Sensors and Actuators B 148 (2010) 610–619

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical

journal homepage: www.elsevier.com/locate/snb

Review

Micro gas preconcentrator in porous silicon filled with a carbon absorbent

E.H.M. Camara a,b , P. Breuil a,∗ , D. Briand b , L. Guillot b , C. Pijolat a , N.F. de Rooij b
a
Ecole Nationale Supérieure des Mines, Centre SPIN, Dpt MICC, LPMG-UMR CNRS 5148, 158 Cours Fauriel, 42023 Saint-Etienne, France
b
Institute of Microtechnology, University of Neuchâtel, Rue Jaquet-Droz 1, P.O. Box 526, CH-2002 Neuchâtel, Switzerland

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents the development of a gas preconcentrator based on a micro-channel in porous silicon
Received 9 October 2008 filled with carbon nanopowders by a micro-fluidic process. The particularity of this device is its applica-
Received in revised form 12 March 2010 bility in the fields of atmospheric pollution monitoring by targeting VOCS (volatiles organic compounds).
Accepted 23 May 2010
Various designs of micro-devices have been investigated and a special focus has been dedicated to the
Available online 4 June 2010
carbon adsorbent. The optimization of the device and its operation were driven by its future application
in outdoor environments. The benefits of using porous silicon to ease the fixing of the carbon absorbent in
Keywords:
micro-channels and to modify the gas desorption kinetic are also investigated. Results on a device based
Preconcentrator
Gas sensors
on a carbon adsorbent powder filled in a porous silicon micro-channel for benzene preconcentration are
Micro-channel reported.
Porous silicon © 2010 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 610
2. Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 611
2.1. Preconcentrator design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 611
2.2. Porous silicon formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 611
2.3. Carbon insertion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612
2.4. Preconcentrator fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612
2.5. Preconcentration experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
3. Results and discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 614
3.1. Porous silicon formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 614
3.2. Carbon insertion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615
3.3. Preconcentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617
4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 618
Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619
Biographies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619

1. Introduction preconcentration in order to increase the sensibility of the detector

[5–7].
In the field of gas detection, one major trend is to miniatur- Indeed, in some applicative environments the concentration
ize analytical techniques. In such technological developments, the of gas is too small and therefore a preconcentration unit at the
gas pre-conditioning is usually very important. This may involve, entrance of the analytical device is needed [8,9]. When a precon-
firstly, the modification of a gas mixture, for example to improve cenrator is used, the gas mixture to be analyzed flows through it
the selectivity of a detection device [1–4], secondly, the effect of and is accumulated during some time, then the mixture is desorbed
by a temperature pulse and brought to the detector [10–14].
Requisite specifications, such as small system size, high perfor-
∗ Corresponding author. Tel.: +33 4 77 42 01 51; fax: +33 4 77 49 36 20. mance, low power budget, and quick time response lead to the
E-mail address: [email protected] (P. Breuil). microfabrication of the key components of the analytical systems.

0925-4005/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.snb.2010.05.054
 E.H.M. Camara et al. / Sensors and Actuators B 148 (2010) 610–619 611

Fig. 1. (a) Top view of a micro-channel in “straight” design, (b) top view of a micro-channel in “zigzag” design, (c) magnification of walls in straight design and (d) magnification
of structures in “zigzag” design.

Recent studies have shown that the fabrication of such micro- experimental conditions and detection system. At this end, exper-
analytical systems on silicon technology offers clear advantages imental conditions chosen in this work have been obtained from
over conventional systems, but the hybrid integration of the com- a series of experimental optimization and theoretical modelling
ponents, using plastic, metal, or silicon/glass manifolds, becomes which are not the main subjects of this paper.
increasingly interesting because of the numerous potential appli-
cations. The benefits of using this approach are the possibility to 2. Experimental
optimize the individual components composing the system and it
enables their replacement in a modular fashion [15,16]. Moreover, 2.1. Preconcentrator design
the miniaturization of the preconcentrators leads to superior heat-
ing rates, reduced power consumption and allows its integration Globally, the micro-channel consists in silicon micro-channels
in portable miniaturized analytical instruments, such as micro gas which can receive the adsorbent material, with inlet and outlet suit-
chromatographs and ion mobility spectrometers. able for fluidic connections using metallic capillaries and are sealed
In a previous study a catalytic microreactor has been developed with a glass cap. The device includes also a platinum micro-heater
to act as a filter to improve the selectivity to methane of a sen- on the backside.
sor placed downstream of the device [17]. The present work aims The depth of the micro-channel, for the inlet, the outlet and the
at developing a silicon preconcentrator device based on this con- preconcentration chamber, was defined at 325 ␮m, correspond-
cept, with new approaches in the design as well as in the choice ing to the external diameter of the capillaries used as fluidics
of materials, replacing the catalyst by a gas absorbent material. interconnections. Micro-channels with three different shapes were
One difference with other work performed on this topic is that the designed in order to ease the introduction of the carbon nanopow-
adsorbent material is here introduced in the form of powder in a ders. Fig. 1 shows two micro-channels respectively in “straight”
fluidic solution [18]. We have reported recently on the feasibility and “zigzag” designs with the magnification of its structures. The
of this concept using a carbon nanopowder [19]. Another strategy “zigzag” design (“V” shaped structures), contributes to improve the
to increase the specific area, and consequently the gas adsorption, carbon deposition by stopping it when injected into the micro-
is to use porous silicon. By varying the parameters used for the for- channel. While the “straight” design (“wall” shaped structures)
mation of porous silicon, micro-, meso- or macropores in silicon leads to an increase of the available area for deposition since the
can be realized [20–22]. The high specific area offers by porous sil- surface of the walls constitute a binding area for adsorbent parti-
icon can be used as an interesting support to increase the quantity cles. And lastly, the “neutral” design without structures was made
of adsorbent materials fixed in the micro-channels when made of in order to have a reference to determine the benefits or not to
this material. Besides this, porous silicon presents in itself a high have structures. At the inlet and outlet, a system of channels was
adsorption capacity of molecules which has been already applied designed targeting a uniform distribution of the gas or liquid flow.
directly to the detection of gases. The association of porous sili- The channels can be made of silicon or porous silicon. The use of
con with the carbon nanopowders deserves to be explored, since it porous silicon offers the possibility of increasing the quantity of the
could lead to some superposition of their gas adsorption capabili- deposited adsorbent material and thus the specific area in a given
ties and results in improved performances. volume. The dimensions of the designs are summarized in Table 1.
The devices in this study have been developed targeting the
preconcentration of benzene. This gas is considered as a relevant 2.2. Porous silicon formation
representative for several types of applications related to atmo-
spheric pollution control (BTEX compounds). As mentioned previously, an objective of this study is to fill in
Except the design of the micro-channel and the adsorbent mate- short time a sufficient quantity of adsorbent material in the micro-
rial proprieties, the performances of the device are closely related to channel. To that end, the use of micro-channels made of porous
612 E.H.M. Camara et al. / Sensors and Actuators B 148 (2010) 610–619

