- Email: [email protected]
Linear hydrogen gas sensors based on bimetallic nanoclusters
Journal of Alloys and Compounds 689 (2016) 1e5
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: http://www.elsevier.com/locate/jalcom
Linear hydrogen gas sensors based on bimetallic nanoclusters Ahmad I. Ayesh Department of Mathematics, Statistics and Physics, Qatar University, Doha, Qatar
a r t i c l e i n f o
a b s t r a c t
Article history: Received 29 April 2016 Received in revised form 14 June 2016 Accepted 29 July 2016 Available online 30 July 2016
This work reports on the fabrication of hydrogen gas sensors based on bimetallic palladium-copper nanoclusters. The nanoclusters were generated by sputtering and inert-gas condensation inside an ultra-high vacuum (UHV) compatible system, and self-assembled on an insulating substrate with a pair of pre-formed interdigitated gold/nichrome electrodes. Nanocluster deposition was stopped once their coverage on the substrate reached the percolation threshold. Electrical properties of the fabricated sensors were investigated by means of electrical conductance measurements, and assigned to charge carrier transport within network of metallic islands that is dominated by tunnelling. The produced devices were utilized as conductometric gas sensors. Herein, a constant voltage was applied across the interdigitated electrodes, and the change in electrical current signal was measured which reflects gas concentration. All fabricated sensors showed increase in the conductance upon exposure to hydrogen which can be assigned to the increase in tunnelling current due to the decrease in the size of the gaps between the nanoclusters or the establishment of conducting paths through the network of percolating nanocluster film. The sensors were found to be sensitive at low concentrations of hydrogen at room temperature, and exhibit a linear relationship between hydrogen concentration and the sensitivity. Therefore, those sensors have the potential to be used for practical life applications. © 2016 Elsevier B.V. All rights reserved.
Keywords: PdCu Bimetallic nanoclusters Nanocluster devices Inertegas condensation Hydrogen sensor
1. Introduction Hydrogen is an important source of clean energy, and it has many advantages such as: it is pollution free and naturally produced by plants and animals [1,2]. However, the utilization of hydrogen fuel requires reliable hydrogen sensing devices [3]. Among the different types of hydrogen sensors, electrical conductivity hydrogen sensors that utilize nanoclusters as hydrogen sensitive elements were found efficient sensors [4,5]. Herein, the development of hydrogen gas sensors that exhibit a linear response signal with hydrogen concentration, an enhanced response time, and operates efficiently at room temperature is essential [6,7]. In this work we report on the fabrication of conductometric hydrogen sensors based on the change of electrical conductivity upon exposure to hydrogen gas. Each hydrogen gas sensor is formed from a percolating Pd-Cu alloy nanocluster film at the percolation threshold deposited on an insulating substrate with a pair of pre-formed interdigitated electrodes. Pd-Cu alloy nanoclusters can be formed by various techniques such as: electroless deposition [8e10], sol-gel polymerized [11],
E-mail address: [email protected]. http://dx.doi.org/10.1016/j.jallcom.2016.07.320 0925-8388/© 2016 Elsevier B.V. All rights reserved.
and ultrasonic-assisted membrane reduction [12]. However, nanoclusters used in the current work were generated by sputtering and inert-gas condensation from a Pd-Cu target inside an ultrahigh vacuum (UHV) compatible system [13e20]. This technique has many advantages over other nanocluster production techniques such as: nanoclusters produced by this technique are of high purity (since they are produced inside an ultra-high vacuum system), the controllability of nanocluster composition, the possibility of nanocluster size selection using a suitable mass filter, and they can be created and self-assembled directly on a desirable substrate [14,19,20]. The present gas sensors may be used in multitude of different applications including, but not limited to, safety sensors, and hydrogen storage devices [4,6,21e24]. For example, hydrogen sensors are used during the hydrogenating cooking oil route, and considered as a cost-effective means to control and quantify the process [25]. 2. Experimental Nanoclusters used for the present work are bimetallic nanoclusters of palladium and copper alloy that were generated by sputtering and inert-gas condensation from a Pd-Cu target inside an UHV system with a base pressure ~108 mbar [19]. The UHV
2
A.I. Ayesh / Journal of Alloys and Compounds 689 (2016) 1e5
system consists of three main chambers [20]: i) source chamber, ii) mass selection chamber, and iii) deposition chamber (see Fig. 1 in Ref. [19]). The nanoclusters are generated inside the source chamber, travel through the mass filter that enables nanocluster size selection, and then they are deposited on a substrate fixed on a temperature controlled sample holder inside the deposition chamber. The sputtering and inert-gas condensation process was established inside the source chamber using Argon (Ar) inert-gas. Herein, Ar was used to: i) generate plasma needed to sputter material from the composite target that consists of 33.3% Cu and 66.6% Pd, ii) form nanoclusters by inert-gas condensation of the sputtered material, and iii) enable nanoclusters to travel from the source through the mass filter to the deposition chamber due to pressure gradient between the two chambers. Nanoclusters used in this work were produced using an Ar gas flow rate (fAr) of 50 sccm and a sputtering discharge power (P) of 62 W. In addition, pure Pd nanoclusters were produced at fAr ¼ 90 sccm and P ¼ 85 W, and used to produce hydrogen gas sensors for comparison. The sputter head and source chamber were water cooled at room temperature. A quadrupole mass filter (QMF) was used to determine nanocluster size distribution [18]. The mass filter consists of four parallel metal rods where each pair of opposite rods is connected electrically together to potentials of (U þ Vcos (ut)) and -(U þ Vcos (ut)), where U is a dc voltage and Vcos (ut) is an ac voltage. For a mass distribution scan, the ratio U/V was fixed and the mass distribution was scanned by varying the frequency, u. Once the nanoclusters' beam leaves the QMF, it moves through a Faraday cup that detects their current signal, thus, the current signal reflects the number of nanoclusters produced inside the source. The size and composition of the produced nanoclusters were confirmed using a Philips CM10 transmission electron microscope (TEM) and energyedispersive Xeray (EDX) measurements, respectively. The percolating gas sensor device was fabricated by depositing the produced nanoclusters on silicon dioxide/silicon (SiO2/Si) substrates with preeformed interdigitated gold/nichrome (Au/ NiCr) electrodes (formed by standard shadow mask technique) that are 50 mm apart [15]. Nanoclusters were also deposited on SiO2/Si substrates with preeformed planer electrodes that are 10 mm apart for electrical conductivity measurements as a function of temperature. The deposition rate of nanoclusters was measured using a quartz crystal monitor (QCM) fixed on a motorized linear translator that enables driving the QCM in front of the substrate, measure the deposition rate, and then drive it back away from the nanocluster beam path. The position of the liner translator holding the QCM could be controlled without venting the system [20]. Fig. 1(a) shows a schematic diagram of the hydrogen gas sensor of the present work. The substrate was mounted on a substrate
