⇦Chinese Physics Letters, 2020, Vol. 37, No. 4, Article code 046101 Effects of Total-Ionizing-Dose Irradiation on Single-Event Burnout for Commercial Enhancement-Mode AlGaN/GaN High-Electron Mobility Transistors * Si-Yuan Chen (陈思远)1,2,3, Xin Yu (于新)1,2, Wu Lu (陆妩)1,2**, Shuai Yao (姚帅)1,2,3, Xiao-Long Li (李小龙)1,2,3, Xin Wang (王信)1,2, Mo-Han Liu (刘默寒)1,2, Shan-Xue Xi (席善学)1,2,3, Li-Bin Wang (王利斌)1,2, Jing Sun (孙静)1,2, Cheng-Fa He (何承发)1,2, Qi Guo (郭旗)1,2 Affiliations 1Key Laboratory of Functional Materials and Devices for Special Environments, Xinjiang Technical Institute of Physics and Chemistry, Chinese Academy of Sciences, Urumqi 830011 2Xinjiang Key Laboratory of Electronic Information Material and Device, Urumqi 830011 3University of Chinese Academy of Sciences, Beijing 100049 Received 17 November 2019, online 24 March 2020 *Supported by the National Natural Science Foundation of China under Grant Nos. U1532261, U1630141, and 61534008, and the West Light Foundation of Chinese Academy of Sciences under Grant No. 2018-XBQNXZ-B-003.
**Corresponding author. Email: luwu@ms.xjb.ac.cn Citation Text: Chen S Y, Yu X, Lu W, Yao S and Li X L et al 2020 Chin. Phys. Lett. 37 046101    Abstract We investigate the synergism effect of total ionizing dose (TID) on single-event burnout (SEB) for commercial enhancement-mode AlGaN/GaN high-electron mobility transistors. Our experimental results show that the slight degradation of devices caused by gamma rays can affect the stability of the devices during the impact of high energy particles. During heavy ion irradiation, the safe working values of drain voltage are significantly reduced for devices which have already been irradiated by $^{60}$Co gamma rays before. This could be attributed to more charges trapped caused by $^{60}$Co gamma rays, which make GaN devices more vulnerable to SEB. Moreover, the electrical parameters of GaN devices after $^{60}$Co gamma and heavy-ion irradiations are presented, such as the output characteristic curve, effective threshold voltages, and leakage current of drain. These results demonstrate that the synergistic effect of TID on SEB for GaN power devices does in fact exist. DOI:10.1088/0256-307X/37/4/046101 PACS:61.80.-x, 85.30.Tv, 84.30.Jc © 2020 Chinese Physics Society Article Text With the pursuit of high performance for semiconductor devices, there is growing interest in AlGaN/GaN high-electron mobility transistors (HEMTs), which have excellent properties including fast switching, high power system efficiency, and small total size.[1,2] GaN power devices are currently offering a viable solution for replacing Si-based devices deployed in harsh environments such as space, in addition to high-reliability military and nuclear applications. However, GaN power devices applied in harsh environments suffer from the bombardment of particles, resulting in performance degradation, which can even lead to premature failure. There are several degradation mechanisms, include total ionizing dose (TID), single-event effects (SEEs), and others that exist simultaneously. Most of the research so far has focused on TID and SEEs with heavy ions for GaN power devices. For the effects of TID, the research at the University of Florida suggested that significant measurable changes of DC parameters of GaN devices were observed for doses up to 300 Mrad (Si).[3] The reason for this phenomenon is that the wide bandgap GaN devices without gate oxide have excellent radiation hardness characteristics for TID,[4] whereas stresses from charge collection or ion damage make GaN power devices sensitive to SEEs. The Japan Aerospace Exploration Agency found two single-event burnout (SEB) catastrophic failure modes: one is the drain-substrate short mode without burnout signature, and the other is the drain-source short mode with the chip surface blown.[5] The experimental results of the California Institute of Technology indicate that SEB may be attributed to the local gate area damage. An ion strike to this area may damage the gate structure such that the device is turned-on for the local gate area.