jo ur nal home page: www.elsev ier .com/ locate /snb
t-decorated In2O3 nanoparticles and their ability as a highly sensitive<10 ppb) acetone sensor for biomedical applications
ohamed Karmaoui a,∗, Salvatore Gianluca Leonardi b, Mariangela Latino b,avid M. Tobaldi a, Nicola Donato b, Robert C. Pullar a,c, Maria P. Seabra a,
oão A. Labrincha a, Giovanni Neri b
Department of Materials and Ceramic Engineering/CICECO—Aveiro Institute of Materials, University of Aveiro, Campus Universitário de Santiago,810-193 Aveiro, PortugalDepartment of Engineering, University of Messina, C. Da Di Dio, 98158 Messina, ItalyDepartment of Materials, Imperial College London, London SW7 2AZ, UK
r t i c l e i n f o
rticle history:eceived 8 December 2015eceived in revised form 16 February 2016ccepted 19 February 2016vailable online 3 March 2016
eywords:cetone sensing
a b s t r a c t
This paper reports on the high sensitivity of sub-spherical In2O3-Pt nanoparticles (NPs) for detectingppb levels of acetone, a biomarker for diabetes. The In2O3-Pt NPs, in the form of monodisperse metal-oxide In2O3 NPs with diameters of 6–8 nm, decorated with 2 wt% Pt metal NPs (2–3 nm) on the surface,were synthesized by a novel non-aqueous sol–gel route. NPs samples were investigated by X-ray powderdiffraction (XRPD), using the advanced whole powder pattern modeling (WPPM) method, and high-resolution transmission electron microscopy (HR-TEM). The advantage of this preparative process is thatit preserves metallic platinum NPs formed during the synthesis. The highly sensitive acetone sensor based
anoparticleslatinum
ndium oxideiomedical application
on these NPs, showed a lower detection limit as low as 10 ppb or less, which is the lowest detection limitever reported for any chemoresistive acetone sensors. This exceptional performance is likely due to thekey role played by very small Pt metal NPs uniformly distributed in the In2O3-Pt nanostructure. Thedeveloped sensor would be suitable for use as a highly sensitive, practical breath acetone checker fordaily diet and diabetes management and diagnosis.
The synthesis of metal oxide nanoparticles (NPs) is one of theost exciting and extensively studied research field in recent years,
ue to the peculiar properties of these nanomaterials which makehem highly suitable for different applications in optical, electricalnd optoelectronic devices, catalysis, chemical sensors, and so on1–3]. In particular, metal oxide nanoparticles, including ZnO, V2O4,nO2, WO3, Fe2O3, In2O3, etc., have been demonstrated to operateery well as sensitive elements in resistive sensors for detectingany gaseous substances such NOx, CO, ozone, CH4, ethanol, etc
4,5].
Many factors can influence the performances of these nanoma-
erials, such as synthesis methods, size and shape of crystalline NPs,
∗ Corresponding author. Present address: School of Chemistry College of Engi-eering and Physical Sciences University of Birmingham, Edgbaston, Birmingham15 2TT, UK. Fax: +351 234 401 470.
crystallinity degree, porosity and especially the presence of metaldopants [6].
Among these, Pt-decorated In2O3 NPs, have recently attractedattention for their improved gas sensing properties for CO, H2 andC2H5OH, compared to undoped/pure In2O3 metal oxide NPs [7].Great efforts have been put in by Neri’s group to investigate the gassensing of Pt-doped In2O3 NPs, aimed to the development of inno-vative and low cost devices capable of real time and fast detectionof gaseous species [8–11].
There are several procedures to synthesize In2O3 NPs andcore–shell Pt@In2O3 NPs, including hydrothermal methods [5],flame spray pyrolysis [12], electrodeposition [13], non-hydrolyticalcoholysis [14] and microwave-assisted synthesis [15]. Mostmethods are carried out in the presence of toxic organic solventsand surfactants, the reaction parameters also involving high tem-perature and long reaction times. These synthesis procedures areoften limited, and it is difficult to accurately control the reaction
parameters.
For industrial applications, the development of feasible syn-thetic processes is of paramount importance, and it is really
mportant to develop a straightforward process for producing In2O3nd Pt@In2O3 nanoparticles for the above applications. To this end,
facile/scalable method to produce In2O3 NPs via a one-pot non-queous sol–gel route at low-temperatures has been developed.n our previous work, we have developed a novel and facile non-queous sol–gel route for producing pure In2O3 metal oxide NPs.n the present study, we further develop this advanced methodol-gy to prepare, easily and at relatively low cost, In2O3-Pt (2 wt%)etal/metal-oxide nanoparticles, in which the larger oxide NPs
re decorated with much smaller metallic NPs on their surface.etailed studies, including synthesis formation mechanism, struc-
ural and physico-chemical characterization, and application of then2O3 NPs as a resistive gas sensor for sevoflurane anesthetic, arevailable in our previous work [16]. In this current paper, the hybridn2O3-Pt (2 wt%) NPs were tested for the gas sensing detection ofcetone in human breath.
Human breath analysis is rapidly gaining ground as a means tooninvasively diagnose and monitor various aspects of metabolism
inked to several diseases [17]. This is possible because there is anlmost instantaneous equilibrium between the pulmonary bloodnd the air in the alveoli of the lung. Several breath componentsre known to be related to specific human diseases: acetone foriabetes [18] ammonia for diseases in amino acids metabolism [19],ydrogen for lactose intolerance [20], etc.
In spite of the potential, breath analysis has not found systematicpplications in routine biological monitoring for diagnostic pur-oses, since the analytical procedures described in literature couldardly be used for routine analysis. Indeed, laboratory analysis ofxhaled air is a complex, expensive and time consuming process,sing GC–MS analysis to detect specific compounds, and thus isot in wide spread use. Sensor-based electronic devices are gain-
ng the main role when specific disease markers have been definedor a new generation of breath test instruments that could, in time,ecome as ordinary and versatile in medical screening as bloodests are today.