holder inside the deposition chamber, and connected electrically to a Keithley 238 source measuring unit. The electrical conductance of the sample is monitored during nanocluster deposition while applying a 100 mV voltage across the electrodes. The electrical current signal normally fluctuates at small current values before the onset of conduction due to the deposition of charged nanoclusters on the substrate. Once the percolation threshold is approached, an onset of conduction is observed and the electrical current increases abruptly (see Fig. 1(b)), thus, nanocluster deposition is suddenly stopped using an automatic shutter. The response of the device to hydrogen gas was characterized inside a temperature controlled a custom-designed Teflon chamber under different concentrations of hydrogen in pure nitrogen or air, as shown in Fig. 1(c) (and Fig. 4 in Ref. [26]). The test chamber was located in a fume hood at atmospheric pressure and 25 C. The gas flow rate in the sensor test chamber was controlled using Bronkhorst mass flow meters. The response signal of the device was measured using a computer controlled Keithley 236 source measuring unit. A constant voltage of 0.1 V was applied to the sensor, and electrical current signal was monitored as a function of time and gas concentration. 3. Results and discussion Pd and Cu contents within the produced nanoclusters were found 77 ± 1% and 23 ± 1%, respectively, as measured using EDX. A representative EDX measurement is shown in Fig. 2(a). The lower Cu contents within nanoclusters compared to that in the target is a result of the lower nanocluster production yield of Cu compared to Pd [18]. The size distribution of the produced nanoclusters was measured using the QMF and TEM as depicted in Fig. 2(b) and (c), respectively. The average nanocluster size as measured using the QMF is ~8.1 ± 1.2 nm, where the error is taken as one standard deviation. The TEM image reveals similar average size to that measured using the QMF. The average size of the pure Pd nanoclusters as measured using the QMF is 6.6 ± 0.7 nm. Electrical conductivity measurement (measured at 1 V) as a function of temperature of Pd0.77Cu0.23 nanoclusters deposited on a substrate with planer electrodes at the percolation threshold as shown in Fig. 3. The figure reveals an exponential dependence of the conductivity as a function of temperature. Similar behaviour was observed previously and assigned to charge carrier transport within network of metallic islands that is dominated by tunnelling through small barriers at the junctions between each pair of islands [15,27]. The electrical conductance of such system can be described using [28]: eV
GðTÞ ¼ s0 ðTÞenkB T
(1)
Fig. 1. (a) Schematic diagram of the hydrogen gas sensor formed from a percolating nanocluster film. (b) Electrical current e time dependence of the sensor at V ¼ 100 mV during nanocluster deposition. (c) Schematic diagram of the test chamber.
A.I. Ayesh / Journal of Alloys and Compounds 689 (2016) 1e5
3
Fig. 2. (a) EDX measurements of the produced nanoclusters. (b) Pd0.77Cu0.23 nanocluster size distribution as measured using QMF. (c) TEM image of Pd0.77Cu0.23 nanoclusters produced at the same conditions as in (a).
Fig. 3. Electrical conductance as a function of temperature. The (bottom) inset is the natural logarithm of the conductance as a function of invers temperature. The solid line is a linear fit of the results according to Eq. (1). The (top) inset reveals a schematic diagram of the planer electrodes with nanoclusters. The solid line is an example of suggested conduction path.
Where V is the applied voltage across the nanocluster network, T is temperature, s0 ðTÞ is the temperature dependant conductance at V ¼ 0, e is the electron charge, kB is Boltzmann constant, and n is the number of barriers between the nanoclusters that are forming the conduction path. The inset of Fig. 3 shows the dependence of the conductance natural logarithm on invers temperature. The slop of a linear fit can be used to estimate the number of tunnelling barriers as well as the nanocluster size. Herein, the slope is 8.29 which gives n ¼ 1399, thus, nanocluster size is 7.15 nm for contact separation of 10 mm. The estimated nanocluster size is close to the values measured using both QMF and TEM image which support that charge carrier transport is dominated by tunnelling through barriers between each two adjacent Pd0.77Cu0.23 nanoclusters. The underestimation of nanocluster size can be assigned to the non-straight conduction path which increases the number of nanoclusters involved in charge conduction as presented by the (top) inset of Fig. 3. Sensor response is defined as [6]:
IH2 I0 S¼ 100% I0
(2)
Where IH2 is the electric current in the presence of hydrogen, and I0 is the reference electrical current in absence of hydrogen gas. The results of sensitivity measurements corresponding to exposure of Pd0.77Cu0.23 sensor for different percentages of hydrogen gas
Fig. 4. Representative response curves for the gas sensor based on Pd0.77Cu0.23 nanoclusters at hydrogen concentrations between 0.5 and 5% (a), and 1e10% (b). The results of the control Pd nanocluster sensor is shown in (a).
relative to air are depicted in Fig. 4(a) and (b). The sensitivity measurements demonstrate that the present sensors provide great degree of sensitivity, with a typical figure of merit S ~30%, for 5% hydrogen relative to air at 25 C. This phenomena is quite different from the conventional hydrogen sensors based on microscopic palladium structure, where the current decreases upon exposure to hydrogen. The invers response of the present sensor is a main base of the sensing behaviour compared to the conventional microscopic hydrogen sensor. It should be noted her that sensitive hydrogen sensors that operate at room temperature were also reported for other nanostructures decorated with Pd [29]. Fig. 4(a) shows also the response of the control sensor formed of pure Pd nanocluster. The response of the control sensor is similar to
4
A.I. Ayesh / Journal of Alloys and Compounds 689 (2016) 1e5
the previously published results [4]. This response is nonlinear unlike the response of the Pd0.77Cu0.23 sensor. The dependence of sensor response on hydrogen concentration for four Pd0.77Cu0.23 sensor devices as well as the control device is shown in Fig. 5. The figure depicts a linear relationship between response and hydrogen concentration of the four Pd0.77Cu0.23 sensors with identical slope of 5.8 ± 0.3, unlike the response of the Pd control sensor. Therefore, it can be concluded that the four Pd0.77Cu0.23 sensors are identical, and their production is a reproducible process. The reproducibility test of a Pd0.77Cu0.23 hydrogen sensor is depicted in Fig. 6. The figure reveals a reproducible response signal of the sensor upon exposure to a fixed concentration of hydrogen of 3%. Furthermore, the response signal returns back to the background once hydrogen is removed. The test was repeated after one month from sensor fabrication and similar results have been observed. Future investigations may include long-term and systematic measurements of the stability as well as the drift of the produced sensors. The response time is defined as the time needed for the response electrical current signal to increase to 90% of the maximum response. Fig. 7 illustrates the dependence of the response time on hydrogen concentration for two different Pd0.77Cu0.23 nanocluster based sensors. The figure depicts a constant response time over the different hydrogen concentrations. The average response time can be calculated for the two devices to 18.6 ± 2.9 s, which is a reasonably fast response time. The sensing behaviour of the present Pd0.77Cu0.23 sensors can be rationalized as follows. A Pd0.77Cu0.23 sensor composes of nanocluster film at the percolation threshold, hence, many tunneling junctions and gaps exist within the film of nanoclusters. Each tunneling barrier or gap functions as a switch [15]. Upon exposure to hydrogen gas, the size of each nanocluster increases which decreases the size of tunneling barriers or gaps [4,30]. Thus, the electrical conductivity of the percolating film of the hydrogen sensor increases. It is well established that exposing Pd to hydrogen causes the expansion of the face centred cubic (fcc) lattice by a maximum of 3.6% due to a phase change in the crystal structure: from a to b phase [31]. The phase expansion occurs of equal compression along each nanocluster axis, and preferably at the grain boundaries. As a result, the intergranular gaps of a Pd nanocluster film are reduced, thus, electrical conductance of the nanocluster film increases. The phase transition is manifested as a plateau in a plot of ambient hydrogen gas pressure versus hydrogen content of the Pd lattice.
