micromachines

Article
A New Miniaturized Gas Sensor Based on Zener Diode
Network Covered by Metal Oxide
Vignesh Gunasekaran 1,2 , Soffian Yjjou 1 , Eve Hennequin 1 , Thierry Camps 1,3 , Nicolas Mauran 1 ,
Lionel Presmanes 2 and Philippe Menini 1,3, *

1 CNRS, LAAS-MICA, 7 Avenue du Colonel Roche, 31400 Toulouse, France; vgunasek@laas.fr (V.G.);
syjjou@laas.fr (S.Y.); ehennequ@laas.fr (E.H.); camps@laas.fr (T.C.); nicolas.mauran@laas.fr (N.M.)
2 CIRIMAT, CNRS-INP-UPS, Université Toulouse 3 Paul Sabatier, 118 Route de Narbonne, CEDEX 9,
31062 Toulouse, France; presmane@chimie.ups-tlse.fr
3 Faculté des Sciences et Ingénierie, Université Toulouse III Paul Sabatier, 118 Rte de Narbonne,
31400 Toulouse, France
* Correspondence: menini@laas.fr

Abstract: The development of “portable, low cost and low consumption” gas microsensors is one
of the strong needs for embedded portable devices in many fields such as public domain. In this
paper, a new approach is presented on making, on the same chip, a network of head-to-tail facing PN
junctions in order to miniaturize the sensor network and considerably reduce the required power for
heating each cell independently. This paper is about recognizing a device that integrates both sensing
and self-heating. This first study aims to evaluate the possibilities of this type of diode network for
use as a gas sensor. The first part concerns the description of the technological process that is based





 on a doped polysilicon wafer in which a thin layer of metal oxide (a gallium-doped zinc oxide in our
case) is deposited by RF sputtering. An electrical model will be proposed to explain the operation
Citation: Gunasekaran, V.; Yjjou, S.;
and advantage of this approach. We will show the two types of tests that have been carried out
Hennequin, E.; Camps, T.; Mauran,
N.; Presmanes, L.; Menini, P. A New
(static and dynamic) as well as the first encouraging results of these electrical characterizations under
Miniaturized Gas Sensor Based on variable atmospheres.
Zener Diode Network Covered by
Metal Oxide. Micromachines 2021, 12, Keywords: miniaturized gas sensor network; p-n junction; metal oxide; RF-sputtering; ZnO:Ga
1355. https://doi.org/10.3390/ thin layer
mi12111355

Academic Editor: Dimitris Tsoukalas

1. Introduction
Received: 14 October 2021
The current sensors, both commercial and those developed in lab, are the subject of
Accepted: 30 October 2021
compromises: size, sensitivity, selectivity, consumption and cost [1]. The optical sensors
Published: 2 November 2021
by direct infrared absorption remain the most efficient and most selective, but they only
address one target gas. However, their size and high power consumption are unfavorable
Publisher’s Note: MDPI stays neutral
to realize embedded multi-gas sensors. Electrochemical sensors, widely present in the
with regard to jurisdictional claims in
market due to their good performance in sensitivity and selectivity, remain penalized by
published maps and institutional affil-
iations.
their size, their significant consumption, and especially, by their limited lifetime. For their
part, electromechanical sensors (MicroElectroMechanical Systems (MEMS), quartz micro
balances (QMB) or electro-acoustic sensors (bulk acoustic wave (BAW), surface acoustic
wave (SAW)) reveal excellent sensitivities and low detection thresholds (at ppb level) but
require a complex electronic system for driving and measuring—not easy to integrate in
Copyright: © 2021 by the authors.
simple embedded devices. Finally, the conductometric sensors still remain relevant for
Licensee MDPI, Basel, Switzerland.
integration of multi-gas detectors in a reduced volume, moderate power consumption, and
This article is an open access article
above all, simple electronics as an interface with good reliability [2]. Due to significant
distributed under the terms and
conditions of the Creative Commons
advances in sensitive materials (nano-structured metal oxides, inorganic materials, etc.),
Attribution (CC BY) license (https://
these sensors now display strong sensitivities and low detection thresholds (several tens
creativecommons.org/licenses/by/ of ppb).
4.0/).

Micromachines 2021, 12, 1355. https://doi.org/10.3390/mi12111355 https://www.mdpi.com/journal/micromachines

 various gas-sensitive elements with low power consumption [3–5]. This kind of platform
(Figure 1) is compatible with various technics of sensing layer integration, such as inkjet
Micromachines 2021, 12, x FOR PEERprinting
REVIEW or microplotter printing [6,7], screen-printing [8,9] or magnetron sputtering 2[10].of 13
Despite the tiny size and the ease of fabrication, due to its smart structure, it takes
Micromachines 2021, 12, 1355 more than five photolithography processes to pattern all the different layers, including a
2 of 12
backside deep reactive ion etching (DRIE) to obtain the suspended round membrane, add-
these sensors now display strong sensitivities and low detection thresholds (several tens
ing, to the whole system, a mechanical weakness [11]. The heating element was designed
of ppb).
to give a homogeneous temperature distribution across the membrane to bring the sens-
More
More specifically,
specifically, microsensors based on microhotplates
microhotplates composed
composed of
of hanging
hanging mem-
mem-
ing layer to the working range. However, the thermal image of the microhotplate showed
branes with sensing electrodes and heating elements
branes with sensing electrodes and heating elements have proven its high application to
that only a small region in the center of the membrane is at the required temperature [12],
various gas-sensitive elements with low power consumption [3–5]. This kind of platform
and there will be always a significant temperature gradient that is not acceptable for the
(Figure 1) is compatible with various technics of sensing layer integration, such as inkjet
whole sensing material at the surface.
printing or microplotter printing [6,7], screen-printing
screen-printing [8,9]
[8,9] or
or magnetron
magnetronsputtering
sputtering[10].
[10].
Despite the tiny size and the ease of fabrication, due to its smart structure, it takes
more than five photolithography processes to pattern all the different layers, including a
backside deep reactive ion etching (DRIE) to obtain the suspended round membrane, add-
ing, to the whole system, a mechanical weakness [11]. The heating element was designed
to give a homogeneous temperature distribution across the membrane to bring the sens-
ing layer to the working range. However, the thermal image of the microhotplate showed
that only a small region in the center of the membrane is at the required temperature [12],
and there will be always a significant temperature gradient that is not acceptable for the
whole sensing material at the surface.
Figure
Figure1.1.(a)
(a)Top
Topview
viewof
ofthe
themicrohotplate;
microhotplate;(b)
(b)Cross
Crosssection
sectionof
ofthe
thesensor.
sensor.

