Reliability of GaN-Based HEMT Devices Roberto Menozzi Department of Information Engineering University of Parma V.le G.P. Usberti 181A, 43100 Parma, Italy Email:
[email protected]
comprehensive way the reliability knowledge database for this technology.
Abstract—This paper reviews the main problems characterizing the past and present of GaN-based HEMT reliability. Some general considerations on the maturity of this technology and published lifetesting extrapolations are followed by a review of physical degradation mechanisms, subdivided between temperature-activated and electrical ones, the latter generally linked with the much-debated “current collapse”. The paper ends with some conclusive remarks on what has been achieved and what still lies ahead. Keywords-gallium nitride; nitride devices; HEMTs; reliability.
I.
INTRODUCTION
For the last few years, GaN-based HEMT technologies have been delivering many of the promises made in their early development stages [1]. As usual, when a technology moves out of the labs and into production lines, reliability becomes a major concern. It is also worth noticing that GaN-based HEMTs owe most of their appeal and success to the ability to operate at higher power densities and temperatures than their competitors, i.e., under conditions that typically exacerbate reliability problems. This paper therefore aims at giving an overview of the development and current status of GaN-based HEMT reliability, highlighting the progress made and the challenges still to be met. II.
Figure 1. Number of papers on AlGaN/GaN HEMT reliability per publication year.
While the fact that the history of GaN-based HEMT reliability knowledge is not long and it is not based on a very wide database can partly be compensated by expertise building up to some extent on GaAs-based FET and HEMT history, one cannot fail to notice that the reliability literature on GaN-based HEMTs still lacks that relatively process-independent uniformity of results and basic consensus on the underlying physics that are distinctive features of more mature technologies.
STATUS OF GaN-BASED HEMT RELIABILITY
While through the 90's and until a few years ago they were mainly research lab material and military pets, AlGaN/GaN HEMTs have recently become commercial products (see for example [2]-[8]). Reliability has therefore become fundamental for product qualification (see for example [9]), and GaN-based HEMTs have had their documented share of three-temperature life-testing, high-temperature storage, humidity testing, thermal cycling, ESD/EOS testing, mechanical testing, and so forth. Thus, it appears that AlGaN/GaN HEMTs are well under way to reach the status of mature technology.
In papers presenting new devices or process improvements over previous generations, it is customary to lump the information on reliability into two figures coming out of threetemperature accelerated life-tests, namely, the activation energy (Ea) and the median time to failure (MTTF) extrapolated to operating conditions (typical junction temperature Tj of 125°C or 150°C). Remarkable values of MTTF > 107 h (i.e., > 1,100 years) at Tj = 150°C [10]-[12] and MTTF = 3.5·109 h (i.e., a bit short of 400,000 years) at Tj = 125°C [13] have indeed been shown for GaN-based HEMTs. Fig. 2 shows a summary of extrapolated MTTF values taken mostly from the 22-year-long history of the Reliability of EIAJEDEC's Compound Semiconductors (ROCS) Workshop
However, if one looks at the scientific literature, the knowledge database on GaN-based HEMT reliability seems to be characterized by a few features indicating that the maturity goal is still somewhat far ahead. For example, Fig. 1 shows the number and temporal distribution of the papers on which the core of this review has been built: while not being an exhaustive catalogue of what has been published in this field, this set of papers nonetheless represents in a fairly
978-1-4244-2717-8/08/$25.00 © 2008 IEEE
44
COMMAD 2008
(formerly known as the GaAs REL Workshop), including all relevant compound semiconductor technologies: AlGaN/GaN HEMTs have clearly joined the mainstream in this respect.
with radiation-induced damage; experiments involving proton and heavy ion [16] and γ-ray irradiation [17] seem to indicate sufficient radiation hardness for space missions. III.
TEMPERATURE-ACTIVATED DEGRADATION MECHANISMS
High-temperature applications were soon recognized as a field where GaN-base HEMTs could offer distinctive advantage over Si and GaAs technologies, thanks to GaN's large band-gap and chemical stability. Consequently, the study of device performance and reliability under high-temperature conditions has accompanied the development of this technology since the early years (see [18] for a review of early processing issues and reliability studies). Different groups in the late 90’s reported on high-temperature stability of GaNbased HEMTs, with particular focus on ohmic and Schottky metallizations, and the use of refractory metals and barriers for enhanced reliability. Here are a few examples. (1)
Ti/WSiN/Au ohmic contacts (annealed at 860°C) and Pt/Au Schottky contacts were shown to be fairly stable up to 800°C during a step stress with 20 min step duration [19]; however, the heterostructure characteristics degraded rapidly between 600°C and 800°C, a range encompassing the Debye temperature for GaN (750°C).
(2)
HEMTs featuring standard Ti/Al/Ti/Au ohmic contacts (annealed by RTA at 850°C) with a WSiN encapsulating barrier and either WSiN/Au or Ir/Au gates showed only marginal degradation after 120 h up to 500°C, while Pt/Ti gates underwent significant degradation under the same conditions [20]-[22].
