Highly tolerant diamond Schottky barrier photodiodes for deep-ultraviolet xenon excimer lamp and protons detection

Abstract The response property and stability of diamond Schottky barrier photodiodes (SBPDs) were investigated for the monitor applications of deep ultraviolet (DUV) light and high-energy radiation particles. The SBPDs were fabricated on the unintentionally doped insulating diamond epilayer grown on a heavily boron-doped p+-diamond (100) conductive substrate by microwave plasma chemical vapor deposition. The vertical-type SBPDs were constructed of semitransparent tungsten carbide (WC) Schottky contact on the top of the device and a WC/titanium ohmic contact on the bottom. The SBPDs were operated to detect the DUV light and protons in zero-bias photovoltaic mode. The spectral response of the SBPDs showed that the peak wavelength was at 182 nm with a sensitivity of 46 ± 1 mA/W. The response speed was shorter than 1 sec, with a negligible charge-up effect and persistent photoconductivity. The SBPDs showed a stable response upon the irradiation by 172-nm xenon excimer lamp with 70 mW/cm2 for 200 hrs and 70 MeV protons for the dose of 10 MGy, corresponding to a non-ionizing energy loss of 1.4 × 1016 MeV neq/cm2.


Introduction
Diamond is a promising semiconductor for detecting deep ultra-violet (DUV) light and high-energy charged particles. A marked improvement in the detection performance is expected from its excellent physical properties, such as large bandgap energy of ∼5.5 eV [1], high radiation resistance, the highest thermal conductivity of 22 W/cm·K, and high thermal stability. The advanced semiconductor integrated circuits require high-resolution photolithography with a shorter wavelength than 193 nm. The DUV-light monitors are needed to calibrate the output power of the light source for the photolithography. On the other hand, radiation monitors are indispensable for high-energy and high-flux radiation places, such as nuclear reactors, accelerators, and cosmic space. In the early development of diamond detectors, the research interest was placed on using poly-crystalline diamonds [1][2][3][4][5]. Later, the research on diamond detectors was carried out using single-crystalline diamonds owing to the great progress in developing device-grade epilayers by a microwave plasma chemical vapor deposition (MPCVD) technique [6][7][8][9][10][11]. Recently, high-performance solar-blind UV photodetectors have been reported based on the heterojunction of graphene/ diamond [12] or β-Ga 2 O 3 /diamond [13]. The research and development team for diamond radiation detectors (RD42), collaborating with many institutes and universities, has long investigated diamond detectors for radiation monitoring and tracking [14][15][16].
The excellent photoresponse properties of singlecrystalline diamond DUV detectors were revealed within a series of our work [17]. The configuration of the photodetectors was planar-type, fabricated on unintentionally-doped epilayers grown on Ib-type nitrogen-doped insulating diamond (100) substrates by MPCVD [18]. The semitransparent tungsten carbide (WC) and annealed WC/titanium (Ti) metals were used as the Schottky and ohmic electrodes, respectively. The thermally stable operation and solar-blind property were obtained owing to the use of a homoepitaxially grown diamond layer and WC metals [19,20]. In particular, the combination of hard materials between diamond and WC was a critical factor in providing excellent thermal stability. One of the unique properties was that the photodetectors exhibited a huge photoconductivity gain, which is substantially larger than unity, producing an enormous photocurrent [21]. However, the colossal photocurrent was accompanied by a persistent electrical conductivity, which induced a slow response speed [22]. One reason was the existence of nitrogen in the substrate, which might act as a deep trap and lead to space charge separation. In addition, the charge-up problem came up under the high-flux irradiation of UV light due to the use of an insulating substrate.
The vertical-type Schottky barrier photodiodes (SBPDs) using single-crystalline diamond were developed as the next step to overcome these problems [23]. The SBPDs were fabricated on the unintentionally-doped epilayer grown on a heavily boron-doped p + -diamond (100) conductive substrate by MPCVD. The semitransparent WC and WC/Ti metals were formed at the top and bottom, respectively. The SBPDs did not provide the photoconductivity gain property and persistent photocurrent due to the reduction of nitrogen [24]. Although the typical photoresponse property from UV and visible light region (220 to 640 nm) has been reported, the stability under high-flux DUV light and radiation has not yet been explored.