Concerning
Despite the the tinysensing
size andelectrodes,
the ease of we can find due
fabrication, fewto different
its smartshapes thatitare
structure, usually
takes more
than five
suited photolithography
to the sensitive material processes to pattern
resistivity. Someallcan thebe different layers, including
a miniaturized a backside
pair of electrodes
deepfew
with reactive
micronsion asetching
a gap (DRIE) to obtain the electrodes,
[5] or interdigitated suspended as round membrane,
an example: adding,
the comb to the
shaped
whole system,
sensing electrodes a mechanical
given in Figure weakness2 [13][11].
withThetight heating elementelectrodes.
interdigitated was designed Theto gapgive
be-a
homogeneous
tween temperature
the two electrodes is L distribution
= 10 µm, andacross the probedthe membrane
length W =to1620 bring
µmthe sensing
gives layer
a number
oftosquare
the working
n□ = L/W range.
of 6.2 However,
× 10−3□, and the thermal
a sheet image of
resistance (R□the
) in microhotplate
a range from 10showedkΩ/□ tothat 10
GΩ/□. If we consider the thickness range of the sensing layer (around 100 nm),[12],
only a small region in the center of the membrane is at the required temperature we canand
there willthe
calculate berange
always ofaresistivity
significantthat temperature gradient that
suits this platform. is not
In this acceptable
case, we have for thethe whole
possibil-
Figure
sensing 1. (a) Top
material view of the microhotplate;
at the surface. (b) Cross section of the sensor.
ity to measure materials with a resistivity from 0.1 Ω·cm to100 kΩ·cm. To be able to char-
acterize Concerning
materials the withsensing
higherelectrodes,
resistivity, we can is
which find
usualfewfordifferent
variousshapes
metal that
oxidesare(MOX),
usually
suited Concerning
to the the sensing
sensitive material electrodes,
resistivity.weSome
can find can few
be adifferent
miniaturizedshapes thatofare
pair usually
electrodes
it could be interesting to significantly reduce the gap L between the two electrodes.
suited
with few to microns
the sensitive
as a gapmaterial
[5] orresistivity.
interdigitatedSome can be aas
electrodes, miniaturized
an example:pair of electrodes
the comb shaped
with few microns as a gap [5] or interdigitated electrodes,
sensing electrodes given in Figure 2 [13] with tight interdigitated electrodes. The as an example: the comb shaped
gap
sensing
betweenelectrodes given in Figure
the two electrodes is L =210 [13]
µm,with
and tight
the interdigitated
probed lengthelectrodes.
W = 1620 The µm gapgives be-a
tween
number theoftwo electrodes
square n = L/W is L =of106.2µm, × and
10−3the
, probed
and a sheet length W = 1620(Rµm
resistance  ) gives
in a a
rangenumber
from
of
10 square
kΩ/ nto□ =10L/W GΩ/ of6.2. If× we
10−3consider
□, and a sheet resistancerange
the thickness (R□) in ofathe
range fromlayer
sensing 10 kΩ/□ to 10
(around
GΩ/□.
100 nm), we can calculate the range of resistivity that suits this platform. In this case,can
If we consider the thickness range of the sensing layer (around 100 nm), we we
calculate
have the the range oftoresistivity
possibility measure that suits this
materials with platform.
a resistivityIn this case,
from Ω·have
0.1we the possibil-
cm to100 kΩ·cm.
ity
To to bemeasure materials with
able to characterize a resistivity
materials withfrom
higher 0.1 resistivity,
Ω·cm to100which kΩ·cm.isTo be able
usual for to char-
various
acterize
metal oxides materials
(MOX), with it higher
could be resistivity,
interesting which is usual forreduce
to significantly variousthe metal
gap oxides
L between (MOX),the
ittwo
could be interesting to significantly reduce the gap L between the two electrodes.
electrodes.
Figure 2. Structure of the comb-shaped sensing electrodes.

The weak point of these “resistive” sensors remains their low intrinsic selectivity re-
lated to the resistance measurement influenced by many gases (and other influencing fac-
tors), regardless of the type of the material used. This drawback can be overcome by the

Figure
Figure 2.
2. Structure
Structure of
of the
the comb-shaped
comb-shaped sensing electrodes.

The weak
The weakpoint
pointofofthese
these“resistive”
“resistive”sensors
sensors remains
remains their
their lowlow intrinsic
intrinsic selectivity
selectivity re-
related to the resistance measurement influenced by many gases (and other influencing
lated to the resistance measurement influenced by many gases (and other influencing fac-
factors),
tors), regardless
regardless of of
thethe type
type ofof
thethe materialused.
material used.This
Thisdrawback
drawbackcancanbebeovercome
overcome byby the
the
use of a multi-sensor (with various materials) associated with an optimized operating mode.
However, their use (in case of numerus sensors in package) in an autonomous portable
system remains limited by their power consumption due to the heating resistor (several
 Micromachines 2021, 12, 1355 3 of 12

tens of mW), which is necessary for the variability of the sensitivity and the reversibility of
the sensor. This indicates the importance to develop new miniaturized selective gas multi-
sensors with low power consumption (<1 mW) and a low limit of detection (several ppb).