Among these biomarkers, acetone is found in anomalous con-entration (threshold limit 1.7 ppm) in breath samples of peoplesith diabetes [21]. The detection of very low concentrations of
cetone detection using chemoresistive sensors (MOx sensors) haseen widely investigated [22–30]. Among the most sensitive soar reported, flame-made Cr-doped �-WO3 nanoparticles display aesponse to low concentration of acetone as low 200 ppb [27]. In2O3ollow microspheres prepared by a hydrothermal and subsequentnnealing process are reported to offer good response with lowetection of <500 ppb [28]. Recently, it has been demonstrated thatu/In2O3 nanorods can effectively detect acetone with a low detec-
ion limit as low as 100 ppb [30]. Here, we report a study focused onhe development of Pt-doped In2O3 nanoparticles as sensing mate-ials for detection of ultra-low concentration of acetone. Resultshowed that a highly sensitive acetone sensor, based on In2O3-Pt2 wt%), was developed, showing an extremely low detection limitf 10 ppb or less, which is the lowest detection limit ever reportedor acetone chemoresistive sensors, making this an unprecedentedesult in the literature.
. Experimental
.1. Chemicals and materials
Indium(III) acetylacetonate [In(OCCH3CHOCCH3)3] (≥99.99%)race metals basis, platinum(II) acetylacetonate [Pt(C5H7O2)2]≥99.99%) trace metals basis and butylamine (99.5%)CH3(CH2)3NH2] were used, all from Aldrich.
uators B 230 (2016) 697–705
2.2. Synthesis of In2O3 nanoparticles and Pt (2 wt%) doped In2O3
The synthesis was carried out in a glove box (O2 andH2O < 1 ppm). In a typical procedure, 1 mmol (0.5 g) of Indium(III)acetylacetonate [In(OCCH3CHOCCH3)3] were added to 15 mL of n-butylamine, the reaction mixture was transferred into a stainlesssteel autoclave, and carefully sealed. The autoclave was taken outof the glove-box and heated in a furnace at temperatures at 140 ◦Cfor 4 h. In the case of the Pt-doped nanoparticles, 2 wt% (the syn-thesis time was only 4 h) indium(III) acetylacetonate was replacedby anhydrous platinum acetylacetonate. The resulting milky sus-pensions were centrifuged, and the precipitates were thoroughlywashed with ethanol and dichloromethane, and dried in air at 60 ◦C.
2.3. Sample characterization
A semi-quantitative phase analysis (QPA) of the crystallinephases in the prepared samples was obtained by X-ray powderdiffraction (XRPD) using the Rietveld method, as implemented inthe GSAS software with its graphical interface EXPGUI [31,32]. Datawere collected on a PANalytical X’Pert Pro (NL) �/2� diffractometerequipped with 0.5◦ divergence slit, 0.5◦ anti-scattering slit, 0.04 radSoller slits, and a 15 mm copper mask in the incident beam pathwayand a fast RTMS detector (PIXcel 1D, PANalytical) on the diffractedarm.
The XRPD data were collected in the 14–115◦ 2� range usingCu K� radiation (45 kV and 40 mA) with a virtual step scan of0.02◦ 2� and virtual time per step of 500 s. The starting atomicparameters for In2O3 described in the SG Ia3 were taken from aprevious work of the present authors [16]; those for InO(OH), SGPnnm, from Christensen et al. [33]. The refined parameters were:scale-factors, zero-point, six coefficients of the Chebyshev back-ground polynomial, unit cell parameters, two Lorentzian (LX andLY) terms, and one angle independent Gaussian term (GW) as theprofile coefficients, and sample displacement effects. Microstruc-tural analysis was exploited via XRPD as well, using the same dataas those for the QPA analysis, and employing the whole powder pat-tern modeling (WPPM) [34,35] implemented in the PM2K software[36]. This novel technique, considered a state-of-the-art, describeseach observed peak profile as a convolution of instrumental andsample-related physical effects, refining the corresponding modelparameters directly on the observed data [37,38]. The instrumentalcontribution was obtained by modeling 14 hkl reflections from theNIST SRM 660b standard (LaB6), according to the Caglioti et al. rela-tionship [39]. Afterward, In2O3 (SG Ia3), and where present InO(OH)(SG Pnnm), were included in the WPPM modeling. The followingparameters were refined: background (modelled using a 4th-orderof the shifted Chebyshev polynomial function), peak intensities,sample displacement, and lattice parameters; crystalline domainswere assumed to be spherical, and distributed according to a log-normal size distribution.
Transmission electron microscopy (TEM) was performed using aJeol-2000 FXII microscope, with point-to-point and line-to-line res-olutions of 0.28 nm and 0.14 nm, respectively. High resolution TEM(HR-TEM) was performed using a JEOL 2200FS microscope with afield emission gun, operated at 200 kV. Samples for TEM/HR-TEMobservations were prepared by dispersing the NPs in ethanol andevaporating the suspension drops on carbon-coated copper grids.
2.4. Sensing tests
The sensing device consists of an alumina substrate (6 × 3 mm2)
with Pt interdigitated electrodes and a Pt heater located onthe backside, on which the sensing layer (1–10 �m thick) wasdeposited by screen printing. A multimeter data acquisition unitAgilent 34970A was used for this purpose, while a dual-channel
M. Karmaoui et al. / Sensors and Actuators B 230 (2016) 697–705 699
Table 1Rietveld agreement factors and crystalline phase composition of the prepared samples.
Sample No. of variables Agreement factors Phase composition (wt%)
R(F2) (%) Rwp (%) �2 In2O3 InO(OH)
In2O3a 19 1.79 3.35 3.58 100 –
In2O3:Pt 2 wt% 21 3.27 2.76 2.35 95.9(1) 4.1(3)
a From Ref. [16].
Table 2WPPM agreement factors, In2O3 unit cell parameters, average crystalline domain diameter, and mode of the size distribution.
