Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 87 (2014) 268 – 271

EUROSENSORS 2014, the XXVIII edition of the conference series

A Tunable Palladium-Based Capacitive MEMS Hydrogen Sensor

Performing High Dynamics, High Selectivity and Ultra-Low Power
Sensing
Thomas Walewynsa, David Spiritoa, Laurent A. Francisa,*
a
Electrical Engineering Department, ICTEAM Institute, Université catholique de Louvain,Place du Levant 3, B-1348 Louvain-la-Neuve, Belgium

Abstract

We developed an ultra-low power MEMS hydrogen (H2) sensor performing high dynamics together with high sensitivity up to
the Lower Explosive Limit (LEL), i.e. 4 % vol. H2 in dry air. The architecture consists of Al clamped-clamped beams (700 nm-
thick, 800 μm-total width, 140 or 240 μm-length) with both ends made of a Pd/Al bimorph (200/700 nm-thick) on a quarter of its
length, acting as actuators, released above a split-bottom electrode. The initial stress defines the initial deflection of the structure
while the Pd-hydriding induces compressive stress variation in the Pd actuator layer and therefore the membrane deflection. A
capacitive transduction is used to continuously measure the gap variation due to the H 2 absorption and hydriding kinetic. Results
show a fast response time of less than 5 s for an effective concentration of about 0.2 % vol. H2 in dry air or N2 mixture, with no
cross sensitivity.
© 2014
© 2014Published
The Authors. Published
by Elsevier Ltd. by Elsevier
This Ltd.access article under the CC BY-NC-ND license
is an open
Peer-review under responsibility of the scientific committee of Eurosensors 2014.
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of the scientific committee of Eurosensors 2014
Keywords: MEMS ; gas sensor ; hydrogen, palladium ; ultra-low power sensing

1. Introduction

Ultra-low power gas sensing is of main interest in the interconnected sensor nodes vision of the Internet-of-
Things. Nowadays, hydrogen (H2) is increasingly used for many applications such as fossil fuel processing and
ammonia production in both petroleum and chemical industries, coolant in power station or pollutant-free energy

* Corresponding author. Tel.: +32 (0)10 47 35 33; fax: +32 (0)10 47 25 98.
E-mail address: laurent.francis@uclouvain.be
David Spirito is currently at Centre de Recherche Public - Gabriel Lippmann, rue du Brill 41, L-4422 Belvaux, Luxembourg

1877-7058 © 2014 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of the scientific committee of Eurosensors 2014
doi:10.1016/j.proeng.2014.11.659
 Thomas Walewyns et al. / Procedia Engineering 87 (2014) 268 – 271 269

carrier in fuel cell. As H2 is flammable or explosive at concentrations between 4 and 75 % in air, its fast detection is
essential to ensure safety. Though, current sensors still show high response time above 5 s, high power consumption
of typically more than 10 mW and poor selectivity, depending on transduction mechanism [1]. Because it features
reversible absorption of H2, palladium (Pd) is often used as active material in H 2 solid-state sensors [2]. Recently,
researches focused on Pd-coated cantilever [3] and Pd-based MEMS resonant devices with increased sensitivity and
low response time, together with high selectivity [4]. In order to reduce power consumption while maintaining high
dynamics and selectivity, we focus on a MEMS-based capacitive transduction with electrostatic actuation electrodes
for further dynamic closed-loop control.

2. Modeling and fabrication

2.1. Device modeling

The sensing mechanism is based on the actuation of double-clamped Al beams by a Pd/Al bimorph on a quarter
of their length. The MEMS structure proposed consists of a top electrode suspended above a triple bottom electrodes
system. The Pd layer acts as actuator of the MEMS structure in presence of H 2. Pd-hydriding induces internal
compressive stress with a specific kinematic depending on the initial tensile stress in the structure [5]. The
membrane deflection is then measured through a capacitive transduction. The initial stress in both Al and Pd layers
gives the initial state of the device, represented in Figure 1, with specific buckling due to the compressive behavior
of the structural layer.