(3)
Mo/Au gates showed good stability up to 2000 h at 340°C, thanks to the refractory nature of molybdenum preventing interdiffusion [23].
(4)
Iridium and TiB2 barrier layers on Ti/Al ohmics (RTA at 900°C) were recently shown to provide good stability up to 350°C over several days of stress [24].
Figure 2. Estimated MTTF values for various compound semiconductor technologies as a function of publication year. GaN-based HEMT data are marked by white circles.
It is worth spending a few words on the significance of these figures. To be sure, process qualification involves much more than three-temperature accelerated life-tests (see for example Table 2 of [12]), but activation energies and median times to failure are often the only numbers that surface out into the literature. Two main problems, however—one quite obvious, another more subtle—make these figures hardly useful as reliability gauges. The obvious one is, high temperatures are required to cause significant degradation in reasonable times (e.g., 1,000 h, or few thousand hours at most), but whatever low-activation-energy degradation mechanism may be limiting the device reliability at operating temperatures will be cancelled out by high-Ea mechanisms likely to be irrelevant at operating temperatures, and the MTTF might be grossly overestimated as a result. A subtler concern is this: lifetesting in most instances involves applying both high temperature and bias to the devices under test, sometimes— GaN-based HEMTs in particular being still mysterious in some respects—without the faintest notion of how temperature and bias conditions may interact. A rather extreme caveat is offered by [14], where AlGaN/GaN HEMT degradation during RF stress is shown to decelerate with increasing temperature—a typical signature of hot-carrier degradation in most devices. Other investigators reported on a substantial independence of HEMT degradation on lifetesting temperature during 3000 h lifetests under different bias conditions [15].
Even standard metallizations, however, seem to warrant sufficient stability for most applications, as shown for instance in [25], where HEMTs with Ti/Al/Pt/Au ohmics (RTA at 860°C) and Pt/Au Schottky were step-stress tested (48 h steps) up to 400°C (estimated) junction temperature: the device degradation is marginal up to 300°C junction temperature. Beyond 300°C the HEMT degrades significantly, but the morphology of the metal contacts still indicates good stability, pointing to material defects as the main limiting factor for high-temperature reliability. Although the mechanism of ohmic contact formation is not completely understood, it is believed that Ti reacts with GaN forming TiN at high temperatures; the N vacancies behave like n-doping and facilitate tunneling. Another possible explanation is the low work function of TiN. High annealing temperatures are also necessary to promote the diffusion of Al into GaN: Al compensates the Ga acceptor-like vacancies that form at high temperatures. The overlying Ti, or Ni, or Pt layer serves as a
For such reasons, it is more instructive to turn one's attention from the extraction of activation energies and MTTFs to the study of physical degradation mechanisms, which will be done in the next two sections, centered on temperatureactivated and electrical degradation mechanisms, respectively. Another interesting reliability aspect of GaN-based HEMTs, which are potentially useful in space applications, is connected
45
barrier to prevent interdiffusion with the top Au, which has the role of reducing the sheet resistance. A detailed structural investigation of Ti/Al/Mo/Au ohmic contacts on AlGaN/GaN structures as a function of annealing temperature can be found in [26].
the maximum RF power the device can output [28]: the more severe the gate-lag, the smaller the output power delivered. 5) Transconductance low-frequency (1 kHz - 1 MHz) dispersion [36].
In conclusion, although some details about ohmic and Schottky contact formation of GaN and AlGaN may not be completely clear yet, the technology appears to be mature enough to provide devices with metal contacts that will not wear out during the operating lifetime at room temperature (and junction temperatures around 150-200°C). For hightemperature applications, specific metallizations featuring refractory metal layers and barriers can be employed. Issues connected with the quality of the starting semiconductor material (e.g., threading dislocations, defects originating from the lattice and thermal mismatch with the substrate, etc.) sometimes contribute to wear-out, but overall the long-term reliability of AlGaN/GaN HEMTs is likely to be limited less by temperature-activated wear-out than by the electrical degradation phenomena to be described in the next section.
It is worth pointing out that the characterization of these effects and the evaluation of HEMT degradation upon stressing is severely complicated by the fact that trapping phenomena in GaN-based HEMTs may have time constants in the range of days [39], which makes it very difficult to discriminate between temporary changes due to trap charge/discharge and permanent device degradation, and calls for very time-consuming procedures for assuring that, at the start of a characterization or stress session, the device is sitting on a steady-state and not slowly drifting towards it. The effect of illumination must also be duly reckoned with [40].
IV.
B. The "virtual gate" model One of the earliest and most successful physical models of these phenomena is the "virtual gate" model [41]. It is believed that surface (donor) states between gate and drain may capture electrons during high-drain-bias stress and cause localized 2dimensional electron gas (2DEG) depletion, as shown in Fig. 3, leading to drain resistance increase [40], and drain current, transconductance, and RF power/gain degradation. These surface states can be charged/discharged depending on the bias applied between gate and drain, thus acting as a second, virtual gate; the dynamics of this virtual gate are that of the surface states, with characteristic times that may be in the range of several seconds [41] - hence the difference between DC and pulsed measurements, the gate-lag, the transconductance frequency dispersion, etc.