Agreement factors
Sample Rwp (%) Rexp (%) �2 Unit cell parameters (nm), a = b = c Average crystalline domain diameter (nm) Mode of the size distribution (nm)
In O a 3.73 1.77 2.11 1.01168(1) 7.0(1) 4.8(1)7.5(2) 5.3(2)
ptattGUe
tacft3
r1mtudut
3
3
ocopeicaAffietsmpP
Fig. 1. Crystalline domain size distribution of In2O3, and In2O3 modified with 2 wt%Pt. In the inset is reported a magnification of the size distribution for domains with
2 3
In2O3:Pt 2 wt% 2.53 1.54 1.64 1.01142(3)
a From Ref. [16].
ower supplier instrument Agilent E3632A was employed to biashe built-in heater of the sensor to perform measurements at super-mbient temperatures. All the equipment mentioned above andhe stream supplying, controlling and regulating ones are linked tohe PC, by means of GPIB (IEEE 488.2) protocol, with a converterPIB/USB. The system is fully automated by means of Graphicalser Interface (GUI) software, custom developed in Agilent VEEnvironment.
The sensors were then introduced in a teflon test chamber forhe sensing tests. The experimental bench for the electrical char-cterization of the sensors allows carrying out measurements inontrolled atmosphere. Gases coming from certified bottles can beurther diluted in air at a given concentration by mass flow con-rollers. The concentration of target gas was varied from 45 ppb to
ppm.Electrical measurements were carried out in the temperature
ange from RT to 350 ◦C, under a synthetic dry air total stream of00 sccm, collecting the sensors resistance data in the four pointode. The sensor response is given by R0/R, where R0 is the resis-
ance baseline in synthetic dry air and R the resistance recordednder different acetone concentrations. The response time, �res, isefined as the time required for the sensor to reach 90% of the sat-ration signal and the recovery time, �rec, the time needed to bringhe signal back to 90% of the baseline signal.
. Results
.1. Synthesis, structural and morphological characterization
In our previous study, we successfully prepared indium metalxide (In2O3) NPs via a non-aqueous sol–gel route. The structuralharacteristics, gas sensing properties and mechanism formationf the obtained NP powders are described in detail in a previousaper [16]. Briefly, the mechanism formation of the In2O3 NPs gen-rally established for the aminolysis of the carbonyl group of thendium acetylacetonate precursor and amine involves as a first stepreating an indium-amine complex, leading to the formation ofn the indium elongate ligand and N-butylacetamide type-species.fterward, the nucleophilic attack of the indium enolate species
or another amine to break the InO complex molecule leads to theormation of a hydroxyl group, and N-butylpropan-2-imine elim-nation compound. Finally, indium acetylacetonate molecules arexposed to nucleophilic attack from the hydroxyl group (formed inhe previous step), giving a bridging oxo group under acetylacetone
pecies elimination which leads to In–O–In bridges (the oxide for-ation). These results encouraged us to investigate whether it is
ossible to use this novel and facile approach to synthesize hybridt-decorated In2O3-Pt (2 wt%) NPs, of a few nanometers in dimen-
size less than 2.8 nm.
sion. The sizes and size distributions were studied by X-ray powderdiffraction (XRPD) and whole powder pattern modeling (WPPM).Quantitative phase analysis (QPA) data are reported in Table 1; agraphical output of a Rietveld refinement is shown in Fig. S1 of theelectronic Supplementary information (ESI).
As can be seen from data reported in Table 1, the unmodifiedIn2O3 is only composed of pure In2O3 metal oxide NPs. Noble metalmodification favored the presence of some hydroxide groups inthe form of indium oxy-hydroxide (InO(OH)), in amounts of about4.1 wt%—most likely, Pt addition retarded the last step of In2O3 NPsformation, resulting in the presence of InO(OH) species. Also, noreflection assignable to In(OH)3 and to Pt◦ were detected from theXRPD diffraction pattern. This latter occurrence is probably due tothe small amount of Pt added (2 wt%), which is below the tech-nique detection limit, but also because the strongest reflections ofPt◦ coincide with those of pure In2O3 metal oxide NPs.
Microstructural data are reported in Table 2 and Fig. 1. A graph-ical output of the WPPM modeling is reported in Fig. S2 of theESI. The unit cell parameters show a slight contraction of the pureIn2O3 metal oxide NPs lattice with the addition of Pt◦, at the Pt◦
weight percentages used (2 wt%) used. However, the addition of◦
this amount of Pt had little effect on either In2O3 NPs particle size
distribution, or on its average crystalline domain diameter, whichhad an average of around 7 nm and a mode of around 5 nm.
700 M. Karmaoui et al. / Sensors and Actuators B 230 (2016) 697–705
Fig. 2. (A) and (B) HR-TEM image of an assembly of the In2O3-Pt (2 wt%) NPs as-synthesized. Inset in (B) shows EDX analysis.
Fig. 3. (A) HR-TEM image of an assembly of the as-synthesized In2O3-Pt (2 wt%). In image (B), the green square highlights an In2O3 NPs, and the yellow circles highlightmetallic Pt NP decorations, which together form the NPs. Image C) shows the area with the green square (In2O3 part of the nanostructure). The inset in (C) depicts the FFTa spaciny of 2.2t .)
tia
nalysis of the area within the black square, and the parallel green lines show the d-ellow circle (Pt metal part of the hybrid NPs), and the green lines show a d-spacingo color in this figure legend, the reader is referred to the web version of this article
From Table 2 and Fig. 1, we can actually observe that the nanos-
ructures of unmodified In2O3 and In2O3: Pt 2 wt% are virtuallydentical, within the experimental error, with NP diameters of 7.0nd 7.5 nm, respectively. They also have narrow size distributions,
g of 3.58 A for the (2 2 0) plane of In2O3. Image (D) shows the area within the lower4 A for the (1 1 1) plane of Pt. Scale bar = 5 nm. (For interpretation of the references
the mode being 4.8 and 5.3 nm, respectively. Furthermore, looking
at the inset of Fig. 1, there are no detected domains with diam-eters ≤ 1.5 nm. High-resolution transmission electron microscopy
M. Karmaoui et al. / Sensors and Actuators B 230 (2016) 697–705 701
F
so
TNs2ahpaNyswivny
cwfPpe2ftaP63(Pd
3
roTw2
Fig. 5. Dynamic response (R0/R) of both sensors to 1.56 ppm acetone at 150 ◦C.
acetone in the range of temperature 100–250 ◦C. As above dis-
ig. 4. Comparison of the tested sensors response as a function of the temperature.
tudies (HR-TEM) were used to evaluate the size and morphologyf the hybrid In2O3-Pt (2 wt%) metal oxide nanoparticles.
