i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e9

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

Selective hydrogen sensor for liquefied petroleum gas steam reforming fuel cell systems Matteo Tonezzer a,*, Thi Thanh Le Dang b,**, Quang Huy Tran c, Van Hieu Nguyen b, Salvatore Iannotta d a

IMEM-CNR, sede di Trento e FBK, Via alla Cascata 56/C, Povo, Trento, Italy ITIMS, Hanoi University of Science and Technology, Dai Co Viet 1, Hanoi, Viet Nam c National Institute of Hygiene and Epidemiology, Yersin Street 1, Hanoi, Viet Nam d IMEM-CNR, Parco Area delle Science 37/a, I-43100, Parma, Italy b

article info

abstract

Article history:

Nowadays miniaturized sensors have become more and more important in order to

Received 23 October 2015

monitor several fields of urban ambient and human life carefully. Low cost synthesis of

Received in revised form

nanostructured metal oxides for gas-sensing application is therefore of crucial importance

14 November 2016

for a mass production. Herein, NiO nanostructures were grown and used to realize sensors

Accepted 16 November 2016

selective to hydrogen gas in presence of liquefied petroleum gas (LPG), to be used on

Available online xxx

portable and isolated fuel cell systems. NiO p-type semiconducting nanowires with polycrystalline structure were prepared via an easy, cheap and scalable hydrothermal method.

Keywords:

Morphology and crystal structure of the NiO nanowires were characterized by scan elec-

Gas sensor

tron microscopy, X-ray diffraction, transmission electron microscopy and selected area

Hydrogen

electron diffraction. The nanostructured material was then tested as hydrogen sensor

LPG

showing very good performance in terms of sensor response (110%), stability, speed of

Nanowire

response and recovery (20 s) and selectivity in presence of LPG (a ratio of 200 times).

Nickel oxide

© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Introduction Nowadays hydrogen is the most promising fuel to produce energy in transportation and domestic applications. Fuel cells (FCs) operating with hydrogen are one of the most attractive clean energy technologies for stationary and mobile applications because of their high efficiency and zero emissions [1e3]. Small stationary fuel cell systems are designed for applications in remote backup power, especially in isolated offgrid settings. Unfortunately, on-site delivery of hydrogen still has high cost and reliability issues, and its storage is still very difficult as the compression or liquefaction of hydrogen

requires high pressure and/or low temperature. Due to these problems, practical applications of FCs require a fuel processing system to convert regular fuels into hydrogen of sufficient purity for FC consumption. As a result, the currently dominant procedure for the hydrogen production is the steam reforming of hydrocarbons [4e7]. Hydrogen production via fuel processing is one of the key issues in the progress of hydrogen-powered fuel cell devices, thus becoming a significant area of catalysis research [8e10]. In the last two decades several light hydrocarbons have been considered for on-board and on-site hydrogen production for fuel cells. Steam reforming of natural gas is currently the most economical process for the supply of hydrogen [11e13], but

* Corresponding author. Fax: þ39 0461 314875. ** Corresponding author. E-mail addresses: [email protected] (M. Tonezzer), [email protected] (T.T.L. Dang). http://dx.doi.org/10.1016/j.ijhydene.2016.11.102 0360-3199/© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Tonezzer M, et al., Selective hydrogen sensor for liquefied petroleum gas steam reforming fuel cell systems, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.102

