NO2 detection in the 1 ppb region with a surface acoustic wave sensor

C. Muller, M. v. Schickfus

Kirchhoff-Institut fur Physik, Universitat Heidelberg

Albert-Ueberle-Str. 3-5, D-69120 Heidelberg, Germany

The detection of NO2 in ambient air requires sensors that are capable of measuringconcentrations in the low ppb range. Usually expensive systems based on gas phasechemoluminescence are used. Here we present a cheap copper phhtalocyanine coatedsurface acoustic wave (SAW) Sensor with the capability to detect NO2 in the 1 ppbconcentration region. We use a cyclic measuring technique that prepares the sensorin a nearly NO2 free initial state and then uses the time derivative of the attenuationas the sensor signal. With this technique we reach a resolution of less than 0,8 ppbin the 1 to 150 ppb concentration range.

Keywords: surface acoustic waves, gas sensor, nitrogen dioxide, phthalocyanine

Introduction

The NO2 concentration of ambient air typically mayvary between 0 and 150 ppb in the course of a day.Therefore air quality monitoring requires NO2 sen-sors which offer a high sensitivity in this concentrationregime. Normally, for environmental monitoring, sys-tems based on chemiluminescence are used. These suf-fer from being quite expensive and large in dimension.

An alternative to this type of detector is to use chemi-cal sensors based on the conductivity change of a semi-conducting material in the presence of the gas to bedetected. Two classes of materials have mainly beeninvestigated for this purpose: Metal oxides (e.g. SnO2)and metal phthalocyanines. Although metal oxides arewidely used for sensing of combustible gases, this sub-stance class has the disadvantage of requiring operatingtemperatures above 300 ◦C, of being not very selectiveto NO2, and of a sensitivity for NO2 which is insuffi-cient for environmental monitoring. In addition, thesesensors suffer from drift and of long response times.The metal phthalocyanines have the advantage of muchhigher selectivity, but have a limited lifetime of a fewdays when operated with outside air as the test gas, andtheir response time is of the order of some 10 minutes.

For a solution of these problems we have developed ameasuring technique which uses the time derivative ofthe attenuation of a SAW Sensor [1], [2] after a desorp-tion cycle. This quantity is proportional to the NO2

concentration [3], [4]. Our new technique is a simplifiedversion of the cyclic method we demonstrated in [5]. Asthe temporal change of the attenuation and not an ab-solute value is used, we avoid drift of the sensor signal.We also achieve a higher data rate for concentrationmeasurements as we do not have to wait for equilib-rium to be reached. The method results in a limit ofdetection (LOD) of about 1 ppb. The only problem

that still has to be overcome is the short lifetime of oneweek when the sensor is operated under ambient air.

Theoretical aspects

Here we will shortly outline the dynamics of adsorptionand desorption of a gas chemically bound by a filmand the resultant change of attenuation if this film iselectrically conducting.

In equilibrium the coverage θeq of the film exposed tothe test gas of concentration c is described by the Lang-muir isotherm

θeq =n

nmax=

K(T )c

1 +K(T )c, (1)

where n is the density of adsorbed molecules in the film,and nmax the density of adsorption sites. K is temper-ature dependent and can be calculated by equating ad-and desorption rates

jads = H (1− θ)cNA

Vmol

√

kBT

2πmA, (2)

jdes = nmaxθ

τAe−Edes/(kBT ) . (3)

Here H is the sticking coefficient, and mA the mass ofthe gas molecule bound to the film with an energy Edes.Desorption is described by an Arrhenius behavior withan attempt frequency τ−1

A . One then finds for K(T )

K(T ) =HτAnmax

√

kBT

2πmA

NA

VMoleEdes/(kBT ) . (4)

Higher temperatures will lead to lower values of K andthus of θeq.

In the non-equilibrium case, ad- and desorption ratesdetermine the behavior of the sensor, the time depen-dence of the coverage θ(t) being given by the resulting

NO2 flow jres = jads − jdes:

dθ

dt=

jresnmax

. (5)

For a step-like change of gas concentration c or of tem-perature at time t0, the coverage will asymptoticallychange from the initial equilibrium θi to the final equi-librium θf . The solution for eq. 5 has the form:

θ = θf − (θf − θi) e−(t−t0)/τ (6)

with the time constant τ given by

τ−1 =1 + cK(T )

τAe−Edes/(kBT ) . (7)

This equation shows that desorption can be enhancedby increasing the operating temperature of the device.