2. Zener Diode-Based Platform

2.1. Design and Model of the Device
We propose here a new type of gas-sensing platform based on an ultra-miniaturized
matrix gas multi-sensor with multiple Zener diodes in series and parallel powered by
a single source. This concept is based on a simple and robust technological approach,
such as the locally doping polysilicon layer, and then it uses a powerful transduction
technique (with a simple current measure), and consequently an ease-of-use electronic
interface. Above all, this approach allows low power consumption (due to the small reverse
current in a diode) well suited for applications focused on embedded systems and Internet
of Things.
As can be seen hereafter, it consists of a simple technological process flow that can be
summarized in only three photolithography steps. From initial p-type polysilicon layer, by
locally doping this polysilicon (to become a n-type region), we obtain a double junction
that is head-to-tail (Figure 3b). When one is in forward conduction (VD ), the other will be
in reverse bias; thus, in Zener conduction (VZ ), the overall conduction will be driven by
the diode in reverse conduction. Due to an insulated SiO2 layer opened above only
Micromachines 2021, 12, x FOR PEER REVIEW
one
4 of 13
junction, we can define a reference diode (covered by SiO2 ) and a sensing diode (covered
at the end by a metal oxide layer) (Figure 3c).

Figure
Figure3. 3.
Basic cross
Basic section
cross of Zener
section diode-based
of Zener diode-basedplatform: (a) the
platform: (a)overall structure
the overall of a single
structure ele-
of a single
ment; (b) the electrical equivalent circuit; (c) the electrical equivalent circuit after the sensitive MOX
element; (b) the electrical equivalent circuit; (c) the electrical equivalent circuit after the sensitive
layer.
MOX layer.

Furthermore, by increasing
Thus, if a current is appliedthe level of
between current flowing
terminals A and B,through the sensing
the reference diode isdiode in
in Zener
Zener conduction,
conduction, theevolution
and the dissipatedofelectrical power
the overall focuses
voltage of theon the sensing
sensor will bearea (delimited
almost the same
byasWthat
T (Figure
of the3c)). Thus
single the localized
reference diode self-heating can reach However,
in Zener conduction. several hundred degrees of
if the polarity with
this
reduced levels of the applied current (around 1 mA), and play the role of self-heating
element in this new type of platform. Thus, the significant reduction in the heating zone
should make it possible to reach high temperatures necessary for gas detection in the
MOX, with a reduced power of typically below 1 mW and with fast thermal transients
(<ms) due to ultra-localized power above the junction (>100 kW/cm2). This approach then
Micromachines 2021, 12, 1355 4 of 12

current is reversed, the measured voltage VAB will almost coincide with that at the terminals
of the sensing diode. Thus, by simple subtraction of the forward and reverse currents, the
conduction in the junctions is removed to keep only that linked to the conduction in the
MOX layer deposited on the open area above the sensing diode.
Effectively, to obtain a miniaturized gas sensor, a gas-sensitive layer (a metal oxide
layer for example) is deposited above the detection diode in order to allow current to pass
through this layer instead of through the junction in reverse conduction (see Figure 3c).
In addition, due to the high doping level, the space charge zone in reverse polarization
is reduced (<1 µm) and corresponds to the inter-electrode distance of conventional inter-
digitated structures. This opens up prospects for characterizing sensitive layers of high
resistivity. The resistances of the lateral access zones (contact resistance and in the N and
P polysilicon layer) are negligible due to the high level of doping of the N and P zones
(ND > 1021 cm−3 , NA > 1018 cm−3 respectively).
Furthermore, by increasing the level of current flowing through the sensing diode in
Zener conduction, the dissipated electrical power focuses on the sensing area (delimited
by WT (Figure 3c)). Thus the localized self-heating can reach several hundred degrees
with reduced levels of the applied current (around 1 mA), and play the role of self-heating
element in this new type of platform. Thus, the significant reduction in the heating zone
should make it possible to reach high temperatures necessary for gas detection in the MOX,
with a reduced power of typically below 1 mW and with fast thermal transients (<ms) due
to ultra-localized power above the junction (>100 kW/cm2 ). This approach then allows the
possibility to modulate the operating temperature of the sensor (well known in classical
MOX sensors) by adjusting the diode reverse current, another way to optimize the sensor
sensitivity for each sensing material.

2.2. Fabrication of the Device

The fabrication of this device uses only basic and standard microelectronic processes.
First of all, a thick SiO2 passivation layer (2 µm) is deposited by low pressure chemical
vapor deposition (LPCVD) onto a standard silicon wafer, followed by deposition of a
polysilicon layer also by LPCVD, and then a full wafer boron implantation permit to adjust
p-type doping level and local n-type doped zones by phosphorus diffusion through a SiO2
mask to obtain a pair of Zener diodes (N++/P+/N++ junctions) (Figure 3a). The whole
structure is passivated by a 300 nm SiO2 layer (also by LPCVD) (in light green on Figure 3a).
Then, the contact zones and the active detection area are opened by RIE etching. Finally,
the contacts pads, in aluminum, deposited by sputtering (500 nm), are patterned by wet
chemical etching. As the N++ regions of polysilicon are degenerated, the contact is ohmic
and has low contact resistance (<1 Ω) (in yellow on Figure 3).
In order to obtain better sensitivity and a large working area, we designed a diode ar-
ray (matrix) (2 × 2 mm2 ), keeping the head-to-tail facing diode configuration
(Figure 4). This first operational platform has a total working length of W = 6.6 cm and a
gap of L = 0.5 µm, which gives a ratio of L/W = 7.6 × 10−6  and thus a sheet resistance
R in the range of 130 MΩ/ to 13 TΩ/ corresponding to sensitive MOx resistivity from
10 kΩ·cm to 1 GΩ·cm. This is large compared with standard interdigitated electrodes on
microhotplate platforms. Thus, this platform is well adapted for high-resistive materials.
The gas sensing ability of this whole system was tested using, first, a 25 nm thick
gallium-doped ZnO (ZnO:Ga). Zinc oxide is already studied for various applications, and
its gas-sensing ability makes it interesting for our work. The choice of gallium-doped zinc
oxide comes from a previous work, where its gas-sensing performance was proven in a
framework of classical resistive sensors [11].
The sensitive layer was deposited by radio-frequency (RF) sputtering. This method
is well suited with the industrial fabrication of miniaturized microelectronics. RF sput-
tering presents many other advantages, such as the possibility of making thin films with
nanometric scale grain sizes and controlling, without difficulty, the inter-granular poros-
ity [14,15] by varying deposition conditions [16]. Controlled nanostructures films, such
 Micromachines 2021, 12, x FOR PEER REVIEW 5 of 13
Micromachines 2021, 12, 1355 5 of 12

This first operational platform has a total working length of W = 6.6 cm and a gap of L =
as0.5 µm, present
these, which gives
greatainterest
ratio of for
L/Wtheir
= 7.6potential
× 10−6□ and thus a sheet
gas-sensing resistance
ability R□ in the onto
once integrated range
gas-sensing
of 130 MΩ/□platforms
to 13 TΩ/□[14]. Films with to
corresponding a controlled nanostructure
sensitive MOx are of10great
resistivity from interest
k Ω·cm to 1 G
because of their
Ω·cm. This potential
is large to act as
compared sensitive
with layers
standard after being integrated
interdigitated electrodes onto gas-sensing
on microhotplate
devices [17].Thus, this platform is well adapted for high-resistive materials.
platforms.

