ALLSENSORS 2018 : The Third International Conference on Advances in Sensors, Actuators, Metering and Sensing

Design and Optimization of Gas Sensor Testing Chamber

Fatima Ezahra Annanouch1, Nicolas Morati1, Virginie

Martini-Laithier1, Tomas Fiorido1, Khalifa Aguir1 and Gilles Bouchet2, Pierre Perrier2
2
Marc Bendahan1, Aix Marseille Univ, CNRS, IUSTI,
1
Aix Marseille Univ, Université́ de Toulon, CNRS, IM2NP, Marseille, France
Marseille, France e-mail: {gilles.bouchet, pierre.perrier}@univ-amu.fr
e-mail: {fatima.annanouch, nicolas.morati, virginie.laithier,
tomas.fiorido, khalifa.aguir, marc.bendahan}@im2np.fr

Abstract—It is important to size and optimize the chamber in fabrication steps including photolithography, metallization
which gas sensors are tested. Indeed, the amplitude, as well as and backside etching (DRIE) of the substrate to define the
the response and recovery times are very dependent on the membrane as the transducer [3]. This latter has a size of
testing chamber. In this study, we show that by optimizing the 400μm × 400 μm. It carries three transducers S1, S2, S3 and
design and reducing the volume of the testing chamber, the
two heaters (Fig. 1). The gap between the electrodes is 4 μm,
responses of metal oxide microsensors are highly enhanced and
faster, and therefore closer to the real answers of the sensor. the resistance of each heater is 100 Ω and the temperature
coefficient is 3 10-3/°C. An SnO2 sensing layer (50 nm) was
Keywords-gas sensor; gas testing chamber; metal oxides; tin directly deposited over the microsensor platform, by reactive
oxide; mathematical modeling; simulation. radio frequency magnetron sputtering technique [4][5].
I. INTRODUCTION
Metal oxide gas sensors have become one of the most
active research areas, owing to their low cost and flexibility
in production, simplicity of their use and their large number
of detectable gases. They are employed in a wide spectrum
of applications, such as environmental monitoring, domestic
safety, disease detection and many more [1].
So far, many works have been reported on the Figure 1. MEMS-based microsensor platform [3].
enhancement of gas sensing performances, either with
catalysts nanoparticles modification, or by the nano- It is worth noting that before depositions, the
structuring of metal oxide sensitive material [2]. However, microsensor platform was cleaned with acetone and then
few works have studied the influence of the gas testing with ethanol, dried with air, and then placed inside the
system on the sensors performances. Achieving a reliable shadow mask.
and highly sensitive sensor, with fast response and recovery
times, cannot be done without an optimized gas testing B. Testing chambers
chamber design. Such a design needs to be modeled and Gas-sensing tests were carried out in two different
simulated in terms of chamber volume, placement of sensor chambers (Fig. 2).
and gas flow direction.
In this work we study and compare the influence of gas
testing chamber design on the sensor performances, namely
Cross chamber (old one) and Boat chamber (new optimized
testing chamber). The paper is structured as follows. Section
II describes the gas microsensor platform and the testing
chamber design. Section III presents and discusses the
obtained results (simulation results and
experimental validation of the optimized chamber). We
conclude the work in Section IV.
Figure 2. Photographs of: (a) cross and (b) boat chamber
II. GAS SENSORS AND TESTING CHAMBERS
A. Gas sensors The first one (Fig. 2.a) was made from stainless steel
The microelectromechanical systems (MEMS) based with total volume of 0.3 L and gas flow range between
microsensor platform was patented by our laboratory and 0.1L/min and 1.5L/min. Besides, it has an inlet, an outlet,
fabricated using clean room facilities and various micro- and the sensor optimal position is at the center, in front of the

Copyright (c) IARIA, 2018. ISBN: 978-1-61208-621-7 15

ALLSENSORS 2018 : The Third International Conference on Advances in Sensors, Actuators, Metering and Sensing

gas flow direction. The second chamber (the new one) is The target gas is injected at the inlet (0.1 L/min during 1
illustrated by Figure 2.b. It was made from polylactic acid min, 50 ppm ethanol in air) and the ethanol concentration is
(PLA) with a small volume of 2.35 10-3L and gas flow range measured on the sensor position. The gas testing transport is
between 0.01 L/min and 0.5 L/min. It has a boat shape with a modeled by a convection-diffusion equation applied to a
planar inlet and outlet. Additionally, the optimal sensor passive scalar. In Figure 4, we can observe that the gas
position is at the center of the boat back-side, in the same concentration in the optimized testing chamber is very
plan as the gas flow direction. In both chambers, the gas flow similar to the setpoint (50 ppm).
was maintained at 0.1 L/min, the sensor response was
defined as R=Ra/Rg, where Ra and Rg are the sensor
resistances at the stationary state in air and after 1 min of
exposure to the target gas, respectively.
III. RESULTS AND DISCUSIION
A. Mathematical modelling and simulation results
The flow is modeled by a finite volume method solving
the Navier-Stokes and the energy equations in the two 3D
geometries, with no-slip and adiabatic boundary conditions
everywhere, except on the inlet and outlet boundaries and on
the sensor. Velocity fields are presented in Figure 3.
Figure 4. Ethanol concentration simulation in both chambers; injection at
t=134s during 1 min