Fig. 1. Left: (a) SEM picture of one fabricated device. Inset: Encapsulated tested device in DIL24 package. Right: Device cross-section schematic
showing the initial deflection with downwards (a) or upwards (b) buckling of the Al layer.

Finite element simulations using COMSOL Multiphysics® have been performed to compute the initial gap and
capacitance as a function of the thicknesses and initial stresses of both Al and Pd layer, as well as the induced
displacement due to the stress modification. Reciprocally, the same method combined with tensile test
microstructures [6,7] has been used to determine the initial stress in the Pd layer given measured initial deformation.

2.2. Device fabrication

The device has been fabricated following a polyimide-based microfabrication process [8]. The parameters of
fabricated devices, depicted in Figures 1 and 2, are detailed in Table 1. Al and Pd are obtained by evaporation and
patterned by lift-off. The bottom electrode is made of Ti/Au 5/150 nm-thick or Al 150 nm-thick above a 300 nm-
thick SiO2 layer. The MEMS transducer is then encapsulated in a DIL-24 package for electrical characterization.
270 Thomas Walewyns et al. / Procedia Engineering 87 (2014) 268 – 271

Table 1. Parameters of fabricated devices.

Device [#] Length Initial gap Al thickness Pd thickness σAl [MPa] σPd [MPa] Central Buckling
[μm] [μm] [nm] [nm] deflection
[μm]
1 240 7.9 800 300 -50 900 2.3 Upwards
2 140 7.9 800 300 -50 900 1.2 Downwards
3 240 5.8 700 200 -50 800 3.3 Upwards
4 140 5.8 700 200 -50 800 3.7 Downwards

White light interferometry measurements using Polytec MSA-500 have been performed in order to measure the
initial gap, as shown in Figure 3.

2.3 μm

9 μm

Fig. 3. White light interferometry characterization using Polytec MSA. The total metal thickness is 900 nm. The buckling of the Al layer is about
2 μm upwards.

3. Characterization

The devices have been electrically characterized in both air and N2 atmosphere, with H2 concentration from 0.5 to
4 % volume, using an Agilent 4284A LCR meter. The results are presented in Figure 4. The capacitance has been
extracted through a Rp/Cp parallel equivalent model. No influence of the carrier gas, i.e. air or N 2, has been
observed. The response time is down to 5 s with a concentration of 2 % H2/N2 (SNR > 5). However, due to the
volume of the characterization cell, about 150 cm3, leading to a gas diffusion time up to 50 s, the effective detected
concentration is only around 0.2 % vol..

Air

H2 desorption
(4)
(3)

H2 absorption
(2)

(1)
H2

a b

Fig. 4. (a) Capacitance variation in 2% H2/dry air mixture. The drift is caused by the plastic deformation induced by the stress due to the Pd-
hydriding. The initial response to dry air flow is due to the forced humidity desorption of the polymer. (b) Maximal phase variation after 5
minutes exposure at 0.5%, 1%, 2%, 3% and 4% H2 using N2 carrier gas.
 Thomas Walewyns et al. / Procedia Engineering 87 (2014) 268 – 271 271

Four phases are successively observed by the device exposed to hydrogen cycle, as depicted in Figure 4(a): (1)
response to the dry carrier air, (2) Pd-hydriding due to H2 absorption and stress modification, (3) H2 desorption and
(4) stabilization with no more carrier gas. As observed in Figure 4(b), the response to H2 is also proportional to the
concentration up to 3 % vol. Saturation in the hydriding mechanism at 4 % vol. seems to be observed after few
cycles.

