influence of electromagnetic interferences on the mass sensitivity of love mode surface acoustic...

Sensors and Actuators A 123–124 (2005) 360–369

Influence of electromagnetic interferences on the masssensitivity of Love mode surface acoustic wave sensors

Laurent A. Francisa,b,∗, Jean-Michel Friedtc, Patrick Bertranda

a PCPM, Universite catholique de Louvain, Croix du Sud 1, B-1348 Louvain-la-Neuve, Belgiumb IMEC, Kapeldreef 75, B-3001 Leuven, Belgium

c LPMO, Universite de Franche-Comte, La Bouloie, 25030 Besancon, France

Received 13 September 2004; received in revised form 2 March 2005; accepted 3 March 2005Available online 25 April 2005

Abstract

Surface acoustic waveguides have found an application for (bio)chemical detection. The mass modification due to surface adsorption leadsto measurable changes in the propagation properties of the waveguide. Among a wide variety of waveguides, the Love mode device hasbeen investigated because of its high mass sensitivity. The acoustic signal launched and detected in the waveguide by electrical transducersi nsor, createsi ustic device.W t identicalf s completedb experimenti©

K

1

fmsertatwmsi

ted an

sub-ma-

Lovekes it

guid-theons.sili-ctors

de-the

cture

0d

s accompanied by an electromagnetic wave; the interaction of the two signals, easily enhanced by the open structure of the senterference patterns in the transfer function of the sensor. The interference peaks are used to determine the sensitivity of the aco

e show that electromagnetic interferences generate a distortion in the experimental value of the sensitivity. This distortion is noor the two classical instrumentation of the sensor that are the open and the closed loop configurations. Our theoretical approach iy the experimentation of an actual Love mode sensor operated under liquid conditions and in an open loop configuration. The

ndicates that the interaction depends on frequency and mass modifications.2004 Elsevier B.V. All rights reserved.

eywords: Surface acoustic wave; Electromagnetic wave; Love mode; Interferences; Gravimetric sensitivity; Biosensor

. Introduction

Acoustic waves guided by the surface of solid structuresorm waveguides used as delay lines and filters in telecom-unications[1]. Waveguides support different modes with

pecific strain and stress fields[2]. The acoustic velocity ofach mode depends on different intrinsic and extrinsic pa-ameters such as the mechanical properties of the materials,he temperature or the applied pressure. Waveguides are useds sensors when the velocity change is linked to environmen-

al changes. For gravimetric sensors, the outer surface of theaveguide is exposed to mass changes. Due to the confine-ent of the acoustic wave energy close to the surface, these

ensors are well suited for (bio)chemical sensors operatingn gas or liquid media. Among a wide variety of waveguides

∗ Corresponding author. Tel.: +32 16 288564; fax: +32 16 281097.E-mail address: [email protected] (L.A. Francis).

used for that purpose, Love mode sensors have attracincreasing interest during the last decade[3,4]. A Love modeis guided by a solid overlayer deposited on top of astrate material. The usual substrates are piezoelectricterials like quartz, lithium tantalate and lithium niobate[5].Associated to specific crystal cut of these substrates, themode presents a shear-horizontal polarization that masuitable for sensing in liquid media.

Current research in Love mode sensors concerns theing materials in order to optimize the sensitivity, that isvariation of the acoustic signal under surface modificatiTypical materials under investigations are dielectrics likecon dioxide and polymers, and more recently semiconduwith piezoelectric properties like zinc oxide[6–8]. Althoughthe dispersion relation for Love mode is well set and thependence of the sensitivity of the liquid loaded sensor tooverlayer thickness has been thoroughly investigated[9–11],little has been devoted to study the role played by the struof the sensor and their transducers.

924-4247/$ – see front matter © 2004 Elsevier B.V. All rights reserved.oi:10.1016/j.sna.2005.03.030

L.A. Francis et al. / Sensors and Actuators A 123–124 (2005) 360–369 361

In this paper, we investigate the role played by the structureof the sensor and by the interferences between the acousticand the electromagnetic waves on the sensitivity. In the firstpart, we present a general model of the transfer function in-cluding the influence of electromagnetic interferences. In thesecond part, we show how these interferences modify the sen-sitivity in open and closed loop configurations of the sensor.Finally, these effects are illustrated experimentally on a Lovemode sensor.

