High-Temperature Tolerance of the Piezoresistive Effect in P-4H-Sic For

Journal of Materials Chemistry C

View Article Online COMMUNICATION View Journal | View Issue

High-temperature tolerance of the piezoresistive eﬀect in p-4H-SiC for harsh environment sensing† Cite this: J. Mater. Chem. C, 2018, 6,8613 Tuan-Khoa Nguyen, *a Hoang-Phuong Phan, a Toan Dinh, a Received 24th June 2018, Abu Riduan Md Foisal, a Nam-Trung Nguyen a and Dzung Viet Daoab Accepted 24th July 2018

DOI: 10.1039/c8tc03094d

rsc.li/materials-c

4H-silicon carbide based sensors are promising candidates for strain/pressure/acceleration sensors, thanks to the simplicity replacing prevalent silicon-based counterparts in harsh environ- of the read-out circuitries and the wide linearity range.2,3 When ments owing to their superior chemical inertness, high stability and mechanical stress/strain is applied to a piezoresistive based sensing reliability. However, the wafer cost and the difficulty in obtaining an element, the energy band is significantly modified, leading to a ohmic contact in the metallization process hinders the use of this large variation in electrical conductivity (or resistivity).4–6 SiC polytype for practical sensing applications. This article presents Due to the different arrangement of Si and C atoms in the the high-temperature tolerance of a p-type 4H-SiC piezoresistor crystal structure, there are more than 200 SiC polytypes which at elevated temperatures up to 600 8C. A good ohmic contact was have been discovered. 3C-SiC, 4H-SiC and 6H-SiC are the most formed by the metallisation process using titanium and aluminium common and commercially available polytypes to date. Over annealed at 1000 8C. The leakage current at high temperatures was the last four decades, many studies have investigated the measured to be negligible thanks to a robust p–n junction. Owing to piezoresistive effect of these polytypes for mechanical sensing the superior physical properties of the bulk 4H-SiC material, a high applications. Shor et al. revealed a gauge factor (GF) of 31.8 gauge factor of 23 was obtained at 600 8C. The piezoresistive effect for n-type single crystalline 3C-SiC grown on a Si substrate with also exhibits good linearity and high stability at high temperatures. a doping concentration of 1017 cm3.7 Phan et al. reported the The results demonstrate the capability of p-type 4H-SiC for the GFs of p-type 3C-SiC to be 25–28 at temperatures varying from development of highly sensitive sensors for hostile environments. 300 K to 573 K using an in situ approach.3 For a-SiC, the GFs of 19 3 Published on 25 July 2018. Downloaded by Griffith University 8/29/2018 2:23:20 AM. highly doped (i.e. 2 10 cm ) n-type and p-type 6H-SiC at The development of hostile environment sensors is hindered by room temperature were found to be 22 and 27, respectively; current technologies relying on silicon (Si) materials. For instance, these GFs decreased by half at 250 1C.8 Akiyama et al. measured Si experiences plastic deformation at 500 1C and instability at the GF of n-type highly doped 4H-SiC to be 20.8 at room high temperatures or in corrosive environments. In contrast, its temperature for transverse piezoresistors.9 Amongst the three superior tolerance to high temperatures, corrosive media, and most common SiC polytypes, 4H-SiC possesses the highest intensive shock positions silicon carbide (SiC) as a promising energy band gap (i.e. 3.26 eV) with a high potential barrier material for a broad range of applications in hostile environments.1 which can effectively minimize the generation of electron–hole Moreover, the progress of wafer growth and the fabrication process pairs with an increase in temperature. This results in the good not only cuts down the cost of SiC wafers, but also greatly improves stability and reliability of 4H-SiC electronic devices in high- the film quality and reliability of SiC devices. In terms of micro- temperature operations.10,11 Additionally, the superior physical machining, SiC can inherit the strong foundation and maturity of properties combined with a high break-down voltage makes Si technology infrastructures, which enables the batch production 4H-SiC the most important SiC polytype for high-power electronics, of SiC sensing devices for corrosive or harsh environment sensing. including Schottky diodes, high power-switching devices, bipolar The piezoresistive effect has been considered as the main junction transistors (BJTs)12–14 and metal oxide semiconductor transduction mechanism for mechanical sensors such as field effect transistors (MOSFETs).10 In terms of sensing devices, to date there are limited studies on 4H-SiC sensors for high temperature applications due to several challenges. a Queensland Micro- and Nanotechnology Centre, Griffith University, Brisbane, Firstly, the development of high quality SiC film growth Queensland 4111, Australia. E-mail: khoa.nguyentuan@griffithuni.edu.au b School of Engineering and Built Environment, Griffith University, Gold Coast, recently enabled the availability of p-4H-SiC wafers, whereas Queensland 4215, Australia engineering-grade n-4H-SiC wafers have been available for a † Electronic supplementary information (ESI) available. See DOI: 10.1039/c8tc03094d decade. Another factor is the difficulty in the formation of an