2

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e9

unfortunately transportation and storage of liquefied natural gas require large energy inputs due to its low liquefaction temperature. Furthermore, transmission and distribution infrastructures for natural gas are generally missing in remote or sparsely populated areas, especially in developing countries. Therefore, an alternative energy source is needed for onsite or on-board hydrogen production for fuel cells [14e16]. LPG is a propane-butane mixture that exists in liquid state at room temperature under moderate pressures (less than 1.5 MPa), readily available from petroleum refineries and conveniently stored and transported [17e19]. With its well established distribution network and safe storage methods, LPG is asserting itself as an attractive fuel for systems in remote areas where a natural gas pipelines are not accessible [20]. Some companies are already active on the market with FC systems working with LPG fuel [21e24], and even using crowdfunding to develop next generation portable LPG-FC systems [25]. This kind of on-board or on-site reforming systems clearly need a feedback on the process efficiency. A low cost, low maintenance miniaturized sensor which can be used to detect hydrogen in atmospheres very rich in LPG is highly useful in such systems: a good control on the steam reforming parameters allows the best combustion conditions to be achieved, fuel economy to be maximized, and exhaust emissions optimized [26e30]. In the last decade metal oxide nanostructures have been extensively studied in order to optimize their physical and chemical properties for many applications, including gas sensing [31,32]. Controlling the size and shape of nanostructured metal oxides is a valid strategy to optimize their performance due to their structure-dependent properties [33e36]. In this paper we are reporting on the growth of nickel oxalate hydrate (NiC2O4$2H2O) nanowires by the hydrothermal method and their thermal decomposition during annealing at 500  C, obtaining polycrystalline NiO nanowires [37,38]. Hydrothermal synthetic method has been chosen for its simplicity and low power consumption, while NiO has been used for its excellent chemical stability and pronounced electrical properties [39,40]. Morphology and structure of the as-grown and annealed materials are shown. The nanosensors are tested for their LPG and hydrogen gas sensing properties, and are proven to be able to detect hydrogen gas [41,42] selectively, even in presence of LPG.

Materials and methods Synthesis of nickel oxide nanowires Nickel oxide nanowires have been grown via a two-step procedure consisting in a hydrothermal method followed by a heating treatment at 500  C. The chemicals used in these experiments were of analytical reagent grade, directly used without further purification. In a typical procedure [43], 0.474 g of [NiCl2$6H2O] (SigmaeAldrich, Germany) were dissolved into a mixture of 32 mL ethylene glycol (EG, SigmaeAldrich) and 18 mL deionized water, in a beaker under constant magnetic stirring at room temperature. After that 0.1206 g of Na2C2O4 were added into the beaker and 30 min of continuous

magnetic stirring were carried out to ensure a good dispersion of Ni2þ ions in the solution. The transparent solution was moved into a teflon-lined stainless steel autoclave of 100 mL, which was sealed and heated for 24 h at 200  C. After the annealing, the autoclave was let cooling down to room temperature naturally. The obtained product was gathered by centrifugation, washed three times using deionized water and absolute ethanol, and then dried in air. A blueegreen powder was obtained as a result. By calcining at 500  C this blueegreen precursor, consisting in nickel oxalate hydrate [NiC2O4$2H2O] long nanowires, polycrystalline NiO nanowires were obtained. The as-grown nanowires and the calcinated polycrystalline nanowires were characterized and analysed by X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM) and selected area electron diffraction (SAED). The XRD analysis was performed with a Bruker A) D5005 X-ray diffractometer with CuKa1 radiation (l ¼ 1.5406  operated at 40 kV and 40 mA. SEM images were obtained using a JEOL7600 scanning electron microscope operated at an accelerating voltage of 20 kV. Transmission electron microscopy (TEM) and selected-area electron diffraction were carried out using a JEOL JEM-2100 transmission electron microscope operated at an accelerating voltage of 200 kV.

Sensing devices fabrication The sensing devices were realized by drop casting nickel oxide nanowires over two interdigitated metal electrodes on a silica substrate. 0.1 mg of NiO nanowires was dispersed in ethanol under ultrasonic vibration for 3 min. The solution was then dropped onto interdigitated comb Pt/Ti electrodes on thermally oxidized piece of silicon wafer. The interdigitated Pt/Ti contacts were deposited onto a thermally oxidized silicon substrate by sputtering and conventional optical lithography technique. The full device is 2 mm  6 mm, while the two metal electrodes consisted in 18 pairs of fingers, each 800 mm long and 20 mm wide. The gap between two adjacent fingers is 50 mm. After the deposition of nickel oxide nanowires, the sensor was heated up to 500  C at a rate of 1.0  C min1 in a furnace and then kept for 2 h in order to increase stability and adhesion between nickel oxide and metal electrodes.

