Accepted Manuscript

Title: Highly selective surface acoustic wave e-nose

implemented by laser direct writing

Authors: Massimiliano Benetti, Domenico Cannatà, Enrico

Verona, Alexandra Palla Papavlu, Valentina Carmen Dinca,
Thomas Lippert, Maria Dinescu, Fabio Di Pietrantonio

PII: S0925-4005(18)32124-5
DOI: https://doi.org/10.1016/j.snb.2018.12.005
Reference: SNB 25766

To appear in: Sensors and Actuators B

Received date: 30 July 2018

Revised date: 22 November 2018
Accepted date: 1 December 2018

Please cite this article as: Benetti M, Cannatà D, Verona E, Palla Papavlu A, Dinca VC,
Lippert T, Dinescu M, Di Pietrantonio F, Highly selective surface acoustic wave e-nose
implemented by laser direct writing, Sensors and amp; Actuators: B. Chemical (2018),
https://doi.org/10.1016/j.snb.2018.12.005

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
 Highly selective surface acoustic wave e-nose implemented by laser direct writing

Massimiliano Benetti1, Domenico Cannatà1, Enrico Verona2, Alexandra Palla Papavlu3,

Valentina Carmen Dinca3, Thomas Lippert4,5, Maria Dinescu3, and Fabio Di Pietrantonio1

PT
1
Institute for Microelectronics and Microsystems, Italian National Research Council, Via del
Fosso del Cavaliere 100, 00133 Rome, Italy

RI
2
Institute for Photonics and Nanotechnologies, Italian National Research Council, Via del
Cineto Romano 42, 00156 Rome, Italy

SC
3
National Institute for Lasers, Plasma, and Radiation Physics, 409 Atomistilor, Zip RO 077125,
Magurele, Romania

U
4
Paul Scherrer Institut, Research with Neutrons and Muons Division, 5232 Villigen PSI,
Switzerland
5
N
ETH Zürich, Department of Chemistry and Applied Biosciences, Laboratory of Inorganic
Chemistry, Zürich, Switzerland
A
M

Highlights

 Surface acoustic wave e-nose fabrication for detection of toxic volatile compounds
D

 Surface acoustic wave resonators are selectively coated by using laser-induced forward
TE

transfer (LIFT).
EP

 The SAW e-nose showed good selectivity for the considered analytes as confirmed by

principal component analysis (PCA).

CC

Abstract
A

In this paper, we present an e-nose for the detection of volatile compounds based on an array of

six surface acoustic wave (SAW) resonators coated with five different polymers (i.e.

polyepichlorohydrin, polyisobutylene, polyethylenimine, (hydroxypropyl)methyl cellulose, and

1
poly(styrene-co-maleic acid) partial isobutyl/methyl mixed ester, plus an uncoated SAW device

used as reference. In particular, matrix assisted pulsed laser evaporation was used to deposit thin

polymer layers which were subsequently used as donor films in the laser induced forward transfer

process to selectively cover each SAW resonators of the array. The SAW e-nose was tested upon

PT
exposure to vapors of ethyl acetate, dimethyl methylphosphonate, dichloromethane,

dichloropentane, and water. The frequency responses showed, for each of the sensors, a different

RI
sensitivity to the selected chemical agents and a good agreement with the theoretical sensitivities

SC
derived from the linear-solvation-energy-relationship between vapors and polymers. Specifically,

the implemented SAW e-nose allowed the discrimination between all the considered vapors in the

U
tested concentrations ranges as highlighted by the results of principal component analysis. Finally,

N
the obtained results indicated that laser deposition of polymers onto SAW resonators is possible
A
without significant modifications of their functionality and with a good reliability and
M

reproducibility.

Keywords (6 max)
D

Surface acoustic wave e-nose, matrix assisted pulsed laser evaporation (MAPLE), laser induced
TE

forward transfer (LIFT), polymer coating

EP
CC
A

2
1. Introduction

For many years, the detection of chemical agents is a major research topic in order to mitigate the

adverse effects on environment and human health [1-3]. In fact, the anthropic activities produce

every year billions of tons of chemical pollutants that are released into air, water, and soil. The

PT
demand for flexible technologies that can effectively be used for developing chemical sensors

becomes even more urgent due to an increased demand for defense against terrorism. Specifically,

RI
the need for detection of volatile organic compounds (VOCs) and warfare agents has stimulated a

SC
wide research activity for the development of sensing devices aiming to obtain high sensitivity,

reversibility, selectivity, low cost, fast response time, and compactness [4].

U
In this context, electronic nose (e-nose) technologies meet all the above requirements combining

N
the data obtained by a sensor array, selected for the target application, with computational
A
intelligence for making an efficient chemical detection system. Generally, the individual sensors
M

in the array are inherently not specific due to the multiplicity of weak chemical interaction

processes involved [5]. Therefore, analytes are discriminated by multivariate statistical methods
D

for data processing and artificial intelligence methods for pattern recognition. A variety of e-noses
TE

has been developed based on different vapor sensing technologies utilizing electrical, optical,
EP

mechanical and gravimetric transductions. Most common vapor sensor types are: metal-oxide

semiconductor, quartz crystal microbalance, and surface acoustic wave (SAW) devices [6-12].
CC

Each sensor shows advantages and disadvantages according to the specific application and the

target VOCs and/or chemical warfare agents [13-15].

A

SAW devices are based on the propagation of acoustic waves generated and detected by interdigital

transducers fabricated on the surface of a piezoelectric substrate. Since the acoustic energy is

strongly confined at the surface, SAW devices are very sensitive to changes in mass, viscosity, or

3
conductivity on the sensor’s surface [16]. In particular, adsorption and desorption of analytes from

a thin sensitive layer (i.e. chemo-selective polymer) coated on the sensor, gives rise to changes in

the velocity and amplitude of the acoustic wave. SAW vapor sensors are fast, highly sensitive, and

reversible, and their fabrication methods are compatible with standard integrated circuit

PT
technology. The latter aspect makes them suitable for volume production at low cost.