2. Modeling

Waveguide sensors consist of a transducing part and asensing part. The transducing part includes the generation andthe reception of acoustic signals and their interfacing to anelectrical instrumentation. The most common transducers arethe widespread interdigital transducers (IDTs) on piezoelec-tric substrates introduced by White and Voltmer in 1965[12].Although the transducing part can be involved in the sensingpart, practical sensing is confined to the spacing between thetransducers. This confinement takes especially place whenliquids are involved since these produce large and unwantedcapacitive coupling between input and output electrical trans-ducers. This coupling dramatically deteriorates the transferfunction and is an important issue for the instrumentation andt

d byt ensorc t oura rationp ovem n-s nceb esh entTww roupv

thes sm s,t itya ala yV pvd

τ

lk be-t edb ustic

Fig. 1. Structure of the acoustic device.

wave and therefore is detected at the output transducer with-out noticeable delay. At the output transducer, the two kindsof waves interact with an amplitude ratio, denoted byα, thatcreates interference patterns in the transfer functionH(ω) ofthe delay line. The transfer function itself is given by the ratioof the output to the input voltages. The transfer function withelectromagnetic interferences is modeled by the followingequation:

H(ω) = HT(ω) exp(−iωτ)︸︷︷︸delay line

+ αHT(ω)︸︷︷︸EM coupling

. (2)

The transfer functionHT(ω) is associated to the design ofthe transducers. The total transfer function can be rewrittenasH(ω) = ‖H(ω)‖ exp(iφ) where expressions for the am-plitude ‖H(ω)‖ and the phaseφ are obtained with help ofcomplex algebra:

‖H(ω)‖ = ‖HT(ω)‖‖√

1 + 2α cos(ωτ) + α2‖; (3)

φ = φ0 − arctan

(sin(ωτ)

α + cos(ωτ)

). (4)

The phaseφ0 corresponds to the packaging of the sensorand is due to different aspects linked to the instrumentation.It will be assumed independent of the frequency and of thesensing event. The synchronous frequencyωT = 2πfT is de-t al tot ho ityλ

het te Inter-f e ob-st

f

he packaging of the sensors.The sensor itself is configured as a delay line forme

wo transducers separated by a certain distance. The san also be configured as a resonator but we will restricpproach to the delay line configuration because the operinciple in these two configurations is not similar. The Lode sensor is sketched inFig. 1. Transducers with a co

tant apodization are identified to their midpoint; the distaetween the midpoints isL and the interdigitated electrodave a periodicityλT. The sensing part is located betwe

he transducers and covers a total lengthD so thatD ≤ L.he guided mode propagates with a phase velocityV = ω/k,hereω = 2πf is the angular frequency andk = 2π/λ is theavenumber. The waveguide is dispersive when the gelocity (Vg = dω/dk) differs from the phase velocity.

The velocity is a function of the frequency and ofurface densityσ = M/A for a rigidly bound and non viscouassM per surface areaA. For an uniformly distributed mas

he surface density is rewritten in terms of material densρnd thicknessd by σ = ρd. The phase velocity for an initind constant massσ0 is denotedV0, and the group velocitg0. In the sensing part, the phase velocity isV and the grouelocityVg. According to this model, the transit timeτ on theelay line is given by

= D

V+ L − D

V0. (1)

Electromagnetic interferences are due to the cross-taween the IDTs[13]. The electromagnetic wave (EM) emitty the input transducer travels much faster than the aco

ermined by the design of the IDTs and is generally equhe maximum amplitude of‖HT(ω)‖ when the wavelengtf the acoustic waveλ0 matches the transducers periodicT.

The relations(3) and (4)are the sources of ripples in transfer function at the ripple frequencyω � 2π/τ, its exacxpression depends also of the dispersion on the line.erence peaks corresponding to the maximum effect arerved at quantified frequenciesfn when cos(2πfnτ) = −1,hat is for frequencies such that

n = 2n + 1

2τ(5)

362 L.A. Francis et al. / Sensors and Actuators A 123–124 (2005) 360–369

wheren ∈ N is the interference mode number. A direct rela-tion to the velocity in the sensing area is obtained from thislatter equation as seen by replacing the transit timeτ by itsdefinition:

V = 2DV0fn

(2n + 1)V0 + 2(D − L)fn

. (6)

The interference mode numbers are determined by con-sidering the uncovered delay line; in such caseV = V0 andD = L, andn for the interference peak located below thesynchronous frequency (i.e. forfn ≤ fT) is given by

n =⌊

L

λT− 1

2

⌋(7)

while the other peaks are labeled subsequently to their posi-tion with respect to the peak referenced by Eq.(7).