Gas-sensing measurements The NiO nanowires sensing performance were measured under dynamic conditions: the dilution (dry air) and the tested gas were always flowing through the sensing chamber at a total flow rate of 500 sccm. The equipment was home-built and incorporated a test chamber, a sensor holder which could be heated up to 500  C and some mass flow controllers (connected to high purity calibrated bottles). The device resistance was measured using a Keithley 2400 multimeter connected to a data acquisition system (LabView, National Instruments). The samples, biased with a DC voltage of 1 V and operated in air, showed a good ohmic behaviour which a negligible metalesemiconductor resistance. The sensor percentage response S% along this paper is defined as S % ¼ (Rgas  Rair)/Rair$100, where Rgas and Rair are the resistance of the device with analyte (reducing) gas or without it,

Please cite this article in press as: Tonezzer M, et al., Selective hydrogen sensor for liquefied petroleum gas steam reforming fuel cell systems, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.102

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e9

respectively. Response and recovery times are defined as the time required to reach 90% of the maximum response value and to get down to 10% of it, respectively. The limit of detection (LoD) has been calculated using the slope of the sensor response as a function of the gas concentration, and three times the standard deviation of the signal noise.

Results and discussion Nanowires characterization The hydrothermal growth generates thin smooth nanowires with a diameter of about 60 nm. Fig. 1a shows a SEM image of the hydrothermally grown material: the nanowires are thin and straight with a constant diameter and smooth surfaces. Their aspect ratio (the ratio between length and diameter) depends on the process temperature, increasing with rising process temperature. Once the nanomaterial (nickel oxalate

3

hydrate nanowires) is calcinated at 500  C for 2 h, the nanomaterial morphology deeply changes, as shown in Fig. 1b. It is evident that the high temperature annealing makes the nanowire material conglomerate in series of nanoparticles which still remind the former nanostructures. The nanoparticles are roundish with a diameter of 20e50 nm. X-ray diffraction patterns, shown in Fig. 2, confirm the composition of the nanowires before and after the calcination. In Fig. 2a, relative to the nanowires before the heating treatment, all the numerous diffraction peaks can be assigned to reflections of monoclinic [NiC2O4$2H2O] (JCPDS 25-0581). Fig. 2b, relative to nanowires after the annealing process, shows instead a well crystallized cubic NiO phase (JCPDS 471049). The three intense diffraction peaks at 37.3, 43.4 and 62.9 can be indexed to its cubic unit cell, with a lattice parameter of 4.1667  A. The absence of any additional peaks in both patterns of Fig. 2 confirms that the nanostructures are pure crystals without impurities, and that the [NiC2O4$2H2O] precursor was entirely transformed to NiO during calcination.

Fig. 1 e SEM images of the nanowires (a) before and (b) after calcination at 500  C.

Fig. 2 e XRD patterns of nanowires (a) before and (b) after the calcination at 500  C. Please cite this article in press as: Tonezzer M, et al., Selective hydrogen sensor for liquefied petroleum gas steam reforming fuel cell systems, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.102

4

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e9

The crystalline structure of the nanowires before and after the calcination was further investigated by TEM analysis and selected area electron diffraction. Fig. 3a shows a TEM image of nickel oxalate hydrate nanowires with diameters ranging from 50 to 70 nm; several nanowires are found forming bundles, and the thickest bundle has a diameter of approximately 450 nm. The inset of Fig. 3a reports a selected area electron diffraction (SAED) pattern displaying two diffused rings of spots, confirming the crystalline structure of [NiC2O4$2H2O] nanowires. These rings are assigned to (332), (510) diffraction lines of monoclinic [NiC2O4$2H2O] phase, respectively. Fig. 3b reports a TEM image of the polycrystalline NiO nanowires, composed of small nanoparticles. It is clear that many nanowires aggregate in bundles. Selected area electron diffraction (SAED) pattern of the NiO nanowires is shown in the inset, displaying four diffused rings of spots. These rings are assigned to (111), (200), (220) and (222) diffraction lines of cubic NiO phase, respectively. These investigations expose the polycrystalline NiO with cubic structure of the nanowires after the heat treatment.