Figure4.4.Example
Figure Exampleofofstructure
structureofofthe
thenew
newsensing
sensingplatform
platformwith
with4646pairs
pairsofofdiodes.
diodes.

Thethin
The gasfilm
sensing ability ofby
was obtained this whole system
sputtering, usingwas tested using,
a homemade first,ceramic
sintered a 25 nm thick
target
ofgallium-doped
ZnO:Ga with aZnO (ZnO:Ga).
density Zinc70%,
of around oxideand
is already studied parameters
the deposition for various applications,
were optimized and
toitspromote
gas-sensing abilityintergranular
maximum makes it interesting forTable
porosity. our work. The choice
1 presents of gallium-doped
the whole parameter setzinc
of
oxide
the comes from
experiment [18].a previous work, where its gas-sensing performance was proven in a
framework of classical resistive sensors [11].
Table 1.The sensitiveparameters
Deposition layer wasofdeposited
thin layers.by radio-frequency (RF) sputtering. This method
is well suited with the industrial fabrication of miniaturized microelectronics. RF sputter-
ing presents manyTarget other
Material
advantages, such as the possibility ZnO:Gaof making thin films with
nanometric scaleMagnetron
grain sizes and controlling, without difficulty, Yes the inter-granular poros-
ity [14,15] by varying deposition
Power conditions [16]. Controlled nanostructures
30 W films, such as
these, present Argon
great pressure
interest for their potential gas-sensing ability 2 Pa once integrated onto
gas-sensing platforms [14]. Films with a controlled nanostructure are of great interest be-
Target to substrate distance 7 cm
cause of their potential to act as sensitive layers after being integrated onto gas-sensing
devices [17]. Thickness 25–50 nm
The thin film was obtained by sputtering, using a homemade sintered ceramic target
of ZnO:Ga with
A first 25 nm athick
density of around
ZnO:Ga 70%,deposited
layer was and the deposition parameters
onto fused silica substratewereforoptimized
morpho-
to promote
logical maximum using
characterization intergranular
the D3000 porosity.
VEECOTable 1 presents
Atomic the whole(AFM).
Force Microscope parameter set of
Tapping
the experiment
Mode images were [18].
obtained using silicon TESP-SS cantilevers with a resonance frequency
of around 300 kHz. The composition of this thin layer was verified by the microprobe
Table 1. Deposition
technique parameters of thin
and the gallium-doping layers.
rate was that of 4% by mass.
The AFM observation shows
Target Material a surface with well-defined circular
ZnO:Gagrain and intergran-
ular porosity (see Figure 5). Using image analysis [19], the mean grain size is centered on
Magnetron Yes
4.8 nm with an error of 0.1 nm, and the measured roughness for this sample is 1.1 nm. This
Power 30 W
is excellent for interactions with gas molecules. This morphology is typical of a sputtering
Argon pressure
deposition and is well known from previous studies [20]. 2 Pa
Target to substrate distance 7 cm
Thickness 25–50 nm

A first 25 nm thick ZnO:Ga layer was deposited onto fused silica substrate for mor-
phological characterization using the D3000 VEECO Atomic Force Microscope (AFM).
 frequency of
frequency of around
around 300
300 kHz.
kHz. The
The composition
composition of of this
this thin
thin layer
layer was
was verified
verified byby the
the mi-
mi-
croprobe technique and the gallium-doping rate was that
croprobe technique and the gallium-doping rate was that of 4% by mass. of 4% by mass.
The AFM
The AFM observation
observation shows
shows aa surface
surface with
with well-defined
well-defined circular
circular grain
grain and
and inter-
inter-
granular porosity (see Figure 5). Using image analysis [19], the mean grain
granular porosity (see Figure 5). Using image analysis [19], the mean grain size is centeredsize is centered
Micromachines 2021, 12, 1355 on 4.8
on 4.8 nm
nm with
with an
an error
error of
of 0.1
0.1 nm,
nm, and
and the
the measured
measured roughness
roughness for for this
this sample
sample is is 1.1
1.1 nm.
6 ofnm.
12
This is excellent for interactions with gas molecules. This morphology is typical
This is excellent for interactions with gas molecules. This morphology is typical of a sput- of a sput-
tering deposition
tering deposition and
and is
is well
well known
known fromfrom previous
previous studies
studies [20].
[20].

Figure5.5.
Figure
Figure 5.AFM
AFMimage
AFM imageof
image ofaaa25
of 25nm
25 nmthick
nm thickZnO:Ga.
thick ZnO:Ga.
ZnO:Ga.

Thisgas-sensitive
This
This gas-sensitivelayer
gas-sensitive layerwas
layer wasdeposited
was depositedonto
deposited ontothe
onto theplatform
the platformusing
platform usingRF
using RFmagnetron
RF magnetronsput-
magnetron sput-
sput-
tering
tering through polymer shadow mask, with the correct opening
tering through polymer shadow mask, with the correct opening above the diodes.The
through polymer shadow mask, with the correct opening above
above the
the diodes.
diodes. The
The
shadow
shadow
shadowmask mask (in
mask(in kapton)
(inkapton)
kapton)was was stuck
wasstuck
stuckononon the
the surface
surface
the surface of
of of the
thethe device,
device, exposing
exposing
device, exposing the
thethe right
right area
areaarea
right to
to deposit,
to deposit,
deposit, and
andand was
waswas then
then
then removed
removed
removed afterafter deposition.
deposition.
after First
First
deposition. prototypes
prototypes
First prototypeswerewere obtained
obtained
were obtained (Figure
(Figure 6),
(Figure
6), and
and
6), and the chips
the chips
the chips were
werewere mounted
mounted in aa TO-8
in a TO-8
mounted in TO-8 package
package with
with with
package the electrodes,
electrodes,
the electrodes,
the and theand
and the package
package
the package
was
was connected
connected with with
wedge wedge
bondingbonding
using using
25 µm 25 µm aluminum
aluminum
was connected with wedge bonding using 25 µm aluminum wires. wires.wires.

