ZnO NWs

Solid-State Electronics 104 (2015) 126–130

Contents lists available at ScienceDirect

Solid-State Electronics journal homepage: www.elsevier.com/locate/sse

Co-axial core–shell ZnMgO/ZnO NWs Abdiel Rivera, Anas Mazady, Mehdi Anwar ⇑ Electrical and Computer Engineering Department, University of Connecticut, 371 Fairfield Way, Storrs, CT 06269, United States

a r t i c l e

i n f o

Article history: Received 8 May 2014 Received in revised form 17 July 2014 Accepted 19 August 2014 Available online 16 September 2014 The review of this paper was arranged by Prof. A. Zaslavsky Keywords: Co-axial ZnMgO/ZnO core–shell Nanorods Gas sensor Traps Activation energy

a b s t r a c t We report the first co-axial Zn0.9Mg0.1O/ZnO core–shell structures, on p-Si substrates, grown using metal–organic chemical vapor deposition (MOCVD). With ZnO buffer serving as a seed layer, vertically aligned ZnMgO NWs with a Mg mole fraction of 10% were grown, serving as the template for the growth of ZnO shell. The core and the shell have c-lattice constants of 5.1868 Å and 5.1996 Å, respectively, with no measurable strain at the core. The core–shell structure is single crystalline with an abrupt ZnMgO–ZnO interface. Performance of ZnMgO/ZnO core–shell structure as a gas sensor is reported. The room temperature response magnitude (RM) was found to be between 2% and 7% for a methanol concentration in the range of 1–15 ppm. RM as high as 37% is measured at 75 °C for higher methanol concentration. Co-axial ZnMgO/ZnO core–shell structures are demonstrated to be a better alternative than ZnO nanorods as gas sensors. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction ZnO is a wide band gap semiconductor with diverse growth morphologies, such as, nanowires (NW), nanorods (NRs), nanoribbons, nanobelts, nanorings and nanotubes. ZnO NWs/NRs have potentials to be used as the building block for solar cells [1,2], UV diodes [3,4], transistors [5,6], and sensors [7] etc. Piezoelectric properties of ZnO NWs/NRs can be exploited to be used as motion or vibration sensors. Zhu et al. reported ZnO-based piezoelectric devices that can generate an open circuit voltage of 2.03 V, sufficient enough to power a LED [8]. A telemetric bridge monitoring system was developed by implementing ZnO-based vibration sensors [9]. Its chemical stability and biocompatibility allows ZnO to be implanted on human body to monitor glucose levels remotely [10]. Besides, applications in the medical and defense industries [11], ZnO NWs/NRs are extensively used to detect humidity [12] and gases such as: ammonia [13], methanol [14] and hydrogen [15]. ZnO NWs/NRs have been synthesized via molecular beam epitaxy (MBE) [16], metal–organic chemical vapor deposition (MOCVD) [17], and hydrothermal synthesis [17], sonochemical process [18], among others. MBE allows monitoring of the structural quality while the NWs are being grown. However, this technique requires the use of catalysts, such as gold, as seed layer [19], which can introduce undesired defects into the structure [20]. Hydrothermal synthesis is a simple, low-cost and low tem⇑ Corresponding author. http://dx.doi.org/10.1016/j.sse.2014.08.010 0038-1101/Ó 2014 Elsevier Ltd. All rights reserved.

