Plasma Nitridation Effect on β-Ga2O3 Semiconductors

The electrical and optoelectronic performance of semiconductor devices are mainly affected by the presence of defects or crystal imperfections in the semiconductor. Oxygen vacancies are one of the most common defects and are known to serve as electron trap sites whose energy levels are below the conduction band (CB) edge for metal oxide semiconductors, including β-Ga2O3. In this study, the effects of plasma nitridation (PN) on polycrystalline β-Ga2O3 thin films are discussed. In detail, the electrical and optical properties of polycrystalline β-Ga2O3 thin films are compared at different PN treatment times. The results show that PN treatment on polycrystalline β-Ga2O3 thin films effectively diminish the electron trap sites. This PN treatment technology could improve the device performance of both electronics and optoelectronics.


Introduction
An ultra-wide-bandgap (UWB) semiconductor, gallium oxide (Ga 2 O 3 ) with a bandgap of~4.9 eV, has been intensively investigated for applications in power electronics and ultraviolet-C (UVC) photodetectors (PDs) [1][2][3][4][5][6][7][8][9][10]. The Ga 2 O 3 semiconductor exhibits five different polymorphs designated as α, β, γ, δ, and ε. Among them, β-Ga 2 O 3 has emerged as the studied phase because of its excellent thermal stability [11]. The electrical and optoelectronic performance of semiconductor devices are mainly affected by the presence of defects or crystal imperfections in the semiconductor. With a metal oxide semiconductor including β-Ga 2 O 3 , oxygen vacancies are one of the most common defects and are known to serve as electron trap sites whose energy levels are below the conduction band (CB) edge [12]. Through the electron trap sites below the CB, electrons are easily trapped and de-trapped during device operation, which significantly degrades device performance for both electronics and optoelectronics. For example, the lifetime of photogenerated charge carriers is one of the important parameters in optoelectronics because it can limit the speed of the PD operation. PDs with slow decay times are less suitable for high-speed applications. The slow decay of the photocurrent is attributed to the trapping and de-trapping of the photogenerated electrons through the electron trap sites. Thus, it is important to suppress those trapping and de-trapping processes by eliminating the trap sites. It has been reported that nitrogen (N) doping in a metal oxide can suppress both hole and electron trapping sites [13][14][15][16][17] by passivating or neutralizing the trapping sites. Thus, the N effect on β-Ga 2 O 3 semiconductors should be studied to improve device performance by suppressing the trap density in the semiconductor.
In the current study, we discuss the effects of plasma nitridation (PN) on polycrystalline β-Ga 2 O 3 thin films. In detail, the electrical and optical properties of polycrystalline β-Ga 2 O 3 thin films are compared at different PN treatment times.

Formation of Polycrystalline β-Ga 2 O 3 and PN Treatment Process
An amorphous Ga 2 O 3 thin film with a thickness of 100 nm was deposited using a polycrystalline Ga 2 O 3 target and radio frequency (RF) sputtering [4]. The deposition process was conducted at a pressure of 1.3 mTorr at 70 W with 5 sccm of Ar gas. Nitrogen (N) atoms were introduced into the surface of the amorphous Ga 2 O 3 surface at a substrate temperature of 370 • C via NH 3 plasma for either 1 min or 5 min. The PN process was performed at a pressure of 0.3 Torr with 200 sccm of NH 3 gas at 200 W. The PN-treated amorphous Ga 2 O 3 turned into β-Ga 2 O 3 after annealing at 900 • C in atmosphere for 1 h.