ELECTRICAL D EGRADATION
A. Current collapse and related phenomena Current collapse is probably the most documented and debated performance and reliability concern for GaN-based HEMTs. It is a performance issue in that it is often observed in as-fabricated devices and therefore it cannot be considered as a degradation mode, but it is also a reliability issue because it tends to be exacerbated by electrical stress. As we will see, the physical mechanisms underlying current collapse are likely to be related to creation and/or charging and discharging of traps (mostly, if not exclusively, at the device surface), the consequences of which may manifest themselves in a variety of ways. Here are some examples. 1) Drain current compression under large-signal microwave operation [27]: the drain current decreases with increasing RF input power, particularly at high (i.e., less negative) gate bias, which indicates that it is likely a surface problem. Degradation of drain current and RF power after RF stress was also reported in [28]-[30]. 2) Degradation of DC output characteristics after long- or short-term stress at high drain voltage [31]-[34]. On-state and off-state stress cycles have been compared in [34], [35]. 3) Drain current compression in pulsed (100 ns - 1 ms) measurements relative to DC I-V curves [28], [36]; this problem is often seen to appear or be aggravated after electrical, high-drain-bias stress [37]. This effect can also be described as 4) gate-lag, i.e., a delayed drain current response following a gate voltage turn-on step [28], [36], [33], [34], [38]; again, this effect may be present in as-fabricated devices, but it may also appear or be enhanced after electrical stress at high drain bias. A trap activation energy of 0.3 eV was extracted in [36] by temperature-dependent gate-lag measurements. The amount of gate-lag (in terms of ratio between the pulsed and DC drain current) may correlate with
Figure 3. Schematic representation of the charge distribution in a GaN-based HEMT and the effect of the “virtual gate” (VG).
C. Evidence of trap creation during stress Surface charge accumulation transients lasting hundreds of seconds were detected by Kelvin probe measurements and found to be aggravated by long-term stress in [42], consistent with the hypothesis of surface damage via trap creation. Deeplevel transient spectroscopy (DLTS) measurements, performed after stress, were able in another instance to detect a new peak (Ea = 1.08 eV) attributed to interface traps [37]. In a few
46
reports [31], [43], [44], 1/f noise measurements also indicate that new traps may be generated during electrical stress.
F. Dependence on gate engineering and HEMT layer structure In spite of some diversity among results and interpretations of current collapse and degradation data, no one doubts that the large gate-drain electric field plays a key role in these phenomena. Consequently, gate engineering efforts, aimed at smoothening the field profile, have accompanied the development of GaN HEMTs. Overlapping gates [57] and field plates [58], [59] effectively relax the field profile in the gatedrain region, thereby prompting significantly larger breakdown voltage without compromising performance. This has beneficial effects on reliability, too [60], [61]. The presence of a leaky dielectric material under the field plate was shown to be particularly effective at reducing current and RF gain collapse [62], [63], a result explained in terms of faster discharge of surface traps via the leaky dielectric.
D. The effect of passivation and surface treatments Passivation, typically by SiN, was shown by several groups to reduce trapping transients and trap-related effects [45], [41], [42] and remarkably improve the HEMT robustness in the presence of electrical stress [42], [32], [46], [40]. DLTS measurements reported in [47] show that SiN passivation results in remarkable reduction of the density of a deep trap with activation energy of about 1.43 eV. MgO and Sc2O3 were also demonstrated to provide effective passivation layers [48]. Surface pre-treatments seem to play an important role in the fabrication of electrically reliable and collapse-free HEMTs. NH3 pre-treatment has been shown by different groups [49][51] to effectively improve the device performance and reliability. Other pre-passivation plasma surface treatments that were reported to improve the HEMT pulse behavior include C2F6, O2, Cl2 [51]. Ozone treatment may also result in improved reliability, as shown in [52]: this effect is believed to be connected with passivation of surface defects caused by dislocations intersecting the device surface, thereby creating localized depletion pockets.
As far as the HEMT layer structure is concerned, good stability under RF stress was attributed in [64] to the presence of an optimized n-doped GaN cap on top of the AlGaN layer, the role of the cap being that of reducing the electric field on the drain side of the gate. An undoped GaN cap was also effective in reducing dispersion effects [34], possibly due to the screening effect provided by surface donors. Decreasing the thickness of the AlGaN barrier was also shown to have a beneficial effect on the HEMT reliability during RF stress [65], a result attributed to reduced strain in the structure (see below for a discussion of the possible reliability implications of stress/strain in these devices).