perature process used to grow vertically oriented NWs and NRs. Nevertheless, low temperature synthesis is often followed by a high temperature annealing process (>700 °C) in order to improve the crystal quality and optical properties of the grown materials [21]. Sonochemical process could provide a novel alternative to hydrothermal synthesis by providing a faster growth rate at room temperature with better crystal quality, as reported by Nayak et al. [18]. In contrast, MOCVD provides control over the morphology and orientation of the NWs/NRs by adjusting growth parameters such as temperature, gas flow, and pressure and does not require pre-treated substrates [22]. Banerjee et al. synthesized ZnO nanorods with diameter and length of 60–70 nm and 500 nm, respectively via chemical bath deposition (CBD) [23]. The sensing properties of these NRs were studied over a temperature range of 27–300 °C in the presence of methanol. The optimum temperature range at a concentration of 1000 ppm of methanol was 150–250 °C with RM exceeding 90%. However, at room temperature, for the same gas concentration the response magnitude was 20–30%. Performance of most gas sensors are reported in an operational temperature range of 150–400 °C. Huang et al. summarized the highest sensitivities reported using ZnO NRs for different gases: for H2S at 50 part per billion (ppb), the room temperature (RT) sensitivity (S) was reported to be 1.7, for methanol at 50 part per million (ppm) S = 3.2 (RT), and for ethanol at 100 ppm S = 20 (RT) [24]. Only a few researchers report sensing performance at low temperatures as the adsorption of gas in much smaller at RT requiring high concentrations of the gases (>1000 ppm).

A. Rivera et al. / Solid-State Electronics 104 (2015) 126–130

Semiconductor heterostructures were reported to enhance the gas sensing properties and thermal stability in comparison with homo-nanostructures [25]. Typically, ZnO core–shell (CS) structures are grown using ZnO NWs as core with multiple NWs (shell) protruding perpendicular to the axis of the core [26] through spin coating [27]. Growth of ZnO/ZnMgO CS structures has been demonstrated using various other methods including hydrothermal and pulsed laser deposition (PLD). Growth using hydrothermal method resulted in flat-hexagonal tops of the NW/NRs which generally show defects in the absorption spectra at long wavelengths [28]. In PLD technique, CS NWs were grown in multiple layers which suffer from the limitation of a low density of NWs to avoid contacts between the outer shells [29]. Growth of CS structures was reported by Choi et al. using SnO2–ZnO nanofibers synthesized via electrospinning and pulsed laser deposition (PLD). The structure was used for O2 sensing at 300 °C with sensitivity in the range 0–3 for O2 concentration of 1–2000 ppm [30]. Wang et al. reported growth of ZnO CS hollow microspheres using hydrothermal synthesis demonstrating a relatively high sensitivity, in the range of 2–14%, with a low concentration of toluene measured at room temperature [31]. We report the growth of co-axial ZnMgO/ZnO core–shell structures using MOCVD on p-Si substrates. We demonstrate for the first time, ZnMgO/ZnO CS gas-sensing performance exceeding a response magnitude (RM) of 2–7% at room temperature with a methanol concentration in the range of 1–15 ppm. A RM of 37% is achieved at 75 °C suggesting that better performance is achievable at even higher temperatures. Finally, we report the growth of ZnO NRs on a flexible substrate, which can be used as a selfpowered gas sensor. 2. Material growth ZnMgO/ZnO CS structures are grown on p-Si substrates using a horizontal First-Nano 3000 MOCVD system. Prior to growth, the substrate was ultrasonically cleaned in acetone and methanol, rinsed in DI water and dried for 15 min. This was followed by the growth of a ZnO thin film used as a seed layer for the succeeding growth of ZnMgO core. The ZnO thin film was grown at 300 °C at a constant pressure of 70 Torrs for 20 min. Flow rates of 50 sccm and 35 sccm were maintained of Diethylzinc (DEZn) and N2O which were used as zinc and oxygen precursors, respectively. Nitrogen was used as the carrier gas during all the growth steps. The temperature was raised to 650 °C and the pressure was decreased to 4 Torrs for the growth of the ZnMgO core structure seeded on the thin film using a ratio of 1/10 of DEZn/Cp2Mg, Cp2Mg (Bis(cyclopentadienyl)magnesium) being the Mg source. The Zn/ Mg ratio can be varied to tune the Mg mole fraction of the core as will be reported in detail elsewhere. A 10 min drive-in of MgO using 200 sccm of Cp2Mg and 500 sccm of N2O was performed to increase the Mg mole fraction. The growth was followed by Mg activation at 800 °C for 10 min in nitrogen ambient. Finally, the ZnO shell was grown at 300 °C using similar parameters used during the growth of the seed layer. 3. Results and discussion Fig. 1A shows the scanning electron microscopy (SEM) images of the ZnMgO/ZnO CS structure grown using MOCVD. The ZnMgO/ZnO CS structures are mostly vertically aligned with respect to the basal plane and have sharp tips (ZnO terminated). Fig. 1B shows the coaxial shape of the CS structure. Energy-dispersive X-ray spectroscopy (EDS) with excitation energy of 10 keV confirms the CS structure by showing 8–10% Mg in the core (inset of Fig. 1B) and 0% Mg in the shell. The Mg concentration of the core can be tuned by adjusting