If θi is assumed to be 0, the initial rate of change of θis given by

dθ

dt|t=0 = θf/τ , (8)

which has its largest value immediately after the changeat time t0 and is proportional to the gas concentration c.

Copper-phthalocyanine (CuPc) is a semiconductor withan intrinsic band gap of 1 eV. When NO2 molecules areadsorbed, they form acceptor sites with a energy Ea =0,16 eV. Therefore, conductivity σ is directly propor-tional to the coverage θ [6] :

σ = θ nmax eµ e−Ea/(kBT ) , (9)

where e is the elementary charge and µ the mobility.

A thin conductive CuPc film on a SAW device causesan attenuation α which is described by the followingequation [7]:

α

k=

K2

2

σSvCS

σ2S + v2C2

S

, (10)

with K2 and CS the electromechanical coupling fac-tor and capacity per unit length, respectively, k thewavevector of the surface wave, and v the velocity ofthe SAW. Sheet conductivity σS is related to conduc-tivity σ and thickness h of the layer by σS = σh.

In our experiments we have worked in a region of theabove parameters where the dependence of α goes lin-ear with surface conductivity σS . Therefore the slopedα/d t is proportional to the NO2 concentration.

Experimental

Our experimental setup consisted of the gas deliverysystem, the measuring cell and the high frequency sig-nal processing system. The experiment was controlledby a personal computer.

Mass flow controllers (Tylan FC 260) were the maincomponent of the gas delivery system which allowed adilution of NO 2 with synthetic air into the 1 ppb range.This mixture was transported through the measuringcell by a membrane suction pump with a gas flow rateof 1 l/min. By switching a magnet valve the gas flow

Fig. 1: The measuring cell: In the center one sees theprimary coil placed about 1 mm above the coupling loopof the SAW device. The interdigital transducers (IDTs)interrupting the coupling loop on the SAW device areindicated by the arrow. A heater foil is pressed againstthe ceramic base plate by a glass plate, which also pro-vides a flat support for the sensor device. Part of theheater foil and the platinum resistance thermometer canbe seen on the right side of the base plate.

could be stopped, resulting in a pressure drop in themeasuring cell.

The vacuum tight measuring cell contained the SAWdevice (Fig.1), a 1.35 µs Rayleigh wave delay line. Theinterdigital transducers (IDTs) were 160 nm thick alu-minium structures on a YZ-LiNbO 3 substrate. Widthand spacing of the fingers were 4,8µm, resulting in acenter frequency of 360 MHz. The device was oper-ated inductively [?]. The primary coil was situated di-rectly above the device. The sensitive layer consisted of15 nm copper phthalocyanine (Aldrich Chemie, Stein-heim), sublimated onto the delay line.

Because the PT100 platinum thermometer for temper-ature control is placed directly at the heating element,temperature is stable within 0.1 ◦C. At the glass platewhere the sensor element is located, however, ratherlarge temperature changes occur, depending on thestate of the experiment. During the evacuation phasetemperature here is 180 ◦ C. At this high temperaturedesorption is enhanced (eq. 3). After admitting the testgas by opening the valve again, the gas flow of 1 l/minlowers the temperature at the copper phthalocyaninesurface to around 140 ◦C. Attenuation of the SAW nowincreases according to the NO2 concentration, but atslower rate.

For laboratory measurements generation and detectionof the SAW was performed with a network analyzer(Hewlett Packard 8752 C). The change of the attenua-tion with time was recorded and evaluated by the PC.For field measurements a small electronic system formeasuring the attenuation has been developed.

Results

In Fig.2 one can see the attenuation change during fivecycles. Each cycle is divided into an adsorption anda desorption phase. During the desorption phase thevalve is closed and the pump lowers the pressure in themeasuring cell and consequently the partial pressure ofNO2. Because of the high temperature of 180 ◦C andthe absence of NO 2, purging of the surface is achievedin a few minutes without using pure air. In the fol-lowing adsorption cycle the test gas is passed throughthe measuring cell. The attenuation peak at the begin-ning of each adsorption phase is caused by the onsetof the test gas flow when opening the valve. Around100 sec after opening the valve, temperature of the de-vice is stabilised again at 140 ◦C and the slope of theattenuation, reflecting the NO 2 concentration, can berecognised.