Overall, these studies indicate that, while other effects (like buffer traps, see below) cannot be ruled out, the device surface conditions are definitely critical as far as electrical stability is concerned. Why passivation is effective in reducing lowfrequency dispersion phenomena, and increasing HEMT robustness during electrical stress, is still a matter of debate: possible explanations are (i) passivation of surface defects, (e.g. by hydrogen during plasma-enhanced CVD), (ii) the formation of leakage paths via the SiN, prompting faster charge/discharge of surface states, (iii) relaxation of the electric field profile [38], or a combination of such effects.
G. Numerical simulation results Numerical two-dimensional device simulations, typically carried out using the drift-diffusion model, have often accompanied and helped the interpretation of the experimental results discussed above. In [36], for example, the experimental activation energy of 0.3 eV was interpreted with the aid of simulations as the energy difference between surface donor traps and the valence band. Simulation may also give indications of the geometrical location of the traps responsible for HEMT degradation: see for example [66], which shows that traps located near the gate edges should be blamed for current collapse. An interesting result of an intensive simulation campaign documented in [33], [67], [38] is the assumption of the existence and relevance of both surface and buffer traps, stemming from the fact that surface traps alone cannot replicate in simulations the whole set of experimental findings (changes of I-V curves, gate leakage, and gate-lag after DC stress).
E. The effect of the source-gate region While, consistent with the "virtual gate" model, most of the attention has been given to the drain-gate surface area, there are studies [53]-[55] indicating that the source-gate surface may be equally important when it comes to current collapse. For sure, off-state bias and stress conditions, as well as the negative cycle of the RF input signal, may cause relatively large reverse bias on the gate-source junction and non-negligible leakage currents interacting with surface states. In addition, surface degradation on the source side (e.g., in terms of access resistance increase) can be expected to have a much larger effect on the HEMT characteristics than that on the drain side, so that much less of a change is required between source and gate to cause appreciable electrical degradation.
H. The role of forward gate current Experiments comparing AlGaN/GaN HEMTs with devices identical save for the existence of a thin gate oxide layer (MOSHEMTS) showed much better stability of the latter upon RF stressing [63], [68]. Since the HEMT and MOSHEMT are identical in all respects but the latter’s gate oxide, and due to the correlation between HEMT RF power degradation and forward gate current increase, the authors conclude that this forward gate current must be the primary cause of device degradation during RF stress.
A circuit model of current collapse based on this sourcegate surface degradation effect is proposed in [56].
47
Forward gate current was also identified as the main concern in a low-noise amplifier reliability investigation [69].
applications include lighting—which does not directly involve HEMTs, but nevertheless requires development of good crystals and reliable metallizations from which HEMT technology can benefit—wireless telecommunication infrastructures, and relatively high-power applications such as automotive converters.
I.
The inverse piezoelectric effect GaN and AlGaN being strongly piezoelectric materials, large electric fields like those applied between gate and drain can be expected to translate into modified strain configuration in the layer structure.
ACKNOWLEDGMENT I am grateful to the ICEM/COMMAD organizing committee for inviting me to present this review, and to Prof. Lorenzo Faraone and the Microelectronics Research Group of the University of Western Australia for hosting and supporting my leave.
A study of gated transmission-line-model structures [70], for example, explained the observed degradation of source and gate resistance following gate pulse stress in terms of modified strain configuration due to the gate field increasing the tensile strain under the gate and decreasing it on the gate sides, thereby lowering the piezoelectric charge and increasing the parasitic resistances.
REFERENCES
In another paper [71], micro-Raman spectroscopy was used to map the strain distribution on the surface of an AlGaN/GaN HEMT; it was found that very large strain exists in the gatedrain surface area, and the strain increases with drain-gate bias. By means of 2D numerical simulations, the strain was found to correlate strikingly with the vertical electric field at the gate edge facing the drain.
[1]
[2] [3] [4] [5] [6] [7] [8] [9] [10]
Consistent with this picture, it was speculated in [72], [73] that HEMT degradation during electrical stress can be due to defect formation through relaxation of this inverse piezoelectric strain; this hypothesis is supported by the observation of a sharp threshold voltage for device degradation, which the authors attribute to some threshold energy for strain relaxation and, therefore, defect formation. V.
[11]
CONCLUSIONS
There is an apparent contradiction between the commercial availability of qualified GaN-based HEMTs and some lingering lack of understanding, uniformity and consistency of reliability data. This contradiction is to some extent reconciled if we realize that the reliability performance achieved by mature technologies is a product of several factors, including: (i) good understanding of wear-out mechanisms and consequently (ii) design of suitable accelerating lifetesting and burn-in procedures for wear-out immunity assurance; (iii) large-volume production and economy of scale, granting meaningful statistics and strenuous attention to production yield; (iv) significant diffusion of technology and deployment in the field—a scenario, characterized by what we could call “the global reliability engineer”, where circuit and system designers and end users (with their field returns data) are, perhaps unwittingly, as much a part of the reliability improvement process as device designers, process and reliability engineers on the vendor side. If this picture is correct, it is obvious that over the last few years GaN-based HEMT technology has made remarkable progress as far as factors (i) and (ii) are concerned, but it is necessarily still in its early infancy if we look at factors (iii) and (iv).