Table 1 have an internal diameter of 220 ␮m and its specific surface close
Dimensions of the preconcentrators.
to 100 m2 /g suitable for the adsorption of same volatile organic
Design of micro-channels Neutral Straight Zigzag compounds (VOCs).
Length × width [mm] 30 × 15 30 × 15 30 × 15 In order to study the fixing of the carbon particles in micro-
Vol. of ␮-micro-channel [mm3 ] 48.75 41.92 44.07 channels, carbon was deposited in a first step on silicon substrates
Walls or pillars width [␮m] No 50 50 by the wetting method which consists in immersing the silicon
chips in the carbon solution. The latter is basically composed of
water, carbon and a few drops of dispersant (Company Roche-
Nonidet P40) to homogenize the mixture. The quantity of carbon
silicon appears particularly attractive. The size, shape and depth of deposited is a function of the substrate nature and the number of
pores depend on several parameters such as the concentration of times this operation was repeated. The porous aspect of silicon
the bath, the doping and the resistivity of the wafer, the applied represents a key parameter to improve the fixing of the carbon
current density and the etching time [20]. In some cases, an illu- particles. In a second step, the carbon was inserted in the sealed sil-
mination or a metallic deposition on the backside of the wafer is icon micro-channels. After testing several methods of filling up the
necessary to facilitate the transfer of charge carriers between the micro-channel with carbon from a fluidic solution, we faced over-
solution and the substrate. However, an unproper choice of these pressure in the micro-channels, especially by injecting the carbon
parameters may have harmful consequences on the pores quality suspension with a push-syringe. To overcome this problem, a new
or even damage the micro-channel since the anodic etching is gov- method was implemented. It consists in putting the device under
erned by three regimes namely the “electropolishing” regime, the vacuum for making easier the circulation of the suspension through
“porosification” regime and at last “transitional” regime. the micro-channels by a difference of pressure between precon-
The porous silicon micro-channel was obtained by anodic etch- centration chamber of the micro-device and liquid suspension.
ing of silicon in a bath of HF (50%) and EtOH in a ratio of 1:1. Changeable capillaries were used in order to reduce the overpres-
This technique is an electrochemical etching during which chemical sure created by their blocking. Results have been characterised
phenomena are linked to reciprocal exchanges of electrical energy using scanning electron microscopy (SEM) and by measuring the
that occur at the interface of two conductor systems (electronic weight of deposited carbon.
and ionic). It is an efficient method to obtain spongy or columnar
layers of silicon. The elaboration of macro and microporous silicon
requires the control of some main parameters such as the silicon 2.4. Preconcentrator fabrication
substrate doping, the concentration of the HF bath, the anodic cur-
rent density and the etching time. The choice of substrate doping is The processing of these preconcentrators is simple. Silicon
a determining factor on the nature of pores formed. The wafer dop- micro-channels are directly dry plasma etched by deep reactive ion
ing determines the movement of charge carriers when a potential etching (DRIE) using a thick photoresist mask. In the case of porous
difference is applied between the electrodes of the HF bath. silicon channels, prior to the DRIE, a silicon nitride film is deposited
To this end, the silicon wafers used for the anodic etching to act as a masking layer during the porous silicon formation. Then,
were n- and p-type with resistivities varying from 0.015 to 5  cm for both types, a thermal oxide is grown prior to their sealing with
and different current densities between 40 and 300 mA/cm2 were a glass cover by anodic bonding. A more detailed description of the
applied on these wafers. This evaluation was done in order to define process is given in the following paragraphs.
two sets of optimum processing conditions to realize micro- and Following a photolithographic step with a thick AZ4562 pho-
macroporous silicon pores. The etching time determines the pen- toresist, the inlet, outlet and preconcentration chamber of the
etration depth of the pores in the substrate, more the latter longer micro-device are DRIE etched in the silicon wafer to a depth of
more the pores formation goes deeper. To avoid the weakening of 325 ␮m (Fig. 2b). If desired, the porous silicon was produced fol-
micro-channels, we targeted a depth of the pores of about 100 ␮m. lowing the procedure described in Section 2.2. In that case, a
Since the pores obtained by anodic etching have a conical growth 500 nm-thick LPCVD low-stress Si3 N4 thin film was first deposited
i.e. the pores diameter increases with the depth, and therefore, after (Fig. 2a) and was used as an etch mask during this step (Fig. 2c).
the pore formation, about few micrometers of silicon were etched Therefore, porous silicon was formed on all the unprotected area
by reactive ion etching (RIE) in order to open these pores. After the of the silicon micro-devices, including the sidewalls of the micro-
RIE stage, the wafers were observed in SEM to determine the pores channels. In order to improve the adhesion of the carbon adsorbent
distribution on the surface and then they were cleaved in order to powder that will be deposited afterwards in the micro-channels
observe the depth of these pores. And following these characteri- made or not of porous silicon, 25 nm of thermal oxide were grown
zations, the wafers went through BET measurements to determine to make them become hydrophilic and also to facilitate the adhe-
their specific area. sion of carbon particles. Finally, the micro-channels were sealed
with a Borofloat glass wafer by anodic bonding (Fig. 2d) and diced
2.3. Carbon insertion according to the dashed line showed in Fig. 2e revealing the inlet
and outlet for the fluidic connectors.
In relation to the size of our carbon particles, the contribution of In order to have an autonomous component, for either the ther-
the silicon micropores on the filling of carbon will be very limited mal treatments of the absorbent materials and the desorption of
since the size of carbon particles has the same order of magni- trapped gases, a platinum heater is integrated on the backside of
tude than the diameter of pores. As a result, only the macroporous the device by screen printing and the electrical connections are
silicon, both on flat substrates and on micro-channels, has been made with gold wires pasted on a platinum lacquer. This heating
used as a support for the carbon deposition while the devices made element is used to heat the device at temperatures up to 500 ◦ C
of microporous silicon were directly applied as preconcentrators, with a very homogeneous distribution of the temperature consid-
without any adsorbent material deposition. ering the good thermal conductivity of the silicon, as it was reported
A carbon nanopowder from the company Sigma–Aldrich (CAS in our previous work on microreactor for gas–solid catalyst [17]
number 7440-44-0) has been chosen as an adequate adsorbent for and as shown in Fig. 3 on the right hand side. The platinum paste
a fluidic deposition because of its distribution of sizes between 30 used for the heater deposition is commercialized by the company
and 100 nm since the metallic capillaries used for fluidic connection Electro-Science Laboratories (ref. 5545).
 E.H.M. Camara et al. / Sensors and Actuators B 148 (2010) 610–619 613