The relative amplitude peak to peak of the perturbationon the amplitude has a maximum effect (in dB) equals to40 log[(1+ α)/(1 − α)]. The amplitude (in dB and normal-ized to have‖HT(ω)‖ = 1) and the phase (in radians) as afunction of the frequency are simulated inFigs. 2–5for dif-ferent values ofα.

Under the influence of the interferences, the phase hasdifferent behaviors function ofα:

(1) whenα = 0 (no interferences), the phase is linear withthe frequency and has a periodicity equal to 2π (Fig. 2);

(2) whenα < 1, the phase is deformed but has still a peri-odicity equal to 2π (Fig. 3);

(3) when α = 1, the phase has a periodicity equal toπ

(Fig. 4);(4) whenα > 1, the periodicity is lower thanπ (Fig. 5);(5) whenα → ∞, the phase is not periodic anymore and its

value tends toφ0.

This specific behavior of the phase under the influence ofthe electromagnetic interferences has to be considered whileevaluating the sensitivity.

Fig. 2. Relative insertion loss (top) and phase (bottom) of the transfer func-tion for α = 0.

Fig. 3. Relative insertion loss (top) and phase (bottom) of the transfer func-tion for α = 1/2.




3. Sensitivity

Changes in the boundary condition of the waveguide dueto the sensing event modify phase and group velocities. Asconsequence, the transit time of the delay line and the phase ofthe transfer function are modified. The sensing event is quan-tified by recording the phase shift at a fixed frequency (openloop configuration) or the frequency shift at a fixed phase(closed loop configuration). This quantification gives rise tothe concept of sensitivity. The sensitivity is not an uniqueconcept for acoustic sensors because various parameters in-fluence the acoustic velocity. As example of such parame-ters, there is the density and the viscosity of liquid solutionsand adsorbed biomolecules film when the device is used asbiosensor. The sensitivity is the most important parameter indesign, calibration and applications of acoustic waveguidesensors. Its measurement must be carefully addressed in or-der to extract the intrinsic properties of the sensor.

3.1. Definitions of the sensitivity

The velocity sensitivity SV is defined by the change ofphase velocity as a function of the surface density change ata constant frequency. Its mathematical expression is given by[10]:

S

∣sing

a od-i ocityVs parth rties,l

oc-i encys theo Thep

S

a

S

3

etici thep thei se oft d oft lves

function of the frequency and of the surface density:

φ(ω, V (ω, σ), V0(ω)) = −ωτ (11)

φ(ω, V (ω, σ), V0(ω)) = −ω

(D

V+ L − D

V0

). (12)

Therefore, the total differential of the phase is

dφ =(

∂φ

∂ω

∣∣∣∣V,V0

+ ∂φ

∂V

∣∣∣∣ω,V0

∂V

∂ω

∣∣∣∣σ

+ ∂φ

∂V0

∣∣∣∣ω,V

dV0

dω

)dω

+ ∂φ

∂V

∣∣∣∣ω,V0

∂V

∂σ

∣∣∣∣ω

dσ (13)

dφ = ∂φ

∂ω

∣∣∣∣σ

dω + ∂φ

∂σ

∣∣∣∣ω

dσ. (14)

The derivative of the phase velocity as a function of thefrequency comes from the definitions of phase and groupvelocities; at constant surface density, we have from[11]:

∂V

∂ω

∣∣∣∣σ

= k−1(

1 − V

Vg

); (15)

dV0 = k−1(

1 − V0)

. (16)

tia-t

a

τ

3

ex-c eeno ationw

V = 1

V

∂V

∂σ

∣∣∣ω

. (8)

The definition reflects the velocity change in the senrea only while outside this area the velocity remains unm

fied. The expression is general because the initial velof the sensing part does not need to be equal toV0; this

ituation occurs in practical situations where the sensingas a selective coating with its own mechanical prope

eading to an initial difference betweenV andV0.To link the sensitivity (caused by the unknown vel

ty shift) to the experimental values of phase and frequhifts, we introduce two additional definitions related topen and the closed loop configurations, respectively.hase sensitivity Sφ is defined by

φ = 1

kD

dφ

dσ, (9)

nd thefrequency sensitivity Sω is defined by

ω = 1

ω

dω

dσ. (10)

.2. Phase differentials without interferences

In order to point clearly the effects of the electromagnnterferences on the different sensitivities presented inrevious section, we calculate the phase differentials in

deal case of no interferences. For that case, the phahe transfer function is a function of the frequency anhe velocities in the different parts of the sensor, themse

dω 0 Vg0

The other partial differentials are obtained by differenion of Eq.(11):