127

the Mg/Zn ratio during growth and by varying the annealing time. High resolution EDS was also carried out under transmission electron microscope (TEM) confirming 8–10% Mg incorporation at the core and 0% Mg incorporation at the shell, as illustrated in Fig. 1C. To assess the co-axial CS interface, bright field images are taken using TEM as shown in Fig. 1D. The TEM sample was prepared by mechanical exfoliation, removing the CS structures from the p-Si substrate and depositing them on TEM copper grids. The diameter of the core (ZnMgO) and the thickness of the shell (ZnO) structures are around 80–100 nm and 20–40 nm, respectively. The total length of the structure is estimated to be in the range of 0.8–1.2 lm. The abrupt interface between the CS is highlighted by arrows. The contrast between the ZnMgO core and ZnO shell is not as pronounced as is observed for typical heterostructures, owing to a similar crystal composition possibly due to similar lattice constants. The inset of Fig. 1C shows the selected area electron diffraction (SAED) measurement indicating the structure is single crystalline. These results suggest that ZnO was epitaxially grown on Zn0.90Mg0.10O NWs resulting in a low or non-defect nano-structure. To further investigate the crystal structure of the ZnMgO/ZnO CS nanowires, X-ray diffraction measurement was performed. Bruker D-8 Advanced X-ray diffractometer with a wavelength k = 1.5406 Å corresponding to the Cu Ka line was used. Fig. 2A shows the XRD pattern for the ZnMgO/ZnO CS structure grown on p-Si substrates using MOCVD. No peak is observed corresponding to cubic MgO crystal or other alloys, such as Zn1.7SiO4 (Zn,Mg)1.7SiO4 [32], suggesting the absence of those impurities in our grown materials. Fig. 2B shows that the dominant peak is related to ZnO (0 0 2) observed at 34.6644° (2h), supporting the vertical alignment of the NWs observed in the SEM images. The peaks corresponding to ZnO and ZnMgO are observed to overlap since their lattice constants are very similar. A peak resulting from a combination of multiple peaks due to diverse crystal planes and/ or materials may be decomposed using a Lorentzian fit to identify the constituent peaks, thereby providing the maximum (where the c-lattice constant is calculated) and the full-width at halfmaximum (FWHM) for each of the peaks. High resolution XRD at 34° shows a split in the diffraction peak confirming the ZnMgO/ ZnO CS structure in agreement with the SEM, EDS and TEM analysis. The peaks at 34.5569° and 34.4699° are attributed to the ZnMgO core and ZnO shell, respectively, from which the c-lattice constants were estimated to be 5.1868 Å and 5.1996 Å. The relative variation of Zn0.90Mg0.10O c-lattice constant is 0.01% (Dc/c) for 14% Mg, according to Ohtomo et al. [33]. Indeed, low or non-threading dislocations are induced as a result of compressive stress from the ZnO shell as observed by Perillat-Merceroz et al. in multiple layers [34]. The c-lattice constant of the core is consistent with ZnMgO NWs grown on p-Si (5.1889 Å, 10%Mg) [35], (5.191 Å, 12% Mg) [36] and on c-sapphire (5.1873 Å, 8% Mg) [37]. The FWHM was estimated to be 0.0818° for the core and 0.0427° for the shell. The FWHM of the ZnMgO core is approximately two times larger than that of ZnO core since FWHM increases with the increment of Mg mole fraction on ZnO. The performance of ZnMgO/ZnO CS structure based gas sensors were compared with that of ZnO NRs based gas sensors. ZnO NRs were grown on a flexible substrate at low temperature using hydrothermal synthesis. The diameters of the NRs ranged from 160 to 210 nm with lengths in the range of 590–660 nm. Fig. 3 shows the XRD of ZnO NRs (blue dots) grown on a flexible substrate along with the seed layer (black line) and ZnO thin film (red dash). Diffraction peaks associated to ZnO NRs oriented along (1 0 1, 1 0 2, 1 0 3 and 1 1 2) planes are observed. In addition to these crystal planes, ZnO along the (0 0 2) is identified with a FWHM of 0.0552 (199 arcsec) and a c-lattice of 5.1923 Å, reflecting out-of-plane strains of 0.22%. The NRs are mostly vertically aligned.