Fig. 2. Fabrication process of the preconcentrator, which is illustrated here with the “straight” design.

Fig. 3. (a) Micro-channel in porous silicon filled with carbon with metallic gas capillaries including the integrated heater, with its electrical connections and (b) optical image
of the screen-printed heater on backside of the micro-channel.

Once the devices filled with carbon, the fluidic connections are 2.5. Preconcentration experiments
made with metallic capillaries (steel) having 220 ␮m of internal
diameter and fixed with ceramic cement using a special procedure Benzene vapor has been chosen as target gas. The weak concen-
so as not to block the connectors during the sealing (Fig. 2f). Fig. 3 trations of benzene between 10 and 500 ppb are generated using a
shows a micro-preconcentrator at the final stage of its fabrication gas cylinder of air with 100 ppm of benzene and a dilution bench
after the carbon deposition, including the integrated heater, with made with two mass-flow controllers. The pressure measurement
its electrical connections appearing on the picture, and the metallic upstream and downstream of the mass-flow controllers allows ver-
capillaries used as fluidic connectors. ifying the good operation of the mass flow in spite of the driving

Fig. 4. Schematic illustration of the micro-preconcentrator test bench.

614 E.H.M. Camara et al. / Sensors and Actuators B 148 (2010) 610–619

head. This device enables to exactly generate very weak concen-

trations in the sub-ppm range (Fig. 4). The first experiments were
dedicated to the validation of the test set-up. No adsorption of ben-
zene occurs in the set-up when tested with micro-devices without
the absorbent. Indeed, when benzene was injected in the channel
(100 ppm), the latter was detected by the sensor without retention
time and no desorption peak was then observed during the heating
step.
Four main parameters are of importance for the optimum oper-
ation of a preconcentrator device with a given geometry and filled
with a fixed quantity of specific absorbent material, the gas flow
rate during the gas adsorption and desorption phases, the heating
rate during the gas desorption, the duration of the gas adsorp-
tion and desorption phases, and the proximity of the gas outlet
to the detector and its detection volume (volume between the
micro-preconcentrator and the detector plus the gas volume of the
detector). Fig. 5. Demonstration of the dependence between the amplitude of the desorption
The miniaturization of such device imposes us a limitation on peak and the heating rate.