working temperatures (200, 250, 300, 350 and 400  C, from bottom to top) is shown in Fig. 4. At all temperatures the resistance of the NiO sample is constant in air and increases abruptly when hydrogen gas is introduced into the system. The sensor resistance decreases to its previous value very quickly once the H2 flow is interrupted and air is restored in the system. This trend completely matches the behaviour of

Gas sensing properties A linearly increasing voltage from 5 V to þ5 V was applied to the devices in order to check the electrical contacts between the nanostructured NiO and the Pt/Ti electrodes. The IeV plot (not shown) confirms a very good ohmic behaviour, which is important for the sensing properties since the sensor response of a conductometric device can be maximized when the metal-semiconductor interface has a negligible junction resistance. The resistance of the samples ranged from 27 kU to 3 MU in air while increasing the temperature from 200 to 400  C. The sensing characteristic of the NiO devices were investigated keeping a voltage of 5 V between the electrodes and flowing different gas concentrations into the apparatus while recording its current.

Hydrogen sensing properties The dynamic resistance of a sensor to different hydrogen concentrations (50, 100, 250, 500 and 1000 ppm) at different

Fig. 4 e Dynamic resistance of a NiO sensor subjected to injections and evacuations of hydrogen at varied concentrations (at 200e400  C from bottom to top).

Fig. 3 e TEM images of the (a) nickel oxalate hydrate nanowires and (b) NiO polycrystalline nanowires. The insets show the SAED patterns relative the two materials. Please cite this article in press as: Tonezzer M, et al., Selective hydrogen sensor for liquefied petroleum gas steam reforming fuel cell systems, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.102

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e9

p-type semiconductor sensors. It is indeed well-known that nickel oxide is typically a p-type semiconductor when grown in standard conditions. When the nanostructured NiO is exposed to air, oxygen molecules will be adsorbed on nanowires surface in the form of O and O2. This superficial layer of adsorbed oxygen drains electrons from the nanostructure, increasing the number of electrical holes, thus enlarging its conductivity in these initial conditions. When hydrogen gas is flown onto the sensors, its molecules interact with the adsorbed oxygen layer releasing the electrons used in the chemical bonds back to the nanowires. This leads to a decrease of holes density and a rise in the sensor resistance. The plots in Fig. 4 show that the response magnitude improves with increasing hydrogen concentration at all working temperatures. At all temperatures the sensor response is evident and sharp, and the recovery is equally good with negligible drifts. From the dynamic resistance plot in Fig. 4, the response values are calculated and reported as a function of H2 concentration in Fig. 5a. It is clear that the sensor response improves at any temperature increasing hydrogen concentration, with a slope (sensitivity) that slowly decreases at higher concentrations (250e1000 ppm), pointing out the beginning of saturation. All the curves show comparable