Fig. 5. The dependence of the response on hydrogen concentration for: four different sensors based on Pd0.77Cu0.23 nanoclusters, and the control sensor based on Pd nanoclusters.
Fig. 6. Response reproducibility for the Pd0.77Cu0.23 nanocluster based gas sensor using a constant hydrogen concentration of 3%.
Fig. 7. Dependence of the response time on hydrogen concentration for two Pd0.77Cu0.23 nanocluster based gas sensors.
Therefore, it can be concluded that the increase in the electrical conductivity of the Pd nanocluster film may occur because of one or a combination of the following points: i) a decrease in the size of the gaps between the nanoclusters which lead to an increase in the tunnelling current; ii) the connection of previously isolated pathways and nanoclusters to a main conducting path through the percolating film; and/or iii) the establishment of main conducting path through the percolating nanocluster film [32]. Exposing a Pd nanocluster film to a pure ambient air induces the b to a phase transition and the contraction of each nanocluster, thus, opening the gaps again within the nanocluster film causing decrease in the electrical conductivity of the nanocluster film. Therefore, Pd sensors can respond for multiple times of hydrogen exposure and removal associated to opening and closing of the gaps between nanoclusters, i.e. this process is a reversible process. Consequently, hydrogen concentration is reflected on the electrical conductivity. Furthermore, the above process is a highly selective process because only hydrogen can be absorbed as discussed above and cause the lattice expansion of Pd. This implies that Pd is resistive to poisoning by humidity and other gaseous species such as: SO2, H2S, and CH4. Production of nanostructures from metal alloys allows to produce gas sensors with tuned sensitivity and selectivity [33,34]. of nanostructure Absorption of hydrogen by Pd/Ag and Pd/Ni alloys has been studied previously by different investigators [35e37]. The main improvement of alloying Pd with other metals was found to
A.I. Ayesh / Journal of Alloys and Compounds 689 (2016) 1e5
enhance the mechanical properties on repeated exposure. In addition, hydrogen was found to be desorbed on Pd-Cu alloy surface at similar temperatures to pure Pd [38], which explains the functionality of the present sensor for hydrogen at room temperature. Alloying Pd with small percentage of Cu has a minimal influence on the binding of hydrogen to Pd surface [39], therefore, it was used in the present work to control the sensitivity of hydrogen sensor because of the difference in the adsorption of hydrogen by the alloyed material. Herein, a main advantage of using Pd-Cu alloy nanoclusters for hydrogen sensor is the linear relationship between the sensitivity and hydrogen concentration, consequently, the easiness of the sensor calibration. Alloying Pd with Cu has the following effects on the chemisorption of H2 [38]: 1) Structural effect: computer simulations of Pd0.7Cu0.3 alloy revealed decrease in relative concentration of triplet sites, and increase in relative concentrations of doublet and singulet sites. The capability of doublet and singulet sites to trap and dissociate hydrogen molecules and the successive diffusion to the triplet Pd sites on Pd-Cu alloy can clarify the sensitivity of Pd0.77Cu0.23 nanocluster based sensor for hydrogen. 2) Dilution effect: the repulsive and attractive forces between the adsorbed H2 and nanocluster surface are different for Pd-Cu than pure Pd and Cu. Herein, Cu repels the d-orbital of Pd which introduces additional normal direction of d-bonds thus the sensitivity of the alloy nanoclusters is enhanced [40]. 3) Electron energy states effect: some electronic states that are occupied for pure Pd and Cu becomes partially free after alloying which enhance the adsorption activity of Pd-Cu nanoclusters [41]. Herein, the d-orbital is a main cause for the formation of chemical bond since it is a long range active which creates a small barrier for adsorption, thus, the reaction cross section (RCS ~ r) and the sensitivity to hydrogen increases [42]. The redistribution of electron density on the Fermi level for Pd (4d10) and Cu (3d104S1) leads to enhancement in the sensitivity according to the striking coefficient equation (CS ) [42]:
CS
Ea Ed 1 Ae kB T
(3)
where A is a constant, Ea is the adsorption energy, and Ed is the desorption energy. For Ed > Ea, activated adsorption (H2 on Cu) has high probability, and for Ed < Ea, a non-activated adsorption (H2 on Pd) has high probability. 4) Spillover effect: this is a result of the accumulation of gas molecules at the interface between nanoclusters and substrate which enhances the sensitivity of the sensors based on Pd-Cu (compared with pure Pd) nanoclusters [41].