Figure 6.
Figure 6. Single
Single device
device of
of Zener
Zener diode
diode array
array with
with MOx
MOx thin
thin layer
layer deposition.
deposition.
Figure 6. Single device of Zener diode array with MOx thin layer deposition.

2.3.Preliminary
2.3.
2.3. PreliminaryTests
Preliminary Testsand
Tests andObservations
and Observations
Observations
2.3.1.
2.3.1.Role
2.3.1. Roleof
Role ofDoping
of DopingLevel
Doping Level
Level
Various
VariousZener
Various Zenerdiodes
Zener diodeswith
diodes withdifferent
with differentp-type
different p-typeinitial
p-type initialdoping
initial dopinglevels
doping levelswas
levels wasfabricated,
was fabricated,and
fabricated, and
and
the characteristics
the characteristicsof
characteristics ofthe
of thewhole
the wholesystem
whole system were
system were first
were first studied
first studied (Figure
studied (Figure 7).
(Figure 7).
7).
the
As shown in Figure 7, the behavior of the reference diode (covered by SiO2 ) and of
the sensing diode (with MOX at the surface) for different doping levels of the P + zone
is distinct.
The forward characteristic shows, as expected, that the Zener voltage decreases when
the doping of the P zone increase. Conversely, the reverse characteristics seem less dis-
persed and show that the overall conduction integrates a non-negligible contribution
through the MOX layer.
To enhance the impact of this conduction in the MOX layer, we opt for the less doped
platform (NA = 3 × 1017 cm−3 ), which corresponds to the highest Zener voltage close to
VZ ~10 V.
Micromachines 2021, 12, 1355 7 of 12
Micromachines 2021, 12, x FOR PEER REVIEW 7 of 13

Figure 7. Electrical characterization of pair of Zener diodes with different initial doping level.

As shown in Figure 7, the behavior of the reference diode (covered by SiO2) and of
the sensing diode (with MOX at the surface) for different doping levels of the P + zone is
distinct.
The forward characteristic shows, as expected, that the Zener voltage decreases when
the doping of the P zone increase. Conversely, the reverse characteristics seem less dis-
persed and show that the overall conduction integrates a non-negligible contribution
through the MOX layer.
To enhance the impact of this conduction in the MOX layer, we opt for the less doped
platform (NA = 3 × 1017 cm−3), which corresponds to the highest Zener voltage close to VZ~10
V.
Figure 7. Electrical characterization of pair of Zener diodes with different initial doping level.
Figure 7. Electrical characterization of pair of Zener diodes with different initial doping level.

2.3.2.
2.3.2. Influence of Surrounding Atmosphere
AsInfluence
shown inofFigure
Surrounding Atmosphere
7, the behavior of the reference diode (covered by SiO2) and of
Figure
the sensing 88 shows
Figurediodeshows the
(with I-V
I-V characteristics
theMOX at the surface)of
characteristics forthe
of the reference
reference
different diode
diode
doping and
and the
levels sensing
thethe
of sensing diode
diode
P + zone is
under ambient atmosphere (with 50% of relative humidity) and then under
under ambient atmosphere (with 50% of relative humidity) and then under dry air flow
distinct. dry air flow in
the probe
in the
The station
probe chamber.
station
forward chamber. shows, as expected, that the Zener voltage decreases when
characteristic
the doping of the P zone increase. Conversely, the reverse characteristics seem less dis-
persed and show that the overall conduction integrates a non-negligible contribution
through the MOX layer.
To enhance the impact of this conduction in the MOX layer, we opt for the less doped
platform (NA = 3 × 1017 cm−3), which corresponds to the highest Zener voltage close to VZ~10
V.

2.3.2. Influence of Surrounding Atmosphere

Figure 8 shows the I-V characteristics of the reference diode and the sensing diode
under ambient atmosphere (with 50% of relative humidity) and then under dry air flow
in the probe station chamber.

Figure 8. Characterization
Figure 8. Characterization of
of the
the diodes
diodes set
set under
under ambient
ambient atmosphere
atmosphereand
anddry
dryair
airflow.
flow.

After
After further
furthertests,
tests,ititisisclear
clearthat
thatonly
onlythethesensing
sensingdiode
diodethreshold
threshold voltage (directly
voltage in
(directly
contact with atmosphere) is sensitive to the humidity variation in the air,
in contact with atmosphere) is sensitive to the humidity variation in the air, contrary to contrary to the
reference diode.
the reference ThatThat
diode. provesprovesthe surface effects,
the surface by the
effects, by shift toward
the shift the high
toward values
the high of the
values of
threshold voltage, of the sensing diode, even at low
the threshold voltage, of the sensing diode, even at low current.current.
A similar qualitative test was performed using isopropanol (IPA) vapor injected
manually into the chamber to observe the behavior of the diodes and to assure that there is
no diffusion of COV through the SiO2 -insulated layer. For that, two sets of diodes were
considered, the first one with a naked sensing diode and the second one with both diodes
covered by a SiO2 layer. Both sets were exposed to the ambient air and a small amount of
IPA vapor.
FigureFigure
8. Characterization
9 shows the of characteristics
the diodes set under ambient
of the atmosphere
two sets. and dry air flow.
The passivated one showed a
classical Zener diode behavior, and the naked sensing diode showed different responses
After
to the further
ambient airtests, it is clear
and under IPA,that only the
proving sensing
again, diode threshold
the concept voltagedue
of gas detection (directly
to the
innaked
contact with atmosphere) is sensitive to
P-N junction compared with the passivated one.the humidity variation in the air, contrary to
the reference diode. That proves the surface effects, by the shift toward the high values of
the threshold voltage, of the sensing diode, even at low current.
 considered, the first one with a naked sensing diode and the second one with both diodes
covered by a SiO2 layer. Both sets were exposed to the ambient air and a small amount of
covered by a SiO2 layer. Both sets were exposed to the ambient air and a small amount of
IPA vapor.
IPA vapor.
Figure 9 shows the characteristics of the two sets. The passivated one showed a clas-
Figure 9 shows the characteristics of the two sets. The passivated one showed a clas-
sical Zener diode behavior, and the naked sensing diode showed different responses to
sical Zener diode behavior, and the naked sensing diode showed different responses to
Micromachines 2021, 12, 1355 the ambient air and under IPA, proving again, the concept of gas detection due to8 of the
12
the ambient air and under IPA, proving again, the concept of gas detection due to the
naked P-N junction compared with the passivated one.
naked P-N junction compared with the passivated one.