Nowadays, however, e-noses based on SAW sensors are used only for some very specific

RI
applications like detection of explosives and warfare agents [17-19]. One of the main reasons for

SC
the lack of a commercial development of SAW e-noses could be related to the available deposition

techniques for the sensitive layers. In fact, SAW sensors require an uniform (i.e. continuous,

U
uniform in thickness and hole free) sensitive layer along the propagation path, to prevent high

N
attenuation and degradation of the Q-factor [20]. The common methods for coating SAW devices
A
are spin coating, spray coating, and inkjet printing. The main drawbacks of these techniques are
M

related with high material utilization like in the case of spin coating, or clogging of nozzles and

creating special inks like in the case of inkjet printing. In addition, the main disadvantage of the
D

spray coating is the limited control over the amount of fluid flow, and in the particular case of
TE

coating SAW devices, macro-molecule “bubbles” at the polymer - substrate interface have been
EP

reported [21]. Finally, these methods lack in offering high spatial resolution when placing the

coating into an array.

CC

In our previous work [22], we have described the use of laser printing techniques, i.e. matrix-

assisted pulsed laser evaporation (MAPLE) and laser-induced forward transfer (LIFT), to fabricate
A

SAW sensors. The main accomplishment of that work was the possibility to use three different

polymers, which could be printed by the laser on SAW resonators (no clogging of nozzles or

4
creating of special inks) which show sensitivities below 100 Hz/ppm for dimethyl

methylphosphonate (DMMP), dichloromethane (DCM) and ethyl acetate (EtOAc).

In this work, we present an e-nose for the detection of VOCs commonly used in the chemical

industry (i.e. ethyl acetate), and simulants of chemical warfare agents (i.e. DMMP, DCM,

PT
dichloropentane (DCP)). The system is based on an array composed of six SAW resonators coated

with five different polymers, plus an uncoated SAW device used as reference. In particular, LIFT

RI
has been applied for the coating of SAW devices with the following polymers:

SC
polyepichlorohydrin (PECH), polyisobutylene (PIB), polyethylenimine (PEI),

(hydroxypropyl)methyl cellulose (HPMC), and poly(styrene-co-maleic acid), partial

U
isobutyl/methyl mixed ester (PScMA-me). As known, polymer coatings provide the chemical

N
interface for sorption of the target vapors and, therefore, the selection of polymers is critical in the
A
design of e-noses [6, 23]. The selected polymers must detect different chemical interaction
M

affinities of the vapors in a manner that pattern recognition methods succeed in discrimination

between target vapors and interferents. In this work, polymers were selected according to the data
D

available in the literature and based on linear-solvation-energy-relationship (LSER) between

TE

vapors and polymers [24]. In particular, PECH, PEI and PIB polymers have been already used for
EP

detection of simulants of warfare agents and other hazardous chemicals [25]. PECH is an attractive

material due to the presence of reactive chlorine groups on the polymer backbone which result in
CC

nucleophilic substitution for side chain modifications [26]. This polymer offers an excellent

balance of properties having a high capability for absorption of DMMP [17]. PIB is a weakly
A

dipolar, weak or no hydrogen bonding polymer, and presents a high sensitivity to

organophosphorus nerve agents [17, 25, 27]. PEI is characterized by largely non-specific weak

interactions (H-bond acceptor) and has been used in a variety of applications ranging from

5
chemical sensing, due to its reactivity with alcohols and hydrazine based compounds, to the

biomedical field, due to its capability as transfection enhancer [28]. HPMC is a versatile polymer,

having numerous applications such as eye drops, contact lenses, lubricant, foam enhancer and

stabilizer, thickener, emulsion stabilizer, etc. HPMC alone has also been used in sensor

PT
applications, in particular for the fabrication of optical sensing membranes for humidity [29] and

the fabrication of paper sensors based on a chitosan – HPMC – silver nanowires [30]. Finally,

RI
PScMA-me is a hydrogen bond acidic polymer [31, 32] which is expected to interact strongly with

SC
hydrogen-bond basic organophosphorous compounds such as DMMP by hydrogen bonding.

A significant accomplishment of this work is the printing of polymers by a laser-based solvent free

U
(“green” approach) method with ~90 % material utilization. For this approach, we utilized a special

N
designed polymer i.e. a triazene polymer (TP), which acts as absorbing/protecting/sacrificial layer
A
for the laser transfer. This could ultimately allow printing on substrates which are thermal-
M

sensitive, e.g. polymers which would in turn result in flexible devices, i.e. sensor for e.g. intelligent

clothing. The laser fabricated sensors show high sensitivities for a variety of analytes around room
D

temperature with fast recovery and reaction times, which is a huge improvement to currently used
TE

chemiresistors based on inorganic semiconductors which are operated at several hundred degrees
EP

(e.g. 400 °C for SnO2 – achieving thus a lower power consumption) [33].

2. Material and methods

CC

2.1 Laser based techniques for the deposition of the active polymer layers

This work aims at the fabrication of a SAW e-nose by using two successive laser based methods:
A

MAPLE and LIFT. A scheme of the system adopted for the deposition of the polymers on SAW

sensors is shown in Fig. 1. First, MAPLE was applied to prepare multi-ribbon polymer strips.

6
Second, the multi-ribbon polymer strips were used as donor materials in LIFT, which allowed

polymer pixels to be placed with high resolution onto the SAW devices.

The polymers used in this work are PIB (Mw = 600 kDa), PEI (Mw = 10 kDa), PECH (Mw = 700

kDa), HPMC (Mw = 26 kDa), and PScMA-me (Mw = 180 kDa). All polymers and solvents used

PT
here are obtained from Sigma-Aldrich, and used as matrices without further purification.