4. Conclusion In conclusion, hydrogen gas sensors based on Pd-Cu alloy nanoparticles were fabricated in this work. The nanoclusters were fabricated by sputtering and inert gas condensation inside an ultrahigh vacuum compatible system. This nanocluster production method enabled the self-assembly of nanoclusters on SiO2/Si substrates with pre-formed electrical electrodes. The results demonstrated that charge carrier transport was dominated by tunnelling through small barriers. The produced sensors were sensitive at low hydrogen concentrations, functional at room temperature with high sensitivity, capable of detecting concentrations of hydrogen as low as 0.5% in air, and exhibit short response time. The main
5
advantage of the present sensors is the linear relationship between hydrogen concentration and the sensitivity, thus, they are easy to calibrate. Consequently, those sensors can be used for different applications including safety and control sensors. In addition, the change in the conductivity of the present sensor is 30% at a hydrogen concentration of 5%, which is much greater than the response of the conventional hydrogen sensors at the same gas concentration. References [1] M. Ismail, F.A.H. Yap, N.N. Sulaiman, M.H.I. Ishak, J. Alloys Compd. 678 (2016) 297. [2] NREL. Technical Report, 2007. [3] T. Tharsika, A.S.M.A. Haseeb, S.A. Akbar, M.F.M. Sabri, Y.H. Wong, J. Alloys Compd. 618 (2015) 455. [4] Lith Jv, A. Lassesson, S.A. Brown, M. Schulze, J.G. Partridge, A. Ayesh, Appl. Phys. Lett. 91 (2007) 181910. [5] M. Okada, J. Nakahigashi, A. Fujita, M. Yamauchi, A. Kamegawa, J. Alloys Compd. 580 (2013) S401. Supplement 1. [6] A.I. Ayesh, S.T. Mahmoud, S.J. Ahmad, Y. Haik, Mater. Lett. 128 (2014) 354. € [7] S. Oztürk, N. Kılınç, J. Alloys Compd. 674 (2016) 179. [8] S. Okada, T. Kamegawa, K. Mori, H. Yamashita, Catal. Today 185 (2012) 109. [9] T. Wang, C. Shannon, Anal. Chim. Acta 708 (2011) 37. [10] D.-Y. Lee, S.-H. Park, D.-J. Qian, Y.-S. Kwon, Curr. Appl. Phys. 9 (2009) e232. [11] M. Ugalde, E. Chavira, M.T. Ochoa-Lara, C. Quintanar, J. Nano Res. 14 (2011) 95. [12] L. Liu, T. Wei, X. Zi, H. He, H. Dai, Catal. Today 153 (2010) 162. [13] H.M. Aldosari, A.I. Ayesh, J. Appl. Phys. 114 (2013) 054305. [14] A.I. Ayesh, Z. Karam, F. Awwad, M.A. Meetani, Sensors and actuators B, Chemical 221 (2015) 201. [15] A.I. Ayesh, Appl. Phys. Lett. 98 (2011) 133108. [16] A.I. Ayesh, H. Ahmed, F. Awwad, S. Abu-Eishah, S.T. Mahmoud, J. Mater. Res. 28 (2013) 2622. [17] A.I. Ayesh, Gas sensor applications using atomic nanoclusters, in: S.S. Naveen Kumar Navani, J.N. Govil (Eds.), Nanosensing 10, Studium Press LLC, Houston, TX 77072, USA, 2012. [18] A.I. Ayesh, N. Qamhieh, S.T. Mahmoud, H. Alawadhi, J. Mater. Res. (2012) 27. [19] A.I. Ayesh, S. Thaker, N. Qamhieh, H.J. Ghamlouche, Nanopart. Res. 13 (2011) 1125. [20] A.I. Ayesh, N. Qamhieh, H.S. Ghamlouche, M.E.-S. Thakera, J. Appl. Phys. 107 (2010) 034317. P. Nemec, R. Andok, A. Has [21] I. Rýger, G. Vanko, T. Lalinský, Ben cúrova cík S, ska, Sensors and actuators a, Physical 227 (2015) 55. M. Toma [22] B. Jang, S. Cho, C. Park, H. Lee, M.-J. Song, W. Lee, Sensors and actuators B, Chemical 221 (2015) 593. [23] B. Kang, Y. Heo, L. Tien, D. Norton, F. Ren, B. Gila, S. Pearton, Appl. Phys. A Mater. Sci. Process. 80 (2005) 1029. [24] J. Brouwer, Curr. Appl. Phys. 10 (2010). S9. [25] S. Prabhu, F. Schweighardt, Hydrogen sensing and detection. Hydrogen fuel: production, transport, and storage, CRC Press Boca Rat. (2009) 495. [26] A.I. Ayesh, A.F.S. Abu-Hani, S.T. Mahmoud, Y. Haik, Sensors and actuators B, Chemical 231 (2016) 593. [27] A.I. Ayesh, S.T. Mahmoud, N. Qamhieh, Z.A. Karam, Acta Metall. Sin. 27 (2014) 156. [28] B. Ozturk, C. Blackledge, B.N. Flanders, D. Grischkowsky, Appl. Phys. Lett (2006) 88. [29] B. Liu, D. Cai, Y. Liu, H. Li, C. Weng, G. Zeng, Q. Li, T. Wang, Nanoscale 5 (2013) 2505. [30] M.Z. Atashbar, S. Singamaneni, Sensors and actuators B, Chemical 111e112 (2005) 13. [31] K.J. Stevens, B. Ingham, M.F. Toney, S.A. Brown, A. Lassesson, Curr. Appl. Phys. 8 (2008) 443. [32] M. Yun, N.V. Myung, R.P. Vasquez, C. Lee, E. Menke, R.M. Penner, NANO Lett. 4 (2004) 419. [33] R. Ab Kadir, W. Zhang, Y. Wang, J.Z. Ou, W. Wlodarski, A.P. O'Mullane, G. Bryant, M. Taylor, K. Kalantar-zadeh, J. Mater. Chem. A 3 (2015) 7994. [34] J.Z. Ou, W. Ge, B. Carey, T. Daeneke, A. Rotbart, W. Shan, Y. Wang, Z. Fu, A.F. Chrimes, W. Wlodarski, S.P. Russo, Y.X. Li, K. Kalantar-zadeh, ACS Nano 9 (2015) 10313. [35] J. Dai, M. Yang, X. Yu, H. Lu, Opt. Fiber Technol. 19 (2013) 26. [36] K. Ohira, Y. Sakamoto, T.B. Flanagan, J. Alloys Compd. 236 (1996) 42. [37] E. Lee, J. Lee, J.-S. Noh, W. Kim, T. Lee, S. Maeng, W. Lee, Int. J. Hydrogen Energy 37 (2012) 14702. [38] A. Noordermeer, G.A. Kok, B.E. Nieuwenhuys, Surf. Sci. 172 (1986) 349. [39] Z. Yin, W. Zhou, Y. Gao, D. Ma, C.J. Kiely, X. Bao, Chem. A Eur. J. 18 (2012) 4887. € der, G.R. Castro, K. Wandelt, [40] R. Linke, U. Schneider, H. Busse, C. Becker, U. Schro Surf. Sci. 307 (1994) 407. [41] V.G. Litovchenko, V.S. Solntsev, Sensing effects in the nanostructured systems, in: J. Bonca, S. Kruchinin (Eds.), NATO Science for Peace and Security Series B: Physics and Biophysics, 2008, p. p.pp 373. [42] V.G. Litovchenko, Condens. Matter Phys. 1 (1998) 383.