[6] However, the synergistic effect of TID on SEB for GaN power devices has not been reported. The mechanism of the synergism effect is attributed to the degradation of Si-based devices caused by gamma rays, since the degradation can severely affect the single-event vulnerability of devices.[7] This raises the question whether TID could aggravate the occurrence of SEB for commercial enhancement-mode AlGaN/GaN HEMTs without gate oxide. Therefore, it is imperative to fully study and understand the synergistic effect of TID on SEB for GaN power devices. In this Letter, to obtain the TID influences on SEB in harsh environments, the performance of GaN power devices, which have already been degraded with increasing gamma ray doses and subsequently irradiated with heavy ions, is first examined. The experimental results mainly include two parts: one is the change of the online off-state leakage current of drain, and the other part includes the offline output characteristic, leakage current $I_{\rm dss}$, and effective threshold voltages $V_{\rm th}$. The failure mechanisms are also briefly discussed. The normal-off GaN devices, epc2037, used in this study were obtained from the Efficient Power Conversion Inc. Figures 1(a) and 1(b) show the detail of the device. The cross section of GaN HEMTs was scanned with a scanning electron microscope as illustrated in Fig. 1(c). A thin AlN layer was grown on a Si substrate to alleviate the lattice mismatch between the Si substrate and the GaN layer. In addition, the GaN and AlGaN layers generated the important two-dimensional electron gas (2DEG) structure with an abundance of electrons due to the piezoelectric and polarization effects. The thickness of the chip was around 700 µm, but the active layer investigated was only 8–11 µm below the chip surface. Ta ions [range: 110.15 µm (Si), energy: 1892 MeV, LET: 76.46 MeV/(mg/cm$^{2}$)] can easily travel through the device's passivation layer to its active region.
cpl-37-4-046101-fig1.png
Fig. 1. EPCs' GaN power transistor: (a) package, (b) schematic, (c) cross section.
cpl-37-4-046101-fig2.png
Fig. 2. Test condition of the SEE. The terminals of gate and drain were monitored and biased by Keithley 2636B.
The devices studied in this work were divided into two groups A and B. The experimental steps were as follows: (1) The group A was exposed to $^{60}$Co gamma rays with dosage rates of 50 rad (Si)/s up to 800 krad (Si)/s, and all pins of the device were grounded. (2) The gamma-irradiated group A and the fresh group B were subjected to irradiation by Ta ions in the heavy ion research facility at Lanzhou. As illustrated in Fig. 2, the Keithley 2636B device was interfaced to the PC running labview software using USB interfaces during the heavy ions irradiation, for both sending commands to bias the gate, drain, and source terminal of the GaN devices, and receiving information. To obtain the safe operating area of drain voltage for the GaN devices in the off bias (the worst case[5]), the drain voltage was raised from 20 V to 100 V in steps of 10 V or 20 V, and the fixed gate bias of $V_{\rm gs} = 0$ V was applied. For each step, around $5 \times 10^{5}$ cm$^{-2}$ of fluence was irradiated. The online test parameter was the drain-source current $I_{\rm ds}$. The BC3193 was used to carry out electrical parameter tests after the $^{60}$Co irradiation and heavy ion irradiation. All the experiments and tests were performed at room temperature.
cpl-37-4-046101-fig3.png
Fig. 3. The magnitude of variation in $I_{\rm ds}$ as a function of Ta ion [LET: 76.46 MeV/(mg/cm$^{2}$)] fluence with increasing drain voltage: (a) device B without TID, (b) device A with 800 krad (Si) TID.
cpl-37-4-046101-fig4.png
Fig. 4. The enlarged area outlined in red in Fig. 3. Here (a), (c), (e) show the changes in $I_{\rm ds}$ of the fresh device B for different values of $V_{\rm ds}$ during heavy ions irradiation; (b), (d), (f) show the changes in $I_{\rm ds}$ of the irradiated device A for different values of $V_{\rm ds}$ during heavy ions irradiation.
