Improved-Performance Amorphous Ga₂O₃ Photodetectors Fabricated by Capacitive Coupled Plasma-Assistant Magnetron Sputtering

Abstract

Ga₂O₃ has received increasing interest for its potential in various applications relating to solar-blind photodetectors. However, attaining a balanced performance with Ga₂O₃-based photodetectors presents a challenge due to the intrinsic conductive mechanism of Ga₂O₃ films. In this work, we fabricated amorphous Ga₂O₃ (a-Ga₂O₃) metal–semiconductor–metal photodetectors through capacitive coupled plasma assisted magnetron sputtering at room temperature. Substantial enhancement in the responsivity is attained by regulating the capacitance-coupled plasma power during the deposition of a-Ga₂O₃. The proposed plasma energy generated by capacitive coupled plasma (CCP) effectively improved the disorder of amorphous Ga₂O₃ films. The results of X-ray photoelectron spectroscopy (XPS) and current-voltage tests demonstrate that the additional plasma introduced during the sputtering effectively adjust the concentration of oxygen vacancy effectively, exhibiting a trade-off effect on the performance of a-Ga₂O₃ photodetectors. The best overall performance of a-Ga₂O₃ photodetectors exhibits a high responsivity of 30.59 A/W, a low dark current of 4.18 × 10⁻¹¹, and a decay time of 0.12 s. Our results demonstrate that the introduction of capacitive coupled plasma during deposition could be a potential approach for modifying the performance of photodetectors.

1. Introduction

Solar-blind photodetectors (SBPDs) based on wide-bandgap (WBG) semiconductors have received extensive attention owing to their compact size, being entirely solid-state, intrinsic solar-blindness, high radiation hardness, and other notable advantages [1,2]. Among many WBG semiconductors, Ga₂O₃ possesses suitable bandgaps of 4.5–5.3 eV without any complex bandgap engineering, making it an ideal material for the preparation of solar-blind photodetectors [3,4,5]. However, defects such as oxygen vacancies and structural dislocations, present both within the material and on its surface, invariably impair the performance of Ga₂O₃ SBPDs [6]. To achieve detection performance enhancement, many effective methods boosting the performance of a-Ga₂O₃ PDs have been reported, including but not limited to increasing oxygen concentration during the deposition of Ga₂O₃ [7], annealing under oxygen conditions [8], plasma treatment [9], using two-dimensional (2D) materials [10], and dielectric passivation [11]. Plasma treatment is widely recognized as an effective method for tailoring the surface properties of various semiconductor materials, such as ZnO, GaN [12,13], and Ga₂O₃ [14,15]. However, plasma treatment can only affect the surface of materials; there is a limited effect on the inside of the material [16]. Hence, introducing plasma into the deposition process of Ga₂O₃ holds the potential to efficiently alter the oxygen vacancy profile across the entire thin film. Capacitance coupled plasma (CCP) technology has effectively improved the quality of thin films in chemical vapor deposition (CVD) and pulsed laser deposition (PLD). Until now, capacitive coupled plasma-assisted magnetron sputtering has not been used in the preparation of high-performance Ga₂O₃-based PDs.

In this paper, we introduce additional plasma in the sputtering process of Ga₂O₃ thin films through the radio frequency bias (RF-bias) power supply connected with a matching network and the corresponding metal–semiconductor–metal (MSM) solar-blind UV photodetectors prepared. The influences of RF-bias power on the morphology and optical response of the amorphous Ga₂O₃ films have been systematically investigated. A series of different plasma powers effectively regulate the concentration of oxygen vacancy in amorphous Ga₂O₃, which exhibits a trade-off effect on the performance of Ga₂O₃ PDs. This work provides a promising way to fabricate high-performance Ga₂O₃ solar-blind ultraviolet detectors.