This journal is © The Royal Society of Chemistry 2018 J. Mater. Chem. C, 2018, 6, 8613--8617 | 8613 View Article Online

Communication Journal of Materials Chemistry C

ohmic contact with p-4H-SiC,15,16 which limits the studies on the piezoresistive effect of p-4H-SiC. Additionally, the deployment of strain configurations which can effectively induce strain to thin- film piezoresistors at high temperatures is challenging. These obstacles result in a lack of understanding of the piezoresistive effect of 4H-SiC at elevated temperatures, which is solely based on theoretical analyses, such as a first principles simulation in SiC nanosheets17 and a deformation potential approximation.18 In this article, we present the first experimental study on the piezoresistive eﬀect of p-type 4H-SiC at high temperatures up to 600 1C. As such, high gauge factors were obtained: 33 at room temperature and 23 at 600 1C. The high sensitivity and good stability at elevated temperatures are attributed to the superior physical properties of 4H-SiC as well as the robust p–n junction which prevents the current from leaking to the substrate. This demonstrates the capability of p-type 4H-SiC for hostile environment sensing where the use of Si or polymer based devices is infeasible. The p-type 4H-SiC piezoresistive element was fabricated from a bulk 4H-SiC wafer which is commercially available from Cree. The wafer consists of epitaxial highly doped (i.e. carrier density of 1018 cm3) p-type 4H-SiC on the top, n-type 4H-SiC lying in the middle and a low doped n-type substrate. The thickness of both the p-type and n-type layers is 1 mm and the thickness of the substrate is 350 mm. The wafer was cleaned Fig. 1 (a) Fabrication process of the 4H-SiC piezoresistors: (1) starting prior to a photolithography process using AZ9245 to form a p-on-n 4H-SiC wafer; lithography process including (2) UV exposer and (3) patterned photoresist mask with a thickness of 4 mm for the photoresist removal; (4) ICP etching of p-type 4H-SiC; (5) metal deposi- subsequent etching of SiC. The patterned mask SiC wafer was tion; (6) pattern metal contact. (b) SEM image of the as-fabricated 4H-SiC piezoresistors. Scale bar, 100 mm. anisotropically etched in an STS SR inductive couple plasma (ICP) system for 10 min with 1500 W coil power. A 50 sccm HCl flow at a pressure of 2 mTorr was supplied. After ICP etching, The contact resistance was measured to be approximately the remaining photoresist mask was removed by acetone and 103 O cm2 at room temperature and then increased about 30% isopropanol, and then thoroughly rinsed in deionized water. An at 600 1C. In the final step, the wafer was diced into 30 3mm2 etch depth of 1.3 mm was measured by a Dektak 150 profiler, beams prior to the piezoresistive eﬀect measurement.