Characterization of β-Ga 2 O 3 after PN Process
The optical bandgaps of the polycrystalline β-Ga 2 O 3 thin films at different PN treatment times were extracted using a UV-VIS spectrophotometer (UV-3600 plus, Shimadzu, Kyoto, Japan) via a Tauc plot with a direct bandgap model [18]. The depth profiles of the N atoms and other atoms in the β-Ga 2 O 3 thin films were characterized using secondary ion mass spectrometry (SIMS, IMS 7F, CAMECA, Gennevilliers, France) in a 200 × 200 µm 2 area using Cs + (6 kV, 10 nA). A photoluminescence (PL) analysis was conducted at 264 nm excitation (Horiba Jobin-Yvon LabRAM HR-800 UV-Visible-NIR, Kyoto, Japan).  Figure 1a shows the optical transmittance and Tauc plots of the β-Ga 2 O 3 thin film at different PN times. The optical bandgap of each sample was extracted from the linear extrapolation of the inset graph, as shown in Figure 1a. The optical bandgap of the intrinsic β-Ga 2 O 3 thin film was~4.9 eV, and its value remained almost constant even after PN treatment. This indicated that the PN treatment did not noticeably alter the energy band structure of the β-Ga 2 O 3 semiconductor. The inset of Figure 1b  However, the ratio of N/O concentration at different PN times showed that more N was introduced in the polycrystalline β-Ga 2 O 3 thin films at higher PN times. In detail, the ratio of N/O at 1 min PN treatment was indistinguishable from that of the intrinsic β-Ga 2 O 3 . However, the ratio of N/O at 5 min PN treatment was much higher compared to that of the intrinsic β-Ga 2 O 3 and β-Ga 2 O 3 at 1 min PN treatment. This indicated that N was noticeably introduced into the β-Ga 2 O 3 in a logical scale after 5 min PN treatment. Figure 2a shows the room temperature PL spectrum of the β-Ga 2 O 3 at different PN treatment times. Regardless of the PN treatment, the sputtered β-Ga 2 O 3 thin film exhibited a broad emission band centered around 510 nm. The emission peak in this region presumably originated from the recombination of electrons on donor sites and holes on acceptor sites where the electrons and holes presumably originated from the oxygen vacancy (V O ) and gallium vacancy (V Ga ), respectively [19][20][21]. The peak intensities of 510 nm at different PN times are compared in Figure 2b. The results showed that the PL intensity of the β-Ga 2 O 3 with 5 min PN treatment significantly decreased compared to that of the intrinsic β-Ga 2 O 3 and the β-Ga 2 O 3 with 1 min PN treatment. It was assumed that the N effect on the β-Ga 2 O 3 was negligible for the 1 min PN treatment. However, after 5 min PN treatment, a noticeable amount of N atoms was introduced to the β-Ga 2 O 3 , as shown in Figure 1b. The introduced N was presumed to effectively suppress the V O -related electron trap sites, which resulted in the reduction in PL intensity, as shown in Figure 2a Figure 2a shows the room temperature PL spectrum of the β-Ga2O3 at different PN treatment times. Regardless of the PN treatment, the sputtered β-Ga2O3 thin film exhibited a broad emission band centered around 510 nm. The emission peak in this region presumably originated from the recombination of electrons on donor sites and holes on acceptor sites where the electrons and holes presumably originated from the oxygen vacancy (VO) and gallium vacancy (VGa), respectively [19][20][21]. The peak intensities of 510 nm at different PN times are compared in Figure 2b. The results showed that the PL intensity of the β-Ga2O3 with 5 min PN treatment significantly decreased compared to that of the intrinsic β-Ga2O3 and the β-Ga2O3 with 1 min PN treatment. It was assumed that the N effect on the β-Ga2O3 was negligible for the 1 min PN treatment. However, after 5 min PN treatment, a noticeable amount of N atoms was introduced to the β-Ga2O3, as shown in Figure 1b. The introduced N was presumed to effectively suppress the VO-related electron trap sites, which resulted in the reduction in PL intensity, as shown in Figure 2a,b.

