Dependence of the Silicon Carbide Radiation Resistance on the Irradiation Temperature

. The effect of high-temperature electron and proton irradiation on SiC-based device characteristics is being investigated. Industrial integrated 4H-SiC Schottky diodes, each with an n-type base and a blocking voltage of either 600 V, 1200 V, or 1700 V, manufactured by Wolfspeed, are being studied. 0.9 MeV electron and 15 MeV proton irradiation were applied. It has been found that the irradiation resistance of silicon carbide Schottky diodes at high temperatures significantly exceeds their resistance at room temperature. This effect is attributed to the annealing of compensating defects induced by high-temperature irradiation. The parameters of radiation-induced defects are determined using the method of deep level transient spectroscopy (DLTS). Under high-temperature ("hot") irradiation, the spectrum of radiation-induced defects introduced into SiC appears to differ significantly from the spectrum of defects introduced at room temperature. It is suggested that approximately half of the compensation is due to radiation-induced defects formed in the bottom part of the bandgap.


Introduction
The first studies of radiation-induced defects (RID) in silicon carbide, carried out in the 1950s-1960s, confirmed the high radiation resistance of this material [1].It is well known that some crystals studied in those years were heavily doped and had a high density of structural defects.However, as more perfect and purer SiC samples were obtained, their measured radiation resistance gradually decreased.There were studies that indicated that SiC not only does not surpass silicon in terms of radiation resistance but is even inferior to it in some parameters [2][3][4].It is also well understood that a decrease in radiation resistance with an improvement in the quality of the material is characteristic of almost every semiconductor material since various structural defects and uncontrolled impurities serve as sinks for RID and thereby reduce the rate of degradation of the material parameters.However, the comparability of the carrier removal rate in Si and silicon carbide under irradiation looks surprising because the band gap of 4H-SiC (3.2 eV) is almost three times greater than that of silicon.In [5], we suggested that this might be due to different annealing temperatures of primary RIDs in Si and SiC.At 300 K, in silicon, there is noticeable annealing of primary RIDs, while in silicon carbide at this temperature, there is practically no annealing process.To verify this assumption, it seems necessary to study the SiC radiation resistance dependence on the irradiation temperature.The objective of this study was precisely to conduct such investigation.

Experiment
The industrial integrated 4H-SiC Junction Barrier Schottky (JBS) diodes, with an n-type base, and blocking voltages of 600 V, 1200 V, and 1700 V respectively, have been studied.Prior to irradiation, the uncompensated donor impurity concentration (Nd-Na) in the devices mentioned above was approximately 6.5×10 15 , 4.5×10 15 , and 3.5×10 15 cm -3 , respectively.Both 0.9 MeV electrons and 15 MeV protons were used for irradiation.The maximum radiation doses were 7×10 16 cm -2 for the electrons and 1×10 14 cm -2 for the protons.

Results and Discussion
Table 1 presents previously obtained data for the carrier removal rate Vd in SiC for irradiation at room temperature.As one can seen, the carrier removal rate in silicon carbide is actually as small as two times lower, than that in silicon.To confirm the hypothesis stated in [5], 4H-SiC JBS structures were irradiated with electrons and protons at temperatures up to 500 0 С.A significant decrease in the carrier removal rate was found with increasing irradiation temperature (see Figure 1).When studying the current-voltage characteristics, it was found that with an increase in the irradiation temperature from 23 to 500 0 C, the base resistance (at an irradiation dose of 1.3x10 17 cm -2 ) decreases by 6 orders of magnitude (see Fig. 2).Similar results were obtained with proton irradiation at elevated temperatures (see Fig. 3 and Fig. 4).Using the DLTS method, the parameters of deep radiation centers formed after irradiation of SiC were studied.
As had been noted, there might be RID rearrangement partial annealing in 4H-SiC while measuring the DLT spectra [8].This paper describes DLT spectra we measured at higher temperatures, which noticeably enhanced the annealing effect (Fig. 5).No doubt that annealing is completed already at a temperature of about 400-450 K.As can be seen from Figure 5, at measurement temperatures > 4500K, the DLT spectrum is practically unchanged.Table 2 presents the detected center parameters.Figure 6 shows the sample VCCs taken prior and posterior to its irradiation after hightemperature DLT spectra measurements.As one can see from the Figure 6, the irradiation leads to compensating defects formation with a concentration of ~ 2.4 х10 14 см -3 .After high-temperature DLTS measurements, approximately half of the compensating defects are annealed and the residual compensation is ~ 1.1 х 10 14 cm -3 .
Solid State Phenomena Vol.361 Let's estimate which of the detected deep levels can be the main compensating defects in n-4H-SiC.If acceptor level is formed during irradiation in the lower half of the n-type zone, then it will be filled with an electron, negatively charged and will contribute to a decrease in the Nd-Na value and a decrease in the n value.
If the acceptor level is formed in the upper half of the zone, then it will reduce the concentration of free electrons (n).But whether it reduces the value of Nd-Na or doesn't depend on its depth.If it is a deep level, then it is filled electronically, charged negatively, and reduce Nd-Na.If it is shallow (that is less deep) enough, then at C-V measurements at room temperature it is emptied, neutral, and doesn't contribute to the value of Nd-Na.
When measuring CVCs, the deep center filling degree is usually differs to that when measuring VCCs.As CVCs is being measured, this level is in the quasi-neutral region.In this case, this level filling is determined by two processes, namely, thermal emission from the center and reverses capture from the C-zone.I.e., if the level is above the Fermi level, it is emptied, while if it is below it is filled.When measuring VCCs, that level is in the bulk charge layer, there are no carriers in the conduction zone, and everything is determined by thermal emissions from that level.In principle, all levels lying in the upper half of the zone should be emptied after waiting for it sufficiently long.Nevertheless, as VCCs is being measured, even during the real time of the experiment, such levels are being emptied, which are deeper as compared to the event at which CVCs is being measured.Annealing performed up to 700 K during DLTS measurements showed that there is a partial restoration of the conductivity of the samples.In this case, all shallow centers in the upper half of the zone are annealed, except for Z1/2, EN5 and E6/7 and shallow boron in the lower half of the zone.
The recharge time constant (τ e ) for each of the detected deep levels can be estimated by the wellknown formula: τ e = [σ n V t N c exp(-E i /kT)] -1 Where V t is the thermal velocity of charge carriers, N c is the density of states in the conduction band, k is the Boltzmann constant, and T is the absolute temperature.Data from Table 2 were used for the calculation.The value of V t at 300 K was assumed to be 10 7 cm/sec, and N c = 8.9 x 10 19 cm -3 [13].
The recharge time constants of all detected deep levels in the upper half of the zone (except E5 and E6/7) are small (see Table 2), significantly less than the time of VCC measurements.Thus, all these centers do not contribute to the decrease in the value of Nd-Na.Only EN5 and E6/7 can give such a

Table 1 .
Carrier removal rates in devices based on SiC and Si.

Table 2 .
Detected deep levels (deep centers) parameters