The morphology of the In2O3-Pt NPs, at lower magnificationEM, is shown in Fig. 2A and B. It can be seen that the individualPs loosely agglomerate into colloidal nanoclusters (CNCs), con-
isting of tens of individual NPs. These CNCs are approximately0–50 nm in diameter. The individual NPs in these CNCs display
regular/spherical morphology with uniform shape, and they areighly monodisperse and homogenous. EDX analysis of the sam-le (see inset in Fig. 2B clearly evidence the presence of Pt. Fig. 3And B shows representative HRTEM images of In2O3-Pt (2 wt%)Ps. Fig. 3B clearly shows the Pt metal decorations (marked withellow circles) on the In2O3 NP surface. The overall size of the as-ynthesized In2O3-Pt (2 wt%) NPs is shown to be between 6–8 nm,hich matches the values extracted from XRPD analysis, confirm-
ng that the In2O3 NPs are single crystals, and the results obtainedia the WPPM method are valid. Fig. 3C and D shows higher mag-ification images of the regions marked by the green square andellow circles, respectively, in Fig. 3B.
These HR-TEM images confirmed that the bulk of the materialonsisted of 6–8 nm diameter In2O3 NPs, but that many of theseere in turn decorated with smaller, 2–3 nm diameter Pt NPs,
orming hybrid NPs, with a lopsided (unbalanced) character. Thet metal NPs are of very small size (2–3 nm) with spherical mor-hology, and are uniformly connected with In2O3 NPs, and do notxist by themselves. The lattice fringe spacing for these Pt NPs is.24 A, which is consistent with the (1 1 1) spacing of Pt, the latticeringe indicating that Pt is present in the face-centred cubic struc-ure (fcc, space group (SG) Fm3m) (Fig. 3D). This clearly shows that,lthough at only 2 wt% they were not enough to be detected by XRD,t nanocrystals were a component of the hybrid NPs. In the larger–8 nm NPs, the measured spacing of the crystallographic planes is.58 A, which corresponds to the (2 2 0) lattice spacing of In2O3 NPsFig. 3C). This also indicates that both components of these In2O3-t NPs are single crystals. The inset of Fig. 3C shows the FFT patternefined by the spot of the In2O3 NPs.
.2. Sensing tests
The In2O3-Pt (2 wt%) nanoparticles were used to fabricate aesistive sensor for acetone. The sensor was first tested at differentperating temperatures in order to find the optimal temperature.
he response of the sensor toward 1.56 ppm of acetone, comparedith the pure In2O3-based one, in the range between 100 ◦C and
50 ◦C, is reported in Fig. 4.
Fig. 6. Response and recovery times to 1.56 ppm acetone for both sensors.
As can be noted, both pure In2O3 and In2O3-Pt NPs sensors showa typical Gaussian temperature behavior of response with a max-imum centered at 150 ◦C, in according to an our previous paper[22]. However, the presence of platinum is able to increase consid-erably the response toward acetone of In2O3-Pt NPs based sensorcompared to other one. The higher sensitivity conferred by plat-inum NPs is capable of ensuring high response to the sensor whileworking at low temperatures, allowing it also to work at 100 ◦C.
Fig. 5 shows the dynamic responses (R0/R) recorded at the opti-mal temperature (150 ◦C) after the exposure to 1.56 ppm acetoneand back to dry reference air. As can be observed, both sensorsshowed fast transient responses from 0 to 1.56 ppm acetone, lowerthan 25 s to reach 90% of final signal. However comparing the tran-sient to recovery the baseline signal after the acetone removal, avery different behavior can be noted for either sensor. Indeed, whilethe In2O3-Pt is able to recovery the baseline signal in relativelyshort time, the other one would take much longer time. The behav-ior observed suggests that platinum, in addition to greatly enhancethe sensitivity of the sensor, is also able to decrease the recoverytime of the device.
Fig. 6 shows the response and recovery times evaluated for bothIn2O3 and In2O3-Pt based sensors after the exposure to 1.56 ppm
cussed, the presence of platinum is capable to improve the dynamicbehavior reducing both response and recovery times comparedto pure In2O3 based sensor. However, while the response times
702 M. Karmaoui et al. / Sensors and Actuators B 230 (2016) 697–705
Ft
rfb
rtcut
aoctcd
cOatats1iic
O
O
C
dafittagrlp
ig. 7. Resistance vs time plot for In2O3-Pt sensor when exposed to 1.56 ppm ace-one at 200 ◦C.
emain low in the whole temperature range investigated (10–20 sor Pt) the recovery times significantly increase at temperatureselow 150 ◦C also for the In2O3-Pt sensor.
In order to preserve the high sensitivity observed, without sac-ificing the rapidity of response and recovery of the signal, theemperature of 200 ◦C was identified as the optimal operationalondition. Fig. 7 plots the variation of resistance after two consec-tive pulses of 1.56 ppm acetone in dry air for In2O3-Pt sensor athis temperature.
The resistance of the sensor decreases as it is exposed to drycetone, which can be explained considering the n-type behaviorf the semiconductor sensing layer. Indeed, acetone reacts with thehemisorbed oxygen ions on the metal oxide surface, resulting inhe liberation of the electrons back to the conduction band thenausing an increasing of carrier concentration and a consequentecreasing of the resistivity of the sensor.