the gas flow rate, but for preconcentration, high flow rates are pre-
ferred in order to increase preconcentration factor for a given time targeted applications as the pollution monitoring, a short precon-
of adsorption. However, a high gas flow can cause significant pres- centration cycle is more appropriate.
sure drop thus weakening the device and modifying the physical So it is very difficult to define a preconcentration factor since,
chemistry phenomena. In this study, the use of metallic gas capillar- for a given preconcentrator, the “amplification” of the concentra-
ies with 220 ␮m of internal diameter as fluidic connectors allowed tion is closely related to these four external parameters. However,
us to generate a flow of 167 mL/min either–or 2 bars of relative we will measure a “practical” preconcentration factor which we
pressure without consequence on the preconcentration. Regard- define as the ration between the maximum of the measured con-
ing the optimization of the preconcentration, the adsorption phase centration during the desorption phase and the concentration of
flow was set at 167 mL/min, which allowed us to inject a signif- the injected gas. This factor will be largely lower than the “theoret-
icant quantity of pollutant in a short time, while the desorption ical” preconcentration factor because of the significant detection
phase flow was maintained at 33 mL/min to avoid an important volume needed by the sensor system which is yet not optimized.
dilution of the desorbed quantity. Likewise, to avoid a consider-
able detection volume, the metallic capillary at the outlet of the 3. Results and discussions
micro-preconcentrator was directly connected to the inside of the
detector through its packaging (Sensor from FIGARO ref. 2620), and 3.1. Porous silicon formation
the latter was then calibrated in order to correct the influence of
the gas flow on the characteristics of its response. The calibration Preliminary experiments on unstructured wafers have shown
curb of the sensor has been realized for different concentrations in that the n-type substrate led mostly to a non-uniform porosity,
the same conditions (flow rate and capillary inside the packaging). unlike the p-type substrate which has an excellent diffusion length
Le detection limit of this sensor is estimated about 15 ppb. of minority charge carriers. The resistivity of the wafer is also a
As mentioned previously, the effectiveness of the preconcen- significant factor because the p-type wafers with low resistivity
tration is closely linked to the physical chemistry properties of the lead often to micropores in spite of high-applied current densities
adsorbent, the desorption protocol, the detection system and its while n-type substrates led to macropores for these same resistivity
dead volume. Therefore, the heating rate is an important element ranges. The current density is an increasing function of the pore
for increasing the preconcentration factor. In fact, it is well known, size but a high current density caused an electropolishing regime
that the maximum of the desorption peak is an increasing function manifesting by the stripping of the substrate.
of the heating rate. Higher the heating rate is, narrower and higher As a first step, macroporous silicon was realized on 5 ·cm,
the desorption peak is i.e. the maximum of the desorption more n-type, (1 0 0)-oriented and 525 ␮m-thick flat substrates with
important and consequently a significant factor of preconcentra- a current density of 40 mA/cm2 but the result was unsatisfy-
tion (Fig. 5). According to this observation, the preconcentration ing because there was a non-uniform distribution of pores with
tests were performed with a heating rate of 160 ◦ C/min, which particular areas of porous high density and non-porous areas conse-
corresponds to the maximum rate that can reach the heating resis- quence of poor diffusion of charge carriers (Table 2). Nevertheless,
tance deposited on the backside of the micro-channels. And also, we chose to test the adhesion of carbon particles on pieces of theses
in order to follow the evolution of the desorption temperature and wafers on which the porosity seemed locally uniform. The results
to determine the activation energy of this desorption characteristic of these tests are described in Section 3.2.
of a given type of micro-preconcentrator, we used different heat- So as to make uniform the distribution of charge carriers across
ing rates of 5, 10 and 20 ◦ C/min. The temperature of desorption the wafer and thus uniform the porosity, we decided to reduce the
has been set at 200 ◦ C after its optimization performed in previous available area structuring the wafers before the anodic etching.
experiments with charges of carbon [19]. The passage of flat substrates to micro-channels wafers requires
Finally, the increase of the adsorption time can lead to an the readjustment of some etching parameters due to the mod-
enhancement of the desorption peak amplitude but in the mean ification of electrical field distribution by the three-dimensional
time cause a longer preconcentration cycle which for some appli- shape structures in the wafer. As a result, macroporous silicon
cations might not be desirable. In addition, the optimization micro-channels have been realized on 5 ·cm, n-type, (1 0 0)-
experiments made on the duration of the desorption phase showed oriented and 525 ␮m-thick silicon wafers. The etching time was
that the desorption is almost complete after 4–5 min. To this end, 45 min at current densities between 60 and 70 mA/cm2 . The diam-
the adsorption and desorption times were maintained constant eters of the pores obtained were between about 900 nm and
all through the experiences and lasted for 5 min since for these 1.2 ␮m with a penetration in the silicon between 100 and 120 ␮m
 E.H.M. Camara et al. / Sensors and Actuators B 148 (2010) 610–619 615

Table 2
Several porous silicon characteristics with their different processing parameters and in some case a RIE etching.