Results and Discussion
Furthermore, the PL decay profiles at 510 nm are presented at different PN treatment times and the results are compared in Figure 2c. All of the decay curves were approximated by a sum of two exponential functions (τp1 and τp2). τp1 is an instantaneous response to light and represents a fast response component, while τp2 corresponds to the β-Ga2O3 defect and represents a slow response component. Figure 2d exhibits a comparison of τp1 and τp2 at different PN treatment times. This comparison showed that τp1 remained unchanged while τp2 continued to decrease with PN treatment time. The reduced τp2 with PN treatment was attributed to the suppression of the oxygen vacancies via N doping. It was presumed that the N atoms effectively passivated or neutralized the trapping sites.   Furthermore, the PL decay profiles at 510 nm are presented at different PN treatment times and the results are compared in Figure 2c. All of the decay curves were approximated by a sum of two exponential functions (τ p1 and τ p2 ). τ p1 is an instantaneous response to light and represents a fast response component, while τ p2 corresponds to the β-Ga 2 O 3 defect and represents a slow response component. Figure 2d exhibits a comparison of τ p1 and τ p2 at different PN treatment times. This comparison showed that τ p1 remained unchanged while τ p2 continued to decrease with PN treatment time. The reduced τ p2 with PN treatment was attributed to the suppression of the oxygen vacancies via N doping. It was presumed that the N atoms effectively passivated or neutralized the trapping sites.  Figure 3b. The results showed that the photocurrent increased with PN treatment time. The improved photocurrent was attributed to the reduction in the trap density and/or enhancement of carrier mobility via the suppression of trapping sites due to the PN treatment. For a quantitative analysis of the photocurrent decay, the transient curves were fitted using two exponential functions (τ c1 and τ c2 ), as shown in Equation (1).
where I 0 is the dark current, A and B are constants, and t is the transient time. τ c1 and τ c2 are the relaxation lifetime decay rate of photo-generated electrons with a fast component (τ c1 ) and a slow component (τ c2 ). The fast component corresponds to the band-to-band transition, while the slow component is attributed to the trapping and de-trapping of photogenerated electrons via defect states. τ c1 and τ c2 are compared at different PN treatment times in Figure 3c. The results showed that τ c1 was less affected by the PN treatment. This was because the band-to-band transition and optical bandgap remained unchanged after PN treatment in these conditions. However, the τ c2 values continued to decrease with PN treatment time, indicating that the PN treatment effectively suppressed the trapping and de-trapping processes of the photogenerated electrons via passivating or neutralizing the trapping sites. This is consistent with the results in Figure 2. Finally, photo-to-dark-current (PDCR) of the MSM PD was compared at different PN treatment times, as shown in Figure 3d. The PDCR was defined using Equation (2).
where I p and I d are the photocurrent and dark current of the PD, respectively. The results showed that the PDCR values continued to enhance with PN treatment time. It was noted that the I d remained unchanged with PN treatment, while the I p continued to increase due to the reduction in oxygen vacancies or/and enhancement of carrier mobilities. Figure 4 shows a schematic drawing of the trap site distribution with donor and acceptor levels before and after PN treatment. The donor and acceptor bands corresponded to the oxygen vacancies and gallium vacancies, respectively. The schematic image demonstrates that the introduced N atoms were able to passivate or neutralize the trap site below the CB and eventually reduce the electron trapping sites. It is believed that the reduced electron trapping sites resulted in a lower PL intensity, fast decay time, and higher PDCR for the β-Ga 2 O 3 MSM PD with a PN treatment of 5 min.
The PDCR and decay time in this work were compared with those of reported β-Ga 2 O 3 based MSM PDs in Figure 5. The results showed that after PN treatment on the polycrystalline β-Ga 2 O 3 thin films, the PDCR value increased but the decay time decreased. It indicated that the device performance of the PDs was enhanced due to the reduction in electron trap sites after PN treatment. This approach could improve the device performance of both electronics and optoelectronics.
where Ip and Id are the photocurrent and dark current of the PD, respectively. The results showed that the PDCR values continued to enhance with PN treatment time. It was noted that the Id remained unchanged with PN treatment, while the Ip continued to increase due to the reduction in oxygen vacancies or/and enhancement of carrier mobilities.  Figure 4 shows a schematic drawing of the trap site distribution with donor and acceptor levels before and after PN treatment. The donor and acceptor bands corresponded to the oxygen vacancies and gallium vacancies, respectively. The schematic image demonstrates that the introduced N atoms were able to passivate or neutralize the trap site below the CB and eventually reduce the electron trapping sites. It is believed that the reduced electron trapping sites resulted in a lower PL intensity, fast decay time, and higher PDCR for the β-Ga2O3 MSM PD with a PN treatment of 5 min.. The PDCR and decay time in this work were compared with those of reported β-Ga2O3 based MSM PDs in Figure 5. The results showed that after PN treatment on the polycrystalline β-Ga2O3 thin films, the PDCR value increased but the decay time decreased. It indicated that the device performance of the PDs was enhanced due to the reduction in electron trap sites after PN treatment. This approach could improve the device performance of both electronics and optoelectronics. The PDCR and decay time in this work were compared with those of reported β-Ga2O3 based MSM PDs in Figure 5. The results showed that after PN treatment on the polycrystalline β-Ga2O3 thin films, the PDCR value increased but the decay time decreased. It indicated that the device performance of the PDs was enhanced due to the reduction in electron trap sites after PN treatment. This approach could improve the device performance of both electronics and optoelectronics.

Conclusions
The effects of plasma nitridation (PN) on polycrystalline β-Ga2O3 thin films were investigated. N atoms were introduced in the polycrystalline β-Ga2O3 thin films via NH3 plasma. The results showed that the PL intensity of the β-Ga2O3 with 5 min PN treatment significantly decreased compared to that of the intrinsic β-Ga2O3 and the β-Ga2O3 with 1 min PN treatment. In addition, the decay time of both the PL and photocurrent decreased after PN treatment for 5 min, which could be explained by the suppression of the trapping and de-trapping processes of the photogenerated electrons via passivating or neutralizing the trapping sites. It was assumed that the introduced N effectively suppressed the VOrelated electron trap sites, which resulted in the reduction in the PL intensity, PL decay time, PDCR, and photocurrent decay time. The above results showed that PN treatment

Conclusions
The effects of plasma nitridation (PN) on polycrystalline β-Ga 2 O 3 thin films were investigated. N atoms were introduced in the polycrystalline β-Ga 2 O 3 thin films via NH 3 plasma. The results showed that the PL intensity of the β-Ga 2 O 3 with 5 min PN treatment significantly decreased compared to that of the intrinsic β-Ga 2 O 3 and the β-Ga 2 O 3 with 1 min PN treatment. In addition, the decay time of both the PL and photocurrent decreased after PN treatment for 5 min, which could be explained by the suppression of the trapping and de-trapping processes of the photogenerated electrons via passivating or neutralizing the trapping sites. It was assumed that the introduced N effectively suppressed the V O -related electron trap sites, which resulted in the reduction in the PL intensity, PL decay time, PDCR, and photocurrent decay time. The above results showed that PN treatment on polycrystalline β-Ga 2 O 3 thin films effectively diminished the electron trap sites. This PN treatment technology could improve the device performance of both electronics and optoelectronics.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare that they have no known competing financial interests or personal relationships that could have influenced the work reported in this paper.