Radiation Resistance of Silicon Carbide Schottky Diode Detectors in D-T Fusion Neutron Detection

Silicon carbide (SiC) is a wide band-gap semiconductor material with many excellent properties, showing great potential in fusion neutron detection. The radiation resistance of 4H-SiC Schottky diode detectors was studied experimentally by carefully analyzing the detectors’ properties before and after deuterium-tritium fusion neutron irradiation with the total fluence of 1.31 × 1014 n/cm2 and 7.29 × 1014 n/cm2 at room temperature. Significant degradation has been observed after neutron irradiation: reverse current increased greatly, over three to thirty fold; Schottky junction was broken down; significant lattice damage was observed at low temperature photoluminescence measurements; the peaks of alpha particle response spectra shifted to lower channels and became wider; the charge collection efficiency (CCE) decreased by about 7.0% and 22.5% at 300 V with neutron irradiation fluence of 1.31 × 1014 n/cm2 and 7.29 × 1014 n/cm2, respectively. Although the degradation exists, the SiC detectors successfully survive intense neutron radiation and show better radiation resistance than silicon detectors.


Neutron radiation. The irradiation was performed at the K600 Neutron Generator in China Institute
Atomic Energy (CIAE) in Beijing, China, which can provide a constant fast neutron beam generated by the deuterium-tritium fusion, with an average energy of 14 MeV and a neutron fluence rate of 4-12 × 10 9 n/cm 2 s. Two SiC chips (Detector #2 and #3, before being packaged) were irradiated by the fast-neutron with fluence of 1.31 × 10 14 n/cm 2 and 7.29 × 10 14 n/cm 2 , respectively. The radiation temperature was at 283 K.

I-V and C-V characteristics. The front and reverse I-V curves were measured with IWATSU CS-3200C
Curve Tracer, and the results are shown in Fig. 2. The detectors were applied with reverse bias voltages in the range of 10 V to 600 V, thus the dark current of the detectors could be expressed by the reverse current. It is found that the dark current (reverse current) of the two detectors being irradiated increased significantly with the increase of neutron fluence. At the low fluence, it increased by three times or more; at the high fluence, it increased even more greatly by more than thirty times.
The forward I-V curve of the un-irradiated diode chip shows the rectification character, but the two chips being irradiated lost that character and the Schottky junction was broken down. The forward I-V characteristics of Schottky barrier diodes can be described by the Bethe's Thermionic emission theory, in which the effective Richardson's constant is 146 A cm −2 K −2 for 4H-SiC [28][29][30][31][32] . According to the forward I-V characteristics and the Bethe equation, the ideality factor was calculated to be 1.14, which indicates the current is dominated by thermionic current. The Schottky barrier height Ф b for the Ni/4H-SiC contact was calculated to be 1.7 eV. Figure 3a shows the C-V curve acquired by Agilent B1500A Semiconductor Parameter Analyzer. Figure 3b shows the curve of 1/C 2 vs. V of the detector #1 (un-irradiated), from which the net doping concentration of 4H-SiC epitaxial layer and the built-in voltage of the Schottky diode can be acquired. From Fig. 3, we find the two samples (#2 and #3) being irradiated lost their C-V characters, and the effective doping concentration (N eff ) of the 4H-SiC epitaxial layer was 7.9 × 10 13 cm −3 and the built-in V bi potential of the Schottky contact was 1.7 V. The Schottky barrier height can be expressed as where N c is the effective density of the states in the conduction band of 4H-SiC, here is taken as 1.7 × 10 19 cm −3 . The barrier height thus was calculated 2.0 eV. The difference between the barrier heights from the I-V and C-V curves are due to the following factors: barrier height obtained from the forward I-V curve was calculated with the current which flows through the Schottky barrier over the entire area where the metal electrode covers, while the one derived from the C-V curve was calculated with the average capacitance related to the whole detector. Besides, the ideality factor we got is deviated from 1, which exposes the spatial inhomogeneity of the surface barrier height.