For the preparation of the polymer multi-ribbon strips (by MAPLE), each of the five polymers is

RI
dispersed in an inert solvent (i.e. the matrix) (see Table I) and then flash frozen in liquid nitrogen

SC
resulting in solid targets which are irradiated with a laser beam. When the target is ablated by the

laser pulses, the solvent evaporates and is pumped out the deposition chamber, while the polymer

U
molecules are collected onto a substrate which is placed parallel and at 4 cm distance from the

N
target. As substrates we used fused silica plates previously spin coated with a triazene polymer
A
(TP) layer (60 nm thick). The specific TP used herein is synthesized following the procedure
M

published previously [34] and its role is discussed later in the section devoted to the LIFT

experiments.
D

In order to achieve a uniform evaporation of the polymers, the target is rotated with a motion feed-
TE

through driven by a motor. For the MAPLE experiments, the deposition system is equipped with
EP

a “Surelite II” pulsed Nd:YAG laser system (Continuum Company) [34], and the main deposition

parameters used in this work are reported in Table I. The multi-ribbon polymer depositions are
CC

carried out in vacuum, at a background pressure of 5×10−3 Pa.

The experimental LIFT setup is presented elsewhere [27, 35]. In this work, the multi-ribbon
A

polymer strips obtained by MAPLE represent the donor layers, while the SAW sensors are the

receiving substrates in the LIFT experiments. In order to coat the SAW sensors with one polymer

pixel, the laser beam from a XeCl 308 nm laser system was guided and imaged at the fused silica

7
substrate – TP layer interface, perpendicular on the donor-receiver system. The role of the TP layer

is to protect the material to transfer (in this case the chemoselective polymer) by absorbing the

laser radiation. The laser is operated at a frequency of 1 Hz. The laser beam impinges on the

polymer film to be transferred through the fused silica substrates. The donor and the receiver are

PT
kept in contact while the laser irradiates the donor from the backside. All LIFT experiments are

performed under ambient pressure at room temperature. The pulse energy at each fluence was

RI
measured by a pyroelectric energy meter and an average of 100 pulses was used. The laser fluence

SC
was varied from 350 mJ/cm2 to 500 mJ/cm2. The sample holder is mounted on a motorized x, y, z

translation stage. The whole system is controlled by a computer running LabVIEW® software

U
which allowed creating a matrix of points for each sample, where the pulse energy and the number

of pulses are varied. N

A
2.2 Characterization of the deposited polymer thin films
M

The morphology of the deposited films is investigated by optical and atomic force microscopy

(AFM). The optical microscopy images are acquired with an Axiovert 200 Microscope coupled to
D

a Carl Zeiss AxioCam MRm camera. The AFM measurements are carried out with a XE 100 AFM
TE

setup from Park systems in non-contact mode on different areas and dimensions, i.e. from 5×5 μm2
EP

to 40×40 μm2.

2.3 SAW e-nose

CC

The SAW e-nose is composed of six SAW devices arranged on the same α-quartz substrates (ST-

cut, x propagation) and used as frequency control element in the feedback path of oscillators. This
A

configuration provides a simple, effective and accurate method for monitoring the small variations

in SAW velocity typical of SAW chemical-based sensors [36]. SAW devices consist of 2-ports

resonators obtained by two interdigital transducers arranged between reflecting gratings and

8
operating at approximately 393 MHz. The fabrication process and the geometrical features of

SAW devices are reported in [22]. The SAW array (Fig. 2 (a)) is mounted on a custom package,

placed inside a stainless steel sealed chamber and connected to the electronic circuits by feed

through pin-connectors. In the system, five oscillators are used for SAW devices coated with the

PT
five selected polymers and the remaining one is used as reference uncoated device. Details on the

electronics oscillators have been previously reported [22]. The schematic drawing of the system is

RI
shown in Fig. 2 (b).

SC
2.4 Setup and measurement

The sensor responses, given by frequency shifts of the oscillators, were measured with a frequency

U
counter (HP 536B) and a multiplexer module (Agilent 34980A and 34941A). Data were acquired

N
using a custom LabVIEW® routine. The frequency response of each sensor was obtained
A
subtracting the frequency of oscillation of the coated resonator from that of the reference uncoated
M

device, allowing, thus, a first order compensation for the effects due to changes in measurement

conditions (i.e. temperature and humidity). In order to obtain different concentrations of analytes
D

in nitrogen (N2), the sensor array was exposed to a total flux of 500 sccm, controlled by two flow
TE

meters: the main for the gas carrier (N2) and the second for the analyte. The vapor concentrations
EP

were obtained fluxing N2 in the liquid analyte (i.e. DMMP, DCM, etc.) by using a bubbler. A flow

switch together with a mixing chamber allowed obtaining the desired gas mixture (see ref. [22]).
CC

Previous to the analyte measurements, the frequency response baselines were obtained exposing

the e-nose to a flux of pure nitrogen (t0). Then, the analyte concentration was added to the system
A

until saturation of the frequency responses was reached (ts). Specifically, the frequency response

of j-th sensor due to exposure of i-th analyte was given by:

𝑅𝑗𝑖 = [(𝑓𝑟𝑒𝑓 (𝑡𝑠 ) − 𝑓𝑟𝑒𝑓 (𝑡0 )) − (𝑓𝑗 (𝑡𝑠 ) − 𝑓𝑗 (𝑡0 )] (1)

9
where: j  PECH , PIB, PEI ; i  DMMP, DCM , H 2O, EtOAc

For each vapor, measurements at four different concentrations were performed to evaluate the

sensitivities 𝑆𝑗𝑖 (evaluated as the angular coefficient of the linear approximation of the frequency

responses), while four measurements in the same conditions and for each concentration were

PT
executed to evaluate the repeatability. The limit of detection (LOD) was calculated as

(3×Noise)/Sensitivity [37] considering a noise level of 10 Hz.

RI
The obtained data were used to perform pattern recognition studies based on principal component

SC
analysis (PCA). Data pre-processing includes the mean centering of the raw data 𝑅𝑖𝑗 by its mean

value determined by averaging over all measured analyte samples, and the normalization of mean-

U
centered data using as weights the inverse of the sample variance. The PCA were implemented in
N
Matlab® using the singular value decomposition (SVD) algorithm.
A
Finally, hierarchical cluster analysis calculating Euclidean distance on normalized variables have
M

been used to identify groups of relatively similar polymers coatings from which subsets can be
D

chosen for inclusion in an optimized SAW sensors array.