Oxygen species from air ambient adsorb on the surface of semi-onducting metal oxide particles in the ionic forms as O2
−(ads),
−(ads) and O2−
(ads). However, oxygen chemisorption is an energyctivated processes and each species requires different tempera-ure to take place on the grains surface. The O2−
(ads), requiring highctivation energy to chemisorb, is normally presented at higheremperatures while the other ones are the probable chemisorbedpecies at lower temperatures. In particular, as know, in the range00–200 ◦C the O−
(ads) is the predominant species [11]. Therefore,n the range of temperature here investigated, the possible chem-cal reaction involved could be written as (1) and (2) for oxygenhemisorption from air and (3) for acetone sensing [30,40]:
2− + e− ↔ 2O− (1)
2 + 2e− ↔ 2O− (2)
H3COCH3 + 9O−(ads) → 3CO2 + 3H2O + 8e− (3)
For metal oxides decorated with noble metal additives, twoifferent models are generally utilized to explain the sensing mech-nism based on catalytic and electronic sensitization [41,42]. Therst one, also known as spillover effect, is a chemical sensitiza-
ion mechanism where the metal phase (Pt nanoparticles) is ableo catalyze the dissociation of oxygen molecules into ions whichre then transferred to the metal oxide surface as adsorbed oxy-
en species which results in an increase of the active sites for theeaction with the analyte. The second one, also known as Fermievel control, considers that the metallic nanoparticles, eventuallyartially oxidized due to the air, having different Fermi level than
Fig. 8. Calibration curves of the In2O3-Pt sensor. Inset shows log–log scale.
metal oxide, will tend to delocalize electrons from metal oxide lead-ing to the formation of a depletion region at the metal/metal oxideinterface then a new Fermi level is created. When the analyte inter-acts directly with the Pt nanoparticles changes its energy state andtherefore the Fermi level energy of Pt/In2O3 composite. In otherwords, the electrons are quickly transferred back from Pt to In2O3with a consequent variation of conductivity of the material [43,44].We hypothesize that both effects are present in the sensing mech-anism, therefore, the spillover effect increases the availability ofactive sites and for another hand, a more rapid electron transferfrom platinum catalyst to indium oxide is obtained thanks to theeffect of the modification the Fermi level. Both effects, in associ-ation with homogeneous and extremely thin nanoparticles thatresult in a high specific surface, lead to the excellent sensitivityobserved. It is worth noting that for Pt/In2O3 sample the responsetimes are slightly improved compared to the pure In2O3 samplewhile a drastic reduction of the recovery time was observed. Toexplain what has been observed it can be assumed that through thespillover effect platinum promotes a faster re-adsorption of oxygenwhen acetone is removed, then decreasing the time to recoverythe baseline. Nevertheless, should be highlighted that during thesensing process intermediate organic species can be formed, thenadsorbed on the surface of the sensitive material. In this case, plat-inum, known to be an excellent catalyst for the breaking of C Cbonds by decomposing these possible intermediate species, wouldfavor their more rapid desorption [44].
The repeatability of the response indicates that the sensor hasalso good reliability. In Fig. 8(a, b) are reported the calibrationcurves for this sensor in a linear and log–log graph. Very low acetoneconcentration can be easily monitored; by a linear interpolation ofthe data, a lower detection limit (at S/N = 3) for this device can beestimated at around 10 ppb.
To better evaluate the characteristics of our sensor, a compar-ison with previous sensors was made (Fig. 9). It is worth notingthat our Pt/In2O3 sensor appears to be the most sensitive deviceso far reported and it is able to detect the very low acetone con-centrations present even in healthy people with unprecedentedaccuracy. Furthermore, most of the previous sensors operate athigher temperatures of 300–400 ◦C. Such high temperatures aretechnologically unfavorable, since they result in greater power con-sumption and reduced device lifetime. The operating temperatureof our sensor is instead much lower, at 200 ◦C, contributing to dras-
tically reduce the limitations found in previous ones.
It should be remarked that, despite the excellent performance ofmetal oxide based sensors, such as high sensitivity and low detec-
M. Karmaoui et al. / Sensors and Actuators B 230 (2016) 697–705 703
Fig. 9. Response of the In2O3-Pt sensor compared with previous reported sensors.
Fo
tbmvt
gmltsowteatttr
hs
Fig. 11. Calibration curves of the In2O3-Pt sensor in dry air and in humid air. Inset
ig. 10. Response of the In2O3-Pt to different gases and vapors. Operative conditionsf the sensors are the same as reported in Fig. 8.
ion limits, their single use for monitoring of biomarkers in humanreath would not provide the accuracy sought. This because, theanifold organic species present in the breath, humidity and their
ariability difficult to predict, could easily interfere with the detec-ion of the species of interest [45].
Exhaled breath consists largely of inorganic gases such as nitro-en, oxygen, carbon dioxide, and water vapor. Along with theseajor gases, there are many endogenous gaseous volatile metabo-
ites present at trace levels, from ppm to ppt, with acetone provedo be among the most abundant [17]. Therefore, proposed sen-ors for breath analysis should have negligible interference fromther gaseous components present in the breath. Sensing testsith different interfering species at higher concentration respect
o acetone were then performed to evaluate any possible interfer-nce in the measurements. Fig. 10 shows the responses calculateds Rair/Rgas and Rgas/Rair for reducing and oxidizing gases respec-ively. From data reported it can be evinced that the response ofhe sensor at these interfering species, in the same operative condi-ions, is lower than the response obtained for acetone. Comparableesponses were obtained only for ethanol and hydrogen.
The effects of humidity had been thoroughly assessed before-and. The effect of water vapor on semiconducting metal oxideensors has been documented [46]. This is believed to be due to
shows dynamic response to 0.29 ppm acetone in humid air (75% RH). Operativeconditions of the sensors are the same as reported in Fig. 8.
displacement of chemisorbed oxygen species by water moleculesand hydroxyl species which form at the sensor surface, and thislead to a decrease of the response to gases. Fig. 11 shows the sensorresponse to acetone as a function of the concentration at 200 ◦C indry air and in humid air, with 75% background relative humidity.As expected, also our sensor was found less sensitive to acetonein humid air, with sensitivity decreasing as the relative humidity(RH) increased to human breath levels. However, a response withhigh signal/noise (S/N) ratio at very low acetone concentration wasregistered in these conditions, (see inset in Fig. 11).