Photoluminescence. Photoluminescence (PL) experiments were performed to detect defects in the epitaxial material of the SiC detectors. Low temperature photoluminescence (LTPL) spectra were acquired in the  spectral region between 380 nm and 800 nm at temperature of 83 K-203 K. A He-Cd laser with a wavelength of 325 nm was used as the excitation light source. Figure 4 shows the integrated pulsed PL spectra of the 20-μm-thick lightly doped epitaxial 4H-SiC layer taken at 83 K, with a nitrogen concentration of about 7.9 × 10 13 cm −3 . As indicated in Fig. 4a, the distribution of the PL intensity of the 4H-SiC layer changes remarkably after neutron irradiation. The PL spectrum of the detector #1 is dominated by the near-band-gap nitrogen bound exciton lines (3.15-3.24 eV) and their associated phonon replicas (LO), but for the detector #2 and #3 being irradiated, the luminescence is completely quenched. This might be attributed to the severe lattice damage induced by neutron radiation.
The PL spectra in lower energy are dominated by a broad PL peak, covering green to yellow-green spectral range, peaking at about 2.12 ~ 2.23 eV, and in the spectra of all the samples, the intense and broad vibronic bands can be observed. By reference to the reported broad PL band by Sridhara et al. 33 , Gao et al 34 . and Sakai et al. 35 , we found this broad PL band might be composed by the peaks at 2.10 eV, 2.35 eV and 2.80 eV. The peaks at 2.35 eV and 2.80 eV might be due to the donor-to-acceptor (DAP) transition from the nitrogen donor (0.1 eV below the conduction band) to the deeper and shallower boron acceptors (0.7 eV and 0.3 eV above the valence band) 36 . The peak at 2.10 eV might be due to the carbon vacancy, which is a candidate for the electron transition, or other unidentified defect level 35 . The average luminance wavelength shifts with neutron fluence, the higher the neutron fluence is, the shorter the average wavelength would be, indicating some non-radiative defects have been produced by neutron irradiation. Figure 5a shows the PL of the samples measured at temperature ranging from 84 K to 203 K. Broadband green luminescence is observed in the figure. As the temperature increases, the intensity of the luminescence decreases and the luminescence band becomes broad. This possibly derives from the lattice vibration and lattice scattering. (Fig. 5b) The broadband green luminescence might be due to both the vacancies of carbon and its extended point   Fig. 6(a).
According to the calculation with SRIM2003 Code 37 , the dead layer of the SiC detectors, comprised of Ni/Au (100 nm/2 μm), can absorb about 1.00 MeV of the incident alpha particles, leaving about 4.16 MeV kinetic energy penetrating into the active layer. Because the projected range of those remnant alpha particles, about 12.2 μm, is smaller than the sensitive thickness of the SiC detector (20 μm), all the remnant energy of the alpha particles would be deposited in the active layer.
The counts of the alpha particles as a function of channel number are shown in Fig. 6(b). The alpha-particle peaks can be clearly observed. The peak centroid of alpha particles is at Channel 665 for detector 1#, Channel 619 for detector 2# and Channel 516 for detector 3#. The higher neutron fluence is, the lower the peak centroid would be. The width of alpha peaks decreases with the increase of the incident neutron fluence. Fitting the peaks with Gaussian function, we got the FWHMs of the three alpha peaks, which are 391 keV for detector 1#, 384 keV for detector 2# and 270 keV for detector 3#. Excluding the influence of electronic noise (10 keV), static broadening (6.0 keV) and energy straggling of dead layer (180 keV) 27 , we got the inherent FWHM of 347 keV for detector 1#, 334 keV for detector 2# and 201 keV for detector 3#. Charge Collection Efficiency. Charge collection efficiency (CCE) is defined as the ratio between the numbers of the electric charges collected by the detectors (Q c ) and the total number of electric charges of all the excited carriers (Q g ) 28,32 . Q g is dependent upon the energy of the alpha particles deposited in the detectors' sensitive volume, about 4.16 MeV. Because all the electronic devices in our alpha detection experiments were kept working stably, the Q g could be determined by the peak centroid of the detector 1# at the reverse bias voltages which could make the detector fully depleted, at Channel 665, and Q c could be determined by the peak centroid of alpha peaks. Figure 7 gives the CCE of the three detectors at different bias voltages with different neutron radiation fluence.