TE

2.5 Frequency domain characterization of SAW sensors

SAW resonators were characterized before and after polymers depositions using microprobes
EP

(Picoprobe 40A-GSG-200-T) by GGB Industries Inc. The frequency response ( S21 scatter
CC

parameter) was measured in both magnitude and phase formats using a Network Analyzer

(HP8753A). The frequency shift, the insertion losses, and the loaded quality factor Q of the SAW
A

resonators were measured and compared before and after the polymer depositions. The thickness

of the polymers coating was evaluated by using the following equation [38]:

f p  (k1  k2 ) f 02h /  (2)

10
where f 0 is the resonant frequency of the uncoated device, k1 and k 2 are material constants of

the piezoelectric substrate, h is the thickness of the polymer coating and  the density of the

polymer. For α-quartz (ST-cut, x propagation), k1  8.7  108 and k2  3.9  108 [39].

3. Results and discussion

PT
3.1 Characterization of the polymer coatings deposited and transferred via laser based methods

onto the SAW devices

RI
In this work, the properties of several polymers in the form of polymer layers, i.e., PIB, PEI, PECH,

SC
HPMC, and PScMA-me are investigated aiming at their integration into a SAW e-nose.

First, PIB, PEI, PECH, HPMC, and PScMA-me polymers were deposited by MAPLE as multi-

U
ribbon polymer layers onto simple quartz substrates. For the sensing applications envisioned here,
N
the polymer films need to have a low roughness, a high uniformity and maintain their chemical
A
structure. Preliminary work [34] on MAPLE of three polymers, i.e. PIB, PEI, and PECH has shown
M

that the surface morphology of the thin films is affected by different experimental parameters,
D

principally by the laser fluence. It has been noticed that 80 mJ/cm2 is the threshold laser fluence
TE

i.e. the minimum laser fluence at which polymer material is deposited onto the substrate, and

continuous, crack and hole free PIB, PEI, and PECH layers can be deposited at laser fluences in
EP

the range 200-400 mJ/cm2. In addition, the study reported in [34] has revealed that the PECH
CC

layers deposited by MAPLE are the roughest, whilst PEI layers are the smoothest (root-mean-

square (RMS) below 1 nm for 5×5 µm2 areas). Therefore, in order to obtain continuous and
A

uniform polymer films, with a low RMS, in this study, PIB and PEI layers were deposited at a

laser fluence of 200 mJ/cm2 and PECH was deposited at 400 mJ/cm2 laser fluence. Fourier

transform infrared spectroscopy has confirmed that the laser fluences applied for the polymer

depositions are appropriate for obtaining thin polymer layers without any structural damages [34].

11
Furthermore, we have deposited HPMC and PScMA-me thin films in similar conditions as for the

other polymers (see Table 1) and found that we can prepare reproducible thin films [40] with a

controlled morphology and an unaltered chemical structure (see Figs. SI1 and SI2 for the FTIR

spectra of HPMC and PScMA-me polymer thin films deposited at 400 mJ/cm2 and 200 mJ/cm2

PT
respectively). In order to assure reproducibility, 20 multi-ribbon polymer layer samples were

produced and characterized.

RI
Following this step, we have successfully deposited all five polymers, onto a quartz substrate

SC
coated with a TP thin film with the final purpose of fabricating donor substrates from which we

could directly print polymer pixels onto SAW devices.

U
Previous work on LIFT for deposition of chemoselective polymers (i.e. PIB, PEI, and PECH) has

N
shown that only for laser fluences in the range 250-500 mJ/cm2 the transferred polymer pixels have
A
sharp edges and are well defined [22, 27, 35]. Thus, we have undertaken additional studies in
M

which we investigated the laser fluences for which regular, functional pixels of HPMC and

PScMA-me could be transferred. We have seen (images not shown here) that the clean transfer of
D

HPMC and PScMA-me occurs only for laser fluences in the range 300-450 mJ/cm2. Below 350
TE

mJ/cm2 only partial pixels are transferred, and above 450 mJ/cm2 the polymer pixels are destroyed.
EP

Optical microscopy images of PIB, PEI, PECH, HPMC, and PScMA-me pixels transferred at

different laser fluences onto the acoustic resonating cavity of SAW devices are shown in Fig. 3. In
CC

addition, another important aspect of the polymer coatings transferred by LIFT is their

morphology, and as it can be seen from Fig. 4, the surfaces of all five transferred polymer pixels
A

are smooth, homogeneous and without droplets.

12
Thus, these investigations reveal that these polymer pixels, obtained by laser based methods, could

compete with more mature and well established techniques (airbrush, spray coating, spin coating,

etc.) and are a good start for being used in sensor applications.

3.2 Polymer coating assessment by frequency domain characterization of SAW resonators

PT
The sensitivity of SAW sensors is significantly dependent on the thickness and morphology of the

coating [38]. Moreover, the polymer coating will directly impact the Q-factor and the insertion

RI
loss of the resonators and, hence, the phase noise and short-term frequency stability of the

SC
oscillators, and, finally, the sensitivity and resolution of the SAW e-nose.

The shifts of resonance frequency were in the range between 150 kHz and 440 kHz and point out

U
the mass loading on the sensor surface due to the polymer deposition, while, as evaluated by the

N
mass sensitivity equation (eq. 1), the thickness of the coating were in the range between 5 nm and
A
22 nm. In Table II, the thickness of the polymer coated sensors used in the following measurements
M

are reported. The behavior of the S21 parameters were not significantly corrupted as a result of
D

depositions, and the increase of insertion loss at the resonance frequencies was lower than 7.8 dB.
TE

After the coatings, the insertion losses at the resonance were below 18 dB. The loaded Q-factor

for the uncoated devices was 7500, while the obtained loaded Q-factors after polymer depositions
EP

were greater than 5000. The obtained results provide suitable frequency stability for the electronic

oscillators and are in line with those obtained for optimized depositions of polymers [25, 41]. The
CC

S21 amplitude response of SAW resonators before and after the deposition by LIFT of PScMA-
A

me polymer is reported in Fig. SI3.