∂φ

∂ω

∣∣∣∣V,V0

= −τ; (17)

∂φ

∂ω

∣∣∣∣σ

= −τ − ωτ

ω

∣∣∣σ

(18)

∂φ

∂ω

∣∣∣∣σ

= −τg; (19)

∂φ

∂V

∣∣∣∣ω,V0

= ωD

V 2 ; (20)

∂φ

∂V0

∣∣∣∣ω,V

= ω(L − D)

V 20

. (21)

The time of flightτg introduced in Eq.(19) is calculateds

g = D

Vg+ L − D

Vg0. (22)

.3. Open loop configuration

In the open loop configuration, the input transducer isited at a given frequency while the phase difference betwutput and input transducers is recorded. This configurith a constant frequency has dω = 0 in Eq. (13); related


phase variations caused by surface density variations are ob-tained by

dφ

dσ= ∂φ

∂V

∣∣∣∣ω,V0

∂V

∂σ

∣∣∣∣ω

(23)

dφ

dσ= ∂φ

∂V

∣∣∣∣ω,V0

VSV . (24)

In the absence of interferences, phase variations obtainedexperimentally are directly linked to velocity changes by theproductkD involving the geometry of the sensor as seen byreplacing Eq.(20) in Eq.(24):

dφ

dσ= kDSV . (25)

In other words:Sφ = SV when there are no interferences.In a first approximationk is assumed equal tokT, an as-sumption valid as long as the phase shift is evaluated closeto the synchronous frequency and for waveguides with lowdispersion. The wavelength is only known when the sensingpart extends over the transducers (D = L). In that case, thetransfer function of the IDTs is modified accordingly to thevelocity changes. In practice, the value of the sensitivity isslightly underestimated to its exact value sincek ≤ kT, thee

ren-t n-t

a attere

S

thep ef isa thet in-g es itiv-i

c sl

α

α

Fig. 6. Phase sensitivity at constant frequency as a function of the relativefrequency for different values of simulated interferences obtained by Eq.(27).

3.4. Closed loop configuration

In the closed loop configuration, the frequency is recordedwhile a feedback loop keeps the phase difference betweenoutput and input transducers constant. The configuration atconstant phase has dφ = 0, the variation of the frequency asa function of the mass change is given by introducing thiscondition in Eq.(14):

dω

dσ=(

∂φ

∂σ

∣∣∣∣ω

)(∂φ

∂ω

∣∣∣∣σ

)−1

. (30)

The upper term is replaced by Eq.(24). The phase slopeas a function of the frequency at constant mass is obtainedby differentiation of Eq.(4):

∂φ

∂ω

∣∣∣∣σ

= −(

1 + α cos(ωτ)

1 + 2α cos(ωτ) + α2

)τg. (31)

We can establish a finalized equation taking into accountthe electromagnetic interferences by combining Eqs.(24),(26) and (31)in Eq.(30):

Sω = DSV

Vτg. (32)

fre-q ces.H i-t en-s line.A andt itcd

rror being less than 5%.In the case where interferences occur, the partial diffe

ial of φ with respect to the velocity is obtained by differeiation of Eq.(4):

∂φ

∂V

∣∣∣∣ω,V0

=(

1 + α cos(ωτ)

1 + 2α cos(ωτ) + α2

)ωD

V 2 , (26)

nd the phase sensitivity is obtained by combining the lquation with Eq.(24):

φ =(

1 + α cos(ωτ)

1 + 2α cos(ωτ) + α2

)SV . (27)

The influence of electromagnetic interferences onhase sensitivity is simulated inFig. 6 versus the relativ

requency for different values ofα. The phase sensitivitylways different compared to the velocity sensitivity. For

hreshold valueα = 1, the phase sensitivity present a sularity and is undefined; for higher values ofα, the phasensitivity is always underestimated to the velocity sensty.

The interference peaks permit a direct evaluation ofα be-ause at these points cos(ωτ) = −1 and Eq.(27) becomeinear withα:

= 1 − V 2

ωD

∂φ

∂V

∣∣∣∣ω,V0

(28)

= 1 − SV

Sφ

. (29)

At the opposite of the open loop configuration, theuency sensitivity is not influenced by the interferenowever, as indicated by Eq.(32), the frequency sens

ivity is strongly dependent of the structure of the sor and the dispersion characteristics of the delays result, the link between the frequency sensitivity

he velocity sensitivity is difficult to exploit althoughan be noticed thatSω ≤ SV sinceVg ≤ V for Love modeevices.