Figure
Figure9.9. Characterization of two
Characterization of two sets
setsofofdiodes
diodeswith
withand
and without
without passivation
passivation under
under ambient
ambient air
air and
Figure
and 9. Characterization
isopropanol vapors. of two sets of diodes with and without passivation under ambient air
isopropanol vapors.
and isopropanol vapors.
In
Inthis
this final preliminary
preliminary test,
test,we
wewanted
wantedtotocompare
compare thethe sensing
sensing diodes
diodes I-VI-V charac-
characteris-
In this final preliminary test, we wanted to compare the sensing diodes I-V charac-
teristics
tics withwith
and and
withoutwithout the MOX
the MOX sensitive
sensitive layer.layer. It is obvious
It is obvious (i.e., Figure
(i.e., Figure 10) that10)
thethat the
reverse
teristics with and without the MOX sensitive layer. It is obvious (i.e., Figure 10) that the
reverse
currentcurrent flow passed
flow passed throughthrough
the P–Nthe P–N junction
junction withoutwithout MOX, revealing
MOX, revealing high re-
high resistance.
reverse current flow passed through the P–N junction without MOX, revealing high re-
sistance.
When the When
MOX the MOX
layer is layer is present,
present, the equivalent
the equivalent resistance
resistance is significatively
is significatively lower lower
at low
sistance. When the MOX layer is present, the equivalent resistance is significatively lower
at low current
current (100which
(100 nA), nA), which
proves proves
that thethat the reverse
reverse current current in thediode
in the Zener Zenerisdiode is well
well shunted
at low current (100 nA), which proves that the reverse current in the Zener diode is well
shunted by the in
by the current current in theThis
the MOX. MOX. This is precisely
is precisely what wewhat we wanted.
wanted.
shunted by the current in the MOX. This is precisely what we wanted.

Figure10.
Figure Characterizationof
10.Characterization ofthe
thesensing
sensingdiode
diodewith
withand
andwithout
withoutMOX
MOXlayer.
layer.
Figure 10. Characterization of the sensing diode with and without MOX layer.
3. Results of Gas-Sensing Tests
3.1. Test Protocol
The apparatus was placed in a chamber with a controlled atmosphere and connected
to a gas dilution bench, where the gas concentration is controlled by a program under
Labwindows CVI, implemented on a computer. It is also possible to control the overall
flow from a few cc/min to 1 L/min, and the relative humidity at room temperature can
vary from a few % to 70%. The micro-sensor is either connected to a source-measure unit
(SMU—NI-PXI) for static measurement or connected to a Hioki impedance analyzer and
driven by LabVIEW for dynamic analysis.
For the static test, dry air was continuously injected into the 250 cc quartz chamber,
and after two hours, 10 ppm of formaldehyde was injected during 30 min. In a second
part, humid air (with 30% relative humidity) was continuously injected, and after 1.5 h,
10 ppm of formaldehyde was again injected into the humid air. The total flow rate was
 flow from a few cc/min to 1 L/min, and the relative humidity at room temperature can
vary from a few % to 70%. The micro-sensor is either connected to a source-measure unit
(SMU—NI-PXI) for static measurement or connected to a Hioki impedance analyzer and
driven by LabVIEW for dynamic analysis.
Micromachines 2021, 12, 1355 For the static test, dry air was continuously injected into the 250 cc quartz chamber,9 of 12
and after two hours, 10 ppm of formaldehyde was injected during 30 min. In a second
part, humid air (with 30% relative humidity) was continuously injected, and after 1.5 h,
10 ppm of formaldehyde was again injected into the humid air. The total flow rate was
continuously
continuously maintained
maintainedat at 200
200 cc/min. First, the
cc/min. First, the static
static response
response of
of the
the sensor
sensor consisted
consisted ofof
measuring the overall resistance when a 1 mA reverse current was injected into the
measuring the overall resistance when a 1 mA reverse current was injected into the sens- sensing
diodes (Figure
ing diodes 11). 11).
(Figure

Figure 11. Static testprotocol:

Static test protocol:equivalent
equivalent resistance
resistance measurement
measurement versus
versus time time
underunder various
various atmos-
atmospheres.
pheres.
In the second step, the dynamic tests consisted of measuring the impedance with a
In thesweep
frequency secondfrom
step,10the
Hzdynamic
to 10 kHztests
andconsisted
with one of measuring
sample the impedance
per second during thewith
samea
frequency
gas sweep
injection fromshown
diagram 10 Hz previously.
to 10 kHz and with one sample per second during the same
gas injection diagram shown previously.
3.2. Results and Analysis
3.2.1. Static and
3.2. Results Measurements
Analysis
3.2.1.The first
Static results of the static test reveal a variation of resistance during the two
Measurements
injections of gas as well in the dry air, such as in humid air, proving that this new device
The first results of the static test reveal a variation of resistance during the two injec-
reveals a significant response to humidity and gas variations (Figure 12). Despite its low
tions of gas as well in the dry air, such as in humid air, proving that this new device reveals
sensitivity (a few % for 10 ppm of formaldehyde), the sensor is also able to detect this target
a significant response to humidity and gas variations (Figure 12). Despite its low sensitiv-
Micromachines 2021, 12, x FOR PEER REVIEW 10 of 13
gas in dry air and at 30% of relative humidity, which demonstrates the proof of concept of
ity (a few % for 10 ppm of formaldehyde), the sensor is also able to detect this target gas
gas detection at room temperature.
in dry air and at 30% of relative humidity, which demonstrates the proof of concept of gas
detection at room temperature.