3.3 E-nose measurements

13
The performances of the fabricated e-nose system were evaluated by exposing the SAW sensors

array to five analytes (DCM, DCP, DMMP, EtOAc, and H2O) in nitrogen atmosphere and

recording the differential frequency shifts as explained in paragraph 2.3.

The SAW sensors showed a fast, remarkable and reversible response as pointed out by the real-

PT
time responses of the sensors to DMMP vapors in the concentration range between 7 and 29 ppm

(Fig. SI4). The responses reached approximately 80% of the saturation value within 100 s. When

RI
the DMMP is removed, the recovery times to return to 80% of the initial baseline values were

SC
within 140 s. Similar results were obtained for the other vapors using all the polymer coated

sensors. The response curves of SAW sensors showed a linear behavior in the tested concentration

U
ranges as reported in Fig. 5, where each point is the average of four independent measurements,

N
while standard deviations are indicated as error bars. The relative standard deviations between
A
responses obtained in the same conditions were within 5%, demonstrating a good repeatability of
M

the system. The obtained sensitivities and the calculated LODs of the sensors for the different

analytes are reported in Table III, while a comparison of the sensitivities to the tested agents for
D

each polymer is shown in Fig. 5 f.

TE

As reported from several authors, the frequency shift response of SAW sensor is proportional to
EP

the partition coefficients K derived from the LSER analysis [42]. The calculated log K values

obtained from data available in the literature are reported in Table II. No data were found for
CC

HPMC and PScMA-me polymers. As expected, the responses to DMMP were quite larger respect

to those of other vapors. In particular, the higher sensitivity to DMMP was obtained for the PECH
A

coated sensor (log K = 4.5), followed by the PIB (log K = 3.56) and, finally, the PEI (log K = 2.85)

sensors [43]. The proportionality between the frequency responses of SAW coated sensors and the

corresponding partition coefficient K seems to be confirmed by the higher sensitivity obtained for

14
the PEI coated sensor to H2O vapors respect to those measured for the PIB and PECH coated

sensors. If the sensitivities of PECH, PIB and PEI coated sensors to DCM and EtOAc vapors are

considered, the obtained results show slight variations from those expected considering the

partition coefficients. Anyway, these small discrepancies are not surprising and can be attributed

PT
to the relatively greater influence of errors in concentration measurements when the responses to

analytes are low for a given polymer. Moreover, it should not be forgotten that the responses

RI
considered in this work were obtained subtracting the frequency response of the uncoated device,

SC
and, hence, the reported sensitivities are not strictly those evaluable by the calculated partition

coefficients. To further verify the accordance between the expected and measured responses,

U
Ex
hierarchical cluster analysis was performed on the experimental sensitivities S ji and the

theoretical data given by S Th

N
ji  4(k1  k2 ) f 0 hK
2
where j  PECH , PIB, PEI ;
A
i  DMMP, DCM , H 2O, EtOAc , and the factor “4” was added to take into account the changes in
M

the polymer modulus due to the swelling induced by the vapor [44]. The dendrograms allocate the
D

coated sensors according to their similarities in response patterns to the four considered vapors:
TE

the sensors connected at a lower point are more similar. As shown in Fig. 6 (a), the cluster analysis

for experimental and theoretical sensitivities identified the same groups for the selected polymers
EP

and confirmed the feasibility of polymer selection based on the LSER analysis for the preparation
CC

of SAW sensors coated by LIFT.

The PCA is reported in Fig. 7 (a) and (b) where the biplots of loadings and scores using the first
A

two principle components (PCs) are shown, respectively, for a system comprising three and five

sensors. In both cases, only minor deviations along the concentration profile were obtained as

expected from the high linearity of the sensor response curves. Considering the PCA results when

only three coated SAW sensors are used (i.e. PECH. PEI and PIB), the selected analytes appear in

15
well-formed and separate clusters with the exception of DMMP and DCP. Clusters are mainly

aligned along PC1 (78.4% variance) and the cumulative variance in the plane PC1-PC2 is

exceeding over 99.6%.

As known, the contributions of the sensors to the first two PCs are specified by the directions and

PT
lengths of the vectors reported in the biplot. In particular, the coefficients on PC2 for the loadings

of the PEI coated device are opposite to that of PECH and PIB coated sensors (Fig. 7 a). This

RI
suggests how the use of the only PEI coated sensor in combination with a PECH or a PIB coated

SC
device could be satisfactory to discriminate between the analytes in the tested concentration ranges

with the exception of DMMP and DCP, as previously mentioned.

U
Better results in terms of selectivity of the system were obtained when all the sensors were

N
considered in the PCA studies. As shown in Fig. 7 b, all analytes are in well separated clusters
A
indicating that our system is able to discriminate the considered vapors in the tested ranges of
M

concentrations. From the hierarchical cluster analysis reported in Fig. 7 b and performed on the

frequency responses of the five polymers coated sensors, an optimized array with the minimum
D

number of sensors should be include coatings chosen from the following three groups: (a) PECH
TE

or PIB; (b) PScMA-me or PEI; (3) HPMC. This finding was also confirmed by the components of
EP

loading vectors reported in Fig. 7 b.

4. Conclusions
CC

An e-nose based on array composed by six SAW resonators has been developed and tested.