4. Discussion and conclusion

As observed in Figure 4(a), plasticity probably occurs in the Pd/Al structure with permanent deformation during
H2 exposure. The etching conditions i.e. wet or dry (barrel O2 or ICP-RIE O2), also influence the stress relaxation
which modify the MEMS structure and consequently the sensing behavior. Moreover, non-uniformities in the
materials thicknesses modify the internal stresses. In order to overcome these effects, a closed-loop feedback can be
implemented keeping the 3D-MEMS capacitance to a reference point through electrostatic actuation. By
continuously maintaining the capacitance to its maximal value such a feedback also increases the sensitivity. Such a
control is allowed by the triple bottom electrode system, represented in Figure 1.
The capacitive sensor is also sensitive to humidity, especially above 60 % RH with exponential behavior, as
measured in climate chamber. In order to avoid this effect, the use of a microporous membrane filter is envisioned
within a dedicated housing. Regarding the temperature, its influence is limited to a linear drift, here measured
between 25 and 85 °C, which can be easily controlled.
The developed sensor shows a fast response time (< 5 s), a high selectivity to hydrogen, and a high sensitivity up
to the Lower Explosive Limit (LEL). Incoming developments will focus on the reliability of the MEMS device,
including stiction, creep and fatigue phenomenon, together with new design improvement. The further analysis of
the hydriding kinetic is essential for increasing the device performance. Such hydrogen sensors associated with a
dedicated ultra-low power capacitive interface are very promising for future low power sensing applications with
high dynamics and selectivity, for higher safety.

Acknowledgements

The authors sincerely thank Pierre Gérard from UCL for the interface measurement software, Carl Emmerechts
from Sirris Wallonie and Serguei Stoukatch from ULg-Microsys for the packaging, as well as Marc Debliquy from
UMons and Driss Lahem from Materia Nova for the characterization in H 2 atmosphere and the technical support.
They are also grateful to the cleanroom facilities WINFAB (Wallonia Infrastructure Nano FABrication) and the
WELCOME (Wallonia ELectrical and COmmunication MEasurements) platform. This research has been supported
by the National Fund for Scientific Research (F.R.S.-FNRS) through a FRIA grant, and by the project MINATIS co-
funded by the European Regional Development Fund (ERDF) and Wallonia, Belgium.

References

[1] T. Hübert, L. Boon-Brett, G. Black, U. Banach, Hydrogen sensors – A review, Sens. Actuators B, 157 (2011) 329-352.
[2] P. Soundarrajan, F. Schweighardt, Hydrogen Sensing and Detection, Hydrogen Fuel: Production, Transport, and Storage, Chap. 15, Ed. R. B.
Gupta, CRC Press (2012).
[3] D. R. Baselt, B. Fruhberger, E. Klaassen, S. Cemalovic, C. L. Britton Jr., S. V. Patel, T. E. Mlsna, D. McCorkle, B. Warmack, Design and
performance of a microcantilever-based hydrogen sensor, Sens. Actuators B, 88 (2003) 120-131.
[4] J. Henriksson, L. G. Villanueva, J. Brugger, Ultra-low power hydrogen sensing based on a palladium-coated nanomechanical beam resonator,
Nanoscale, 4 (2012) 5059-5064.
[5] R. Delmelle, S. Michotte, M. Sinnaeve, J. Proost, Effect of internal stress on the hydriding kinetics of nanocrystalline Pd thin films, Acta
Materialia, 61 (2013) 2320-2329.
[6] M.-S. Colla, B. Wang, H. Idrissi, D. Schryvers, J.-P. Raskin, T. Pardoen, High strength-ductility of thin nanocrystalline palladium films with
nanoscale twins: On-chip testing and grain aggregate model, Acta Materialia, 60 (2012) 1795-1806.
[7] B. Amin-Ahmadi, H. Idrissi, M. Galceran, M.-S. Colla, J.-P. Raskin, T. Pardoen, S. Godet, D. Schryvers, Effect of deposition rate on the
microstructure of electron beam evaporated nanocrystalline palladium thin films, Thin Solid Films, 539 (2013), 145-150.
[8] T. Walewyns, N. Reckinger, S. Ryelandt, T. Pardoen, J.-P. Raskin, L. A. Francis, Polyimide as a versatile enabling material for microsystems
fabrication: surface micromachining and electrodeposited nanowires integration, J. Micromech. Microeng., 23 (2013) 095021.