5

trends, but 200 and 250  C plots are evidence for the highest response values. Increasing the sensor working temperature makes its response decrease along the whole hydrogen concentration range. A plot of the response values as a function of the working temperature is illustrated in Fig. 5b, confirming that the highest response is always found at 200  C (50 ppm H2) or 250  C (all other hydrogen concentrations). The lowest sensor response values of NiO nanowires based sensors go from 30.4% to 81.6% at 400  C. The best values go from 56.8% (200  C) to 106.9% (at 250  C). These values are good when compared with literature [44] (see Table S1 in Supplementary Information). Limit of detection, the lowest gas concentration that a device can detect, has been calculated as 300 ppb at 200  C and 50e70 ppb at higher temperatures. The response and recovery speed of the sensor to different gas concentrations is presented in Fig. 6 as a function of working temperature. We use here the definition of response time as the time to reach 90% of the complete response, and recovery time as the time to recuperate 90% of it (getting down to 10% of response signal). Sensor response times are shown in Fig. 6a, while recovery times are illustrated in Fig. 6b. Both response and recovery time values strongly decrease while the sensor working temperature is increasing. There is no apparent dependency on the hydrogen gas concentration. The response time goes 440e725 s at 200  C to 20e75 s at 300e400  C. The decrease in Fig. 6b plot is even steeper (please notice that the Y-axis is logarithmic in Fig. 6b): the recovery times go from 185 to 900 s at 200  C down to 16e23 s at 400  C. Recovery times are lower than 100 s already at 250  C. Table S1 in Supplementary Information compares the speed of the sensors presented in this work with those present in literature, confirming their merit.

LPG sensing properties

Fig. 5 e Percentage response as a function of a) hydrogen concentration and b) working temperature.

Fig. 7 shows the dynamic resistance of a sensor to different liquefied petroleum gas concentrations (2500, 5000, 10,000 and 20,000 ppm) at different working temperatures (200, 250, 300, 350 and 400  C). The plot at the bottom of Fig. 7, collected at 200  C, is quite noisy and presents a weak drift. All the other plots show that

Fig. 6 e a) Response time and b) recovery time as a function of sensor working temperature. Please cite this article in press as: Tonezzer M, et al., Selective hydrogen sensor for liquefied petroleum gas steam reforming fuel cell systems, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.102

6

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e9

Fig. 7 e Dynamic resistance of a sensor at different LPG concentrations.

the resistance of the NiO nanosensor is stable in air and immediately increases when LPG gas is injected into the system. When the LPG flow is suspended and replaced with air, the sensor resistance quickly returns to its earlier value. LPG is

a mixture of propane and butane, which are both reductive gases: they act analogously to hydrogen, reacting with the superficial layer of oxygen particles and releasing electrons to the NiO nanowires. This leads to a decrease of holes that enlarges the sensor resistance. As it can be seen in Fig. 7, at all working temperatures the intensity of the sensor response improves with increasing LPG concentration. The sensor response is always sharp and clear, and the recovery is complete without any drifts at temperatures higher than 200  C. In order to get an easier quantitative view of the response trend, the values calculated from Fig. 7 are reported in Fig. 8a as a function of LPG concentration. The plots recorded at all temperatures show that the response increases quite linearly with increasing LPG concentration. All the curves show a similar trend, with 200 and 250  C plots showing the highest response values along the whole gas concentration range. In order to quantify this behaviour, a plot of the response values towards the working temperature is reported in Fig. 8b, showing that the best response is always obtained at 200  C (2500 and 10,000 ppm LPG) or 250  C (5000 and 20,000 ppm LPG). The lowest sensor response of NiO sensors to LPG goes from 9.8% to 13.0% at 400  C and 350  C, respectively. The best values go from 14.2% (200  C) to 16.7% (at 250  C). These values are clearly much smaller than those calculated for hydrogen response. The speed of the response and recovery of the sensors at different LPG concentrations and working temperatures, calculated from Fig. 7, are presented in Fig. 9. Fig. 9a shows response times, whereas Fig. 9b shows recovery times. It is apparent that both response and recovery times decrease while increasing the sensor working temperature. The quickest times are obtained when the sensor is operated at 350  C. There is no evident dependency on the liquefied petroleum gas concentration. The response time goes from 50 to 85 s at 200  C to 5e21 s at 350  C. The recovery time goes from 20 to 56 s at 200  C down to 4e6 s at 350  C.

Fig. 8 e Percentage response as a function of a) LPG concentration and b) sensor working temperature. Please cite this article in press as: Tonezzer M, et al., Selective hydrogen sensor for liquefied petroleum gas steam reforming fuel cell systems, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.102

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e9

7

Fig. 9 e a) Response time and b) recovery time to LPG as a function of sensor working temperature.