MAPLE has been used to prepare a donor substrate as a multi-ribbon of polymers strips and LIFT
A

has been successfully applied to print polymers as sensitive films onto SAW resonators. The

deposition parameters for five polymers (i.e. PEI, PECH, PIB, HPMC, and PScMA-me) were

optimized to minimize scattering, diffraction and attenuation of the SAWs. The frequency

16
responses of SAW resonators after the polymers depositions showed a low increase of insertion

loss with a limited degradation of Q-factors, proving the feasibility of LIFT for the deposition of

the selected polymers. The e-nose was tested by exposing the sensors array to five analytes (DCM,

DCP, EtOAc, DMMP, and H2O) in nitrogen atmosphere. The obtained results showed a good

PT
agreement between the theoretical sensitivities, derived from the linear-solvation-energy-

relationship between vapors and polymers, and the experimental data. In particular, the responses

RI
to DMMP were quite larger, as expected, respect to those of other vapors. The calibration curves

SC
were linear for all tested vapors and sensors. The different sensitivities demonstrated by the PEI,

PECH, and PIB coated SAW sensors and the results of PCA indicated the capability of the system

U
to discriminate between the tested analytes with the exception of DMMP and DCP. Otherwise, the

N
use of SAW sensors arrays comprising five polymers allowed the discrimination between all the
A
considered vapors in the tested concentrations ranges. The cluster analysis showed that an
M

optimized array (i.e. comprising the minimum number of sensors) should include coatings chosen

from the following three groups: (a) PECH or PIB; (b) PScMA-me or PEI; (3) HPMC. Finally, the
D

obtained results indicated that laser deposition of polymers onto SAW resonators is possible
TE

without significant modifications of their functionality and with a good reliability and
EP

reproducibility.
CC

Acknowledgements

Financial support from the European Commission – 7th Framework Programme through the (FP7-
A

ICT project n° 247868) e-LIFT project, the Romanian National Program 4N/2016 is gratefully

acknowledged. A. Palla-Papavlu, V. Dinca, and M. Dinescu were supported by a grant of the

Romanian Ministry of Research and Innovation, CCCDI-UEFISCDI, project number PN-III-P1-

17
1.2-PCCDI-2017-0172 TESTES (P1), within PNCDI III. V. Dinca and M. Dinescu acknowledge

support from the project PN-III-P2-2.1-PED-2016-0880 (PED 214/2017).

PT
RI
SC
U
N
A
M
D
TE
EP
CC
A

18
References

[1] R.K. Ibrahim, M. Hayyan, M.A. AlSaadi, A. Hayyan, S. Ibrahim, Environmental application
of nanotechnology: air, soil, and water, Environmental Science and Pollution Research, 23(2016)
13754-88.
[2] D.S. Schmeller, A. Loyau, K. Bao, W. Brack, A. Chatzinotas, F. De Vleeschouwer, et al.,
People, pollution and pathogens – Global change impacts in mountain freshwater ecosystems,
Science of The Total Environment, 622-623(2018) 756-63.

PT
[3] S. Vucinic, B. Antonijevic, A.M. Tsatsakis, L. Vassilopoulou, A.O. Docea, A.E. Nosyrev, et
al., Environmental exposure to organophosphorus nerve agents, Environmental Toxicology and
Pharmacology, 56(2017) 163-71.

RI
[4] A. Antonacci, M.D. Lambreva, F. Arduini, D. Moscone, G. Palleschi, V. Scognamiglio, A
whole cell optical bioassay for the detection of chemical warfare mustard agent simulants,
Sensors and Actuators B: Chemical, 257(2018) 658-65.

SC
[5] B. Wang, E.V. Anslyn, Chemosensors: Principles, Strategies, and Applications2011.
[6] F. Röck, N. Barsan, U. Weimar, Electronic Nose:  Current Status and Future Trends,
Chemical Reviews, 108(2008) 705-25.
[7] K. Arshak, E. Moore, G.M. Lyons, J. Harris, S. Clifford, A review of gas sensors employed

U
in electronic nose applications, Sensor Review, 24(2004) 181-98.
[8] B.C. Muñoz, G. Steinthal, S. Sunshine, Conductive polymer-carbon black composites-based
N
sensor arrays for use in an electronic nose, Sensor Review, 19(1999) 300-5.
[9] S.K. Vashist, P. Vashist, Recent advances in quartz crystal microbalance-based sensors,
A
Journal of Sensors, 2011(2011).
[10] C. Viespe, Surface Acoustic Wave Sensors Based on Nanoporous Films for Hydrogen
M

Detection, Key Engineering Materials, 605(2014) 331-4.

[11] A. Marcu, I. Nicolae, C. Viespe, Active surface geometrical control of noise in nanowire-
SAW sensors, Sensors and Actuators B: Chemical, 231(2016) 469-73.
D

[12] M. Puiu, A.M. Gurban, L. Rotariu, S. Brajnicov, C. Viespe, C. Bala, Enhanced sensitive
love wave surface acoustic wave sensor designed for immunoassay formats, Sensors
TE

(Switzerland), 15(2015) 10511-25.

[13] A.D. Wilson, Review of Electronic-nose Technologies and Algorithms to Detect Hazardous
Chemicals in the Environment, Procedia Technology, 1(2012) 453-63.
EP

[14] A. Wilson, Recent progress in the design and clinical development of electronic-nose
technologies, Nanobiosensors in Disease Diagnosis, 5(2016) 15-27.
[15] S. Chen, Y. Wang, S. Choi, Applications and Technology of Electronic Nose for Clinical
CC

Diagnosis, Open Journal of Applied Biosensor, 2(2013) 39-50.

[16] J.W. Grate, S.J. Patrash, M.H. Abraham, C.M. Du, Selective vapor sorption by polymers
and cavitands on acoustic wave sensors: Is this molecular recognition?, Analytical Chemistry,
68(1996) 913-7.
A

[17] B.-S. Joo, J.-S. Huh, D.-D. Lee, Fabrication of polymer SAW sensor array to classify
chemical warfare agents, Sensors and Actuators B: Chemical, 121(2007) 47-53.
[18] T. Alizadeh, S. Zeynali, Electronic nose based on the polymer coated SAW sensors array
for the warfare agent simulants classification, Sensors and Actuators B: Chemical, 129(2008)
412-23.

