

the same value.

To further investigate the dependence ofthe sensing behavior

on the thicknessofthe active layer, wemodel the reflectance ofthe

layeredsystemusing thetransfermatrixmethod, incorporating the

complex permittivity ofeach individual layer. The change in per-

mittivity ofthe active layer around 680nm due to gas interaction

in the active layer is calculated usinga Lorentzianmodel following

the equation:

() = b()+F20

20 2 i

(4)

where b is the complex permittivity ofthe background, 0 is thefrequency of the absorption line, is the damping rate of the

Fig. 4. (a) ATR reflectanceat 680nm as a function ofthe NO2 concentration at equilibrium. (b) Resonancewavelength as a function ofNO2 concentration at an angle of53.5 .

The grey lines are guides to the eye.

7/31/2019 Berrier SAB 2011

6/8

186 A.Berrieret al. / Sensors and Actuators B 160 (2011) 181188

Fig. 5. (a) Experimental spectrallyand angularly resolvedATR reflectance (SARR) plot for an activelayer thicknessh=65nmafter exposure to 350ppb NO2; (b) experimental

SARR plot for an active layer thickness h=130nm after exposure to 350ppb NO2; (c) angular cuts ofthe SARR plotsfor the two thicknesses (at the position ofthe respective

dashed lines in (a) and (b)); (d) calculated SARR plot for an active layer thickness of130 nm with F=40104. (For interpretation ofthe references to color in this figure

legend, the reader is referred to the web version ofthe article.)

transition and the frequency. Fis a parameterproportional to the

oscillator strength ofthe active layer, whichdepends on the num-

berofNO2porphyrininteraction sitesand,by extension,on thegasconcentration. The damping rate and the value ofFare obtainedfromthefit oftheexperimentaldata fora layer thicknessof130nm

and gasconcentrationof350ppb. TheFfactor is subsequently var-

ied proportionally to the gas concentration, whereas is, for sakeofsimplicity, kept constant (=3.21013 rads1). Fig. 5d presentsthe result of the modeling for the 130nm thick active layer with

F=40104. The agreement with the experimental SARR plot is

verygood, indicatingthevalidityofthemodel.Further, calculations

for various layerthicknesses andgasconcentrations are performed

in order to gain further insight into the sensing behavior and opti-

mize the sensing layer.

The plots in Fig. 6 are obtained point by point: for each thick-

ness and gas concentration we calculate the SARR plot fromwhich

is directly measured from the angular cross section withreflectance dips reaching down to the same value. Fig. 6 displays

the dependence of as a functionofthe layer thickness for threedifferent Fparameters. It also compares the calculations with the

experiments. Theexperimental points, correspondingto data from

Figs. 3b and 5a and b, are indicatedby the black diamondsand the

hexagon. From the modelwe find that increaseswith the layerthickness, in agreement with experimental observation. startsto level offas the thicknessoftheactive layerincreases furtherthan

about 300nm.

This dependence is explained by the overlap between the SPP

fieldprofileandthe activelayer. increaseswith increasingover-lap ofthe SPPfield profilewith the active layerandsaturateswhen

the SPP field profile decays fully within the active layer. We may

expect that the thickness for optimum sensitivity will depend on

the porosity of the embedding matrix ofthe active layer as it is

expected to be a trade-offbetween increased gas diffusion times

and increasedfield overlap with the active layer.We note that thediffusionofNO2 throughthe activelayerandits interactionwith the

porphyrinmoleculesmayhaveadirect impact onthespatialdepen-

dence ofthe imaginary partofthe refractive index k=k(z,),whichis not taken into account in the presentmodeling. Themodeling of

the diffusion ofNO2 molecules through the matrix, their adsorp-

tion and reactionwith theporphyrin molecules is out ofthe scope

Fig. 6. Evolution ofthe calculated as a function ofthe active layer thicknessfor three different values ofthe oscillator coefficient F, the blackdots correspond to

experimentalpoints.The diamondscorrespond toanexposureto 350ppbofNO2 and

a layer thickness of65 and 130nm.The black hexagon corresponds to an exposure

to 6000ppb ofNO2.

7/31/2019 Berrier SAB 2011

7/8

http://dx.doi.org/10.1021/nn201077r

7/31/2019 Berrier SAB 2011

8/8

188 A.Berrieret al. / Sensors and Actuators B 160 (2011) 181188

the Sensors andActuatorsgroupwithin theWirelessAutonomous TransducerSolu-tions program at Holst Centre/IMEC, where she is currently the program managerfor the ultra lowpower sensors group.

Sywert Brongersmaobtained his Ph.D. at the FreeUniversity ofAmsterdam in thefield ofsuperconductivity. After a postdoc at the University ofWestern Ontario

(Canada) concerning clustering phenomena on semiconductor surfaces, hejoinedthe Advanced Silicon Processing division ofIMEC (Leuven, Belgium) in 1998. Since2004 he was a principal scientist in both the Cu/Low-k integration and the Nano-technology industrial affiliation programs. In 2006, he transferred to the WirelessAutonomousTransducer Systemsprogram at Holst Centre/IMEC,wherehe is a cur-rently a Sr. principal researcher.