Wafer type Wafer resistivity Current density Etching time Pore diameter PS thickness Etch rate Specific area
[ cm] [mA/cm2 ] [min] [nm] [␮m] [␮m/min] [m2 /g]

Unstructured wafers
p 0.015 150 6 8–13 1–20.4 3.40 3.07
p 5 300 30 11–25 NA NA 0.10
n 5 40 20 348–1060 1–136 6.80 2.33
n 5 40 20 + RIE 871–1920 2–130 6.80 2.58

Silicon micro-channels
p 5 70 60 + RIE 53–252 NA NA 27.27
n 5 60 45 + RIE 1023–2350 140 2.66 44.5

(Fig. 6). For microporous silicon micro-channels (pore diame- These macroporous silicon micro-channels with pores diameter
ter between 10 and 200 nm) they were developed on 5  cm, between 0.9 and 1.2 ␮m have been used for the carbon insertion
(1 0 0)-oriented, 525 ␮m-thick, p-type silicon wafers with current described in Section 3.2, while the microporous silicon micro-
densities between 50 and 70 mA/cm2 and an etching time of about channels with pores diameter between about 0.05 and 0.25 ␮m
60 min. As previously noted, a fallback of 50 and 5 ␮m on Si3 N4 were especially dedicated to the preconcentration tests without
layer was observed around the preconcentration chamber and at adsorbent.
the inlet and outlet of the channels, respectively, in order to obtain
a significant porosity of the silicon side walls, as shown in Fig. 7. 3.2. Carbon insertion
The porous structures involved in the different designs of micro-
devices are mechanically stable under pressure and even during As mentioned before, the test on the carbon deposition consisted
the adsorbent filling. of two sets of experiments, the first of which was devoted to the
Table 2 summarizes the effect of the various process parame- deposition on silicon flat substrates by the wetting method and the
ters investigated on the nature, depth and specific surface of the second part dedicated to the filling in micro-channels from a fluidic
pores formed in silicon. This table shows that for the p-type micro- solution.
channels, there was a significant reduction of the applied current The first experiments have shown that for the same car-
density and a longer etching time compared with p-type flat sub- bon solution and an identical wetting time, a thicker and more
strates. However, for the n-type micro-channels, it was noticed an homogeneous layer of carbon is obtained on n-type macro-
increase in both the etching time and the current density. porous silicon compared to non-porous silicon, as presented

Fig. 6. SEM images of micro-channels in porous silicon: (a) surface image of the micro-channel, (b) view of the cross section after cleavage of the device, (c) magnification
of the surface image and (d) magnification of the cross section image.
616 E.H.M. Camara et al. / Sensors and Actuators B 148 (2010) 610–619

Fig. 7. (a) Optical image of a silicon wafer including porous micro-channels, (b) SEM image of the cross section of a wall in “straight” design micro-channel and (c) magnification
of the last wall image.

in Fig. 8. This result reflects the fact that carbon particles fit The second part of experiments have shown that the quan-
into the silicon macropores forming a second binding surface tity of carbon deposited in the silicon micro-channels is increased
for new particles and thus increasing the deposited quan- with the presence of structures with an efficiency that seems more
tity. important with the “zigzag” design and on porous silicon (Table 3).

Fig. 8. Top view and cross section SEM images of the carbon deposition on non-porous silicon (a and b) and porous silicon (c and d).
 E.H.M. Camara et al. / Sensors and Actuators B 148 (2010) 610–619 617

Table 3
Quantity of carbon filled in silicon micro-channels with three designs.

Silicon micro-channel Straight Zigzag Neutral

Number of deposition 3 3 3
Weight of deposited carbon (mg) 0.69 ± 0.1 3.67 ± 0.1 1.85 ± 0.1

It was also observed that there is a significant overpressure from

3 to 4 bars in the micro-channels “straight” design after some car-
bon depositions unlike other designs which the loss charge is less
than 1 bar. This overpressure can be caused by the higher surface
covered in the preconcentration chamber by the structures in the
straight design compared to other designs. In short, the overpres-
sure leads to slowdown the progression of particles and to decrease
the quantity of deposited carbon.
Regarding the distribution of the carbon solution in the micro-
channels, the presence of structures in the preconcentration Fig. 10. Breakthrough time determined with the response to benzene of the gas
chamber is a key factor to homogenize the distribution of flows sensor placed after a micro-channel filled with 0.90 mg of carbon.

through the entire device. This was confirmed by the observation

of a uniform deposition of carbon in the case of the designs with
ably and it is set at one for porous silicon micro-channels and two
structures in opposition to the one without. Following these results,
for those non-porous silicon microsytems (see Fig. 9), simplifying
better homogeneity of the carbon deposited in the channels with
the overall filling procedure, which is not negligible if we consider
structures, but appearance of an overpressure with the straight
the manufacturing issue.
structures, we can conclude that the zigzag design is more adapted
The first experiments consisted in determining the break-
to the realization of micro-preconcentrators filled with a carbon
through time with silicon micro-channels filled with the carbon
powder from a fluidic solution.
powder. With 250 ppb of benzene and a flow of 167 mL/min, we
It should be noted that the porous silicon aspect is an important
obtained at 25 ◦ C a breakthrough time of 28.6 min with a micro-
factor for depositing carbon in large quantities in micro-channels.
channel filled with 0.90 mg of carbon (Fig. 10). This breakthrough
Indeed, we have made several experiments of carbon deposition on
time depends on the concentration and flow of gas for a given adsor-
micro-channels which are summarised in Fig. 9. It indicates that the
bent mass. The dependence between the breakthrough time and
contribution of porous silicon is estimated to increase the quantity
the concentration of benzene is not linear since for our operating
of carbon deposited by about a factor 2 compared to non-porous
conditions, the breakthrough time is quite inversely proportional
silicon and that for whatever the design of the micro-channel is.
to the concentration of the pollutant gas and that whatever the
mass of the adsorbent material as shown in Fig. 11 obtained with
3.3. Preconcentration a micro-channel filled with 10.2 mg of carbon and tested in the
same experimental conditions than the previous one. This result
Following the conclusions drawn from the previous experi- shows that for the detection of very low concentration as in the
ments presented in the earlier sections, the preconcentration tests case of 250 ppb of benzene, a large quantity of carbon is not neces-
have been performed on silicon and porous silicon micro-channels sary since, firstly, the breakthrough time is an increasing function
with a “zigzag” design both without and with carbon. As previously of the deposited carbon mass and secondly, the breakthrough time
reported only macroporous silicon was used for filling adsorbent obtained with 0.90 mg of carbon deposition is long enough to cover
while microporous silicon was dedicated solely to the precon- the adsorption phase, with a duration fixed at 5 min, of the precon-
centration test without adsorbent. In order to study in similar centration cycle.
conditions the contribution of porous silicon on the preconcentra- On the one hand, the non-porous silicon micro-channels filled
tion, the quantity of deposited carbon on micro-channels with and with the carbon powder have been tested under benzene. The
without porous silicon will be sensibly equal to about 1 mg. To this demonstration of the preconcentration with a non-porous silicon
effect, the number of deposits required will be reduced consider-