This study investigates the response property in DUV region and the stability of diamond-based vertical-type SBPDs upon the radiation by a 172-nm xenon excimer lamp with a power density of 70 mW/cm 2 and a highenergy 70 MeV protons. The appropriate device operation mode for the monitors is discussed based on these results.

Experimental
The unintentionally-doped diamond epilayers with a thickness of 0.5 or 1.6 μm were homoepitaxially grown on p + -diamond (100) substrate by MPCVD system (CORNES Technologies, AX5200S). The p + -diamonds with 2.5 × 2.5 × 0.5 mm 3 in size were used as the substrate from Technological Institute for Super-hard and Novel Carbon Materials (TISNCM). The nominal boron concentration of the p + -diamond substrate was ∼10 20 cm −3 . The resistivity, hole carrier concentration, and hole mobility evaluated by Hall-effect measurement were 0.5 ± 0.2 Ωcm, 8.7 × 10 18 -1.2 × 10 19 cm −3 , and 3 ± 1.5 cm 2 /Vs, respectively. The hole mobility decreased with increasing the boron concentration [25]. The source gases are H 2 and CH 4 for the diamond growth. The ratio of CH 4 and H 2 was 0.05%, and the corresponding flow rates of CH 4 and H 2 were 0.5 and 500 sccm, respectively. The reactor pressure and growth temperatures were 80 Torr and 800-850 °C, respectively. The boron concentration of the epilayer was controlled at ∼10 15 cm −3 or less [26]. Prior to fabricating the SBPD, the samples were oxidized for 1 hr in a boiling acid solution of H 2 SO 4 /HNO 3 (1:1). The semitransparent WC (5 nm, 1 mmΦ) and WC (30 nm)/Ti (40 nm) electrodes were deposited for the formation of Schottky and ohmic contacts by sputter deposition on the top of diamond epilayer and back side of the p + -diamond substrate, respectively. The details of the sample structures are described in Figure 1.
The macroscopic crystalline quality of the diamond epilayer grown on p + -diamond substrate was investigated using an X-ray diffraction (XRD) system (Bruker, D8 Discover) with a four-crystal monochromator (4 × Ge[220]). The crystallinity and lattice constant were evaluated by rocking-curve and triple-axis methods [27].
The microscopic crystalline quality of the diamond epilayer was investigated using a confocal Raman system (WITec, α-300 R). The deuterium D2 lamp (Hamamatsu Photonics, L9026) and DUV spectrometer (Bunkoukeiki, VUV-200) were used to investigate the spectral response. The 172-nm xenon excimer lamp with a power density of 70 mW/cm 2 (Hamamatsu Photonics) was used to evaluate the stability of the device. The 185-nm lowpressure mercury lamp (Hamamatsu Photonics, L937-02) was used to evaluate the angle-dependent photoresponse of the light source. The xenon lamp (USHIO, UXL-500D) with monochromatic 220-nm light was used to evaluate the response speed. The power density of the 220-nm light was ∼10 μW/cm 2 . The SBPDs were packed in the nitrogen atmosphere to evaluate the photoresponse properties in the DUV region. The sensitivity (S λ ) was obtained by the output signal per radiant power, which is expressed as follows: where λ, I photon , and P are light wavelength, measured photocurrent, and radiant power, respectively. While the S λ can be expressed as follows: where QE, G, h, c, and q are quantum efficiency, gain, Planck constant, light velocity, and elementary charge, respectively. The QE can be considered as the number of free carriers generated by one photon, estimated by Eqs. (1) and (2) where the G was assumed to be 1 [28].
The protons with an energy of 70 MeV were generated by cyclotrons at Cyclotron and Radioisotope Center (CYRIC) in Tohoku University. The temperature of the device was controlled at -15 °C by a cold nitrogen gas flow. The proton beam with the size of 3.5 mmΦ was irradiated on the sample surface using the scanning system to investigate the response of the protons. The bare SBPDs were used to evaluate the radiation response properties. The 500-nA beam current (I beam ) was used, producing a flux of 3 × 10 12 protons per sec. The I beam was given by where N proton is the number of protons per sec. The number of electron-hole pairs (N e-h ) generated by one proton is expressed as follows: where η, E i. , and E creat.e-h are efficiency, energy induced by high-energy charged particles, and energy for creating one electron-hole pair, respectively. The η is normally 0.7-0.8, where the charge collection and conversion from electron-hole pair to free carrier efficiencies. Therefore, the current (I proton ) obtained by protons is expressed as follows: In the case of a 1.6-μm-thick depletion layer of diamond, the E i was estimated to be 4.9 keV by GEANT4 [29][30][31]. In addition, the E creat.e-h of diamond was reported to be ∼13 eV [32]. Therefore, the N e-h was assumed to be ∼283 pairs for η = 0.75, producing a I proton of ∼141 μA. Considering the spot size of the proton beam (3.5 mmΦ) and the size of the Schottky metal (1 mmΦ), the I proton was evaluated to be ∼11.5 μA in this study. The non-ionizing energy loss was estimated to be 1.4 × 10 16 MeV n eq /cm 2 , corresponding to the total ionizing dose of 10 MGy for Si.