Published on 25 July 2018. Downloaded by Griffith University 8/29/2018 2:23:20 AM. which ensured that the unmasked p-type regions were com- The 4H-SiC beams were placed in a temperature-controlled pletely removed, as shown in Fig. 1(a). To form the metal probe system (i.e. Linkam HFS600E-PB2). The temperature of contact, 100 nm-thick Ti and Al layers were subsequently the probe chamber was precisely regulated by a built-in closed- sputtered in an SNS Sputterer. Another photolithography step loop control in the temperature controller unit. The upper limit was performed to create a contact patterned mask prior to the of temperature in this work is restricted by the equipment. wet etching of Al and Ti by an Al etchant (i.e. H3PO4)anda The small temperature tolerance (i.e. 0.1 1C) minimizes the mixture of hydrogen fluoride (HF) and hydrogen peroxide deviation of the resistance change by the temperature variation. (H2O2), respectively. Subsequently, the wafer was thoroughly The sensitivity of the 4H-SiC piezoresistor was defined by the rinsed and cleaned with deionized water. After the metallisation, fractional change of the resistance versus the applied strain: the wafer was annealed by a rapid thermal annealing (RTP) DR=R process at 1000 1C for 3 min to create the ohmic contact. The GF ¼ð1 þ 2uÞþ (1) e ohmic contact with p-type 4H-SiC has been extensively investigated in previous studies,15,16,19 and the ohmic contact for- where e is the strain induced to the piezoresistor and u is mation of Ti/Al to p-type 4H-SiC annealed at 1000 1C occurs in the Poisson ratio of 4H-SiC. For SiC materials, the first term 16 two stages. In the early stage, Ti and Al react to form TiAl2, of eqn (1) represents the change due to the geometrical

TiAl3 and Ti3Al, leaving unreacted C atoms at the Ti/SiC effect under strain which is inferior to the second term of the interfaces. Subsequently, in the later stage, Ti, SiC and C react piezoresistive effect. To characterise the piezoresistive effect of

to form a ternary Ti3SiC2 compound. The use of thin Al and Ti 4H-SiC, uniaxial strain was induced to the piezoresistor using a layers ensures that they react with C and form the compound bending method.9,28,29 The analytical model is a double-layer during the two stages. Subsequently, the reaction of SiC and the cantilever with one clamped end, and a static force was applied contact materials increases the doping concentration in the region to the other end. Subsequently, uniaxial strain was eﬀectively close to the SiC/metal interface and forms the ohmic contact. transferred to the piezoresistor lying on the top surface of the

8614 | J. Mater. Chem. C, 2018, 6, 8613--8617 This journal is © The Royal Society of Chemistry 2018 View Article Online

Journal of Materials Chemistry C Communication

cantilever (see ESI† for the analysis of the straining model). The bending configuration is depicted in the Fig. 4 inset in which one end of the 4H-SiC beam was fixed on the hot plate by probes while the other end was deflected by a static force. For Si based devices the p–n junction cannot withstand temperatures above 200 1C due to the generation of thermally activated electron–hole pairs, leading to instability and unrelia- bility in their operation. In contrast, the large energy band gap in 4H-SiC forms a high potential barrier, enabling stability at high temperatures. We confirmed this property in 4H-SiC by characterizing the rectification behaviour of the 4H-SiC p–n junction at elevated temperatures up to 600 1C using a Keithley 2602. Fig. 2 shows the current–voltage characteristics of the p-type piezoresistor and the current leak through the p–n junction at a temperature range from 23 1C to 600 1C. The measured leakage current was approximately three orders of magnitude smaller then the current flowing in the p-type Fig. 3 Fractional change of resistance of the p-type 4H-SiC piezoresistor sensing layer. For example, at an applied voltage of 5 V at with various applied strains of up to 215 ppm, showing the good linearity of 600 1C, the current in the p-type was approximately 460 mA the piezoresistive eﬀect at high temperatures. while the leakage current was measured to be below 1 nA. This is attributed to the robust back-to-back p/n diode preventing This is equivalent to a reduction of sensitivity of approximately the current from leaking into the substrate.9,30 The very small 30% in the given temperature range. This result agrees well current leaking into the substrate demonstrates a significant with the decreasing trend of the GF with an increase in the advantage of 4H-SiC over other low band gap materials for ambient temperature in other reports for SiC polytypes mechanical sensing at elevated temperatures. (Table 1). It should be noted that p-type 4H-SiC is able to retain The relative resistance change versus the applied strain a relatively high GF at 600 1C, which is ten times higher than ranging from 0 to 215 ppm at temperatures ranging from the GF of conventional metal strain gauges at room tempera- 23 1C to 600 1C is illustrated in Fig. 3. Evidently, the p-type ture. When the temperature increases, the ionization eﬀect 4H-SiC piezoresistors exhibited good linearity in the chosen raises the carrier concentration as well as the electrical con- strain range. The measured GFs at varying temperatures are ductance of p-type 4H-SiC (Fig. 2). As a consequence, the shown in Fig. 4. The GF at room temperature was approxi- increase in carrier concentration and the carrier scattering will mately 33, then it gradually decreased to about 23 at 600 1C. reduce the GF in p-type 4H-SiC. This hypothesis was confirmed by the experimental results, as shown in Fig. 3 and 4. However, the GF did not significantly decrease under the wide range of Published on 25 July 2018. Downloaded by Griffith University 8/29/2018 2:23:20 AM.