[12]
[13]
[14]
[15]
[16]
[17]
Overall, I believe the GaN community can look at the near future with sound optimism, thanks to present and foreseeable applications that should provide the necessary leverage for achieving the stage characterized by factors (iii) and (iv). These
[18]
48
U. K. Mishra, L. Shen, T. E. Kazior, and Y.-F. Wu, "GaN-based RF power devices and amplifiers," Proc. IEEE, vol. 26, pp. 287-305, Feb. 2008. www.nitronex.com/ www.rfmd.com/adocs/bf/GaN_bro.pdf www.eudyna.com/e/products_e/newproducts_e/gan_hemt.html www.hrl.com/pages/mel.html triquint.com/prodserv/new/index.cfm www.cree.com/products/wireless_products.asp#gan www3.toshiba.co.jp/snis/ovs/des/micro/technical/tech20071009.htm www.nitronex.com/reliability.html S. Singhal, T. Li, A. Chaudhari, A. W. Hanson, R. Therrien, J. W. Johnson, W. Nagy, J. Marquart, P. Rajagopal, E. L. Piner, K. J. Linthicum, "Reliability of large periphery GaN-on-Si HFETs," Proc. Reliab. Compound Semicond. Workshop (ROCS), pp. 135-149, 2005. S. Singhal, J. C. Roberts, P. Rajagopal, T. Li, A. W. Hanson, R. Therrien, J. W. Johnson, I. C. Kizilyalli, K. J. Linthicum, "GaN-on-Si failure mechanisms and reliability improvement," Proc. 44th Annual Int. Reliab. Phys. Symp. (IRPS), pp. 95-98, 2006. S. Singhal, A. Chaudhari, A. W. Hanson, J. W. Johnson, R. Therrien, P. Rajagopal, T. Li, C. Park, A. P. Edwards, E. L. Piner, I. C. Kizilyalli, K. J. Linthicum, "GaN-on-Si reliability: a comparative study between process platforms," Proc. Reliab. Compound Semicond. Workshop (ROCS), pp. 21-24, 2006. A. M. Conway, M. Chen, P. Hashimoto, P. J. Willadsen, and M. Micovic, "Accelerated RF life testing of GaN HFETs," Proc. 45th Annual Int. Reliab. Phys. Symp. (IRPS), pp. 472-475, 2007. R. Coffie, Y. Chen, I. P. Smorchkova, B. Heying, V. Gambin, W. Sutton, Y.-C. Chou, W.-B. Luo, M. Wojtowicz, and A. Oki, "Temperature and voltage dependent RF degradation study in AlGaN/GaN HEMTs," Proc. 45th Annual Int. Reliab. Phys. Symp. (IRPS), pp. 568-569, 2007. A. Sozza, C. Dua, E. Morvan, B. Grimbert, and S. L. Delage, "A 3000 hours DC life test on AlGaN/GaN HEMT for RF and microwave applications," Microelectron. Reliab., vol. 45, pp. 1617-1621, 2005. G. Sonia, E. Richter, F. Brunner, A. Denker, R. Lossy, M. Mai, F. Lenk, J. Bundesmann, G. Pensl, J. Schmidt, U. Zeimer, L. Wang, K. Baskar, M. Weyers, J. Würfl, G. Tränkle, “2 MeV ion irradiation effects on AlGaN/GaN HFET devices,” Soli-State Electron., vol. 52, pp. 10111017, 2008. G. A. Umana-Membreno, J. M. Dell, G. Parish, B. D. Nener, L. Faraone, U. K. Mishra, “Total-dose radiation-induced effects in AlGaN/GaN high electron mobility transistors,” Int. Conf. Electronic Materials (ICEM 2008), Sydney, Australia, 28th July – 1st Aug. 2008, paper A2-S2.2. J. Würfl, V. Abrosimova, J. Hilsenbeck, E. Nebauer, W. Rieger, G. Tränkle, "Reliability considerations of III-nitride microelectronic devices," Microelectron. Reliab., vol. 39, pp. 1737-1757, 1999.