Comparison and selectivity A very important parameter for a hydrogen sensor that must work in probable or sure LPG presence, is its selectivity to H2 towards LPG. Fig. 10a reports the response values of the NiO nanogranular sensors to different hydrogen concentrations (100, 250, 500 and 1000 ppm, black bars) and LPG concentrations (2500, 5000, 10,000 and 20,000 ppm, grey bars, red online) while working at different working temperatures (200e400  C). It is apparent that hydrogen gives rise to a very high response while compared to LPG, even without considering the different gas concentrations. The strong difference is

always present, for different gas concentrations, and at all temperature values (important parameter for steam reforming). In order to compare the response values more precisely, the percentage response are plotted in Fig. 10b as a function of the gas concentration. Please notice that the x-axis is logarithmic, thus the gas concentrations are more different than they appear. Fig. 10b confirms the good selectivity of the NiO sensor: 1000 ppm of hydrogen give response values of 80e120% at different working temperatures, while 20 times higher LPG concentration gives a response of only 12e17%. In order to compare quantitatively the response of the sensor to different gas concentrations, we define the “response per

Fig. 10 e a) Bar plot showing the different percentage response to hydrogen (black) and LPG (grey, red online), b) graph showing the selectivity to hydrogen toward LPG (please notice that the X-axis is logarithmic). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Please cite this article in press as: Tonezzer M, et al., Selective hydrogen sensor for liquefied petroleum gas steam reforming fuel cell systems, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.102

8

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e9

ppm” as the percentage response divided by the gas concentration expressed in ppm. Using this quantity to compare the sensitivity of the NiO sensors to H2 and LPG, we find that the values span 0.08e1.14 for hydrogen, while the range covered by LPG is only 7e56  104. The ratio between the such sensitivities goes from 120 to 200 times, thus proving that the NiO polycrystalline nanowires can be used as selective hydrogen sensors in an ambient very rich of LPG.

Conclusions We have grown p-type NiO polycrystalline nanowires via a simple and cheap two-steps process consisting in hydrothermal growth followed by high temperature calcination. The thin polycrystalline nanowires, composed by NiO monocrystalline nanoparticles with an average diameter of 20e50 nm, demonstrate very good hydrogen sensing performance with response values that exceed 100%, quick response and recovery times (around 20 s at high working temperature) and a very low limit of detection (50e70 parts per billion at high working temperature). They also demonstrate a very good selectivity to hydrogen against liquefied petroleum gas (120e200 times), which makes them ideal candidate as hydrogen sensors in rich LPG environments like steam reforming systems.

Acknowledgements The authors acknowledge the Italian Ministry of Foreign Affairs and International Cooperation (MAECI) for funding the bilateral project HyMN and the Laboratory of Geology, Geoengineering, Geoenvironment and Climate Change at Vietnam National University, Hanoi for HRTEM characterizations.

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.ijhydene.2016.11.102.

references

[1] Tollefson J. Fuel of the future? Nature 2010;464:1262e4. [2] Trimm DL, Onsan ZI. Onboard fuel conversion for hydrogenfuel-cell-driven vehicles. Catal Rev Sci Eng 2001;43:31e84. [3] Takehira K, Shishido T, Wang P, Kosaka T, Takaki K. Autothermal reforming of CH4 over supported Ni catalysts prepared from Mg-Al hydrotalcite-like anionic clay. J Catal 2004;221:43e54. [4] Ahmed S, Krumpelt M. Int J Hydrogen Energy 2001;26:291. [5] Haryanto A, Fernando S, Murali N, Adhikari S. Energy Fuel 2005;19:2098. [6] Kohn MP, Castaldi MJ, Farrauto RJ. Auto-thermal and dry reforming of landfill gas over a Rh/gAl2O3 monolith catalyst. Appl Catal B Environ 2010;94:125e33.