Fig. 2 Current–voltage characteristics of the p-type piezoresistor with a voltage ranging from 5 V to 5 V at various temperatures from 23 1Cto Fig. 4 Measured gauge factor of p-type 4H-SiC at a temperature ranging 600 1C, showing the linear ohmic characteristics of the Ti/Al contact to the from 23 1C to 600 1C; the GF at 600 1C decreases about 30% in p-type layer. Inset: Schematic sketch of the measurement of the current comparison with the GF at room temperature. Inset: Schematic sketch flowing in the p-type layer and through the p–n junction. of the bending beam experiment in a Linkam chamber (not to scale).

This journal is © The Royal Society of Chemistry 2018 J. Mater. Chem. C, 2018, 6, 8613--8617 | 8615 View Article Online

Communication Journal of Materials Chemistry C

Table 1 Comparison of the gauge factor in various SiC polytypes at high temperatures

Doping Maximum Gauge factor Material Carrier type concentration (cm3) temperature (room temp. - max temp.) 3C-Si20 N 2.9 1018 573 K 32 - 14 3C-SiC7 N1016–1017 450 1C 31.8 - 17 3C-SiC21 N Unintentional 400 1C 18 - 7 3C-SiC22 N1019 400 1C 24 - 11 3C-SiC23 N — 400 1C 16 - 12.5 3C-SiC24 N — 200 1C 18 - 10 3C-SiC25 P — 300 1C28- 25 6H-SiC26 N 1.8 1017–3 1018 250 1C 29.4 - 17 6H-SiC8 N2 1019 250 1C 22 - 11.5 P2 1019 250 1C27- 11.5 Poly-SiC27 N1018–1020 200 1C10- 7 4H-SiC (this work) P 1018 600 1C33- 23

temperature. This result is attributed to the high carrier p2, and pSOSO are the hole concentrations of the LH, HH, and concentration of the p+ layer used in this work, which stabi- SOSO bands, respectively. This means that the number of holes lized the Fermi–Dirac integral at high temperatures and occupying the LH band is much larger than those in the HH resulted in a relatively stable GF. and SOSO bands and the majority of the holes are located in the The GFs of different SiC polytypes including 3C-, 6H-, and LH band, leading to the dominant effect of the LH band over 4H-SiC at high temperatures are summarised in Table 1. There the HH and SOSO bands in the piezoresistivity of p-type 4H-SiC. are a number of factors for the variation in the GF amongst Therefore, the change in the electrical conductivity of p-type different studies, including the difference in the SiC polytypes, 4H-SiC can be considered solely dependent on the redistribu- the impurity concentrations, and the quality/growth processes tion of the holes in the LH band under strain and ambient of the SiC films. The dopant type (n or p) typically dictates the temperature T as33 sign of the GF (i.e. ‘‘’’ sign for longitudinal n-type or trans- Ds Dp 1 DE verse p-type piezoresistors and ‘‘+’’ sign for transverse n-type or 1 ¼ V (2) s p k T 3=2 longitudinal p-type piezoresistors). Moreover, the GF normally 1 B 1 þ ðÞm1 =m2 decreases with an increase in carrier concentration and with an where m * and m * are the effective masses of the hole in the increase in temperature.31 The positive (or negative) values of 1 2 LH and HH bands, respectively; DE is the band splitting the GF indicate that the conductivity of the SiC piezoresistors V energy as a function of the induced stress to the crystal lattice. decreases (or increases) with an increase in stress/strain. In From eqn (2), it can be realized that the relative resistance terms of sensing devices, only the magnitude of the GF is change (or conductivity variation) is almost inversely propor- significant and not the sign. However, in the case of mobility tional to the increase in ambient temperature, which is in good enhancement, it is possible to significantly increase the electron agreement with the experimental data (Fig. 4).