260,0 280,0 300,0

20

25

30

AttenuationDesorption Adsorption

NO

2 concentration [ppb]

Atte

nuat

ion

[dB

]

Time [min]

0

50

100

150

200

NO2concentration

Fig. 2: Three adsorption/desorption cycles: signal(SAW attenuation) and test gas concentration (dottedline). Note that the attenuation slope in the adsorptionphase differs for the three NO2 concentrations.

To see the lower limit in resolution and concentration,we applied a concentration of 800 ppt. Fig.3 shows theattenuation slope in the adsorption phase together withthe corresponding test gas concentration. The massflow controllers worked at their lower limit. They donot close totally but just to 2% of the maximal flow,resulting in a concentration uncertainty of around 350ppt. Therefore, in this concentration range the error isprimarily caused by the test gas preparation.

Conclusion

We have presented a method to detect NO2 concentra-tions with 800 ppt resolution down to the 1 ppb concen-tration range. This method is based on a simplificationof the cyclic measuring technique reported by us ear-lier [5]. Instead of using active heating cycles, we nowtake advantage of heating the sensor by switching offthe gas flow so that the sensor temperature is raised

1 2 3 4 5 6 70,030

0,035

0,040

0,045

Attenuation slope

concentration steps

NO

2 concentration [ppb]

Atte

nuat

ion

slop

e [d

B m

in-1]

0

1

2

NO2 concentration

Fig. 3: Detection of NO2 concentrations in the rangefrom 800 ppt to 2 ppb. Concentration was varied insteps of 400 ppt. The detected level is indicated bysquares, the applied NO2 concentration by solid lines.

40 ◦C above the operating temperature (140 ◦C). To-gether with the evacuation we get a quite rapid signal(less than 10 minutes) without drift problems. The onlyproblem which still has to be overcome is the short life-time of the sensor when operating it in ambient air.We found that ozone can oxidize the copper atom irre-versibly. Also NO seems to play an important role inthis desensitization. Therefore we are planning to usefilters that selectively remove these components.

Acknowledgment

This work was supported by Deutsche Forschungsge-meinschaft (Contract Nr. Hu 359/9). We would like tothank the Gesellschaft fur Umweltmessungen und Erhe-bungen ( UMEG) for making their chemiluminescencedata available to us. We also are grateful for support bythe Fonds der Chemischen Industrie and by the CentralResearch Laboratory of Siemens AG.

REFERENCES

[1] A. Rugemer, S. Reiss, A. Geyer, M. v. Schickfus,and S. Hunklinger: Surface acoustic wave NO2

sensing using attenuation as the measured quan-tity, Sensors and Actuators B, 56 ( 1999)45-49.

[2] M. S. Nieuwenhuizen, and A. J. Nederlof: Asilicon-based SAW chemical sensor for NO2 byapplying a silicon nitride passivation layer, Sen-sors and Actuators B , 9 (1992) 100–176.

[3] J. D. Wright: Gas Adsorption on Phthalocya-nines and its Effect on Electrical Properties,Prog. Surf. Sci., 31 (1989) 1-60.

[4] A. Sczurek, and K. Lorenz: Copper Phthalocya-nine Layer as an Organic Semiconductor Sensorof NO2 in Air, Int. J. Environ. Anal. Chem., 41(1990) 57-63.

[5] C. Muller, T. Nirmaier, A. Rugemer,M.v.Schickfus: Sensitive NO2 detection with

surface acoustic wave devices using a cyclicmeasuring technique , Accepted for publicationin Sensors and Actuators B

[6] S. Dogo, J.-P. Germain, C. Maleyssson , and A.Pauly: Interaction of NO2 with Copper Phthalo-cyanine thin Films II: Application to Gas Sens-ing, Thin Solid Films, 219 (1992) 251-256.

[7] S. Datta, Surface Acoustic Wave Devices:Prentice-Hall, Englewood Cliffs, NJ, 1985.

[8] K. Beck, M. Wohlfahrt, A. Rugemer, S. Reiss,M. v. Schickfus, and S. Hunklinger: Inductivelycoupled Surface Acoustic Wave Device for Sen-sor Application, IEEE Transactions UFFC onUltrasonics, Ferroelectrics, and Frequency Con-trol, 45(5) (1998) 1140-1144.