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: http://www.elsevier.com/locate/jalcom
Linear hydrogen gas sensors based on bimetallic nanoclusters Ahmad I. Ayesh Department of Mathematics, Statistics and Physics, Qatar University, Doha, Qatar
a r t i c l e i n f o
a b s t r a c t
Article history: Received 29 April 2016 Received in revised form 14 June 2016 Accepted 29 July 2016 Available online 30 July 2016
This work reports on the fabrication of hydrogen gas sensors based on bimetallic palladium-copper nanoclusters. The nanoclusters were generated by sputtering and inert-gas condensation inside an ultra-high vacuum (UHV) compatible system, and self-assembled on an insulating substrate with a pair of pre-formed interdigitated gold/nichrome electrodes. Nanocluster deposition was stopped once their coverage on the substrate reached the percolation threshold. Electrical properties of the fabricated sensors were investigated by means of electrical conductance measurements, and assigned to charge carrier transport within network of metallic islands that is dominated by tunnelling. The produced devices were utilized as conductometric gas sensors. Herein, a constant voltage was applied across the interdigitated electrodes, and the change in electrical current signal was measured which reflects gas concentration. All fabricated sensors showed increase in the conductance upon exposure to hydrogen which can be assigned to the increase in tunnelling current due to the decrease in the size of the gaps between the nanoclusters or the establishment of conducting paths through the network of percolating nanocluster film. The sensors were found to be sensitive at low concentrations of hydrogen at room temperature, and exhibit a linear relationship between hydrogen concentration and the sensitivity. Therefore, those sensors have the potential to be used for practical life applications. © 2016 Elsevier B.V. All rights reserved.
Keywords: PdCu Bimetallic nanoclusters Nanocluster devices Inertegas condensation Hydrogen sensor
1. Introduction Hydrogen is an important source of clean energy, and it has many advantages such as: it is pollution free and naturally produced by plants and animals [1,2]. However, the utilization of hydrogen fuel requires reliable hydrogen sensing devices [3]. Among the different types of hydrogen sensors, electrical conductivity hydrogen sensors that utilize nanoclusters as hydrogen sensitive elements were found efficient sensors [4,5]. Herein, the development of hydrogen gas sensors that exhibit a linear response signal with hydrogen concentration, an enhanced response time, and operates efficiently at room temperature is essential [6,7]. In this work we report on the fabrication of conductometric hydrogen sensors based on the change of electrical conductivity upon exposure to hydrogen gas. Each hydrogen gas sensor is formed from a percolating Pd-Cu alloy nanocluster film at the percolation threshold deposited on an insulating substrate with a pair of pre-formed interdigitated electrodes. Pd-Cu alloy nanoclusters can be formed by various techniques such as: electroless deposition [8e10], sol-gel polymerized [11],
E-mail address: [email protected]. http://dx.doi.org/10.1016/j.jallcom.2016.07.320 0925-8388/© 2016 Elsevier B.V. All rights reserved.
and ultrasonic-assisted membrane reduction [12]. However, nanoclusters used in the current work were generated by sputtering and inert-gas condensation from a Pd-Cu target inside an ultrahigh vacuum (UHV) compatible system [13e20]. This technique has many advantages over other nanocluster production techniques such as: nanoclusters produced by this technique are of high purity (since they are produced inside an ultra-high vacuum system), the controllability of nanocluster composition, the possibility of nanocluster size selection using a suitable mass filter, and they can be created and self-assembled directly on a desirable substrate [14,19,20]. The present gas sensors may be used in multitude of different applications including, but not limited to, safety sensors, and hydrogen storage devices [4,6,21e24]. For example, hydrogen sensors are used during the hydrogenating cooking oil route, and considered as a cost-effective means to control and quantify the process [25]. 2. Experimental Nanoclusters used for the present work are bimetallic nanoclusters of palladium and copper alloy that were generated by sputtering and inert-gas condensation from a Pd-Cu target inside an UHV system with a base pressure ~108 mbar [19]. The UHV
2
A.I. Ayesh / Journal of Alloys and Compounds 689 (2016) 1e5
system consists of three main chambers [20]: i) source chamber, ii) mass selection chamber, and iii) deposition chamber (see Fig. 1 in Ref. [19]). The nanoclusters are generated inside the source chamber, travel through the mass filter that enables nanocluster size selection, and then they are deposited on a substrate fixed on a temperature controlled sample holder inside the deposition chamber. The sputtering and inert-gas condensation process was established inside the source chamber using Argon (Ar) inert-gas. Herein, Ar was used to: i) generate plasma needed to sputter material from the composite target that consists of 33.3% Cu and 66.6% Pd, ii) form nanoclusters by inert-gas condensation of the sputtered material, and iii) enable nanoclusters to travel from the source through the mass filter to the deposition chamber due to pressure gradient between the two chambers. Nanoclusters used in this work were produced using an Ar gas flow rate (fAr) of 50 sccm and a sputtering discharge power (P) of 62 W. In addition, pure Pd nanoclusters were produced at fAr ¼ 90 sccm and P ¼ 85 W, and used to produce hydrogen gas sensors for comparison. The sputter head and source chamber were water cooled at room temperature. A quadrupole mass filter (QMF) was used to determine nanocluster size distribution [18]. The mass filter consists of four parallel metal rods where each pair of opposite rods is connected electrically together to potentials of (U þ Vcos (ut)) and -(U þ Vcos (ut)), where U is a dc voltage and Vcos (ut) is an ac voltage. For a mass distribution scan, the ratio U/V was fixed and the mass distribution was scanned by varying the frequency, u. Once the nanoclusters' beam leaves the QMF, it moves through a Faraday cup that detects their current signal, thus, the current signal reflects the number of nanoclusters produced inside the source. The size and composition of the produced nanoclusters were confirmed using a Philips CM10 transmission electron microscope (TEM) and energyedispersive Xeray (EDX) measurements, respectively. The percolating gas sensor device was fabricated by depositing the produced nanoclusters on silicon dioxide/silicon (SiO2/Si) substrates with preeformed interdigitated gold/nichrome (Au/ NiCr) electrodes (formed by standard shadow mask technique) that are 50 mm apart [15]. Nanoclusters were also deposited on SiO2/Si substrates with preeformed planer electrodes that are 10 mm apart for electrical conductivity measurements as a function of temperature. The deposition rate of nanoclusters was measured using a quartz crystal monitor (QCM) fixed on a motorized linear translator that enables driving the QCM in front of the substrate, measure the deposition rate, and then drive it back away from the nanocluster beam path. The position of the liner translator holding the QCM could be controlled without venting the system [20]. Fig. 1(a) shows a schematic diagram of the hydrogen gas sensor of the present work. The substrate was mounted on a substrate