Published on 25 July 2018. Downloaded by Griffith University 8/29/2018 2:23:20 AM. mobility in p-type/n-type SiC by utilizing transverse/longitudinal In conclusion, the high gauge factors of p-type 4H-SiC were strains (with a negative GF). Akiyama et al. reported GFs for n-type obtained with good linearity and stability at high temperatures 4H-SiC, with a highest value of 20.8 for transverse resistors and up to 600 1C. At 600 1C the gauge factor was found to be 10 for longitudinal resistors at room temperature. In compar- approximately 23 which is at least ten times higher than the ison, our measured GFs for p-type 4H-SiC are significantly higher conventional metal strain gauges. The measured piezoresistivity (i.e. 33 at room temperature and 23 at 600 1C). This means that of p-type 4H-SiC with temperature is in good agreement with the the use of p-type 4H-SiC will yield a higher sensitivity compared hole transfer model in p-type SiC. The high gauge factor of p-type to the n-type counterpart. Additionally, the as-fabricated p-type 4H-SiC demonstrates its potential for mechanical sensing in 4H-SiC piezoresistors are able to retain a relatively high GF hostile environments where the use of Si or polymer counter- at elevated temperatures where the use of other common semi- parts is infeasible. conductors (e.g. silicon and polymers) is not possible. The variation in the electrical conductivity of p-type 4H-SiC is due to the increased ionization eﬀect by temperature. Under Conflicts of interest strain, there is a hole transfer between the top valence bands including the light hole (LH), heavy hole (HH), and spin–orbit There are no conflicts to declare. split-off (SOSO) bands. In the energy band of 4H-SiC, the band gaps between the light hole band versus the heavy hole and Acknowledgements SOSO bands are 0.15 eV and 0.65 eV, respectively.32 Addition- ally, the hole concentration ratio between the valence bands is This work has been partially supported by Australian Research DE/kBT e . By substituting the Boltzmann constant kB = 1.38 Council grants LP150100153 and LP160101553. This work 1023 JK1 and the temperature range T = 298–873 K, we derive was also supported by the Queensland node of the Australian 3 10 p1/p2 = 7.3–300 and p1/pSOSO = 5.6 10 –9.7 10 , where p1, National Fabrication Facility, a company established under the