Figure 3 denotes the changes of $I_{\rm ds}$ for increasing voltages during heavy ions irradiation. In order to see clearly the change of $I_{\rm ds}$ under low voltages, Fig. 4 shows the details of the area outlined in red in Fig. 3. As shown in Fig. 4, the device with TID irradiation has higher leakage currents of $I_{\rm ds}$ than the fresh device at all drain voltages. In the case of $V_{\rm ds} = 40$ V, a significant SEB phenomenon for device A appears. After the SEB phenomenon, the $I_{\rm ds}$ of device A at approximately 30 nA is considerably higher than that of device B at 1 nA. Device B does not appear to exhibit the SEB phenomenon until $V_{\rm ds}$ reaches 60 V. This shows that the safe working area of drain voltage is significantly reduced for device A. In other words, the coupling of TID and SEE jointly causes degradation of the device parameters. It must be mentioned that a step-progressive increase of $I_{\rm ds}$ is depicted in Fig. 3(b), whereas a few relatively narrow pulses of $I_{\rm ds}$ are shown in Fig. 3(a). The degradation mechanisms of device A differ markedly from device B during irradiation. This shows that the synergism effect of TID on SEB does exist.
cpl-37-4-046101-fig5.png
Fig. 5. The leakage current $I_{\rm dss}~(V_{\rm ds} = 80$ V, $V_{\rm gs} = 0$ V) with irradiation for devices A and B. The area in red marks the values of $I_{\rm dss}$ after SEE.
cpl-37-4-046101-fig6.png
Fig. 6. Output characteristics measured under the initial $V_{\rm gs} = 2$ V (lower curve) with the increment of 1 V.
As shown in Fig. 5, device A's leakage current increases slowly after the gamma irradiation, which is consistent with Fig. 4. More importantly, the increment of $I_{\rm dss}$ is significant for device A after heavy ion irradiation, while $I_{\rm dss}$ decreases slightly for device B. Figure 6 illustrates the changes of output characteristics as irradiation continues. In the case of $V_{\rm gs} = 4$ V and $V_{\rm ds} = 3$ V, the drain current firstly declines by 220 mA at a dose of 100 krad (Si), before increasing sharply by 260 mA at a dose of 800 krad (Si), and finally slightly dropping by 20 mA after heavy ions irradiation. Furthermore, there is also a change in the slope of the output characteristics curve, which indicates a change in carrier mobility or carrier density. In Fig. 7, starting with a dramatic surge, the threshold voltage reaches its peak from 1.353 V to 1.517 V, followed by a significant decline back to 1.304 V at a dose of 500 krad (Si). After that, the threshold voltage is much more steady, albeit with a slight fluctuation. It would therefore appear that there are multiple damage mechanisms causing simultaneous degradation of device performance, which is similar to the competition between oxide trapped charges and the interface trap for Si-based devices.
cpl-37-4-046101-fig7.png
Fig. 7. Curves of threshold voltage versus irradiation dose amounts.