128

A. Rivera et al. / Solid-State Electronics 104 (2015) 126–130

Fig. 1. (A) SEM of co-axial ZnMgO/ZnO core–shell structure showing a clean surface and sharp tips grown using MOCVD. (B) SEM shows the core–shell structure. The diameter was measured to be 80–100 nm for the ZnMgO core and 20–40 nm for the ZnO shell. The inset shows EDS of the core and shell, reflecting 10% Mg at the core, 0% at the shell. (C) EDS of ZnMgO/ZnO core–shell structure with nm scale resolution carried out under TEM confirming the Mg mole fraction at the core of the CS. (D) Bright field TEM image shows the co-axial core–shell structure; the inset shows SAED pattern suggesting single crystalline structure.

A

6000

60000

B

Intensity (a.u.)

Intensity (a.u.)

5000 4000 3000 2000 1000

40000

20000

0 10

20

30

40

50

60

70

80

2θ (deg.)

90

34.2

34.4

34.6

34.8

2θ (deg)

Fig. 2. XRD pattern of the Zn0.90Mg0.10O/ZnO core–shell structure with the dominant peak associated to ZnO (0 0 2); the inset shows a split on the diffraction pattern of the (0 0 2) peak as a result of the core–shell structure.

Fig. 4 shows the room temperature dynamic response of the ZnMgO/ZnO CS structure over time with/without the presence of methanol. The ZnMgO/ZnO CS detectors were fabricated by depositing copper on p-Si, serving as the bottom metal contact, and mounting the sample on a chip holder. Indium dots were deposited on the ZnMgO/ZnO surface using a home-built indium-soldering station, serving as the top metal contact. During measurement a load resistance was connected in series with the gas sensor (1 kO for the CS-based sensor and 1 MO for the ZnO NR-based sensor) forming a voltage divider. The rather large variation of the load resistances, in series, is to match the intrinsic resistances of the devices as ZnO NRs based sensors were grown on high resistivity flexible substrates

whereas CS structures were grown on Si-substrates. The sensor was biased with 2 V, and the output voltage was measured across the load resistor. The resistance of the device changes for varying concentration of gas molecules, affecting the output voltage of the sensor. With gas reaching the sensor surface, methoxy species and formaldehyde molecules are formed by dissociative chemisorption of methanol [38]. Subsequently, the mentioned species gives rise to formic acid liberating an electron and adsorbed at the surface of the gas sensor. This electron enters into the conduction band and lowers the resistivity of the sensor material [39]. It should be pointed out that compared to ZnO NRs based detector, the CS structure exhibited 3 times higher sensitivity at room temperature for

129

desorption of the gas molecules, two transitions are observed for ZnO NRs based sensor while a single transition for CS based sensor. The activation energy of the defects are calculated from these transition times using the following equation [40]:

ZnO (110)

ZnO (112)

ZnO (102)

Seed Layer Thin Film Nanorods ZnO (103)

td ¼

20000

10000

35

40

45

50

55

60

65

70

75

80

2θ (deg.)

1.0 0.9 0.8 0.7 0.6 0.5 0.4

Ga sI n

t Ou

15

Core-Shell

30

s Ga

5

45

n sI Ga

t Ou

1.02 1.00 0.98 0.96 0.94 0.92 0.90 0.88 0.86

s Ga

Voltage (Normalized Units)

Fig. 3. XRD of the seed layer, ZnO thin film and ZnO NRs grown on flexible substrate.