Fig. 9. Quantity of carbon filled in porous and non-porous silicon micro-channels Fig. 11. Breakthrough time as a function of the concentration of benzene for a micro-
by fluidic solution depending on the design. channel filled with 10.2 mg of carbon.
618 E.H.M. Camara et al. / Sensors and Actuators B 148 (2010) 610–619

obtained which the maximum reaches up to 16 ppm at 170 ◦ C, cor-

responding to a “practical” factor of preconcentration of about 64
(Fig. 12).
Unlike the preconcentrator made of non-porous silicon, the des-
orption of porous silicon preconcentrator was almost complete
before the end of the heating cycle since the gas sensor reaches its
normal baseline during this desorption phase. Thus, this result was
confirmed when the heating rate applied was 5 ◦ C/min (Fig. 13).
From a calculation of activation energies of desorption from the
different heating rates, the slow desorption of the pair non-porous
silicon-carbon seems to be due to its activation energy of desorption
bigger than the pair porous silicon–carbon one (60 versus 55 kJ).

4. Conclusions
Fig. 12. Comparison of gas sensor response between a porous silicon micro-channel
filled with 0.83 mg carbon and a non-porous silicon micro-channel filled with The effect of gas preconcentration with silicon micro-channels
0.90 mg carbon after an adsorption step of 250 ppb and desorption step at 200 ◦ C filled with carbon nanopowders was demonstrated. At the moment,
with a heating rate of 160 ◦ C/min. with our experimental conditions and detection system, an adsorp-
tion time of 5 min provides a “practical” preconcentration factor of
55. This factor is largely dependent on the preconcentrator external
micro-channel containing 0.90 mg of carbon nanopowder was real- parameters such as the detection system which we did not try to
ized at very low concentrations under 250 ppb of benzene. During optimize in this paper.
the adsorption phase of benzene no breakthrough was observed One easy way to improve this result would be to remove the
proving that all the injected quantity of pollutant was absorbed by limitation linked to the flow rate. The device can withstand under
the micro-preconcentrator. Then, in the desorption phase, during high flow rate by increasing the diameter of the fluidic connectors
which the micro-channel was heated up to 200 ◦ C with a heat- at the inlet and outlet of the preconcentrator. The analytical calcu-
ing rate of 160 ◦ C/min, the maximum of concentration obtained lations on the pressure drop levels show that a slight increase of
at 196 ◦ C is about 10 ppm corresponding to a “practical” factor of this internal diameter to 385 ␮m would lead to a decrease of the
preconcentration (as defined above) of about 40 (Fig. 12). pressure drop below 200 mbar and allowed a significant increase
On the other hand, we tested a microporous silicon micro- of the flow rate and improved preconcentration factor.
preconcentrator. First, this device without carbon was tested in The use of porous silicon has increased the quantity of deposited
order to validate our set-up. Interestingly, a low and incomplete carbon in the micro-channels because of the presence of pores on
adsorption was observed when benzene was injected in the micro- the surface of the preconcentration chamber constitutes an impor-
channels (1 ppm) and was confirmed during the desorption since a tant asset to get a good binding layer and thus to increase the
small desorption peak appeared at 80 ◦ C when the micro-channel quantity of adsorbent filled in the micro-channel compared to non-
was heated up to 200 ◦ C. We believe that this low gas adsorption porous silicon. The tests of these porous silicon micro-channels
capacity of porous silicon in itself has some potential application in with carbon deposition have shown that the porous aspect of the
the field of preconcentration and deserves more consideration. It micro-channels constitutes an important contribution to the per-
could be used directly as a gas adsorbent material. Following that, formances of the preconcentrator. In fact, we have observed a total
a macroporous silicon micro-preconcentrator containing 0.83 mg desorption of the pollutant in the case of the porous silicon pre-
was tested in similar conditions as those of non-porous silicon concentrator filled with carbon in opposition to the non-porous
micro-preconcentrator. Indeed, such as in the case of non-porous silicon preconcentrator filled also with carbon. This result shows
silicon, no breakthrough is observed during the adsorption phase that the porous silicon, besides increasing the quantity of deposited
and when the micro-channel is heated up to 200 ◦ C with a heat- carbon, contributes to the effectiveness of the gas preconcen-
ing rate of 160 ◦ C/min, (desorption phase), a desorption peak is tration by making the desorption easier and thereby reducing
the desorption temperature and at last the duration of the pre-
concentration cycle. This complete desorption of the pair porous
silicon-carbon/benzene is, may be, due to its activation energy of
desorption lower than the pair non-porous silicon-carbon/benzene
one. In addition, the low gas adsorption capacity of porous silicon
micro-channels without adsorbent is no less important. But the
use of porous silicon itself as adsorbent material requires specific
experimental conditions because its lower temperature of desorp-
tion reflects a fragile thermodynamic equilibrium with benzene
molecules at room temperature (25 ◦ C).
Currently we are evaluating the influence of water vapor on the
preconcentration cycle, which we believe is an essential element
insofar as this type of detector is intended for ambient air analysis
which contains a relatively high humidity. In this type of appli-
cation, carbon nanotubes and polymers appear to be promising
materials in such environmental conditions. The interest of carbon
nanotubes is related not only to their high specific area close to
Fig. 13. Comparison of gas sensor response between a porous silicon micro-channel 1300 m2 /g but also to the possibility to be functionalized by ther-
filled with 0.83 mg carbon and a non-porous silicon micro-channel filled with
0.90 mg carbon after an adsorption step of 250 pb and desorption step at 200 ◦ C
mal or chemical treatment according to the compound to detect
with a heating rate of 5 ◦ C/min. and to reduce their affinity to water.
 E.H.M. Camara et al. / Sensors and Actuators B 148 (2010) 610–619 619