Results & discussion
The crystalline quality of the diamond epilayer grown on p + -diamond substrate in macro-and micro-scales was evaluated by XRD and Raman measurements. Figure 2 shows the (a) X-ray rocking curve and (b) 2θ-ω scan profiles of 004 diamond reflection. Due to that the thickness of the diamond epilayer being too thin, the reflection from the substrate was predominant. The peak width (full-width at half maximum: FWHM value) of the rocking curve profiles was ∼70 arcsec. The absolute lattice constant of the substrates was estimated to be 3.567 Å, the same value as the bulk diamond [33]. The increase of boron concentration in the substrate would produce a larger lattice constant [34,35]. Figure 3 shows the Raman spectra of p + -diamond substrate (black line), 0.5-μm thick diamond epilayer (blue line), and 1.6-μm thick diamond epilayer (red line). The p + -diamond substrate exhibits a considerably downshifted peak position from the bulk value (1332 cm −1 ) [36] and a broadened FWHM value. In contrast, in the case of the diamond epilayer, the peak position approaches the bulk value, and the FWHM values become narrow with increasing the epilayer thickness.
Next, the electric property of the SBPD was investigated. Figure 4 shows the current-voltage (I-V) characteristics of the SBPD. Here the 0.5-μm-thick diamond epilayer grown on p + -diamond substrate was used for the SBPD. The I-V characteristics in the dark and 220nm DUV light illumination are shown by black and violet circles, respectively. The forward and reverse voltages correspond to negative and positive biases, respectively. The dark I-V characteristics exhibit excellent rectifying properties with a current on/off ratio of 10 12 at ±5 V and the apparent threshold voltage of 1.2-1.5 V. The SBPDs were operable in either photovoltaic or reverse-bias mode. The dark I-V curves were fitted by a conventional equivalent circuit model (inset) using a diode, series resistance (R s ), and parallel resistance (R p ). The equation based on the thermionic emission model [37] is given by where I 0 , A*, V SBH , n, k B , and T are the saturation current, Richardson constant, Schottky barrier height, ideality factor, Boltzmann constant, and temperature, respectively. The A* was assumed to be 90 A/cm 2 [38]. The R s and R p were resistance components that affect forward and reverse bias characteristics, respectively. Hence, the R s and R p were expected to be small and large, respectively. The fitted results are shown by the blue line in Figure 2. The V SBH , n, and R s were estimated to be 1.48-1.50 eV, 1.04-1.06, and 0.77 Ω cm 2 , respectively. Here, the R p was ignored because of the extremely small leakage current in reverse bias mode. The experimentally obtained R s of 0.77 Ω cm 2 from the I-V curve was much higher than 3.12 mΩ cm 2 , which was estimated from the film thickness, carrier concentration, and mobility of 0.5 μm, ∼1 × 10 14 cm −3 , and 1000 cm 2 /V s, respectively. The high R s may be due to the inhomogeneous boron concentration and crystal defects in the p + -diamond substrate [39]. Next, the spectral response characteristics of DUV light and the stability of the SBPDs were evaluated at zero bias in photovoltaic mode [40]. Figure 5 (a) shows the sensitivity as a function of wavelength evaluated by   the irradiation of the deuterium D2 lamp. The SBPDs have an obvious response in the DUV region. The peak wavelength was 182 nm with sensitivity at 46 ± 1 mA/W, corresponding to a QE of ∼31%. The QE is 10 times higher than SBPDs for the illumination of 220-nm UV light [23], which is attributed to the large bandgap energy of ∼6.8 eV with a direct transition at Γ point [1]. The devices exhibit a fast response speed of less than 1 sec for the 220-nm UV light without the charge-up effect and persistent photoconductivity owing to the highly conductive diamond substrates ( Figure 5(b)). The angle dependence of the 185-nm DUV light on the photocurrent revealed a reduction of ∼35% at a take-off angle of 45°, where 0° was surface-normal geometry ( Figure 5(c)). Figure 6 shows the stability of the device evaluated by 70 mW/cm 2 xenon excimer lamp at zero bias in photovoltaic mode. The intensity was obtained by the output current normalized by the initial photocurrent, considering the reduction of the intensity of the light source. We confirmed that the device showed highly stable operation for 200 hrs without degradation.