[7] Karatzas X, Dawody J, Grant A, Svensson EE, Pettersson LJ. Zone-coated Rh-based monolithic catalyst for autothermal reforming of diesel. Appl Catal B Environ 2011;101:226e38. [8] Li Y, Wang X, Xie C, Song C. Influence of ceria and nickel addition to alumina-supported Rh catalyst for propane steam reforming at low temperatures. Appl Catal A 2009;357:213e22. [9] Strohm JJ, Zheng J, Song C. Low-temperature steam reforming of jet fuel in the absence and presence of sulfur over Rh and Rh-Ni catalysts for fuel cells. J Catal 2006;238:309e20. [10] Jeong H, Kang M. Hydrogen production from butane steam reforming over Ni/Ag loaded MgAl2O4 catalyst. Appl Catal B 2010;95:446e55. [11] Zhu T, van Grootel PW, Filot IAW, Sun SG, van Santen RA, Hensen EJM. Microkinetics of steam methane reforming on platinum and rhodium metal surfaces. J Catal 2013;297:227e35. [12] Nikolla E, Schwank J, Linic S. Comparative study of the kinetics of methane steam reforming on supported Ni and Sn/Ni alloy catalysts: the impact of the formation of Ni alloy on chemistry. J Catal 2009;263:220e7. [13] Seo JG, Youn MH, Park S, Jung JC, Kim P, Chung JS, et al. Hydrogen production by steam reforming of liquefied natural gas (LNG) over nickel catalysts supported on cationic surfactant-templated mesoporous aluminas. J Power Sources 2009;186:178e84. [14] Lu Y, Chen J, Liu Y, Xue Q, He M. Highly sulfur-tolerant Pt/ Ce0.8Gd0.2O1.9 catalyst for steam reforming of liquid hydrocarbons in fuel cell applications. J Catal 2008;254:39e48. [15] Rakib MA, Grace JR, Lim CJ, Elnashaie SSEH. Steam reforming of heptane in a fluidized bed membrane reactor. J Power Sources 2010;195:5749e60. [16] Matsumura Y, Ishibe H. Suppression of CO by-production in steam reforming of methanol by addition of zinc oxide to silica-supported copper catalyst. J Catal 2009;268:282e9. [17] Zeman H, Url M, Hofbauer H. Autothermal reforming of hydrocarbon fuels. Chem Eng Trans 2011;24:169e74. [18] Demirbas A. Fuel properties of hydrogen, liquefied petroleum gas (LPG), and compressed natural gas (CNG) for transportation. Energy Source 2002;24:601e10. [19] Wichert M, Men Y, O'Connell M, Tiemann D, Zapf R, Kolb G, et al. Self-sustained operation and durability testing of a 300 W-class micro-structured LPG fuel processor. Int J Hydrogen Energy 2011;36:3496e504. [20] Ahmed K, Gamman J, Foger K. Demonstration of LPG-fueled solid oxide fuel cell systems. Solid State Ionics 2002;152e153:485e92. [21] Voller energy VE100 v2: http://www.gizmag.com/go/2870/. [22] Eneramic (Fraunhofer Institute for Ceramic Technologies and Systems IKTS): http://153.96.53.17/wordpress/?page_id¼18. [23] eZelleron: http://www.ezelleron.eu/en/eZell2.html. [24] Tropical SA: http://tropicalsa.en.forbuyers.com/product/ 9829985. [25] kraftwerk on kickstarter: https://www.kickstarter.com/ projects/ezelleron/kraftwerk-highly-innovative-portablepower-plant/description. [26] Park N, Park M-J, Baek S-C, Ha K-S, Lee Y-J, Kwak G, et al. Modeling and optimization of the mixed reforming of methane: maximizing CO2 utilization for non-equilibrated reaction. Fuel 2014;115:357e65. [27] Saebea D, Authayanun S, Patcharavorachot Y, Paengjuntuek W, Arpornwichanop A. Use of different renewable fuels in a steam reformer integrated into a solid oxide fuel cell: theoretical analysis and performance comparison. Energy 2013;51:305e13.