Acknowledgement [19] C. Pijolat, M. Camara, J. Courbat, J.-P. Viricelle, D. Briand, N.F. de Rooij, Appli-
cation of carbon nanopowders for gas micro-preconcentrator, Sensors and
Actuators B 127 (2007) 179–185.
Part of this work has been financially supported by the Rhône- [20] H. Föll, M. Christophersen, J. Cartensen, G. Hasse, Formation and application of
Alpes region through Cluster «chimie». porous silicon, Materials Science and Engineering R39 (2002) 93–141.
[21] J. Salonen, J. Tuura, M. Björkqvist, V.-P. Lehto, Sub-ppm trace moisture detection
with a simple thermally carbonized porous silicon sensor, Sensor and Actuators
References B 114 (2006) 423–426.
[22] P. Bettotti, L. Dal Negro, Z. Gaburro, L. Pavesi, P-type macroporous silicon for
[1] Th. Becker, St. Mühlberger, Chr. Bosch, V. Braunmühl, G. Müller, Th. Ziemann, two-dimensional photonic crystals, Journal of Applied Physics 92 (12) (2002)
K.V. Hechtenberg, Air pollution monitoring using tin-oxide-based microreactor 6966–6972.
systems, Sensors and Actuators B 69 (2000) 108–119.
[2] B. Alfeeli, M. Agah, Selective preconcentration by temperature manipulation of
cascaded micro-preconcentrators, in: Transducers 2009 Conference, Denver, Biographies
CO, USA, June, 2009, pp. 1662–1665.
[3] S.M. Cho, Y. Jun Kim, G.S. Heo, S.-M. Shin, Two step preconcentration for analysis
of exhaled gas of human breath with electronic nose, Sensors and Actuators B El Hadji Malick Camara received his master in Nanosciences and Nanotechnolo-
117 (2006) 50–57. gies from the University of Bourgogne (France). He is currently a PhD student
[4] W.A. Groves, E.T. Zellers, G.C. Frye, Analyzing organic vapors in exhaled breath at both the Ecole des mines of Saint-Etienne (France) and the institute of
using a surface acoustic wave sensor array with preconcentration: selection Microtechnology-Neuchâtel (Switzerland). The objective of his work is to develop a
and characterization of preconcentrator adsorbent, Analytica Chimica Acta 371 micro-preconcentrator for the detection of chemical substances in trace level.
(1998) 131–143.
Philippe Breuil is a research engineer and teacher of Instrumentation at Ecole des
[5] F. Bender, N. Barié, G. Romoudis, A. Voigt, M. Rapp, Development of a precon-
Mines of Saint-Etienne (France) where he received his PhD in chemical engineering
centration unit for a SAW sensor micro array and its use for indoor air quality
in 1989. He developed first some instruments based on UV–vis spectrophotometry
monitoring, Sensors and Actuators B 93 (2003) 135–141.
and chemometrics. Since 2000, he works on the utilization of semiconductors sen-
[6] W.A. Groves, E.T. Zellers, G.C. Frye, Analyzing organic vapors in exhaled breath
sors in the MICC department (Microsystems, Instrumentation and Chemical Sensors)
using a surface acoustic wave sensor array with preconcentration: Selection
attached to the SPIN research center (Natural and Industrial Process Sciences).
and characterization of preconcentrator adsorbent, Analytica Chimica Acta 371
(1998) 131–143. Danick Briand received his B.Eng. degree and M.A.Sc. degree in engineering physics
[7] G. Serrano, H. Chang, E.T. Zellers, A micro gas chromatograph for high-speed from École Polytechnique in Montréal, Canada, in collaboration with the Institut
determinations of explosive vapors, in: Transducers 2009 Conference, Denver, National Polytechnique de Grenoble (INPG), France, in 1995 and 1997, respectively.
CO, USA, June, 2009, pp. 1654–1657. He obtained his PhD degree in the field of micro-chemical systems from the Institute
[8] M. Martin, M. Crain, K. Walsh, R.E. McGill, E. Houser, J. Stepnowski, S. Step- of Microtechnology (IMT), University of Neuchâtel, Switzerland in 2001. His PhD led
nowski, H.-D. Wu, S. Ross, Microfabricaded vapor preconcentrator for portable to the successful technological transfer of micromachined transducing platforms
ion mobility spectroscopy, Sensors and Actuators B 126 (2007) 447–454. to the gas-sensing industry. He is currently a team leader at IMT in the field of
[9] K. Persaud, G.H. Dodd, Analysis of discrimination mechanisms in the mam- MEMS and Microsystems. He is in charge of European and industrial projects, of the