19
[19] S. Liu, T. Ponrathnam, H. Sun, R. Nagarajan, J. Kumar, Z. Gu, et al., Detection of explosive
vapors by surface acoustic wave sensors containing novel siloxane based coatings, Journal of
Macromolecular Science, Part A: Pure and Applied Chemistry, 47(2010) 1172-5.
[20] F. Di Pietrantonio, D. Cannatà, M. Benetti, E. Verona, A. Varriale, M. Staiano, et al.,
Detection of odorant molecules via surface acoustic wave biosensor array based on odorant-
binding proteins, Biosensors and Bioelectronics, 41(2013) 328-34.
[21] M. Bruening, D. Dotzauer, Polymer films: Just spray it, Nat Mater, 8(2009) 449-50.
[22] F. Di Pietrantonio, M. Benetti, D. Cannatà, E. Verona, A. Palla-Papavlu, V. Dinca, et al.,

PT
Volatile toxic compound detection by surface acoustic wave sensor array coated with
chemoselective polymers deposited by laser induced forward transfer: Application to sarin,
Sensors and Actuators, B: Chemical, 174(2012) 158-67.

RI
[23] R.D.S. Yadava, Modeling, simulation, and information processing for development of a
polymeric electronic nose system, in: G. Korotcenkov (Ed.) Chemical Sensors – Simulation and
Modeling, Momentum Press, New York, 2012, pp. 411-502.

SC
[24] A. Hierlemann, E.T. Zellers, A.J. Ricco, Use of Linear Solvation Energy Relationships for
Modeling Responses from Polymer-Coated Acoustic-Wave Vapor Sensors, Analytical
Chemistry, 73(2001) 3458-66.
[25] D. Matatagui, J. Martí, M.J. Fernández, J.L. Fontecha, J. Gutiérrez, I. Grcia, et al., Chemical

U
warfare agents simulants detection with an optimized SAW sensor array, Sensors and Actuators,
B: Chemical, 154(2011) 199-205.
N
[26] F. Bekkar, M. Belbachir, Chemical Modification of Poly(epichlorohydrin) Using
Montmorillonite Clay, Chinese Journal of Chemistry, 27(2009) 1174-8.
A
[27] V. Dinca, A. Palla Papavlu, A. Matei, C. Luculescu, M. Dinescu, T. Lippert, et al., A
comparative study of DRL-lift and lift on integrated polyisobutylene polymer matrices, Applied
M

Physics A: Materials Science and Processing, 101(2010) 429-34.

[28] X. Ke, C. Yang, W. Cheng, Y.Y. Yang, Delivery of NF-κB shRNA using carbamate-
mannose modified PEI for eliminating cancer stem cells, Nanomedicine: Nanotechnology,
D

Biology and Medicine, 14(2018) 405-14.

[29] M.D. Fernández-Ramos, Y.F. Ordóñez, L.F. Capitán-Vallvey, I.M.P.d. Vargas-Sansalvador,
TE

J. Ballesta-Claver, Optical humidity sensor using methylene blue immobilized on a hydrophilic

polymer, Sensors and Actuators B: Chemical, 220(2015) 528-33.
[30] X. Duohua, Q. Jingshen, W. Yucheng, Y. Jiajun, L. Gui-Shi, Y. Bo-Ru, Chitosan-assisted
EP

buffer layer incorporated with hydroxypropyl methylcellulose-coated silver nanowires for paper-
based sensors, Applied Physics Express, 10(2017) 065002.
[31] A.V. Shevade, M.A. Ryan, M.L. Homer, H. Zhou, A.M. Manfreda, L.M. Lara, et al.,
CC

Chemical Sensor Array Response Modeling Using Quantitative Structure-Activity Relationships

Technique, in: M.A. Ryan, A.V. Shevade, C.J. Taylor, M.L. Homer, M. Blanco, J.R. Stetter
(Eds.), Computational Methods for Sensor Material Selection, Springer US, New York, NY,
2010, pp. 167-92.
A

[32] M.A. Ryan, A.V. Shevade, C.J. Taylor, M.L. Homer, M. Blanco, J.R. Stetter,
Computational Methods for Sensor Material Selection: Springer-Verlag New York; 2010.
[33] A. Palla Papavlu, T. Mattle, S. Temmel, U. Lehmann, A. Hintennach, A. Grisel, et al.,
Highly sensitive SnO2 sensor via reactive laser-induced transfer, Scientific Reports, 6(2016)
25144.

20
[34] A. Palla-Papavlu, V. Dinca, M. Dinescu, F. Di Pietrantonio, D. Cannata, M. Benetti, et al.,
Matrix-assisted pulsed laser evaporation of chemoselective polymers, Applied Physics a-
Materials Science & Processing, 105(2011) 651-9.
[35] V. Dinca, A. Palla-Papavlu, M. Dinescu, J.S. Stewart, T.K. Lippert, F. Di Pietrantonio, et
al., Polymer pixel enhancement by laser-induced forward transfer for sensor applications,
Applied Physics a-Materials Science & Processing, 101(2010) 559-65.
[36] R.D.S. Yadava, R. Kshetrimayum, M. Khaneja, Multifrequency characterization of
viscoelastic polymers and vapor sensing based on SAW oscillators, Ultrasonics, 49(2009) 638-

PT
45.
[37] A. Shrivastava, V. Gupta, Methods for the determination of limit of detection and limit of
quantitation of the analytical methods, Chronicles of Young Scientists, 2(2011) 21-5.

RI
[38] H. Wohltjen, Mechanism of operation and design considerations for surface acoustic wave
device vapour sensors, Sensors & Actuators, 5(1984) 307-25.
[39] J.W. Grate, M. Klusty, Surface acoustic-wave vapor sensors based on resonator devices,

SC
Analytical Chemistry, 63(1991) 1719-27.
[40] A. Palla-Papavlu, L. Rusen, V. Dinca, M. Filipescu, T. Lippert, M. Dinescu,
Characterization of ethylcellulose and hydroxypropyl methylcellulose thin films deposited by
matrix-assisted pulsed laser evaporation, Applied Surface Science, 302(2014) 87-91.