holder inside the deposition chamber, and connected electrically to a Keithley 238 source measuring unit. The electrical conductance of the sample is monitored during nanocluster deposition while applying a 100 mV voltage across the electrodes. The electrical current signal normally fluctuates at small current values before the onset of conduction due to the deposition of charged nanoclusters on the substrate. Once the percolation threshold is approached, an onset of conduction is observed and the electrical current increases abruptly (see Fig. 1(b)), thus, nanocluster deposition is suddenly stopped using an automatic shutter. The response of the device to hydrogen gas was characterized inside a temperature controlled a custom-designed Teflon chamber under different concentrations of hydrogen in pure nitrogen or air, as shown in Fig. 1(c) (and Fig. 4 in Ref. [26]). The test chamber was located in a fume hood at atmospheric pressure and 25 C. The gas flow rate in the sensor test chamber was controlled using Bronkhorst mass flow meters. The response signal of the device was measured using a computer controlled Keithley 236 source measuring unit. A constant voltage of 0.1 V was applied to the sensor, and electrical current signal was monitored as a function of time and gas concentration. 3. Results and discussion Pd and Cu contents within the produced nanoclusters were found 77 ± 1% and 23 ± 1%, respectively, as measured using EDX. A representative EDX measurement is shown in Fig. 2(a). The lower Cu contents within nanoclusters compared to that in the target is a result of the lower nanocluster production yield of Cu compared to Pd [18]. The size distribution of the produced nanoclusters was measured using the QMF and TEM as depicted in Fig. 2(b) and (c), respectively. The average nanocluster size as measured using the QMF is ~8.1 ± 1.2 nm, where the error is taken as one standard deviation. The TEM image reveals similar average size to that measured using the QMF. The average size of the pure Pd nanoclusters as measured using the QMF is 6.6 ± 0.7 nm. Electrical conductivity measurement (measured at 1 V) as a function of temperature of Pd0.77Cu0.23 nanoclusters deposited on a substrate with planer electrodes at the percolation threshold as shown in Fig. 3. The figure reveals an exponential dependence of the conductivity as a function of temperature. Similar behaviour was observed previously and assigned to charge carrier transport within network of metallic islands that is dominated by tunnelling through small barriers at the junctions between each pair of islands [15,27]. The electrical conductance of such system can be described using [28]: eV
GðTÞ ¼ s0 ðTÞenkB T
(1)
Fig. 1. (a) Schematic diagram of the hydrogen gas sensor formed from a percolating nanocluster film. (b) Electrical current e time dependence of the sensor at V ¼ 100 mV during nanocluster deposition. (c) Schematic diagram of the test chamber.
A.I. Ayesh / Journal of Alloys and Compounds 689 (2016) 1e5
3
Fig. 2. (a) EDX measurements of the produced nanoclusters. (b) Pd0.77Cu0.23 nanocluster size distribution as measured using QMF. (c) TEM image of Pd0.77Cu0.23 nanoclusters produced at the same conditions as in (a).
Fig. 3. Electrical conductance as a function of temperature. The (bottom) inset is the natural logarithm of the conductance as a function of invers temperature. The solid line is a linear fit of the results according to Eq. (1). The (top) inset reveals a schematic diagram of the planer electrodes with nanoclusters. The solid line is an example of suggested conduction path.
Where V is the applied voltage across the nanocluster network, T is temperature, s0 ðTÞ is the temperature dependant conductance at V ¼ 0, e is the electron charge, kB is Boltzmann constant, and n is the number of barriers between the nanoclusters that are forming the conduction path. The inset of Fig. 3 shows the dependence of the conductance natural logarithm on invers temperature. The slop of a linear fit can be used to estimate the number of tunnelling barriers as well as the nanocluster size. Herein, the slope is 8.29 which gives n ¼ 1399, thus, nanocluster size is 7.15 nm for contact separation of 10 mm. The estimated nanocluster size is close to the values measured using both QMF and TEM image which support that charge carrier transport is dominated by tunnelling through barriers between each two adjacent Pd0.77Cu0.23 nanoclusters. The underestimation of nanocluster size can be assigned to the non-straight conduction path which increases the number of nanoclusters involved in charge conduction as presented by the (top) inset of Fig. 3. Sensor response is defined as [6]:
IH2 I0 S¼ 100% I0
(2)
Where IH2 is the electric current in the presence of hydrogen, and I0 is the reference electrical current in absence of hydrogen gas. The results of sensitivity measurements corresponding to exposure of Pd0.77Cu0.23 sensor for different percentages of hydrogen gas
Fig. 4. Representative response curves for the gas sensor based on Pd0.77Cu0.23 nanoclusters at hydrogen concentrations between 0.5 and 5% (a), and 1e10% (b). The results of the control Pd nanocluster sensor is shown in (a).
relative to air are depicted in Fig. 4(a) and (b). The sensitivity measurements demonstrate that the present sensors provide great degree of sensitivity, with a typical figure of merit S ~30%, for 5% hydrogen relative to air at 25 C. This phenomena is quite different from the conventional hydrogen sensors based on microscopic palladium structure, where the current decreases upon exposure to hydrogen. The invers response of the present sensor is a main base of the sensing behaviour compared to the conventional microscopic hydrogen sensor. It should be noted her that sensitive hydrogen sensors that operate at room temperature were also reported for other nanostructures decorated with Pd [29]. Fig. 4(a) shows also the response of the control sensor formed of pure Pd nanocluster. The response of the control sensor is similar to
4
A.I. Ayesh / Journal of Alloys and Compounds 689 (2016) 1e5
the previously published results [4]. This response is nonlinear unlike the response of the Pd0.77Cu0.23 sensor. The dependence of sensor response on hydrogen concentration for four Pd0.77Cu0.23 sensor devices as well as the control device is shown in Fig. 5. The figure depicts a linear relationship between response and hydrogen concentration of the four Pd0.77Cu0.23 sensors with identical slope of 5.8 ± 0.3, unlike the response of the Pd control sensor. Therefore, it can be concluded that the four Pd0.77Cu0.23 sensors are identical, and their production is a reproducible process. The reproducibility test of a Pd0.77Cu0.23 hydrogen sensor is depicted in Fig. 6. The figure reveals a reproducible response signal of the sensor upon exposure to a fixed concentration of hydrogen of 3%. Furthermore, the response signal returns back to the background once hydrogen is removed. The test was repeated after one month from sensor fabrication and similar results have been observed. Future investigations may include long-term and systematic measurements of the stability as well as the drift of the produced sensors. The response time is defined as the time needed for the response electrical current signal to increase to 90% of the maximum response. Fig. 7 illustrates the dependence of the response time on hydrogen concentration for two different Pd0.77Cu0.23 nanocluster based sensors. The figure depicts a constant response time over the different hydrogen concentrations. The average response time can be calculated for the two devices to 18.6 ± 2.9 s, which is a reasonably fast response time. The sensing behaviour of the present Pd0.77Cu0.23 sensors can be rationalized as follows. A Pd0.77Cu0.23 sensor composes of nanocluster film at the percolation threshold, hence, many tunneling junctions and gaps exist within the film of nanoclusters. Each tunneling barrier or gap functions as a switch [15]. Upon exposure to hydrogen gas, the size of each nanocluster increases which decreases the size of tunneling barriers or gaps [4,30]. Thus, the electrical conductivity of the percolating film of the hydrogen sensor increases. It is well established that exposing Pd to hydrogen causes the expansion of the face centred cubic (fcc) lattice by a maximum of 3.6% due to a phase change in the crystal structure: from a to b phase [31]. The phase expansion occurs of equal compression along each nanocluster axis, and preferably at the grain boundaries. As a result, the intergranular gaps of a Pd nanocluster film are reduced, thus, electrical conductance of the nanocluster film increases. The phase transition is manifested as a plateau in a plot of ambient hydrogen gas pressure versus hydrogen content of the Pd lattice.