Please cite this article in press as: Tonezzer M, et al., Selective hydrogen sensor for liquefied petroleum gas steam reforming fuel cell systems, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.102

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e9

[28] Damle AS. Hydrogen production by reforming of liquid hydrocarbons in a membrane reactor for portable power generationdmodel simulations. J Power Sources 2008;180:516e29. [29] Livio D, Donazzi A, Beretta A, Groppi G, Forzatti P. Experimental and Modeling Analysis of the thermal behavior of an autothermal C3H8 catalytic partial oxidation reformer. Ind Eng Chem Res 2012;51:7573e83. [30] Liu Z, Mao Z, Xu J, Hess-Mohr N, Schmidt VM. Operation conditions optimization of hydrogen production by propane autothermal reforming for PEMFC application. Chin J Chem Eng 2006;14:259e65. [31] Tonezzer M, Iannotta S. H2 sensing properties of twodimensional zinc oxide nanostructures. Talanta 2014;122:201e8. [32] Tonezzer M, Dang TTL, Tran QH, Iannotta S. Dual-selective hydrogen and ethanol sensor for steam reforming systems. Sens Actuators B 2016;236:1011e9. [33] San X, Wang G, Liang B, Ma J, Meng D, Shen Y. Flower-like NiO hierarchical microspheres self-assembled with nanosheets: surfactant-free solvothermal synthesis and their gas sensing properties. J Alloys Compd 2015;636:357e62. [34] Tonezzer M, Hieu NV. Size-dependent response of singlenanowire gas sensors. Sens Actuators B 2012;163:146e52. [35] Zhang J, Zeng D, Zhu Q, Wu J, Xu K, Liao T, et al. Effect of grain-boundaries in NiO nanosheet layers room temperature sensing mechanism under NO2. J Phys Chem C 2015;119:17930e9.

9

[36] Tonezzer M, Dang TTL, Bazzanella N, Nguyen VH, Iannotta S. Comparative gas-sensing performance of 1D and 2D ZnO nanostructures. Sens Actuators B 2015;220:1152e60. [37] Choi J-M, Byun J-H, Kim SS. Influence of grain size on gassensing properties of chemiresistive p-type NiO nanofibers. Sensors Actuators B 2016;227:149e56. [38] Miao B, Zeng W, Lin L, Xu S. Characterization and gassensing properties of NiO nanowires prepared through hydrothermal method. Phys E 2013;52:40e5. [39] Hassan ES, Saeed AA, Elttayef AK. Doping and thickness variation influence on the structural and sensing properties of NiO film prepared by RF-magnetron sputtering. J Mater Sci Mater Electron 2016;27:1270e7. [40] Lu Y, Ma YH, Ma SY, Jin WX, Yan SH, Xu XL, et al. Synthesis of cactus-like NiO nanostructure and their gas-sensing properties. Mater Lett 2016;164:48e51. [41] Sta I, Jlassi M, Kandyla M, Hajji M, Koralli P, Krout F, et al. Surface functionalization of sol-gel grown NiO thin films with palladium nanoparticles for hydrogen sensing. Int J Hydrogen Energy 2016;41:3291e8. [42] Jamal RK, Aadim KA, Al-Zaidi QG, Taaban IN. Hydrogen gas sensors based on electrostatically spray deposited nickel oxide thin film structures. Photonic Sensors 2015;5:235e40. [43] Dang TTL, Tonezzer M. Polycrystalline NiO nanowires: scalable growth and ethanol sensing. Procedia Eng 2015;120:427e34. [44] Dang TTL, Tonezzer M, Nguyen VH. Hydrothermal growth and hydrogen selective sensing of nickel oxide nanowires. J Nanomater 2015;2015:785856.

Please cite this article in press as: Tonezzer M, et al., Selective hydrogen sensor for liquefied petroleum gas steam reforming fuel cell systems, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.11.102