In addition, the speed of the filling and emptying in boat

chamber is much faster compared to the cross one. In
contrast, the ethanol concentration simulated for the cross-
chamber does not reach 50 ppm with a flow rate of 0.1 L/min
during 60 seconds, due to the greater volume of the cross-
chamber and the presence of turbulences. The simulations
show that an injection lasting 10 minutes is necessary to
Velocity : magnetude m/s
reach 50 ppm with such flow rate; at the same time, an
0.000 0.053 0.106 0.160 0.213 0.267 injection with a flow rate of 0.5 L/min for 120 seconds
allows us to reach the same level of ethanol concentration
(50 ppm).
These results will be experimentally validated in the next
paragraph by measuring the electrical sensor response
towards 50 ppm of ethanol for an exposer of 1 minute, in
both chambers.
B. Validation of the optimized gas testing chamber
In order to validate the simulation results, we have
exposed SnO2 sensor (S2) to 50 ppm of ethanol, using the
same measurements parameters, in both testing chambers
(Fig. 5).
Velocity : magnetude m/s

0.000 0.201 0.403 0.605 0.807 1.009

Figure 3. Velocity field of the converged flow in the Cross Chamber (a)
and in the Boat chamber (b)

As we can see, in the cross chamber, at the sensor

surface, the gas flow velocity is very important (0.1 m/s),
which leads to the creation of many turbulences and
negatively affects the sensor detection. However, in the boat
chamber, it is almost zero and the gas flow direction is linear
with the sensor position, which eliminates the creation of gas
turbulences. Figure 5. Experimental sensor response toward 50 ppm of ethanol in both
chambers - injection during 1 min

Copyright (c) IARIA, 2018. ISBN: 978-1-61208-621-7 16

ALLSENSORS 2018 : The Third International Conference on Advances in Sensors, Actuators, Metering and Sensing

The results show that the sensor performances are highly in regional economic development associated with
enhanced using the optimized chamber. For instance, the innovation and Mr. A. Combes for his technical support.
response and recovery times in the boat chamber are 4 s and
89 s, respectively. However, they are five times higher when REFERENCES
using the cross chamber. [1] G. F. Fine, L.M. Cavanagh, A. Afonja, R. Binions, “Metal
oxide semi-conductor gas sensors in environmental
IV. CONCLUSION monitoring”, Sensors, vol. 10, pp. 5469–5502, June 2010.
[2] F. Annanouch et al., “Aerosol-Assisted CVD-Grown PdO
In this study, we have highlighted the strong influence of Nanoparticle-Decorated Tungsten Oxide Nanoneedles
the test chamber design with respect to the electrical Extremely Sensitive and Selective to Hydrogen”, ACS Appl.
response of the sensor towards ethanol. To achieve our goals, Mater. Interfaces, vol. 8, pp. 10413−10421, April 2016.
we have reduced the dimensions of the test chamber while [3] K. Aguir, M. Bendahan, V. Laithier, “Gas sensor based on
adapting the geometry. Besides, we have eliminated dead heated sensitive layer”, patent N° FR 13 59494, 2013,
international extension 2016.
volumes, obtained a homogeneous gas concentration, and
reduced the gas flow velocity at the sensor surface. The [4] M. Kaur et al., “RF sputtered SnO2: NiO thin films as sub-
ppm H2S sensor operable at room temperature”, Sens.
experimental results are in agreement with the mathematical Actuators B-Chem, vol. 242, pp. 389−403, April 2017.
modelling and simulation results. [5] G. Sberveglieri, S. Groppelli, G. Coccoli, “Radio frequency
magnetron sputtering growth and characterization of indium-
ACKNOWLEDGMENT tin oxide (ITO) thin films for NO2 gas sensors”, Sens.
The authors would like to acknowledge NANOZ, a Actuators, vol. 15, pp. 235−242, November 1988.
company specialized in gas sensors; SATT Sud-Est,
"Accelerator of Technology Transfer", which is a key player

Copyright (c) IARIA, 2018. ISBN: 978-1-61208-621-7 17