[19] I. Daumiller, C. Kirchner, M. Kamp, K. J. Ebeling, and E. Kohn, "Evaluation of the temperature stability of AlGaN/GaN heterostructure FETs," IEEE El. Dev. Lett., vol. 20, pp. 448-450, 1999. [20] J. Würfl, J. Hilsenbeck, E. Nebauer, G. Tränkle, and H. Obloh, "Technology and thermal stability of AlGaN/GaN HFETs," Proc. GaAs Applications Symp. (GAAS), pp. 430-434, 1999. [21] J. Hilsenbeck, E. Nebauer, J. Würfl, G. Tränkle, and H. Obloh, "Aging behavious of AlGaN/GaN HFETs with advanced ohmic and Schottky contacts," El. Lett., vol. 36, pp. 980-981, 2000. [22] J. Würfl, J. Hilsenbeck, E. Nebauer, G. Tränkle, and H. Obloh, W. Österle, "Reliability of AlGaNGaN HFETs comprising refractory ohmic and Schottky contacts," Microelectron. Reliab., vol. 40, pp. 1689-1693, 2000. [23] A. Sozza, C. Dua, E. Morvan, B. Grimbert, M. A. di Forte-Poisson, S. L. Delage, and E. Zanoni, "Thermal stability of Mo-based Schottky contact for AlGaN/GaN HEMT," El. Lett., vol. 41, no. 16, pp. 61-62, 2005. [24] R. Khanna, L. Stafford, L. F. Voss, S. J. Pearton, H. T. Tang, T. Anderson, S.-C. Hung, and F. Ren, "Aging and stability of GaN high electron mobility transistors and light-emitting diodes with TiB2 - and Irbased contacts," IEEE Trans. Dev. Mat. Reliab., vol. 8, pp. 272-276, 2008. [25] Y. C. Chou, D. Leung, I. Smorchkova, M. Wojtowicz, R. Grundbacher, L. Callejo, Q. Kan, R. Lai, P. H. Liu, D. Eng, A. Oki, "Degradation of AlGaN/GaN HEMTs under elevated temperature lifetesting," Microelectron. Reliab., vol. 44, pp. 1033-1038, 2004. [26] L. Wang, F. M. Mohammed, and I. Adesida, "Formation mechanism of ohmic contacts on AlGaN/GaN heterostructure: electrical and microstructural characterizations," J. Appl. Phys., vol. 103, 093516, 2008. [27] C. Nguyen, N. X. Nguyen, and D. E. Grider, "Drain current compression in GaN MODFETs under large-signal modulation at microwave frequencies," El. Lett., vol. 35, pp. 1380-1382, 1999. [28] S. C. Binari, P. B. Klein, and T. E. Kazior, "Trapping effects in GaN and SiC microwave FETs," Proc. IEEE, vol. 90, pp. 1048-1058, 2002. [29] K. S. Boutros, P. Rowell, and B. Brar, "A study of output power stability of GaN HEMTs on SiC substrates," Proc. 42nd Annual Int. Reliab. Phys. Symp. (IRPS), pp. 577-578, 2004. [30] D. C. Dumka, C. Lee, H. Q. Tserng and P. Saunier, "RF reliability performance of AlGaN/GaN HEMTs on Si substrates at 10 GHz," El. Lett., vol. 40, no. 24, 2004. [31] D. K. Sahoo, R. K. Lal, H. Kim, V. Tilak, and L. F. Eastman, "Highfield effects in silicon nitride passivated GaN MODFETs," IEEE Trans. El. Dev., vol. 50, pp. 1163-1170, 2003. [32] J. A. Mittereder, S. C. Binari, P. B. Klein, J. A. Roussos, D. S. Katzer, D. F. Storm, D. D. Koleske, A. E. Wickenden, and R. L. Henry, "Current collapse induced in AlGaN/GaN HEMTs by short-term DC bias stress," 41st Annual Int. Reliab. Phys. Symp. (IRPS), pp. 320-323, 2003. [33] M. Faqir, A. Chini, G. Verzellesi, F. Fantini, F. Rampazzo, G. Meneghesso, E. Zanoni, J. Bernat, P. Kordos, "Physical investigation of high-field degradation mechanisms in GaN/AlGaN/GaN HEMTs," Proc. Reliab. Compound Semicond. Workshop (ROCS), pp. 25-28, 2006. [34] G. Meneghesso, F. Rampazzo, P. Kordos, G. Verzellesi, and E. Zanoni, "Current collapse and high-electric field reliability of unpassivated GaN/AlGaN/GaN HEMTs," IEEE Trans. El. Dev., vol. 53, pp. 29322941, 2006. [35] G. Meneghesso, G. Verzellesi, F. Danesin, F. Rampazzo, F. Zanon, A. Tazzoli, M. Meneghini, and E. Zanoni, "Reliability of GaN highelectron-mobility transistors: state of the art and perspectives," IEEE Trans. Dev. Mat. Reliab., vol. 8, pp. 332-343, 2008. [36] G. Meneghesso, G. Verzellesi, R. Pierobon, F. Rampazzo, A. Chini, U. K. Mishra, C. Canali, and E. Zanoni, "Surface-related drain current dispersion effects in AlGaN-GaN HEMTs," IEEE Trans. El. Dev., vol. 51, pp. 1554-1560, 2004. [37] H. Kim, A. Vertiatchikh, R. M. Thompson, V. Tilak, T. R. Prunty, J. R. Shealy, L. F. Eastman, "Hot electron induced degradation of undoped AlGaN/GaN HFETs," Microelectron. Reliab., vol. 43, pp. 823-827, 2003.