Figure
Figure 12.
12. Static
Static test:
test: resistance
resistance measurement
measurement versus
versus time
time under
under various
various atmospheres.
atmospheres. Injection
Injection of
of
10
10 ppm
ppm of
of formaldehyde
formaldehyde underunder dry
dry air
air and
and then
then under
under humid
humid air
air (30%RH).
(30%RH).

3.2.2. Dynamic
3.2.2. Dynamic Measurements
Measurements
Figure 13a,b shows the
Figure the results
resultsof
ofthe
thedynamic
dynamictests.
tests.There
Thereisisa ashift toto
shift lower
lowervalues
valuesof
impedance when injecting target gas at low frequencies, confirming the results
of impedance when injecting target gas at low frequencies, confirming the results of the of the static
static test, and at higher frequencies (at 10 kHz). Although the real part (resistance) reveals
small variations, the imaginary part (capacitance) is significantly affected by gas and hu-
midity.
 Figure 12. Static test: resistance measurement versus time under various atmospheres. Injection of
10 ppm of formaldehyde under dry air and then under humid air (30%RH).

3.2.2. Dynamic Measurements

Micromachines 2021, 12, 1355 10 of 12
Figure 13a,b shows the results of the dynamic tests. There is a shift to lower values
of impedance when injecting target gas at low frequencies, confirming the results of the
static test, and at higher frequencies (at 10 kHz). Although the real part (resistance) reveals
test, and
small at higherthe
variations, frequencies
imaginary (atpart
10 kHz). Although
(capacitance) is the real part (resistance)
significantly affected byreveals
gas andsmall
hu-
variations,
midity. the imaginary part (capacitance) is significantly affected by gas and humidity.

Micromachines 2021, 12, x FOR PEER REVIEW 11 of 13

(a)

(b)
Figure 13.
Figure 13. Results of dynamic tests
tests under
under the
the same
same atmospheres
atmospheres versus
versus in
in static
statictest:
test: (a)
(a) impedance
impedance
at low
at low frequencies;
frequencies; (b)
(b) resistance
resistance and
and capacitance
capacitanceat
at1010kHz.
kHz.

dynamic test
The dynamic test makes
makes itit possible
possible to
to measure
measure two
two characteristics
characteristics to
to distinguish
distinguish the
the
target gas, which
which can
canhopefully
hopefullybe beused
usedtotoimprove
improve
thethe selectivity
selectivity of this
of this new new type
type of
of gas
gas sensor.
sensor.

4.
4. Conclusions
Conclusions
We
We have
havepresented
presenteda anew
newtypetypeofof
miniaturized
miniaturized gasgas
sensor based
sensor on on
based a surface conduc-
a surface con-
tion of aof
duction Zener diode
a Zener coated
diode with
coated withporous ZnO:Ga
porous ZnO:Ga thin
thinlayer
layerusing
usingRFRFsputtering.
sputtering.This
This
technology
technologyallows
allowsaahigh
highlevel
levelofofintegration inin
integration terms
termsof of
thethe
probed
probedsurface of the
surface detection
of the detec-
material (level less than ppm) and then allows the analysis of conduction in
tion material (level less than ppm) and then allows the analysis of conduction in materials materials
with
with high
high resistivity.
resistivity.This
This approach
approachalso alsomakes
makesitit possible
possibletoto operate
operate with
with the
the reverse
reverse bias
bias
current: at low level (few µA) during measurement and few mA to exploit
current: at low level (few µA) during measurement and few mA to exploit self-heating self-heating and
thus control
and thus the sensitivity
control of theofsensor.
the sensitivity the sensor.
We proved the effects of the surface on the characteristics of the diode with and
We proved the effects of the surface on the characteristics of the diode with and with-
without the MOX layer, which allowed us to use this head-to-tail diode configuration as a
out the MOX layer, which allowed us to use this head-to-tail diode configuration as a
platform for gas-detection materials.
platform for gas-detection materials.
Two types of tests were carried out on the use of this micro-platform, a static electrical
Two types of tests were carried out on the use of this micro-platform, a static electrical
test and a dynamic electrical test with a frequency sweep from 10 Hz to 10 kHz. The
test and a dynamic electrical test with a frequency sweep from 10 Hz to 10 kHz. The results
show a resistance response to 10 ppm of formaldehyde gas in dry air and at 30% humidity.
The dynamic test reveals a variation of real part (resistance) at low frequency and a vari-
ation of imaginary part (capacitance) at high frequency, which is promising in the per-
spective of improving selectivity.
Micromachines 2021, 12, 1355 11 of 12

results show a resistance response to 10 ppm of formaldehyde gas in dry air and at 30%
humidity. The dynamic test reveals a variation of real part (resistance) at low frequency
and a variation of imaginary part (capacitance) at high frequency, which is promising in
the perspective of improving selectivity.
Despite the low sensitivity obtained by these first, and not optimal, experimental
results, this new device revealed encouraging results, and this device can be used as a
gas sensor.
In perspective, to make this new device more reliable and more sensitive, we must
carry out more tests under various gases and study the effect of (i) the design (number
of diodes in parallel), (ii) the temperature (by playing with different reverse bias current)
on the entire structure and of course, and (iii) other sensitive materials to consider the
possibility to realize a multigas sensor platform.