malian olfactory system using a model nose, Nature 299 (1982) 352–355. supervision of doctoral students and of teaching assignments. He is a member of
[10] K. Song, S.-K. Lee, Development of a compact sample pré-concentration system the technical program committee of the IEEE Sensors and PowerMEMS conferences
for the detection of trace amount of volatile organic compounds (VOCs), Sensors and of the steering committee of the Eurosensors conference. He has been author or
and Actuators B 125 (2007) 173–179. co-author on more than 100 papers published in scientific journals and conference
[11] Edward T. Zellers, K.D. Wise, K. Najafi, R.B. Brown, Q.J. Cai, J. Driscoll, Determina- proceedings. His research interests in the field of microsystems include polymeric
tion of complex vapor mixtures in ambient air with a wireless microanalytical and Power MEMS, MEMS packaging and reliability, and the development of micro-
system: vision, progress, and homeland security applications, in: Technical analytical instruments and sensors for gas detection.
Digest of the IEEE Conference on Technologies for Homeland Security, Waltham
MA, IEEE, Boston, November 13–14, 2002, pp. 92–95. Christophe Pijolat is a professor of chemical engineering and microsystems at Ecole
[12] C. Pijolat, D. Briand, Micro-preconcentrator for trace level detection of gases, des Mines of Saint-Etienne (France), he manages the MICC department (Microsys-
MST News No. 4/07, 2007, pp 15–17. tems, Instrumentation and Chemical Sensors) attached to SPIN research center
[13] J.W. Gardner, P.N. Bartlett, Electronic noses: principles and applications, Mea- (Natural and Industrial Process Sciences). Since 1980, he has been working in the
surement Science & Technology 11 (2000) 1087–11087. field of electrical properties of solids and on the development of potentiometric
[14] I. Gràcia, P. Ivanov, F. Blanco, N. Sabaté, X. Vilanova, X. Correig, L. Fonseca, and semiconductor sensors. He has contributed to several technological transfers of
E. Figueras, J. Santander, C. Cané, Sub-ppm gas sensor detection via spiral ␮- sensors into industrial applications.
preconcentrator, Sensors and Actuators B 132 (2008) 149–154.
[15] P.R. Lewis, R.P. Manginell, D.R. Adkins, R.J. Kottenstette, D.R. Wheeler, Recent Nicolaas F. de Rooij received a PhD degree from Twente University of Technology,
advancements in the gas-phase MicroChemLab, IEEE Sensors Journal 6 (3) The Netherlands, in 1978. From 1978 to 1982, he worked at the Research and Devel-
(2006) 784–795. opment Department of Cordis Europa N.V., The Netherlands. In 1982, he joined the
[16] C. Lu, W.H. Steinecker, W.-C. Tian, M.C. Oborny, J.M. Nichols, M. Agah, J.A. Institute of Microtechnology of the University of Neuchâtel, Switzerland (UniNE-
Potkay, H.K.L. Chan, J. Driscoll, R.D. Sacks, K.D. Wise, S.W. Pang, E.T. Zellers, First- IMT), as professor and head of the Sensors, Actuators and Microsystems Laboratory.
generation hybrid MEMS gas chromatograph, The Royal Society of Chemistry, Since October 1990 till October 1996, he was acting as director of the UniNE-IMT.
Lab Chip 5 (2005) 1123–1131. Since 1987, he has been a lecturer at the Swiss Federal Institute of Technology,
[17] C. Roumanie, V. Pijolat, C. Meille, P. Bellefon, C. Pouteau, Delattre, Deposi- Zurich (ETHZ), and since 1989, he has also been a professor at the Swiss Federal Insti-
tion of Pt-catalyst in a micro-channel of a silicon reactor: application to gas tute of Technology, Lausanne (EPFL). His research activities include microfabricated
micro-TAS working at high temperature, Sensors and Actuators B 118 (2006) sensors, actuators and microsystems. He is the director of the Sensors, Actuators
297–304. and Microsystems Laboratory, with more than 40 staff members, more than 500
[18] W.-C. Tian, H.K.L. Chan, C.-J. Lu, S.W. Pang, E.T. Zellers, Multiple-stage microfab- referred publications and several patents. In 2007, he was the recipient of the IEEE
ricated preconcentrator-focuser for micro gas chromatography system, Journal Jun-ichi Nishizawa Medal for “pioneering contributions to microsystem technology
of Micro-Electro-Mechanical Systems 14 (3) (2005) 498–507. and effective transfer into industrial products and applications”.