Next, the response property under protons irradiation and the stability of the SBPDs were evaluated at zero bias in photovoltaic mode. Figure 7 shows the response property and stability of the device under the 70 MeV protons irradiation at zero bias in photovoltaic mode. Here the 1.6-μm-thick diamond epilayer was used for the SBPD. The discrete response signal was obtained due to the sample scanning (Figure 7(a)). Under the huge amount of radiation flux of 10 MGy, the device exhibited stable response properties without charge-up effect and persistent photoconductivity, similar to the photodetectors for excimer lamp. The detected current values of ∼1.5 μA indicated that the collection efficiency was estimated to be ∼13%. Although the estimation of the normalized intensity was not accurate well due to the sample scanning, a significant reduction in the detection efficiency was not observed up to 10-MGy dose of protons corresponding to the non-ionizing energy loss of 1.4 × 10 16 MeV n eq /cm 2 (Figure 7(b)).
Finally, the dark I-V characteristics of the SBPDs before and after the irradiation of 70 MeV protons was investigated to discuss the appropriate device operation mode for the monitors. Figure 8(a) shows the dark I-V characteristics of the SBPDs before and after the irradiation by high-energy protons. The forward and reverse voltages correspond to negative and positive biases, respectively. Before the irradiation of protons, the device exhibited excellent rectifying properties with the current on/off ratio of ∼10 12 , showing the ideality factor of 1.16 and R s of 1.2 Ω cm 2 . After the irradiation of protons, the forward current exhibited two-step profiles, and the reverse current increased up to ∼3 × 10 −4 A/cm 2 at 10 V. These results indicated that defects were generated by irradiating protons, which increased the On-resistance at forward bias mode and leakage current path at reverse bias mode. Due to the two-step profiles, we used the equivalent circuit model shown in Figure 8(b), which consists of two parallel diodes with R s and R p components for fitting [41]. The parameters obtained by fitting were inset in Figure 8(a). The model explained the  forward I-V characteristics with two diode components (diode_1 and diode_2). Therefore, the diode_2 is dominant at biases lower than 1 V, while diode_1 is dominant at biases higher than 1 V. However, this model could not fit the reverse I-V characteristics well. Therefore, another mechanism for the leakage current path was dominant in this case. Several fitting models, such as thermionic-field emission [42,43] and Pool-Frenkel emission [44], were adopted to fit the reverse I-V curve. More than two equations were required to fit the experimental data. Hence, at least two electrical transport mechanisms should be considered. The mechanisms for the leakage current flow were not clear at this moment. Although the protons caused the radiation damage, the response property at zero bias in photovoltaic mode was insensitive to the irradiation damage. Possibly, the reason is due to that no electric field induced the carrier injection occurs at zero bias, which does not modify the photo-carriers generation and recombination. Therefore, the zero bias photovoltaic mode is appropriate for detecting DUV light and protons, which applies to the monitors.

Conclusion
We investigated the response property and stability of diamond-based SBPDs by irradiating DUV light with a 172nm xenon excimer lamp having a power density of 70 mW/cm 2 and high-energy protons with an energy of 70 MeV. Using a thin insulating diamond epilayer, electrically conductive diamond substrate, and thermally stable WC-based electrodes were the critical factors in obtaining the vertical-type SBPDs with negligible charge-up effect and persistent photoconductivity. The radiation stability indicates that the zero bias photovoltaic operation mode is appropriate as the monitors for high flux excimer lamps and high-energy protons.   Dr. Yasuo Koide is now a senior scientist with special mission and a group leader of the next-generation semiconductor group at NIMS, Japan. He received a Ph.D. in Crystal Growth from Nagoya University (Nagoya, Japan) in 1988. He was a vice president at NIMS from 2016-2019. His research interest includes wide bandgap semiconductor materials (mainly diamond and nitrides) and metal/semiconductor contact properties.