The interferences of ethanol, hydrogen and humidity during ace-tone monitoring in the breath has been assessed by other authors,providing simple solutions for overcoming this trouble [30,47].Xing et al. showed a compensation method to monitor low acetoneand ethanol concentrations counteracting the effect of high level ofhumidity [30]. Toyooka et al. developed a portable breath acetoneanalyzer equipped with only two semiconductor-based gas sensorswith different sensitivity characteristics. This enables the acetoneconcentration to be easily calculated while taking into account thepresence of ethanol, hydrogen and humidity [47].
4. Conclusion
In this work, hybrid In2O3-Pt NP material was used to fabricatea highly sensitive acetone sensor. The developed sensor shows alower detection limit as low as 10 ppb or less, which is an unprece-dented result in the literature, much lower than any previouslyreported chemoresistive acetone sensors by a factor of 20 or more.
A novel non-hydrolytic sol–gel route, taking place undersolvothermal conditions (using n-butylamine as solvent) at lowtemperatures (140 ◦C) for few hours, was introduced to synthesizethe hybrid In2O3-Pt NP sensing material. The short-chain amineacts as a reactive solvent during the reaction, not only producingthe indium metal oxide nanoparticles, but also favoring the forma-tion of metallic platinum. Overall, hybrid dumb-bell-like In2O3-PtNPs, consisting of 6–8 nm In2O3 NPs decorated with much smaller2–3 nm Pt NPs, were easily produced by this method.
These characteristics are likely at the origin of the exceptional
sensitivity observed toward acetone. Then, it can be suggested thatthese hybrid In2O3-Pt NPs could have a promising practical applica-tion in chemoresistive devices for monitoring acetone at ppb levels
7 nd Act
ia
A
am
A
ndMfbai
A
t
R
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
04 M. Karmaoui et al. / Sensors a
n human breath, which is essential for examples in the diagnosisnd monitoring of diabetes.
uthor contributions
The manuscript was written through contributions of alluthors. All authors have given approval to the final version of theanuscript. These authors contributed equally.
cknowledgments
Mohamed Karmaoui thanks Fundac ão para a Ciência e a Tec-ologia (FCT) for Grant N◦. SFRH/BPD/74477/2010. This work waseveloped in the scope of the project CICECO-Aveiro Institute ofaterials (Ref. FCT UID/CTM/50011/2013), financed by national
unds through the FCT/MEC and when applicable co-financedy FEDER under the PT2020 Partnership Agreement. Authorscknowledge the PEstC/CTM/LA0011/2013 programme. R.C. Pullars supported by the FCT grant SFRH/BPD/97115/2013.
ppendix A. Supplementary data
Supplementary data associated with this article can be found, inhe online version, at http://dx.doi.org/10.1016/j.snb.2016.02.100.
eferences
[1] X. Sun, H. Hao, H. Ji, X. Li, S. Cai, C. Zheng, Nanocasting synthesis of In2O3 withappropriate mesostructured ordering and enhanced gas-sensing property,ACS Appl. Mater. Interfaces 6 (2013) 401–409.
[3] M. Epifani, R. Díaz, J. Arbiol, E. Comini, N. Sergent, T. Pagnier, et al.,Nanocrystalline metal oxides from the injection of metal oxide sols incoordinating solutions: synthesis, characterization, thermal stabilization,device processing, and gas-sensing properties, Adv. Funct. Mater. 16 (2006)1488–1498.
[4] M. Hjiri, L. El Mir, S. Leonardi, N. Donato, G. Neri, CO and NO2 selectivemonitoring by ZnO-based sensors, Nanomaterials 3 (2013) 357–369.
[5] H. Jiang, J. Hu, F. Gu, W. Shao, C. Li, Hydrothermal synthesis of novel In2O3
microspheres for gas sensors, Chem. Commun. (2009) 3618–3620.[6] C. Wang, D. Chen, X. Jiao, C. Chen, Lotus-root-like In2O3 nanostructures:
fabrication, characterization, and photoluminescence properties, J. Phys.Chem. C 111 (2007) 13398–13403.
[7] X. Zou, J. Wang, X. Liu, C. Wang, Y. Jiang, Y. Wang, et al., Rational design ofsub-parts per million specific gas sensors array based on metal nanoparticlesdecorated nanowire enhancement-mode transistors, Nano. Lett. 13 (2013)3287–3292.
[8] G. Neri, A. Bonavita, G. Micali, G. Rizzo, N. Pinna, M. Niederberger, et al., Effectof the chemical composition on the sensing properties of In2O3–SnO2
nanoparticles synthesized by a non-aqueous method, Sens. Actuators B:Chem. 130 (2008) 222–230.
[9] G. Neri, A. Bonavita, G. Micali, G. Rizzo, S. Galvagno, M. Niederberger, et al., Ahighly sensitive oxygen sensor operating at room temperature based onplatinum-doped In2O3 nanocrystals, Chem. Commun. (2005) 6032–6034.
10] G. Neri, A. Bonavita, G. Micali, G. Rizzo, E. Callone, G. Carturan, Resistive COgas sensors based on In2O3 and InSnOx nanopowders synthesized viastarch-aided sol–gel process for automotive applications, Sens. Actuators B:Chem. 132 (2008) 224–233.
11] G. Neri, A. Bonavita, G. Micali, G. Rizzo, N. Pinna, M. Niederberger, In2O3 andPt-In2O3 nanopowders for low temperature oxygen sensors, Sens. Actuators B127 (2007) 455–462.
12] S. Singkammo, A. Wisitsoraat, C. Sriprachuabwong, A. Tuantranont, S.Phanichphant, C. Liewhiran, Electrolytically exfoliated graphene-loadedflame-made Ni-doped SnO2 composite film for acetone sensing, ACS Appl.Mater. Interfaces 7 (2015) 3077–3092.
13] J. Gan, X. Lu, J. Wu, S. Xie, T. Zhai, M. Yu, et al., Oxygen vacancies promotingphotoelectrochemical performance of In2O3 nanocubes, Sci. Rep. 3 (2013).