4. Experimental results

For the practical consideration of the described and mod-eled behavior, we investigated a Love mode sensor. It wasfabricated and tested under liquid condition in the open loopconfiguration to evaluate the influence of the electromagneticinterferences. In a first part, the sensor fabrication and in-strumentation is described, followed in a second part by theapplication of the model to these results to demonstrate theinfluence of the interferences on the sensitivity of the sensor.

4.1. Sensor fabrication and instrumentation

The Love mode was obtained by conversion of a surfaceskimming bulk wave (SSBW) launched in the direction per-pendicular to the crystallineX axis of a 500�m thick ST-cut(42.5◦ Y-cut) quartz substrate. The conversion was achievedby a 1.2�m thick overlayer of silicon dioxide deposited onthe top side of the substrate by plasma enhanced chemicalvapor deposition (Plasmalab 100 from Oxford Plasma Tech-nology, England). Via were etched in the silicon dioxide layerusing a standard SF6/O2 plasma etch recipe. This processstopped automatically on the aluminum contact pads of thetransducers.

The transducers consist of split fingers electrodes etchedin 200 nm thick sputtered aluminum. The fingers are 5�mw tyλ rlapo fe ntero

e leftb n andl x-p Thefi sen-s ng1 rtzg tofi

poxyp dedo s tot cov-e Thes ,r wayo asr dur-i intso ini-ta the2 hings xi-m

Fig. 7. Initial aspect of the experimentally recorded transfer function of theLove mode sensor with (dashed line) and without (solid line) an overlayerof 200 nm of gold. This device presents an initial phaseφ0 = π, leading toa vertical offset byπ compared to the simulated phase curve represented inFig. 3.

4.2. Correlation of the results with the model

The correlation of the experimental results with the modelis presented in two steps. In the first step, we show the cal-culation of the phase velocity from the interference peaks;and in the second step, we evaluate the mass sensitivityin the open loop configuration by the delay phase angleand the phase velocity variations recorded during the goldetching.

The record of the interference peaks frequencyfn duringa sensing event permits to follow the evolution of the phasevelocity in the sensing area either for constant and integervalues of the interference mode numbersnas given by Eq.(6),

F dif-f andd d1

ide and equally spaced by 5�m. This defines a periodiciT of 40�m. The acoustic aperture defined by the ovef the fingers is equal to 80λT (=3.2 mm), the total length oach IDT is 100λT (=4 mm) and the distance center to cef the IDTs is 225λT (L = 9 mm,D = 5 mm).

The sensing area was defined by covering the spacetween the edges of the IDTs by successive evaporatio

ift-off of 10 nm of titanium and 50 nm of gold in a first eeriment, and 200 nm of gold in a second experiment.ngers were protected against liquid by patterning photoitive epoxy SU-8 2075 (Microchem Corp., MA) defini20�m thick and 80�m wide walls around the IDTs. Qualasses of 5 mm× 5 mm were glued on top of the wallsnalize the protection of the IDTs[14].

The device was mounted and wire-bonded to an erinted circuit board and its transfer function was recorn a HP4396A Network Analyzer. This setup correspond

he open loop configuration. Epoxy around the devicered and protected it and defined a leak-free liquid cell.ensing area was immersed in a solution of KI/I2 (4 and 1 gespectively, in 160 ml of water) that etched the gold af the surface[15]. The transfer function of the device wecorded every 4 s (limited by the GPIB transfer speed)ng the etching of the gold with a resolution of 801 pover a span of 2 MHz centered around 123.5 MHz. Theial transfer function of the device is presented inFig. 7withnd without gold. The transfer function during etching of00 nm is shown at two moments (44 and 356 s after etctart) inFig. 8. The total time for this etching was approately 620 s.

ig. 8. Aspect of the experimentally recorded transfer function at twoerent moments of the etching of 200 nm of gold (solid line after 44 sashed line after 356 s). The solid line shows a value ofα close to 1 aroun23.5 MHz.


Fig. 9. Interferences mode in the amplitude of the transfer function as afunction of time and frequency.

either by sampling the mode numbers at a constant frequency.Fig. 9plots the interference peaks versus time and frequencyfor the etching of the 200 nm thick gold layer; the interferencemode numbersn were attributed according to Eqs.(6) and (7)with V0 = 4940 m/s (given by the synchronous frequency offT = 123.5 MHz times the transducers periodicityλT).