In GaN devices the pit/crack shaped defects are more easily formed for stressed devices in the top AlGaN layer under the drain-side edge of the gate.[8,9] When a particle penetrates this sensitive area, the electron-hole pairs are produced along its track. These charges may be accumulated in the gate region such that the drain source is turned on, which leads to an increase in drain current. Even, small line-shaped crystal defects introduced by incident ions in the substrate and buffer layers can form leakage paths.[5,10] Therefore, in the case of high voltage and high leakage current, the device burns out. Gamma rays can influence the crystallinity of GaN film[11] and lead to the creation of traps on the AlGaN surface.[12,13] You can easily find the areas in Fig. 1. These defects introduced by gamma rays get worse following heavy ion irradiation, which easily forms leakage paths, resulting in a lower safe operating area of drain voltage for the GaN devices. Therefore, as the drain-source voltage increases, lines of leakage paths will be easily induced for device A by heavy ions and the electric stress, causing the leakage current to rise continuously. However, compared to device A, device B cannot easily form a large number of leakage paths. More charges are accumulated on the gate region such that the drain source is turned on, which causes a leakage current spike. The burn signs of the two devices after irradiation are inspected. For the AlGaN/GaN HEMT, the basic formula of $I_{\rm ds}$ is given as[14] $$ I_{\rm ds} =qw\mu E n_{\rm 2DEG},~~~ E < E_{\rm c},~~ \tag {1} $$ where $q$ is the electron charge, $w$ is the width of the gate area, $\mu$ is the electron mobility, $E$ is the electric field and $n_{\rm 2DEG}$ is the concentration of two-dimensional electron gas (2DEG). The concentration of 2DEG is obtained from[15] $$ n_{\rm 2DEG}= \frac{\varepsilon }{qd}\Big({V_{\rm gs}-V_{\rm th}-\frac{E_{\rm F}}{q}} \Big),~~ \tag {2} $$ where $\varepsilon$ is the dielectric constant of AlGaN, $d$ is the doped AlGaN layer thickness, $V_{\rm gs}$ is the applied gate-source voltage, $V_{\rm th}$ is the threshold voltage, $E_{\rm F}$ is the Fermi energy, and $q$ is the electron charge. The slope of the output characteristics curve can be expressed as $$ \frac{dI_{\rm ds}}{dV_{\rm ds}}=\frac{d}{dV_{\rm ds}}qw\mu En_{\rm 2DEG},~~ \tag {3} $$ From formula (1)-(3), both the drain current and the slope of the output characteristic curve show a positive correlation with the carrier concentration and carrier mobility. There is a negative correlation between the threshold voltage and drain current, which is consistent with the observations from Figs. 6 and 7. All the irradiated devices show an increase or decrease in $V_{\rm th}$ corresponding to a decrease or increase in the drain-source current, respectively. Therefore, we only need to determine why $I_{\rm ds}$ changes. The increase of $I_{\rm ds}$ is due to electron emission from defects. Conversely, defects are filled by abundant electrons, leading to a decrease of $I_{\rm ds}$. To sum up, studying how $I_{\rm ds}$ changes due to electrons and holes trapped or released by deep level traps can be a useful means of measuring the influence of defects induced by irradiation. At low gamma irradiation doses (below 100 krad), deep level traps are introduced, which causes a reduction in the saturation current and a positive shift in the threshold voltage. It gives rise to increased trapping of carriers and dispersion of charge, so that the concentration of 2DEG is decreased.[16,17] However, the saturation current is increased at the higher doses. Some researchers have reported that the saturation currents can be enhanced due to nitrogen vacancies and the relaxation of strain after gamma irradiation.[18,19] Therefore, it can be speculated that nitrogen vacancies acting as donors and the relaxation of strain dominate over the trapping of carriers. After heavy-ion irradiation, there is a decreased saturation current that is the result of reduced electron mobility and density.