Author Contributions: Data curation, S.Y., E.H., T.C. and N.M.; Investigation, P.M.; Methodology,
T.C. and P.M.; Software, N.M.; Supervision, P.M.; Writing—original draft, V.G. and T.C.; Writing—
review & editing, L.P. and P.M. All authors have read and agreed to the published version of
the manuscript.
Funding: This research received no external funding.
Acknowledgments: This work was supported by LAAS-CNRS micro and nanotechnologies platform,
a member of the Renatech french national network.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Zhou, X.; Lee, S.; Xu, Z.; Yoon, J. Recent Progress on the Development of Chemosensors for Gases. Chem. Rev. 2015, 115, 7944–8000.
[CrossRef] [PubMed]
2. Nazemi, H.; Joseph, A.; Park, J.; Emadi, A. Advanced Micro- and Nano-Gas Sensor Technology: A Review. Sensors 2019, 19, 1285.
[CrossRef] [PubMed]
3. Parret, F.; Menini, P.; Martinez, A.; Soulantica, K.; Maisonnat, A.; Chaudret, B. Improvement of micromachined SnO2 gas sensors
selectivity by optimised dynamic temperature operating mode. Sens. Actuators B Chem. 2006, 118, 276–282. [CrossRef]
4. Long, H.; Harley-Trochimczyk, A.; He, T.; Pham, T.; Tang, Z.; Shi, T.; Zettl, A.; Mickelson, W.; Carraro, C.; Maboudian, R. In Situ
Localized Growth of Porous Tin Oxide Films on Low Power Microheater Platform for Low Temperature CO Detection. ACS Sens.
2016, 1, 339–343. [CrossRef]
5. Long, H.; Turner, S.; Yan, A.; Xu, H.; Jang, M.; Carraro, C.; Maboudian, R.; Zettl, A. Plasma assisted formation of 3D highly
porous nanostructured metal oxide network on microheater platform for Low power gas sensing. Sens. Actuators B Chem. 2019,
301, 127067. [CrossRef]
6. Sendi, A.; Besnard, G.; Menini, P.; Talhi, C.; Blanc, F.; Franc, B.; Fau, P. Detection and Discrimination Methods of Sub-ppm
Nitrogen Dioxide (NO2) with a Copper Oxide Sensor Operated with a Pulsed Temperature Modulation. Sens. Transducers 2018,
222, 24–30.
7. Mokrushin, A.S.; Nagornov, I.A.; Simonenko, T.L.; Simonenko, N.P.; Gorobtsov, P.Y.; Khamova, T.V.; Kopitsa, G.P.; Evzrezov, A.N.;
Simonenko, E.P.; Sevastyanov, V.G.; et al. Chemoresistive gas-sensitive ZnO/Pt nanocomposites films applied by microplotter
printing with increased sensitivity to benzene and hydrogen. Mater. Sci. Eng. B 2021, 271, 115233. [CrossRef]
8. Chapelle, A.; Charlot, S.; Conédéra, V.; Jonca, J.; Kahn, M.; Fajerwerg, K.; Menini, P. Selective gas sensor based on screenprinted
ZnO nanorods. In Proceedings of the Sensors, Energy Harvesting, Wireless Network &Smart Objects Conference (SENSO),
Marseille, France, 14–16 October 2015.
9. Nagornov, I.; Mokrushin, A.; Simonenko, E.; Gorobtsov, P.; Sevastyanov, V.; Kuznetsov, N. Zinc oxide obtained by the solvothermal
method with high sensitivity and selectivity to nitrogen dioxide. Ceram. Int. 2020, 46, 7756–7766. [CrossRef]
10. Presmanes, L.; Gunasekaran, V.; Thimont, Y.; Sinnarasa, I.; Barnabe, A.; Tailhades, P.; Menini, P. Sub-ppm NO2 sensing in
temperature cycled mode with Ga doped ZnO thin films deposited by RF sputtering. Multidiscip. Digit. Publ. Inst. Proc. 2019,
14, 48. [CrossRef]
11. Presmanes, L.; Thimont, Y.; Chapelle, A.; Blanc, F.; Talhi, C.; Bonningue, C.; Barnabé, A.; Menini, P.; Tailhades, P. Highly Sensitive
Sputtered ZnO:Ga Thin Films Integrated by a Simple Stencil Mask Process on Microsensor Platforms for Sub-ppm Acetaldehyde
Detection. Sensors 2017, 17, 1055. [CrossRef] [PubMed]
12. Dufour, N.; Chapelle, A.; Mesnilgrente, F.; Conédéra, V.; Menini, P. Technological Improvements of A Metal Oxide Gas Multi-
Sensor Based On A Micro-Hotplate Structure And Inkjet Deposi-Tion For An Automotive Air Quality Sensor Application. In
Proceedings of the European Conference MME2014, Istanbul, Turquey, 31 August–3 September 2014.
Micromachines 2021, 12, 1355 12 of 12

13. Yoboue, P.; Konaté, A.; Asseu, O.; Tety, P.; Ménini, P. An Efficient Metal oxide Gas Sensor using Nanostructured SnO2 for
Industrial Applications. Int. J. Mater. Eng. Technol. 2014, 11, 41–56.
14. Oudrhiri-Hassani, F.; Presmanes, L.; Barnabé, A.; Tailhades, P. Microstructure, porosity and roughness of RF sputtered oxide thin
films: Characterization and modelization. Appl. Surf. Sci. 2008, 254, 5796–5802. [CrossRef]
15. Sandu, I.; Presmanes, L.; Alphonse, P.; Tailhades, P. Nanostructured cobalt manganese ferrite thin films for gas sensor application.
Thin Solid Films 2006, 495, 130–133. [CrossRef]
16. Thornton, J. High Rate Thick Film Growth. Annu. Rev. Mater. Res. 1977, 7, 239–260. [CrossRef]
17. Chapelle, A.; El Younsi, I.; Vitale, S.; Thimont, Y.; Nelis, T.; Presmanes, L.; Barnabé, A.; Tailhades, P. Improved semiconducting
CuO/CuFe2O4 nanostructured thin films for CO2 gas sensing. Sens. Actuators B Chem. 2014, 204, 407–413. [CrossRef]
18. Shang, C.; Thimont, Y.; Barnabe, A.; Presmanes, L.; Pasquet, I.; Tailhades, P. Detailed microstructure analysis of as-deposited and
etched porous ZnO films. Appl. Surf. Sci. 2015, 344, 242–248. [CrossRef]
19. Nečas, D.; Klapetek, P. Gwyddion: An open-source software for SPM data analysis. Open Phys. 2012, 10, 181–188. [CrossRef]
20. Nimbalkar, A.; Patil, N.; Ganbavle, V.; Mohite, S.; Madhale, K.; Patil, M. Sol-gel derived aluminium doped zinc oxide thin films: A
view of aluminium doping effect on physicochemical and NO2 sensing properties. J. Alloys Compd. 2019, 775, 466–473. [CrossRef]