10

ZnO NRs

15

20

Time (s) Fig. 4. Sensing performance of the core–shell structure based gas sensor is compared with ZnO NRs based gas sensor over time with/without the presence of methanol at room temperature. The CS based sensor has a sensitivity almost 3 times higher than ZnO nanorods, with a faster desorption rate.

high concentration of methanol (>5000 ppm). The relative sensitivGas Þ, where RAir and RGas are the ity or response magnitude ðRM ¼ RAirRR Gas

Intensity (Normalized Units)

resistances of the sensor in air and in the methanol gas environment, respectively, equals 50.79% for CS, with RAir = 767 X and RGas = 379 X, while for ZnO NRs, RM = 13.71%, with RAir = 445 kX and RGas = 384 kX. The response time for adsorption and desorption of gas molecules is faster for CS based gas sensor than for ZnO NRs. During Laser

707nm

Core-Shell ZnO NRs

0.2 587nm

668nm

0.1

728nm

537nm

0.0 510 540 570 600 630 660 690 720

Wavelength (nm)

1

rn v th Nc eðEc ET Þ=kT

where td (=0.130 s for CS and, 3.3 and 5 s for NRs) is the detrapping time, rn (=1.6  1013 cm2 [41]) is the electron capture cross secpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi tion, vth = thermal velocity ð 3kT=me Þ, Nc (=1  1018 cm3) is the effective density of states, Ec and ET are the conduction and trap energy levels, and me (=0.734 for CS and 0.138 for NR [42]) is the electron effective mass. The activation energy for CS-gas sensor and ZnO NRs is approximately 1.16 eV and, 1.25–1.26 eV, respectively. The energy trap for CS, with a band gap of 3.48 eV, correspond to the energy level of neutral oxygen vacancies (VO) [43], while the energy traps for ZnO NRs-gas sensor with a band gap of 3.27 eV correspond to single charged oxygen vacancy (V+O) [43]. The energy level of VO traps is located at a shallower level compared to V+O, allowing faster adsorption and desorption in CS-based sensor than for ZnO NRs as observed in Fig. 4. In addition to a faster adsorption and desorption rate, sensitivity is higher for CS-gas sensor compared to ZnO NR-based sensor. The rate of adsorption can be defined as the product of the incident flux of methanol molecules, and the sticking probability, which depends upon the coverage (h = number of adsorption sites occupied/number of adsorption sites available) and the presence of any activation barrier for adsorption. For simplicity, assuming the incident flux of methanol molecules to be constant for both cases (CS and ZnO NRs), the rate of adsorption is determined by the sticking probability. The number of adsorbed methanol and oxygen ions on the ZnO surface is directly related to the surface-to-volume ratio [44]. Comparing the intensity of the photoluminescence (PL) peaks shown in Fig. 5, CS structure has more oxygen vacancies and interstitial defects than ZnO NRs, resulting in higher sensitivity. In addition, the enthalpy of chemisorption is possibly lower for CS than ZnO NRs contributing to larger adsorption rates. However, further investigation is needed to confirm this hypothesis. Fig. 6 compares the response magnitude (RM) of NRs and CS-based gas sensors at room temperature and at 75 °C under different concentrations of methanol. Response magnitude of the CS-based sensor varies from 1.72% to 7.9% at room temperature for methanol concentration varying in the range of 1–15 ppm. At room temperature the adsorbed O ions that interact with methanol are limited resulting in lower sensitivity compared to higher temperature. Most of the reported gas sensors required an operational temperature between 150 and 300 °C. For example, the CNT/Sn2O CS based gas sensor was reported to show a sensitivity of 11 with

Intensity (Normalized Units)

Intensity (a.u.)

30000

ZnO (002) ZnO (101)

A. Rivera et al. / Solid-State Electronics 104 (2015) 126–130

Core-Shell ZnO NRs

1083nm

0.18

0.09 1039nm

806nm

1175nm

853nm

0.00 800

900

1000

1100

Wavelength (nm)

Fig. 5. Photoluminescence (PL) of ZnMgO/ZnO core–shell structure and ZnO NRs at room temperature showing the oxygen vacancy related defects at long wavelengths (>500 nm).