[38] M. Faqir, G. Verzellesi, A. Chini, F. Fantini, F. Danesin, G. Meneghesso, E. Zanoni, and C. Dua, "Mechanisms of RF current collapse in AlGaN-GaN high electron mobility transistors," IEEE Trans. Dev. Mat. Reliab., vol. 8, pp. 240-247, 2008. [39] P. Valizadeh, "High-temperature very low frequency noise-based investigation of slow transients in AlGaN/GaN MODFETs," IEEE Trans. Dev. Mat. Reliab., vol. 8, pp. 265-271, 2008. [40] T. Mizutani, Y. Ohno, M. Akita, S. Kishimoto, and K. Maezawa, "A study on current collapse in AlGaN/GaN HEMTs induced by bias stress," IEEE Trans. El. Dev., vol. 50, pp. 2015-2020, 2003. [41] R. Vetury, N. Q. Zhang, S. Keller, and U. K. Mishra, "The impact of surfacte states on the DC and RF characteristics of AlGaN/GaN HFETs," IEEE Trans. El. Dev., vol. 48, pp. 560-566, 2001. [42] G. Koely, V. Tilak, L. F. Eastman, and M. G. Spencer, "Slow transients observed in AlGaN/GaN HFETs: effects of SiNx passivation and UV illumination," IEEE Trans. El. Dev., vol. 50, pp. 886-893, 2003. [43] [valizadeh06] P. Valizadeh and D. Pavlidis, "Low-frequency noisebased degradation prediction of AlxGa1-xN/GaN MODFETs," IEEE Trans. Dev. Mat. Reliab., vol. 3, pp. 479-485, 2006. [44] A. Sozza, A. Curutchet, C. Dua, N. Malbert, N. Labat and A. Touboul, "AlGaN/GaN HEMT reliability assessment by means of low frequency noise measurements," Microelectron. Reliab., vol. 46, pp. 1725-1730, 2006. [45] B. M. Green, K. K. Chu, E. M. Chumbes, J. A. Smart, J. R. Shealy, and L. F. Eastman, "The effect of surface passivation on the microwave characteristics of undoped AlGaN/GaN HEMTs," IEEE El. Dev. Lett., vol. 21, pp. 268-270, 2000. [46] H. Kim, R. M. Thompson, V. Tilak, T. R. Prutny, J. R. Shealy, and L. F. Eastman, "Effect of SiN passivation and high-electric field on AlGaNGaN HFET degradation," IEEE El. Dev. Lett., vol. 24, pp. 421-423, 2003. [47] A. V. Vertiatchikh, L. F. Eastman, W. J. Schaff and T. Prutny, "Effect of surface passivation of AlGaN/GaN heterostructure field-effect transistor," El. Lett., vol. 38, pp. 388-389, 2002. [48] B. Luo, R. M. Mehandru, J. Kim, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, S. J. Pearton, R. C. Fitch, J. Gillespie, R. Dellmer, T. Jenkins, J. Sewelll, D. Via, A. Crespo, "The role of cleaning conditions and epitaxial layer structure on reliability of Sc2O3 and MgO passivation on AlGaN/GaN HEMTs," Solid-State Electron., vol. 46, pp. 2185-2190, 2002. [49] T. Kunii, M. Totsuka, Y. Kamo, Y. Yamamoto, H. Takeuchi, Y. Shimada, T. Shiga, H. Minami, T. Kitano, S. Miyakuni, S. Nakatsuka, A. Inoue, T. Oku, T. Nanjo, T. Oishi, T. Ishikawa, and Y. Matsuda, "A high reliability GaN HEMT with SiN passivation by cat-CVD," Proc. IEEE Comp. Semicond. IC Symp. (CSICS), 2004. [50] A. P. Edwards, J. A. Mittereder, S. C. Binari, D. S. Katzer, D. F. Storm, and J. A. Roussos, "Improved reliability of AlGaN-GaN HEMTs using a NH3 plasma treatment prior to SiN passivation," IEEE El. Dev. Lett., vol. 26, pp. 225-227, 2005. [51] D. J. Meyer, J. R. Flemish, and J. M. Redwing, "Prepassivation surface treatment effects on pulsed and dc I-V performance of AlGaN/GaN high-electron-mobility transistors," Appl. Phys. Lett., vol. 92, 193505, 2008. [52] D. W. DiSanto, H. F. Sun and C. R. Bolognesi, "Ozone passivation of slow transient current collapse in AlGaN/GaN field-effect transistors: the role of threading dislocations and the passivation mechamism," Appl. Phys. Lett., vol. 88, 013504, 2006. [53] D. W. DiSanto and C. R. Bolognesi, "Effect of gate-source access region stress on current collapse in AlGaN/GaN HFETs," El. Lett., vol. 41, no. 8, 2005. [54] H. F. Sun and C. R. Bolognesi, "Impact of selective Al2O3 passivation on current collapse in AlGaN/GaN HEMTs," El. Lett., vol. 43, no. 23, 2007. [55] S. Arulkumaran, G. I. Ng, Z. H. Liu, "Effect of gate-source and gatedrain Si3N 4 passivation on current collapse in AlGaN/GaN high-electron mobility transistors on silicon," Appl. Phys. Lett., vol. 90, 173504, 2007. [56] G. L. Bilbro and R. J. Trew, "RF knee walkout and source access region of unpassivated HFETs," El. Lett., vol. 42, no. 24, 2006.