The evolution of the velocity in the sensing area with timeis representative of the etching rate of the gold layer andis plotted for three different frequencies (123.5, 123.75 and124 MHz) inFig. 10. At three different frequencies, the valuesof velocity should differ as a function of the group velocity.This effect is seen better when the probing frequencies aretaken far away from each others and for a strongly dispersivedelay line, which is not the case for the experimental devicepresently used.

At constant frequency, the peaks are spaced by an unitvariation ofn, therefore the velocity difference measured be-

F g thee cy.

Fig. 11. Evaluation ofα at the position of the interference peak.

tween two peaks is obtained by differentiation of Eq.(6)withrespect ton:

∂V

∂n

∣∣∣∣fn,V0

= − 4DV 20 fn

[(2n + 1)V0 + 2(D − L)fn]2, (33)

which gives a variation roughly equals to−40 m/s betweentwo peaks at the three sampling frequencies. From the acous-tic velocity variation (4610 m/s for 200 nm gold to 4940 m/swhen all the gold is etched) and by assuming that gold hasa density ofρ = 19.3 g/cm3, we have an evaluation ofSVequals to−173 cm2/g. Because the phase loses its period-icity for the thick gold layer, we were not able to deter-mine a value for the phase variation and consequently wehave no value forSφ. The Eq.(28) was employed to esti-mate the value ofα at the interference peak; the result isdisplayed inFig. 11that demonstrates a variation ofα withthe frequency. Around the synchronous frequency,α equals0.33 and the phase has a periodicity of 2π; but as the fre-quency is far from the synchronous frequency,α clearlychange above the critical value of 1 (in the present case,α = 5.7). The consequence is seen in the phase that presentsat this point of calculation a positive slope and a periodicitybelowπ.

We applied the same procedure to the thinner gold layerof 50 nm.Fig. 12 shows the transfer function recorded be-f um-b rom4− tch-i thes ergyi

n-c ati ottedv s-t e

ig. 10. Evaluation of the acoustic velocity on the sensing area durintching of the gold as a function of time for different values of frequen

ore and after the gold etching; the interference mode ner 225 has been followed and give a velocity varying f876.5 to 4940 m/s. The resulting velocity sensitivity isSV =96 cm2/g. This value is lower than the one obtained by e

ng of the thick gold layer since a thicker layer enhancesensitivity due to a better entrapment of the acoustic enn the top guiding layer.

The phase sensitivitySφ could be calculated for frequeies whereα remained inferior to the critical value of 1, ths close to the synchronous frequency. The result is plersus the frequency inFig. 13and compared with the eimated value ofSV while the values ofα indicated on th


Fig. 12. Transfer function before and after the etching of 50 nm of gold. Thearrows indicate the interference mode 225 followed to estimate the velocitysensitivity.

Fig. 13. Phase sensitivity relative to the velocity sensitivity as a functionof the frequency and computed from the experimental data obtained byetching 50 nm of gold. Oscillations are attributed to the electromagneticinterferences.

graph have been estimated at the interference peaks thanks toEq.(29). The graphs shows that the interferences modify thevalue of the sensitivity as given by Eq.(27). A comparison oftheFigs. 6 and 13shows the correlation between the theoret-ical modeling of the effects of electromagnetic interferenceson the sensitivity of the surface acoustic waveguide sensorand the experimental results.

5. Discussion

Electromagnetic interferences have a clear effect on thetransfer function of the acoustic device because of the ripplesthey cause. The interaction modeled as a constant factorα

is specific to each device and must be identified by a care-

ful inspection of the transfer function. The amplitude of thetransfer function peak to peak is supposed to be the productbetween the transfer function of the transducers and the in-terference, and therefore an evaluation ofα is possible if thetransfer function of the transducers only is known. However,the experiment shows thatα is a function of the frequencyand the surface density, indicating that finding its exact valueis not straightforward. Only the phase indicates whetherα ishigher or lower than one.

In term of sensitivity, whenα ≥ 1 the phase has a period-icity P in the range 0–π. We suggest the following correctionto the experimental phase sensitivity:

Sφ = 2π

P

1

kD

dφ

dσ. (34)

This modification gives a better evaluation of the velocitysensitivity by stretching the phase of the transfer function to2π. Only the extraction ofP is not immediate since it dependsuponα. From a physical point of view,α indicates the strengthof the electromagnetic wave in comparison with the acousticwave. For a constant amplitude of the EM wave, a higherα

stands for a larger attenuation of the acoustic wave; its precisevalue is an indication of the actual attenuation of the acousticwave along the delay line.