[20] Although some conclusions are given here, the evolution mechanism of defects and lattice stress during gamma and heavy ion irradiation remains to be further analyzed. In summary, TID and subsequent heavy-ion irradiation of enhancement-mode AlGaN/GaN HEMTs have been performed with the aim of studying the synergism effect of TID on SEB. As the irradiation proceeded, radiation damage causes the degradation of the device parameters such as output characteristic curves, threshold voltage, and leakage current. These observations show that the synergistic effect of TID on SEB for GaN power devices does exist. Furthermore, TID significantly reduces the safe working area of drain voltage of GaN devices during heavy ions irradiation, which is a great safety hazard for the power supply system. This also poses a challenge for new power devices used in extreme radiation environments and thus the radiation damage of GaN devices needs to be further evaluated. References AlGaN/GaN HEMTs on Silicon Substrate With 206-GHz $F_{ \rm MAX}$High breakdown voltage AlGaN-GaN power-HEMT design and high current density switching behaviorInfluence of 60Co γ-rays on dc performance of AlGaN/GaN high electron mobility transistors60Co gamma radiation effects on DC, RF, and pulsed I–V characteristics of AlGaN/GaN HEMTsSingle-Event Damage Observed in GaN-on-Si HEMTs for Power Control ApplicationsDetermination of Single-Event Effect Application Requirements for Enhancement Mode Gallium Nitride HEMTs for Use in Power Distribution CircuitsThe Increased Single-Event Upset Sensitivity of 65-nm DICE SRAM Induced by Total Ionizing DoseTEM Observation of Crack- and Pit-Shaped Defects in Electrically Degraded GaN HEMTsA model for the critical voltage for electrical degradation of GaN high electron mobility transistorsSingle-Event Damages Caused by Heavy Ions Observed in AlGaN/GaN HEMTsEffects of gamma-ray irradiation on the microstructural and luminescent properties of radio-frequency magnetron-sputtered GaN thin filmsThe effects of 60 Co γ-ray irradiation on the DC characteristics of enhancement-mode AlGaN/GaN high-electron-mobility transistorsImpact of low gamma radiation dose on electrical trap related effects in AlGaN/GaN HEMTsInfluence of heavy ion irradiation on DC and gate-lag performance of AlGaN/GaN HEMTsAn accurate charge control model for spontaneous and piezoelectric polarization dependent two-dimensional electron gas sheet charge density of lattice-mismatched AlGaN/GaN HEMTsLow and moderate dose gamma-irradiation and annealing impact on electronic and electrical properties of AlGaN/GaN high electron mobility transistorsGamma irradiation impact on electronic carrier transport in AlGaN/GaN high electron mobility transistorsElectrical characterization of 60 Co gamma radiation-exposed InAlN/GaN high electron mobility transistorsEffects ofγ-irradiation on AlGaN/GaN-based HEMTsDegradation mechanisms of AlGaN/GaN HEMTs under 800 MeV Bi ions irradiation
[1] Bouzid-Driad S et al 2013 IEEE Electron Device Lett. 34 36
[2] Saito W et al 2003 IEEE Trans. Electron Devices 50 2528
[3] Luo B et al 2002 Appl. Phys. Lett. 80 604
[4] Aktas O et al 2004 Solid-State Electron. 48 471
[5] Mizuta E et al 2018 IEEE Trans. Nucl. Sci. 65 1956
[6] Scheick L 2014 IEEE Trans. Nucl. Sci. 61 2881
[7] Zheng Q et al 2018 IEEE Trans. Nucl. Sci. 65 1920
[8] Chowdhury U et al 2008 IEEE Electron Device Lett. 29 1098
[9] Joh J, Gao F, Palacios T and del Alamo J A 2010 Microelectron. Reliab. 50 767
[10] Kuboyama S et al 2011 IEEE Trans. Nucl. Sci. 58 2734
[11] Wang C W et al 2000 J. Appl. Phys. 88 6355
[12] Chen C et al 2012 Chin. Phys. B 21 078503
[13] Berthet F et al 2012 Electron. Lett. 48 1078
[14] Lei Z F et al 2015 Chin. Phys. B 24 056103
[15] Rashmi, Kranti A, Haldar S and Gupta R S 2002 Solid-State Electron. 46 621
[16] Yadav A et al 2015 Radiat. Eff. Defects Solids 170 377
[17] Schwarz C et al 2013 Appl. Phys. Lett. 102 062102
[18] Kim H Y et al 2013 J. Vac. Sci. Technol. B Microelectron. Nanom. Struct 31 051210
[19] Vitusevich S A et al 2003 Phys. Status Solidi A 195 101
[20] Lei Z F et al 2018 Microelectron. Reliab. 80 312