As shown in Fig. 7, the CCE decreases with the increase of the neutron fluence, which is consistent with early researches of other scientists 28,32 . The 4H-SiC detectors couldn't work without power supply after neutron irradiation. After the irradiation, the Schottky barrier of the detector was broken down, the 4H-SiC detector would not be depleted even if it is applied with reverse bias voltages within 300 V. The detectors are still alive after neutron irradiation but need high reverse bias voltages to avoid steep decrease of signal intensity. The CCEs of the SiC detectors decrease about 7.0% and 22.5% at 300 V when the neutron fluence reaches to 1.31 × 10 14 n/cm 2 and 7.29 × 10 14 n/cm 2 , respectively. But the CCE of a silicon detector decreased much more, over 75%, at lower fast-neutron irradiation fluence of 9.98 × 10 12 n/cm 2 38 . It can be concluded that the radiation resistance of SiC detector is much better than silicon detector.
Theoretically, the CCE of the un-irradiated detector is expressed as 32 where E ion is the energy of the incident alpha particles, 5157 MeV; d is the depletion width at a given bias; dE/dx is the rate of loss of energy of the implanted alpha particles as they penetrate the 4H-SiC epilayer; R is the projected range of the incident particles with an energy of E ion , 16 μm; L d is the diffusion length of the minority carriers, 6.8 μm; ε and ε r are dielectric and relative dielectric constant of the dead layer, 9.7 and 8.85 × 10 −14 F/cm, respectively; q is 1.6 × 10 −19 C; N eff is 7.9 × 10 13 cm −3 . The CCEs of the SiC detectors being irradiated are expressed as 28 where E is the electric field, E = V/d; T is the thickness of the 4H-SiC epitaxial layer, T = 20 μm; λ p and λ n are the mean free path of the holes and electrons, respectively, λ µ τ λ µτ  1.5 × 10 −8 cm 2 /V and 1.1 × 10 −8 cm 2 /V with the high fluence of 7.29 × 10 14 cm −2 . Compared with the value of μτ (over 1000 × 10 −8 cm 2 /V) for the holes of the detector not being irradiated 28 , it decreases significantly after neutron irradiation. The decrease of μτ can be atrributed to the neutron irradiation defects, which act as trapping centers for carriers, reducing the probability of the carrier transportation through the SiC material.
The calculation result of the CCEs is plotted in Fig. 7 with open dots. They are well consistent with the experimental results acquired in the alpha particle detection.

Conclusions
We compared the properties and performance of 4H-SiC Schottky diode detectors before and after the irradiation of deuterium-tritium fusion neutrons with total fluence of 1.31 × 10 14 n/cm 2 and 7.29 × 10 14 n/cm 2 at room temperature. We found that the 4H-SiC Schottky diode detectors being irradiated survived the intense neutron radiation, and were still effective and could be used in radiation detection even though the detector performance was degraded by the increase of dark current, reduction of CCE, decrease of μτ and movement of alpha peaks' centroid. The degradation can be attributed to the lattice damage, non-radiative defects and other defects induced by neutron irradiation.
It is known that the silicon detector is hard to operate above neutron fluence of 1 × 10 14 n/cm 2 and the degradation of its CCE is worse than SiC detector after fast-neutron irradiation 12,38 . Hence it can be concluded that the 4H-SiC Schottky diode detectors have a better neutron resistance than silicon detector and could be expected to be well used in fusion neutron detection.