Fig. 5. The dependence of the response on hydrogen concentration for: four different sensors based on Pd0.77Cu0.23 nanoclusters, and the control sensor based on Pd nanoclusters.
Fig. 6. Response reproducibility for the Pd0.77Cu0.23 nanocluster based gas sensor using a constant hydrogen concentration of 3%.
Fig. 7. Dependence of the response time on hydrogen concentration for two Pd0.77Cu0.23 nanocluster based gas sensors.
Therefore, it can be concluded that the increase in the electrical conductivity of the Pd nanocluster film may occur because of one or a combination of the following points: i) a decrease in the size of the gaps between the nanoclusters which lead to an increase in the tunnelling current; ii) the connection of previously isolated pathways and nanoclusters to a main conducting path through the percolating film; and/or iii) the establishment of main conducting path through the percolating nanocluster film [32]. Exposing a Pd nanocluster film to a pure ambient air induces the b to a phase transition and the contraction of each nanocluster, thus, opening the gaps again within the nanocluster film causing decrease in the electrical conductivity of the nanocluster film. Therefore, Pd sensors can respond for multiple times of hydrogen exposure and removal associated to opening and closing of the gaps between nanoclusters, i.e. this process is a reversible process. Consequently, hydrogen concentration is reflected on the electrical conductivity. Furthermore, the above process is a highly selective process because only hydrogen can be absorbed as discussed above and cause the lattice expansion of Pd. This implies that Pd is resistive to poisoning by humidity and other gaseous species such as: SO2, H2S, and CH4. Production of nanostructures from metal alloys allows to produce gas sensors with tuned sensitivity and selectivity [33,34]. of nanostructure Absorption of hydrogen by Pd/Ag and Pd/Ni alloys has been studied previously by different investigators [35e37]. The main improvement of alloying Pd with other metals was found to
A.I. Ayesh / Journal of Alloys and Compounds 689 (2016) 1e5
enhance the mechanical properties on repeated exposure. In addition, hydrogen was found to be desorbed on Pd-Cu alloy surface at similar temperatures to pure Pd [38], which explains the functionality of the present sensor for hydrogen at room temperature. Alloying Pd with small percentage of Cu has a minimal influence on the binding of hydrogen to Pd surface [39], therefore, it was used in the present work to control the sensitivity of hydrogen sensor because of the difference in the adsorption of hydrogen by the alloyed material. Herein, a main advantage of using Pd-Cu alloy nanoclusters for hydrogen sensor is the linear relationship between the sensitivity and hydrogen concentration, consequently, the easiness of the sensor calibration. Alloying Pd with Cu has the following effects on the chemisorption of H2 [38]: 1) Structural effect: computer simulations of Pd0.7Cu0.3 alloy revealed decrease in relative concentration of triplet sites, and increase in relative concentrations of doublet and singulet sites. The capability of doublet and singulet sites to trap and dissociate hydrogen molecules and the successive diffusion to the triplet Pd sites on Pd-Cu alloy can clarify the sensitivity of Pd0.77Cu0.23 nanocluster based sensor for hydrogen. 2) Dilution effect: the repulsive and attractive forces between the adsorbed H2 and nanocluster surface are different for Pd-Cu than pure Pd and Cu. Herein, Cu repels the d-orbital of Pd which introduces additional normal direction of d-bonds thus the sensitivity of the alloy nanoclusters is enhanced [40]. 3) Electron energy states effect: some electronic states that are occupied for pure Pd and Cu becomes partially free after alloying which enhance the adsorption activity of Pd-Cu nanoclusters [41]. Herein, the d-orbital is a main cause for the formation of chemical bond since it is a long range active which creates a small barrier for adsorption, thus, the reaction cross section (RCS ~ r) and the sensitivity to hydrogen increases [42]. The redistribution of electron density on the Fermi level for Pd (4d10) and Cu (3d104S1) leads to enhancement in the sensitivity according to the striking coefficient equation (CS ) [42]:
CS
Ea Ed 1 Ae kB T
(3)
where A is a constant, Ea is the adsorption energy, and Ed is the desorption energy. For Ed > Ea, activated adsorption (H2 on Cu) has high probability, and for Ed < Ea, a non-activated adsorption (H2 on Pd) has high probability. 4) Spillover effect: this is a result of the accumulation of gas molecules at the interface between nanoclusters and substrate which enhances the sensitivity of the sensors based on Pd-Cu (compared with pure Pd) nanoclusters [41].