14] A. Singhal, M. Kaur, K.A. Dubey, Y.K. Bhardwaj, D. Jain, C.G.S. Pillai, et al.,Polyvinyl alcohol-In2O3 nanocomposite films: synthesis, characterization andgas sensing properties, RSC Adv. 2 (2012) 7180–7189.
15] G. Bühler, D. Thölmann, C. Feldmann, One-pot synthesis of highly conductiveindium tin oxide nanocrystals, Adv. Mater. 19 (2007) 2224–2227.
16] M. Karmaoui, S.G. Leonardi, D.M. Tobaldi, N. Donato, R.C. Pullar, M.P. Seabra,et al., Novel nanosynthesis of In2O3 and its application as a resistive gassensor for sevoflurane anesthetic, J. Mater. Chem. B (2015).
uators B 230 (2016) 697–705
17] C. Di Natale, R. Paolesse, E. Martinelli, R. Capuano, Solid-state gas sensors forbreath analysis: a review, Anal. Chim. Acta 824 (2014) 1–17.
18] Z. Wang, C. Wang, Is breath acetone a biomarker of diabetes? A historicalreview on breath acetone measurements, J. Breath Res. 7 (2013) 037109.
19] M. Holecek, Branched-chain amino acids and ammonia metabolism in liverdisease: therapeutic implications, Nutrition 29 (2013) 1186–1191.
20] M. Lomer, G. Parkes, J. Sanderson, Review article: lactose intolerance inclinical practice-myths and realities, Aliment. Pharmacol. Ther. 27 (2008)93–103.
21] C. Deng, J. Zhang, X. Yu, W. Zhang, X. Zhang, Determination of acetone inhuman breath by gas chromatography–mass spectrometry and solid-phasemicroextraction with on-fiber derivatization, J. Chromatogr. B 810 (2004)269–275.
22] G. Neri, A. Bonavita, G. Micali, N. Donato, Design and development of a breathacetone MOS sensor for ketogenic diets control, Sens. J. IEEE 10 (2010)131–136.
23] S. Ryabtsev, A. Shaposhnick, A. Lukin, E. Domashevskaya, Application ofsemiconductor gas sensors for medical diagnostics, Sens. Actuators B 59(1999) 26–29.
25] B. Zhu, C. Xie, W. Wang, K. Huang, J. Hu, Improvement in gas sensitivity of ZnOthick film to volatile organic compounds (VOCs) by adding TiO 2, Mater. Lett.58 (2004) 624–629.
26] R. Khadayate, J. Sali, P. Patil, Acetone vapor sensing properties of screenprinted WO3 thick films, Talanta 72 (2007) 1077–1081.
27] L. Wang, A. Teleki, S. Pratsinis, P. Gouma, Ferroelectric WO3 nanoparticles foracetone selective detection, Chem. Mater. 20 (2008) 4794–4796.
28] H. Dong, Y. Liu, G. Li, X. Wang, D. Xu, Z. Chen, et al., Hierarchically rosette-likeIn2O3 microspheres for volatile organic compounds gas sensors, Sens.Actuators B 178 (2013) 302–309.
29] K.-W. Kao, M.-C. Hsu, Y.-H. Chang, S. Gwo, J.A. Yeh, A sub-ppm acetone gassensor for diabetes detection using 10 nm thick ultrathin InN FETs, Sensors 12(2012) 7157–7168.
30] R. Xing, L. Xu, J. Song, C. Zhou, Q. Li, D. Liu, et al., Preparation and gas sensingproperties of In2O3/Au nanorods for detection of volatile organic compoundsin exhaled breath, Sci. Rep. 5 (2015).
31] A.C. Larson, R.B.V. Dreele, General Structure Analysis System (GSAS), LosAlamos National Laboratory, 2004.
32] B.H. Toby, EXPGUI, a graphical user interface for GSAS, J. Appl. Crystallogr. 34(2001) 210–213.
33] A.N. Christensen, R. Grønbæk, S.E. Rasmussen, The crystal structure of InOOH,Acta Chem. Scand. 18 (1964) 1261–1266.
34] P. Scardi, M. Leoni, Whole powder pattern modelling, Acta Crystallogr. Sect. A58 (2002) 190–200.
35] P. Scardi, M. Leoni, Whole powder pattern modelling: theory andapplications, in: E. Mittemeijer, P. Scardi (Eds.), Diffraction Analysis of theMicrostructure of Materials, Springer, Berlin, Heidelberg, 2004, pp. 51–91.
36] M. Leoni, T. Confente, P. Scardi, PM2K: a flexible program implementingwhole powder pattern modelling, Z. Kristallogr. 23 (2006) 249–254.
37] P. Scardi, Y.H. Dong, M. Leoni, Line profile analysis in the Rietveld method andwhole-powder-pattern fitting, Mater. Sci. Forum 378–381 (2001) 132–141.
38] P. Scardi, M. Leoni, Line profile analysis: pattern modelling versus profilefitting, J. Appl. Crystallogr. 39 (2006) 24–31.
39] G. Caglioti, A. Paoletti, F.P. Ricci, On resolution and luminosity of a neutrondiffraction spectrometer for single crystal analysis, Nucl. Instrum. Methods 9(1960) 195–198.
40] L. Yadava, R. Verma, R.S. Singh, Detection and sensing mechanism of acetonewith modeling using Pd/TiO2/Si structure, Thin Solid Films 520 (2012)3039–3042.
41] M. Hübner, D. Koziej, J.-D. Grunwaldt, U. Weimar, N. Barsan, An Au clustersrelated spill-over sensitization mechanism in SnO2-based gas sensorsidentified by operando HERFD-XAS, work function changes, DC resistance andcatalytic conversion studies, Phys. Chem. Chem. Phys. 14 (2012)13249–13254.
42] G.-J. Sun, S.-W. Choi, S.-H. Jung, A. Katoch, S.S. Kim, V-groove SnO2 nanowiresensors: fabrication and Pt-nanoparticle decoration, Nanotechnology 24(2012) 025504.
43] J. Fu, C. Zhao, J. Zhang, Y. Peng, E. Xie, Enhanced gas sensing performance ofelectrospun Pt-functionalized NiO nanotubes with chemical and electronicsensitization, ACS Appl. Mater. Interfaces 5 (2013) 7410–7416.