The observation of the interference peaks in the experi-mental part was facilitated by the large velocity change in-d ondst nc ppli-c rdero hes rs ofm sen-s lm nt de-p .e.a ichi hee tiono sure-m stantv e oft cep Eq.(

t e.T cmr ichi sticw

pos-s inga aves

uced by the gold coating. Indeed, 50 nm of gold correspo a surface density of 96.5�g/cm2, a relatively large shift iomparison to the targeted (bio)chemical recognition aation where molecules films surface density are in the of hundreds of ng/cm2 and even lower. The calibration of tensitivity is best recorded by adding or etching thin layeaterials and that under the operating conditions of the

or, especially if liquids are involved[16]. In (bio)chemicaeasurements, the precision on the velocity measuremeends upon the assessments on the initial conditions (iV0ndn) but also on the induced variation of velocity, wh

s function of the velocity sensitivity of the waveguide. Tvaluation of the mass sensitivity by the frequency variaf an interference peak is identical to a closed loop meaent locked on the interference peak instead on a con

alue of the phase. For the detection of a minimum valuhe surface densityσ, the frequency shift of an interfereneak must be measured with a precision estimated from32):

fn = DSVfn

Vτgσ, (35)

hat givesfn/σ � −6.5 cm2 Hz/ng in the present cashe detection of a monolayer of proteins, about 400 ng/2,equires to detect a frequency variation of 2.6 kHz, whs compatible with the instrumentation of surface acouaveguide sensors.One benefit of our calculation method resides in the

ibility to still measure the acoustic velocity in the sensrea even when the electromagnetic and the acoustic w


are strongly interfering, in particular forα greater than one.Strong interferences are unwanted in an experimental set-upbecause they prevent the correct electrical measurement ofthe sensor. It must be noticed that for these high values ofinterference, the electromagnetic wave amplifies the acous-tic signal as seen inFig. 5, which could be of interest forthe operation of the device even in conditions where theacoustic signal is weak. Finally, the presented method op-erates directly on the raw signal of the acoustic device thusavoiding a lost or a modification of the physical informationit carries.

6. Conclusion

We have proposed a model for surface acoustic waveg-uides used as sensors. The model shows the influence of elec-tromagnetic interferences caused by interdigital transducerson the velocity sensitivity in open and closed loop configu-rations. In both cases, the dimensions of the delay line andthe sensing part influence the experimental value of phase orfrequency shifts.

The interference peaks in the transfer function offer anunique possibility to access the information about the acous-tic phase velocity in the sensing area. The velocity sensitivitywas calculated directly from these peaks.

, thep er-e ongim o ob-t

, thef ro-p h oft ucerI sivep

n thet uidc peri-m h thef

f thee s act-i nceo sur-f

A

portw theg ndt p-

ported by the Fonds pour la Formationa la Recherche dansl’Industrie et dans l’Agriculture (F.R.I.A., Belgium).

References

[1] C. Campbell, Surface Acoustic Wave Devices and Their Signal Pro-cessing Applications, Academic Press, San Diego, 1989.

[2] B.A. Auld, Acoustic Fields and Waves in Solids, vol. 2, Wiley, NewYork, 1973.

[3] G.L. Harding, J. Du, P.R. Dencher, D. Barnett, E. Howe, Love waveacoustic immunosensor operating in liquid, Sens. Actuator A Phys. 61(1997) 279–286.

[4] E. Gizeli, Design considerations for the acoustic waveguide biosensor,Smart Mater. Struct. 6 (1997) 700–706.

[5] F. Herrmann, M. Weinacht, S. Buttgenbach, Properties of sensorsbased on shear-horizontal surface acoustic waves in LiTaO3/SiO2 andQuartz/SiO2 structures, IEEE Trans. Ultrason. Ferroelectr. Freq. Con-trol 48 (2001) 268–273.

[6] A. Rasmusson, E. Gizeli, Comparison of poly(methylmethacrylate) andNovolak waveguide coatings for an acoustic biosensor, J. App. Phys.90 (2001) 5911–5914.

[7] G.L. Harding, Mass sensitivity of Love-mode acoustic sensors incor-porating silicon dioxide and silicon-oxy-fluoride guiding layers, Sens.Actuator A Phys. 88 (2001) 20–28.

[8] K. Kalantar-Zadeh, W. Wlodarski, Y.Y. Chen, B.N. Fry, K. Galatsis,Novel Love mode surface acoustic wave based immunosensors, Sens.Actuator B-Chem. 91 (2003) 143–147.