Journal of Materials Chemistry C Communication

National Collaborative Research Infrastructure Strategy to provide 17 K. Nakamura, T. Toriyama and S. Sugiyama, Jpn. J. Appl. nano and micro-fabrication facilities for Australian researchers. Phys., 2011, 50, 06GE05. 18 T. Toriyama, J. Micromech. Microeng., 2004, 14, 1445–1448. Notes and references 19 H. Vang, M. Lazar, P. Brosselard, C. Raynaud, P. Cremillieu, J.-L. Leclercq, J.-M. Bluet, S. Scharnholz and D. Planson, 1 Z. He, L. Wang, F. Gao, G. Wei, J. Zheng, X. Cheng, B. Tang Superlattices Microstruct., 2006, 40, 626–631. and W. Yang, CrystEngComm, 2013, 15, 2354. 20 R. Ziermann, J. von Berg, W. Reichert, E. Obermeier, 2 T.-K. Nguyen, H.-P. Phan, J. Han, T. Dinh, A. R. M. Foisal, M. Eickhoff and G. Krotz, Proc. Inter. Solid State Sens. S. Dimitrijev, Y. Zhu, N.-T. Nguyen and D. V. Dao, RSC Adv., Act. Conf. (Transducers), 1997. 2018, 8, 3009–3013. 21 C.-H. Wu, C. Zorman and M. Mehregany, IEEE Sens. J., 2006, 3 H.-P. Phan, T. Dinh, T. Kozeki, T.-K. Nguyen, A. Qamar, 6, 316–324. T. Namazu, N.-T. Nguyen and D. V. Dao, IEEE Electron Device 22 M. Eickhoff and M. Stutzmann, J. Appl. Phys., 2004, 96, Lett., 2016, 37, 1029. 2878–2888. 4 J. Bi, G. Wei, L. Wang, F. Gao, J. Zheng, B. Tang and 23 J. Strass, M. Eickhoff and G. Kroetz, Proc. Inter. Solid State W. Yang, J. Mater. Chem. C, 2013, 1, 4514. Sens. Act. Conf. (Transducers), 1997. 5 F. Gao, J. Zheng, M. Wang, G. Wei and W. Yang, Chem. 24 M. Eickhoff, H. Mo¨ller, G. Kroetz, J. v. Berg and Commun., 2011, 47, 11993. R. Ziermann, Sens. Actuators, A, 1999, 74, 56–59. 6 X. Li, S. Chen, P. Ying, F. Gao, Q. Liu, M. Shang and 25 H.-P. Phan, T. Dinh, T. Kozeki, A. Qamar, T. Namazu, W. Yang, J. Mater. Chem. C, 2016, 4, 6466–6472. S. Dimitrijev, N.-T. Nguyen and D. V. Dao, Sci. Rep., 2016, 7 J. Shor, D. Goldstein and A. Kurtz, IEEE Trans. Electron 6, 28499. Devices, 1993, 40, 1093–1099. 26 J. Shor, L. Bemis and A. Kurtz, IEEE Trans. Electron Devices, 8 R. Okojie, A. Ned, A. Kurtz and W. Carr, IEEE Trans. Electron 1994, 41, 661–665. Devices, 1998, 45, 785. 27 T. Homma, K. Kamimura, H. Y. Cai and Y. Onuma, Sens. 9 T. Akiyama, D. Briand and N. F. de Rooij, J. Micromech. Actuators, A, 1994, 40, 93–96. Microeng., 2012, 22, 085034. 28 H.-P. Phan, T. Dinh, T. Kozeki, T.-K. Nguyen, A. Qamar, 10 N. G. Wright, A. B. Horsfall and K. Vassilevski, Mater. Today, T. Namazu, N.-T. Nguyen and D. V. Dao, Appl. Phys. Lett., 2008, 11, 16–21. 2016, 109, 123502. 11 B. J. Baliga, Silicon carbide power devices, World scientific, 2006. 29 T.-K. Nguyen, H.-P. Phan, T. Dinh, J. Han, S. Dimitrijev, 12 L. Lanni, R. Ghandi, B. G. Malm, C.-M. Zetterling and P. Tanner, A. R. M. Foisal, Y. Zhu, N.-T. Nguyen and M. Ostling, IEEE Trans. Electron Devices,2012,59, 1076–1083. D. V. Dao, IEEE Electron Device Lett., 2017, 38, 955–958. 13 S. Singh and J. A. Cooper, IEEE Trans. Electron Devices, 2011, 30 T.-K. Nguyen, H.-P. Phan, T. Dinh, T. Toriyama, 58, 1084–1090. K. Nakamura, A. R. M. Foisal, N.-T. Nguyen and D. V. Dao, 14 A. Siddiqui, H. Elgabra and S. Singh, IEEE J. Electron Devices Appl. Phys. Lett., 2018, 113, 012104. Soc., 2018, 6, 126–134. 31 T. Toriyama and S. Sugiyama, J. Microelectromech. Syst.,

Published on 25 July 2018. Downloaded by Griffith University 8/29/2018 2:23:20 AM. 15 L. Fursin, J. Zhao and M. Weiner, Electron. Lett., 2001, 2002, 11, 598–604. 37, 1092. 32 C. Persson and U. Lindefelt, J. Appl. Phys., 1997, 82, 5496–5508. 16 R. Konishi, R. Yasukochi, O. Nakatsuka, Y. Koide, M. Moriyama 33 P. Kleimann, B. Semmache, M. L. Berre and D. Barbier, and M. Murakami, Mater. Sci. Eng., B, 2003, 98, 286–293. Phys. Rev. B: Condens. Matter Mater. Phys., 1998, 57, 8966.