130

A. Rivera et al. / Solid-State Electronics 104 (2015) 126–130

References

Fig. 6. Measured response magnitude (RM) of ZnMgO/ZnO core–shell structure at low methanol concentration (solid black line with left y-axis and bottom x-axis) carried out at room temperature. Top x-axis and right y-axis (red-dash line-square) shows the RM of the CS at larger methanol concentration carried at 75 °C. The blue triangles are the RM for ZnO nanorods (NR) recorded at 75 °C with methanol concentration of 3000 and 5000 ppm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

an ethanol concentration of 10 ppm at 300 °C [45]. At a temperature of 75 °C with methanol concentration below 1000 ppm, the RM of the ZnMgO–ZnO co-axial structures remained almost constant at 1–2%. For a gas concentration in between 1000 and 3000 ppm, the RM varied from 10% to 37% (Fig. 6 right side, top axis). ZnMgO/ZnO CS detectors provide better response both at room and elevated temperatures compared to the ZnO NRs based detectors with similar geometry and size as shown in Fig. 6. The response magnitude of standard ZnO NR based gas sensor are 22% and 25% for methanol concentrations of 3000 and 5000 ppm, respectively, that are lower than ZnMgO/ZnO CS structure based detectors. It is worth noting that the sensor performance was consistent over multiple cycles and over a temperature range of 25–100 °C. 4. Conclusion Co-axial ZnMgO/ZnO CS structures are reported using MOCVD on p-Si with a Mg mole fraction of 10%. The thickness of the core (ZnMgO) and shell (ZnO) structures are 80–100 nm and 20–40 nm, respectively, with a total length of the structure being in the range of 0.8–1.2 lm. The morphology of the co-axial CS structure is established via XRD analysis, showing a split at 34° suggesting the presence of both ZnMgO and ZnO. We demonstrated for the first time the sensing property of the co-axial ZnMgO–ZnO CS structure with response magnitude as high as 7% for a methanol concentration of 15 ppm at room temperature. At an operational temperature of 75 °C, a RM of 37% was achieved with the CS structures for a methanol concentration of 3000 ppm, larger when compared to ZnO NWs based sensor with a RM of 22%.