U
[41] R.A. McGill, R. Chung, D.B. Chrisey, P.C. Dorsey, P. Matthews, A. Pique, et al.,
Performance optimization of surface acoustic wave chemical sensors, Ultrasonics, Ferroelectrics
N
and Frequency Control, IEEE Transactions on, 45(1998) 1370-80.
[42] R.D.S. Yadava, R. Chaudhary, Solvation, transduction and independent component analysis
A
for pattern recognition in SAW electronic nose, Sensors and Actuators, B: Chemical, 113(2006)
1-21.
M

[43] D.S. Ballantine, S.L. Rose, J.W. Grate, H. Wohltjen, Correlation of surface acoustic wave
device coating responses with solubility properties and chemical structure using pattern
recognition, Analytical Chemistry, 58(1986).
D

[44] S.J. Patrash, E.T. Zellers, Characterization of polymeric surface acoustic wave sensor
coatings and semiempirical models of sensor responses to organic vapors, Analytical Chemistry,
TE

65(1993) 2055-66.
EP
CC
A

21
TABLE CAPTION

TAB. I Deposition parameters used for the obtainaing multi-ribbon polymer strips which were

subsequently used as donor materials in the LIFT experiments to coat the SAW sensors.

TAB. II Thickness of the polymer coating on the sensors and partition coefficients log K derived

PT
from the LSER analysis.

TAB. III Experimental sensitivities and calculated LODs of the sensors for the different analytes.

RI
SC
U
N
A
M
D
TE
EP
CC
A

22
FIGURE CAPTION

Fig.1 Scheme of the laser based methods applied for coating the SAW devices: MAPLE is used

to fabricate in one step a polymer multi-ribbon, which is afterwards used as a donor for

printing by LIFT of polymer pixels onto the SAW resonator.

PT
RI
SC
U
N
A
M
D
TE

Fig.2 (a) Diagram of the six SAW resonator array obtained on a quartz substrate. (b) Schematic
EP

drawing of the e-nose system composed of the SAW sensor array and the electronic

oscillators embedded in the measuring chamber.

CC
A

23
 A
CC
EP
TE
D
M
A
N
U
SC
RI
PT

24
Fig.3 Optical microscopy images of the SAW devices printed by LIFT with different polymer

pixels (HPMC, PScMA, PECH, PIB, and PEI). The arrows indicate the edge of the polymer

pixels. Scale bar is 400 µm. The laser fluences used for the transfers are: 500 mJ/cm2 for

HPMC, 350 mJ/cm2 for PScMA-me, 400 mJ/cm2 for PECH, 450 mJ/cm2 for PIB, and 400

PT
mJ/cm2 for PEI.

RI
SC
U
N
A
M
D
TE
EP
CC
A

25
Fig.4 AFM images on 5 × 5 µm2 areas of a) HPMC, b) PScMA, c) PECH, d) PIB, and e) PEI

polymers transferred by LIFT onto the active area of the SAW devices; f) statistical

parameters, i.e. root-mean-square (RMS), Sp (peak height), and Sv (valley height)

corresponding to each of the AFM images shown in a) – e).

PT
RI
SC
U
N
A
M
D
TE
EP
CC
A

Fig.5 Response curves for the SAW sensors upon exposure to different concentrations of: (a)

EtOAc; (b) DCM; (c) DMMP; (d) H2O; (e) DCP. Error bars indicate the repeatability. (f)

Comparison between sensitivities to the tested agents for each polymer coated sensor.

26
 PT
RI
SC
U
Fig.6 Dendrograms of hierarchical cluster analysis calculated by using: (a) experimental and

N
theoretical sensitivities for PEI, PECH and PEI coated sensors to EtOAc, DCM, DMMP,
A
and H2O; (b) experimental sensitivities for all sensors and vapors.
M
D
TE
EP
CC
A

27
Fig.7 Biplots of loadings and scores of PCA using the first two principle components: (a) for the

system comprising the PEI, PECH and PIB sensors; (b) for the system comprising all the

five coated sensors.

PT
RI
SC
U
N
A
M
D
TE
EP
CC
A

28
 TABLE I

Polymer PIB PECH PEI HPMC PScMA-me

Methanol-
Solvent Toluene Acetone Ethanol Dimethyl-sulfoxide
Isopropanol
Laser* fluence
200 400 200 400 200
(mJ/cm2)

PT
Repetition rate (Hz) 10 10 10 10 10
Number of pulses 20000 20000 25000 50000 25000

RI
Distance
4 4 4 4 4
target-substrate (cm)
Spot area (mm2)

SC
0.75 0.75 0.75 0.75 0.75
*
laser wavelength 266 (nm)

U
N
A
M
D
TE
EP
CC
A

29
 TABLE II

EtOAc DCM DMMP H2O DCP

Polymer f0 fp Coating thickness
log K log K log K log K log K
(MHz) (kHz) (nm)

PEI 393.545 440 17.75 0.35 1.73 2.85 5.06 -

PT
PECH 393.535 150 5.65 1.60 2.07 4.50 1.64 -

PIB 393.540 210 12.81 1.59 1.48 3.56 -0.19 -

RI
HPMC 393.725 215 22.00 - - - - -
PScMA-me 393.705 300 15.36 - - - - -

SC
U
N
A
M
D
TE
EP
CC
A

30
 TABLE III

EtOAc DCM DMMP H2O DCP

Polymer S LOD S LOD S LOD S LOD S LOD
(Hz/ppm) ppm (Hz/ppm) ppm (Hz/ppm) ppm (Hz/ppm) ppm (Hz/ppm) ppm
PEI 0.34 88 0.48 63 31.37 1.0 2.04 15 1.54 19

PT
PECH 0.38 79 0.34 88 86.71 0.3 0.57 53 5.54 5

PIB 0.27 111 0.30 100 69.15 0.4 0.54 56 3.88 8

RI
HPMC 0.11 273 0.10 300 33.17 0.9 0.20 150 16.57 2
PScMA- 0.32 94 0.67 45 25.44 1.2 1.78 17 5.46 5

SC
me

U
N
A
M
D
TE
EP
CC
A

31