49
[57] N.-Q. Zhang, S. Keller, G. Parish, S. Heikman, S. P. DenBaars, and U. K. Mishra, "High breakdown GaN HEMT with overlapping gate structure," IEEE El. Dev. Lett., vol. 21, pp. 421-423, 2000. [58] J. Li, S. J. Cai, G. Z. Pan, Y. L. Chen, C. P. Wen and K. L. Wang, "High breakdown voltage GaN HFET with field plate," El. Lett., vol. 37, pp. 196-197, 2001. [59] S. Karmalkar and U. K. Mishra, "Enhancement of breakdown voltage in AlGaN/GaN high electron mobility transistors using a field plate," IEEE Trans. El. Dev., vol. 48, pp. 1515-1521, 2001. [60] C. Lee, H. Tserng, L. Witkowski, P. Saunier, S. Guo, B. Albert, R. Birkhahn and G. Munns, "Effects of RF stress on power and pulsed IV characteristics of AlGaN/GaN HEMTs with field-plate gates," El. Lett., vol. 40, no. 24, 2004. [61] V. Adivarahan, J. Yang, A. Koudymov, G. Simin, and M. Asif Khan, "Stable CW operation of field-plated GaN-AlGaN MOSHFETs at 19 W/mm," IEEE El. Dev. Lett., vol. 26, pp. 535-537, 2005. [62] A. Koudymov, V. Adivarahan, J. Yang, G. Simin, and M. Asif Khan, "Mechanism of current collapse removal in field-plated nitride HFETs," IEEE El. Dev. Lett., vol. 26, pp. 704-706, 2005. [63] G. Simin, V. Adivarahan, J. Yang, A. Koudymov, S. Rai and M. Asif Khan, “Stable 20W/mm AlGaN-GaN MOSHFET,” El. Lett., vol. 41, no. 13, 2005. [64] T. Kikkawa, T. Maniwa, H. Hayashi, M. Kanamura, S. Yokokawa, M. Nishi, N. Adachi, M. Yokoyama, Y. Tateno, and K. Joshin, "An over 200-W output power GaN HEMT push-pull amplifier with high reliability," IEEE MTT-S Dig., pp. 1347-1350, 2004. [65] C. Lee, L. Witkowski, H.-Q. Tserng, P. Saunier, R. Birkhahn, D. Olson, D. Olson, G. Munns, S. Guo and B. Albert, "Effects of AlGaN/GaN HEMT structure on RF reliability," El. Lett., vol. 41, no. 3, 2005.
[66] N. Braga, R. Mickevicius, R. Gaska, M. Shur, M. Asif Khan, G. Simin, "Simulation of gate lag and current collapse in GaN heterojunction field effect transistors," Proc. IEEE Comp. Semicond. IC Symp. (CSICS), 2004. [67] M. Faqir, G. Verzellesi, G. Meneghesso, E. Zanoni, and F. Fantini, "Investigation of high-electric-field degradation effects in AlGaN/GaN HEMTs," IEEE Trans. El. Dev., vol. 55, pp. 1592-1602, 2008. [68] A. Koudymov, C. X. Wang, V. Adivarahan, J. Yang, G. Simin, and M. Asif Khan, “Power stability of AlGaN/GaN HFETs at 20 W/mm in the pinched-off operation mode,” IEEE El. Dev. Lett., vol. 28, pp. 5-7, 2007. [69] M. Rudolph, R. Behtash, R. Doerner, K. Hirche, J. Würfl, W. Heinrich, and G. Tränkle,” Analysis of the survivability of GaN low-noise amplifiers,” IEEE Trans. Microw. Th. Tech., vol. 55, pp. 37-43, 2007. [70] G. Simin, A. Koudymov, A. Tarakji, X. Hu, J. Yang, M. Asif Khan, M. S. Shur and R. Gaska, “Induced strain mechanism of current collapse in AlGaN/GaN heterostructure field-effect transistors,” Appl. Phys. Lett., vol. 79, pp. 2651-2653, 2001. [71] A. Sarua, H. Ji, M. Kuball, M. J. Uren, T. Martin, K. J. Nash, K. P. Hilton, and R. S. Balmer, “Piezoelectric strain in AlGaN/GaN heterostructure field-effect transistors,” Appl. Phys. Lett., vol. 88, 103502, 2006. [72] J. Joh and J. A. del Alamo, “Mechanisms for electrical degradation of GaN high-electron mobility transistors,” IEEE IEDM Tech. Dig., DOI 10.1109/IEDM.2006.346799, 2006. [73] J. Joh and J. A. del Alamo, “Critical voltage for electrical degradation of GaN high-electron mobility transistors,” IEEE El. Dev. Lett., vol. 29, pp. 287-289, 2008.
50