[1] Bi D, Boschloo G, Schwarzmuller S, Yang L, Johansson EM, Haqfeldt A. Nanoscale 2013;5:11686–91. [2] Ko SH, Lee D, Kang HW, Nam KH, Yeo JY, Hong SJ, et al. Nano Lett 2011;11:666–71. [3] Lupan O, Pauporte T, Viana B, Tiginyanu IM, Ursaki VV, Cortes R. Appl Mater Interf 2010;2:2083–90. [4] Kim J, Yun J-H, Kim CH, Park YC, Woo JY, Park J, et al. Nanotechnology 2010;21:115205. [5] Bashir A, Wobkenberg PH, Smith J, Ball JM, Adamopoulos G, Bradley DDC, et al. Adv Mater 2009;21:2226–31. [6] Bong H, Lee WH, Lee DY, Kim BJ, Cho JH, Cho K. Appl Phys Lett 2010;96:192115. [7] Hongsith N, Wongrat E, Kerdcharoen T, Choopun S. Sens Actuat B 2010;144:67–72. [8] Zhu G, Yang R, Wang ZL. Nano Lett 2010;10:3151–5. [9] Yu A, Jiang P, Wang ZL. Nano Energy 2012;1(3):418–23. [10] Yu R, Pan C, Chen J, Zhu G, Wang ZL. Adv Funct Mater 2013. http://dx.doi.org/ 10.1002/adfm.201300593. [11] Zhou J, Gu Y, Hu Y, Mai W, Yeh P-H, Bao G, et al. Appl. Phys. Lett. 2009;94:191103. [12] Chang S-P, Chang S-J, Lu C-Y, Li M-J, Hsu C-L, Chiou Y-Z, et al. Supperlattices Microstruct 2010;47:772–8. [13] Renganathan B, Sastikumar D, Gobi G, Yogamalar NR, Bose AC. Opt Laser Technol 2011;43:1398–404. [14] Faisal M, Khan SB, Rahman MM, Jamal A, Abdullah MM. Appl Surf Sci 2012;258:7515–22. [15] Al-Hardan NH, Abdullah MJ, Aziz AA. Int J Hydrogen Energy 2010;35:4428–34. [16] Heo YW, Varadarjan V, Kufman M, Kim K, Norton DP, Ren F, et al. Appl Phys Lett 2002;81:3046–8. [17] Rivera A, Zeller J, Sood A, Anwar M. J Electronic Mater 2013;42:894–900. [18] Nayak AP, Lin T-C, Lam D, Kaya S, Islam MS. Nanosci Technol Lett 2012;4:977–82. [19] Falyouni F, Benmamas L, Thiandome C, Barjon J, Lusson A, Galter P, et al. J Vac Sci Technol B 2009;87:1662. [20] Kim SW, Fujita S. Appl Phys Lett 2005;86:153119. [21] Heo YH, Varadarjan V, Kaufman M, Kim K, Norton DP, Ren F, et al. Appl Phys Lett 2002;81:3046. [22] Jeong MC, Oh BY, Lee W, Myoung JM. J Cryst Growth 2004;268:149. [23] Banerjee N, Roy S, Sarkar CK, Bhattacharyya P. IEEE Sens J 2013;13:1669–76. [24] Huang J, Wan Q. Sensors 2009;9:9903–24. [25] Wang L, Lou Z, Fei T, Zhang T. J Mater Chem 2011;21:19331. [26] Pan C, Guo W, Dong L, Zhu G, Wang ZL. Adv Mater 2012;24:3356–61. [27] Guillén E, Azaceta E, Vega-Poot A, Idígoras J, Echebarría J, Anta JA, et al. J Phys Chem C 2013;117:13365–73. [28] Liu W. J Phys Chem C 2010;114:16148. [29] Cao BQ. Nanotechnology 2009;20:305701. [30] Choi S-W, Park JY, Kim SS. Nanotechnology 2009;20:465603. [31] Tang W, Wang J, Yao P, Li X. Sens Actuat B 2014;192:543–9. [32] Shimpi P, Ding Y, Suarez E, Ayers J, Gao P-X. Appl Phys Lett 2010;97:103104. [33] Ohtomo A, Kawasaki M, Koida T, Masubuchi K, Koinuma H, Sakurai Y, et al. Appl Phys Lett 1998;72:2466. [34] Perillat-Merceroz G, Thierry R, Jouneau P-H, Ferret P, Feuillet G. Appl Phys Lett 2012;100:1731102. [35] Chang V-S, Chien C-T, Chen C-W, Chu T-Y, Chiang H-H, Ku C-H, et al. J Appl Phys 2007;101:033502. [36] Zhi M, Zhu Z, Ye Z, Wang F, Zhao B. J Phys Chem B 2005;109:23930. [37] Wassner TA, Laumer B, Maier S, Laufer A, Meyer BK, Stutzman M, et al. J Appl Phys 2009;106:051111. [38] Bowker M, Houghton H, Waugh KC. J Chem Soc Faraday Trans 1981;1(77):3023–36. [39] Gao Q, Zheng W-T, Wei C-D, Lin H-W. J Nanomater 2013. Article ID 263852. [40] Islam SS, Anwar AFM, Webster RT. IEEE Trans Electron Devices 2004;51:846–53. [41] Chicot G, Murret P, Pernot J, Condensed Matter 2014arXiv1401.6851C. [42] Yarar Z. J Electron Mater 2011;40:466–72. [43] Djurisic AB, Ng AMC, Chen XY. Prog Quantum Electron 2010;34:191–259. [44] Hongsith N, Wongrat E, Kerdcharoen T, Choopun S. Sens Actuat B 2010;144:67. [45] Chen Y, Zhu C, Wang T. Nanotechnology 2006;17:3012–7.