4. Conclusion In conclusion, hydrogen gas sensors based on Pd-Cu alloy nanoparticles were fabricated in this work. The nanoclusters were fabricated by sputtering and inert gas condensation inside an ultrahigh vacuum compatible system. This nanocluster production method enabled the self-assembly of nanoclusters on SiO2/Si substrates with pre-formed electrical electrodes. The results demonstrated that charge carrier transport was dominated by tunnelling through small barriers. The produced sensors were sensitive at low hydrogen concentrations, functional at room temperature with high sensitivity, capable of detecting concentrations of hydrogen as low as 0.5% in air, and exhibit short response time. The main
5
advantage of the present sensors is the linear relationship between hydrogen concentration and the sensitivity, thus, they are easy to calibrate. Consequently, those sensors can be used for different applications including safety and control sensors. In addition, the change in the conductivity of the present sensor is 30% at a hydrogen concentration of 5%, which is much greater than the response of the conventional hydrogen sensors at the same gas concentration. References [1] M. Ismail, F.A.H. Yap, N.N. Sulaiman, M.H.I. Ishak, J. Alloys Compd. 678 (2016) 297. [2] NREL. Technical Report, 2007. [3] T. Tharsika, A.S.M.A. Haseeb, S.A. Akbar, M.F.M. Sabri, Y.H. Wong, J. Alloys Compd. 618 (2015) 455. [4] Lith Jv, A. Lassesson, S.A. Brown, M. Schulze, J.G. Partridge, A. Ayesh, Appl. Phys. Lett. 91 (2007) 181910. [5] M. Okada, J. Nakahigashi, A. Fujita, M. Yamauchi, A. Kamegawa, J. Alloys Compd. 580 (2013) S401. Supplement 1. [6] A.I. Ayesh, S.T. Mahmoud, S.J. Ahmad, Y. Haik, Mater. Lett. 128 (2014) 354. € [7] S. Oztürk, N. Kılınç, J. Alloys Compd. 674 (2016) 179. [8] S. Okada, T. Kamegawa, K. Mori, H. Yamashita, Catal. Today 185 (2012) 109. [9] T. Wang, C. Shannon, Anal. Chim. Acta 708 (2011) 37. [10] D.-Y. Lee, S.-H. Park, D.-J. Qian, Y.-S. Kwon, Curr. Appl. Phys. 9 (2009) e232. [11] M. Ugalde, E. Chavira, M.T. Ochoa-Lara, C. Quintanar, J. Nano Res. 14 (2011) 95. [12] L. Liu, T. Wei, X. Zi, H. He, H. Dai, Catal. Today 153 (2010) 162. [13] H.M. Aldosari, A.I. Ayesh, J. Appl. Phys. 114 (2013) 054305. [14] A.I. Ayesh, Z. Karam, F. Awwad, M.A. Meetani, Sensors and actuators B, Chemical 221 (2015) 201. [15] A.I. Ayesh, Appl. Phys. Lett. 98 (2011) 133108. [16] A.I. Ayesh, H. Ahmed, F. Awwad, S. Abu-Eishah, S.T. Mahmoud, J. Mater. Res. 28 (2013) 2622. [17] A.I. Ayesh, Gas sensor applications using atomic nanoclusters, in: S.S. Naveen Kumar Navani, J.N. Govil (Eds.), Nanosensing 10, Studium Press LLC, Houston, TX 77072, USA, 2012. [18] A.I. Ayesh, N. Qamhieh, S.T. Mahmoud, H. Alawadhi, J. Mater. Res. (2012) 27. [19] A.I. Ayesh, S. Thaker, N. Qamhieh, H.J. Ghamlouche, Nanopart. Res. 13 (2011) 1125. [20] A.I. Ayesh, N. Qamhieh, H.S. Ghamlouche, M.E.-S. Thakera, J. Appl. Phys. 107 (2010) 034317. P. Nemec, R. Andok, A. Has [21] I. Rýger, G. Vanko, T. Lalinský, Ben cúrova cík S, ska, Sensors and actuators a, Physical 227 (2015) 55. M. Toma [22] B. Jang, S. Cho, C. Park, H. Lee, M.-J. Song, W. Lee, Sensors and actuators B, Chemical 221 (2015) 593. [23] B. Kang, Y. Heo, L. Tien, D. Norton, F. Ren, B. Gila, S. Pearton, Appl. Phys. A Mater. Sci. Process. 80 (2005) 1029. [24] J. Brouwer, Curr. Appl. Phys. 10 (2010). S9. [25] S. Prabhu, F. Schweighardt, Hydrogen sensing and detection. Hydrogen fuel: production, transport, and storage, CRC Press Boca Rat. (2009) 495. [26] A.I. Ayesh, A.F.S. Abu-Hani, S.T. Mahmoud, Y. Haik, Sensors and actuators B, Chemical 231 (2016) 593. [27] A.I. Ayesh, S.T. Mahmoud, N. Qamhieh, Z.A. Karam, Acta Metall. Sin. 27 (2014) 156. [28] B. Ozturk, C. Blackledge, B.N. Flanders, D. Grischkowsky, Appl. Phys. Lett (2006) 88. [29] B. Liu, D. Cai, Y. Liu, H. Li, C. Weng, G. Zeng, Q. Li, T. Wang, Nanoscale 5 (2013) 2505. [30] M.Z. Atashbar, S. Singamaneni, Sensors and actuators B, Chemical 111e112 (2005) 13. [31] K.J. Stevens, B. Ingham, M.F. Toney, S.A. Brown, A. Lassesson, Curr. Appl. Phys. 8 (2008) 443. [32] M. Yun, N.V. Myung, R.P. Vasquez, C. Lee, E. Menke, R.M. Penner, NANO Lett. 4 (2004) 419. [33] R. Ab Kadir, W. Zhang, Y. Wang, J.Z. Ou, W. Wlodarski, A.P. O'Mullane, G. Bryant, M. Taylor, K. Kalantar-zadeh, J. Mater. Chem. A 3 (2015) 7994. [34] J.Z. Ou, W. Ge, B. Carey, T. Daeneke, A. Rotbart, W. Shan, Y. Wang, Z. Fu, A.F. Chrimes, W. Wlodarski, S.P. Russo, Y.X. Li, K. Kalantar-zadeh, ACS Nano 9 (2015) 10313. [35] J. Dai, M. Yang, X. Yu, H. Lu, Opt. Fiber Technol. 19 (2013) 26. [36] K. Ohira, Y. Sakamoto, T.B. Flanagan, J. Alloys Compd. 236 (1996) 42. [37] E. Lee, J. Lee, J.-S. Noh, W. Kim, T. Lee, S. Maeng, W. Lee, Int. J. Hydrogen Energy 37 (2012) 14702. [38] A. Noordermeer, G.A. Kok, B.E. Nieuwenhuys, Surf. Sci. 172 (1986) 349. [39] Z. Yin, W. Zhou, Y. Gao, D. Ma, C.J. Kiely, X. Bao, Chem. A Eur. J. 18 (2012) 4887. € der, G.R. Castro, K. Wandelt, [40] R. Linke, U. Schneider, H. Busse, C. Becker, U. Schro Surf. Sci. 307 (1994) 407. [41] V.G. Litovchenko, V.S. Solntsev, Sensing effects in the nanostructured systems, in: J. Bonca, S. Kruchinin (Eds.), NATO Science for Peace and Security Series B: Physics and Biophysics, 2008, p. p.pp 373. [42] V.G. Litovchenko, Condens. Matter Phys. 1 (1998) 383.