[9] Z. Wang, J.D.N. Cheeke, C.K. Jen, Sensitivity analysis for Love mode942.s in

aveys. 92

ace

itivi-amics,1996e In-996,

id) 353–

York,

m-cation.

L ori-e eL greea hes nsinga

J S,F ousticw dy ont ion oft is nowa

In an open loop configuration and with interferenceshase shift is disturbed and the sensitivity is over- or undstimated to the value of the velocity sensitivity. For str

nterferences, the phase has a periodicity lower than 2π thatust be considered when normalizing the phase shift t

ain a correct figure of the sensitivity.In a closed loop configuration and with interferences

requency shift is not disturbed. The frequency shift is portional to the sensitivity by the ratio between the lengt

he sensing area and the distance separating the transdn addition, the frequency shift is influenced by the disperroperties of the waveguide.

The influence of the electromagnetic interferences oransfer function of a Love mode sensor operating in liqonditions was presented for a comparison. From the exent it appears that the interferences are function of bot

requency and the surface density.For future investigations, an analytical expression o

lectromagnetic-acoustic interaction and the parameterng on it have to be identified in order to reduce the influer, on the opposite, to enhance the velocity sensitivity of

ace acoustic waveguides.

cknowledgements

The authors are thankful to N. Posthuma for the supith the PECVD tool, to R. De Palma for the help withold etching, to C. Bartic for the SU8 walls fabrication, a

o A. Campitelli for fruitful discussions. L. Francis is su

s.

acoustic gravimetric sensors, Appl. Phys. Lett. 64 (1994) 2940–2[10] B. Jakoby, M. Vellekoop, Properties of Love waves: application

sensors, Smart Mater. Struct. 6 (1997) 668–679.[11] G. McHale, F. Martin, M.I. Newton, Mass sensitivity of acoustic w

devices for group and phase velocity measurements, J. Appl. Ph6 (2002) 3368–3373.

[12] R.M. White, F.W. Voltmer, Direct piezoelectric coupling to surfelastic waves, Appl. Phys. Lett. 7 (1965) 314–316.

[13] G. Feuillard, Y. Janin, F. Teston, L. Tessier, M. Lethiecq, Sensties of surface acoustic wave sensors based on fine grain cerin: Instrumentation and Measurement Technology Conference(IMTC-96), Conference Proceedings ‘Quality Measurements: Thdispensable Bridge Between Theory and Reality’, vol. 2, IEEE, 1pp. 1211–1215.

[14] L.A. Francis, J.-M. Friedt, C. Bartic, A. Campitelli, An SU-8 liqucell for surface acoustic wave biosensors, Proc. SPIE 5455 (2004363.

[15] J.L. Vossen, W. Kern, Thin Film Processes, Academic Press, New1978

[16] J.-M. Friedt, L. Francis, K.-H. Choi, F. Frederix, A. Campitelli, Cobined atomic force microscope and acoustic wave devices: applito electrodeposition, J. Vac. Sci. Technol. A21 (2003) 1500–1505

aurent A. Francis received his BE in materials science with a minorntation in electrical engineering in 2001 from the Universite catholique douvain (UCL, Belgium). He is working since then towards the PhD det the PCPM unity of UCL in collaboration with IMEC (Belgium) on ttudy and the realization of acoustic waveguides aiming mainly at biosepplications.

ean-Michel Friedt obtained his PhD in engineering from LPMO (CNRrance) in 2000 after working on scanning probe microscopies and acave sensors. He spent 3 years at IMEC (Belgium) to complete a stu

he combination of acoustic and optical sensors for the characterizathe physical properties of organic layers adsorbed on substrates. Hepostdoctoral researcher at the LPMO in Besanc¸on (France).


Patrick Bertrand is Professeur ordinaire at the Universite catholique deLouvain (UCL, Belgium). Ingenieur civil physicien, 1971, UCL; PhD, 1976,UCL. Professor at the Facultes Universitaires St. Louis (Belgium). VisitingProfessor, University of Houston (USA), 1992. Invited professor at the In-stitut National de la Recherche Scientifique (INRS)Energie et Materiauxde Varennes, Universite du Quebec, Canada, since 1997. President of the

Department of Materials Science and Processes, UCL (since 09/1999). Re-search domain: physical-chemistry of surfaces and interfaces—ion–solidinteraction. P. Bertrand has published more than 250 papers in scientificjournals with referees, and he has given 40 invited keynote lectures at inter-national conferences.

influence of electromagnetic interferences on the mass sensitivity of love mode surface acoustic...

Documents