44] A.A. Daryakenari, A. Apostoluk, J.J. Delaunay, Effect of Pt decoration on the gasresponse of ZnO nanoparticles, Phys. Status Solidi C 10 (2013) 1297–1300.
45] G. Konvalina, H. Haick, Effect of humidity on nanoparticle-basedchemiresistors: a comparison between synthetic and real-world samples, ACSAppl. Mater. Interfaces 4 (2011) 317–325.
46] Q. Qi, T. Zhang, X. Zheng, H. Fan, L. Liu, R. Wang, et al., Electrical response ofSm2O3-doped SnO2 to C2H2 and effect of humidity interference, Sens.Actuators B 134 (2008) 36–42.
47] T. Toyooka, S. Hiyama, Y. Yamada, A prototype portable breath acetoneanalyzer for monitoring fat loss, J. Breath Res. 7 (2013) 036005.
Mohamed Karmaoui obtained his Chemical Engineerdegree from the Sciences and Technology University ofOran-Oran-Algeria) in 1998. In 2001, he went to NationalInstitute of Applied Sciences of Lyon (France) where heobtained his Advanced Diploma and Master Degree 2002.In 2005, he joined the Martin Luther University, Halle-Wittenberg (Germany), as a Ph.D. student and 2006 hehas continued his Ph.D. at the Department of Chemistryand CICECO of the University of Aveiro (Portugal). Hewas visiting scholar at the departement of chemistry andCatalysis, in GAP lab—University of Oslo in 2009–2010. In2011, he started his post-doc at University of Aveiro from2011–2014. He joined University of Birmingham (UK) in
015, where he is currently a research associate fellow at the School of Chemistry. Hisesearch activity is focused on developing synthesis for nanomaterials and nanopar-icles using non-aqueous sol–gel routes for the synthesis of metal and metal oxideanoparticles.
alvatore Gianluca Leonardi was born in Messina, Italy. He received the M.S. degreen Materials Engineering and the Ph.D. degree in Engineering and Chemistry of
aterials from the University of Messina, Messina, Italy in 2011 and 2015 respec-ively. His research interests include the development of chemical sensors based onanostructured materials.
Mariangela Latino received the M.S in Physics and Ph.D.degrees in ‘Materials for Energy and Environment’ fromUniversity of Messina and University of Roma Tor Vergatarespectively. Her current research interests at Universityof Messina are focused on development processes andcharacterization techniques of solid-state gas sensors. Sheis junior member of Italian Electric & Electronics Measure-ment Association.
avid Maria Tobaldi obtained his M.Sc. in Geology and Ph.D. in Materials Sci-nce both from the University of Bologna, Italy. He was visiting scholar at theepartment of Chemical Engineering, University of Patras, Greece (2007), Slove-ian National Building and Civil Engineering Institute, Ljubljana (2008), Shanghai
nstitute of Ceramics, China (2009). He joined University of Aveiro in 2011,
here he is currently a post-doc fellow at the Department of Materials and
eramic Engineering/CICECO—Aveiro Institute of Materials. His research inter-sts cover structural and microstructural analyses via advanced X-ray methods,multi) functional application of titania-based nanomaterials—for photocatalysis,hotochromism, antibacterial and light-to-energy functionalisations.
uators B 230 (2016) 697–705 705
Nicola Donato was born in Messina, Italy, in 1971. Hereceived the M.S. in Electronic Engineering and Ph.D.Degrees from University of Messina and University ofPalermo respectively. His research interests are focusedon sensor characterization and modeling, developmentof measurement system for sensors, characterization ofelectronic device up to microwave range. Actually, he isassociate professor and head of the laboratory of Elec-tronics for Sensors and for System of Transduction atUniversity of Messina. He is the author of more than 100papers on international journals (Scopus) and member ofIEEE.
Robert C. Pullar is an Investigador (Researcher) in theUniversity of Aveiro, Portugal. He received his Ph.D. inMaterials Engineering from the University of Warwick,and was also a Research Associate there, and was then aResearch Fellow/Senior Research Fellow at London SouthBank University and Imperial College London. Dr Pullarhas published 94 papers, of which 58 are published in jour-nals in the top SCI quartile and 28 in journals the top 3 oftheir field on SCI, and 5 book chapters. He has a large num-ber of citations (>2000), an h index of 24 and a g index of42.
Maria Paula Seabra (Research Scientist): Author/co-author of 68 papers belongingto the SCI (h-index 15, ISI Web of Science) and 2 patents. Supervised (or supervises)3 post-docs, 1 Ph.D. student, 10 M.Sc. students and 4 research fellows. Participationin 6 R&D projects (one running FP7 project). Expertise on: (i) materials process-ing; (ii) development of inorganic pigments from industrial wastes, promoting, insome cases, the inertization of hazardous materials; (iii) deposition and study ofphotocatalytic layers and (iv) formulation and testing of aerial lime based mortars.
João A. Labrincha (Associate Professor): Participation in 25 R&D projects (8as leader/responsible) and in 15 contracts/projects financed/in cooperation withindustries. Author/co-author of 22 patent applications (three as International PTC),and over 300 publications (230 papers belonging to the SCI, h-factor of 30 on Sco-pus). Expertise on the development of innovative and sustainable solutions forrecycling hazardous materials. Knowledge on the development of functionalisedeco-materials (lime based and/or containing secondary raw materials).
Giovanni Neri was born in Reggio Calabria, Italy, in 1956. He received the M.S.degree in Chemistry from the University of Messina, in 1980. Since 2002 he is fullprofessor of Chemistry at the University of Messina. From 2004 to 2007, he wasHead of the Department of Industrial Chemistry and Materials Engineering, Uni-
versity of Messina. He was visiting professor at the University of Michigan (USA)and University of Alagappa (India). His research interests include the synthesis andcharacterization of nanostructured materials for chemical sensors, applied in a widerange of sectors from medical diagnostics to automotive, industrial processes andenvironmental control.