2. Materials and Methods

Ga₂O₃ films were deposited on a c-plane sapphire substrate at room temperature utilizing the customized CCP-assistant magnetron sputtering system (PVD Products Co. Ltd., Wilmington, MA, USA), which is schematically illustrated in Figure 1b. A 4 in. high-purity (99.99%) Ga₂O₃ ceramic target was used. Before deposition, the vacuum system achieved a base pressure of 1 × 10⁻⁷ torr. Subsequently, 20 sccm high-purity argon (Ar) sputtering gas was introduced into the chamber and maintained at a pressure of 5 mtorr. The thickness of the Ga₂O₃ film deposited on the sapphire substrate was controlled to about 300 nm, employing a sputtering power of 200 W. During the deposition of the Ga₂O₃ films, the CCP RF-bias power was fixed at 0 W, 20 W, 40 W, and 60 W. The corresponding samples were labeled S0, S20, S40, and S60, respectively. Finally, 100 nm thick interdigital Au electrodes were deposited on the Ga₂O₃ films via vacuum evaporation to form a Ga₂O₃ MSM photodetector, where the length of the interfinger electrode was 900 μm, the electrode width was 80 μm, and the spacing of the electrode fingers was 100 μm. A microscopy image of the Ga₂O₃ photodetectors is shown in Figure 1a. The photoelectric performances of Ga₂O₃ photodetectors were measured using Keithley 2636B. A 150 W UV-enhanced Xe lamp (GLORIA-X150A, Zolix, Beijing, China), in conjunction with a monochromator (Omni-λ3047i, Zolix), served as the light source. The transmittance spectrum was measured using a Shimadzu UV-1700 spectrophotometer (Shimadzu, Tokyo, Japan). To investigate the material morphology, X-ray diffraction (XRD) measurements (TTRIII, Rigaku, Tokyo, Japan) were conducted with a scanning range of 10–80°. The chemical status of the thin films was analyzed via X-ray photoelectrons (XPS, Nexsa, Thermo Scientific, Waltham, MA, USA), with all the XPS data calibrated based on the C1s binding energy at 284.8 eV.

3. Results and Discussion

The XRD patterns of S0–S60 deposited on sapphire substrates at different RF-bias power were measured and are shown in Figure 2a. The diffraction peaks at 2θ = 21.0° and 41.7° correspond to the (0003) and (0006) plants of c-plant sapphire. The peak at 2θ = 37.5° (marked with an asterisk) represents the K_β line associated with the 00.6 Bragg reflection [17]. Beyond that, no other observable diffraction peaks can be found in Figure 2a, indicating the amorphous feature of these four samples. Atomic force microscopy (AFM) topographies of the a-Ga₂O₃ film are shown in Figure 2b. The images indicate that the a-Ga₂O₃ film without RF-bias exhibits numerous protrusions. As the RF-bias power increases, both the number and size of these protrusions decrease. The roughness of the a-Ga₂O₃ film without RF-bias shows a higher roughness value of R_rms ~0.98 nm. As RF-bias power increases from 0 W to 60 W, the R_rms significantly decreases from 0.98 nm to 0.13 nm. While the alternative electric field is applied on the substrate through the RF-power supply and match network, a DC self-bias potential difference builds up at the surface of the substrate. This DC potential difference accelerates positively charged ions, which then collide with the atoms sputtered from the targets, transferring essential energy for the structural modification of the sputtered particles. Consequently, the a-Ga₂O₃ films deposited with RF-bias power exhibit smooth and featureless surfaces.

Figure 3a presents the optical transmittance spectra of Ga₂O₃ samples deposited under various substrate RF-bias powers ranging from 0 W to 60 W. A distinct absorption edge at approximately 280 nm demonstrates the high visible light transmittance and selective absorption against solar-blind light of Ga₂O₃. The Tauc relation was used to determine the E_g of Ga₂O₃ films by the linear extrapolation of the absorption edge, which is show in Figure 3b. The values of E_g were extracted using Equation (1) [18]:

$$\mathsf{\alpha }\mathrm{h}\mathsf{\nu }=\mathrm{A}{(\mathrm{h}\mathsf{\nu }-{\mathrm{E}}_{\mathrm{g}})}^{1/2}$$

(1)

where α is the absorption coefficient, hν is the photon energy, and A is a constant. In Figure 3b, the plot illustrates the variation in Eg as a function of the RF-bias powers, revealing values close to 5.05 eV with a minor variation within 0.1 eV. In addition, the Urbach tail (E_U) serves as a crucial indicator for analyzing the structures of amorphous semiconductors, and it is usually associated with the structural disorder of amorphous semiconductors. Greater disorder in the structure corresponds to a broader local electron band tail within the pseudogap, leading to an increase in the magnitude of E_U [19], which is described by

$$\ln \mathsf{\alpha }=\mathrm{h}\mathsf{\nu }/{\mathrm{E}}_{\mathrm{U}}+\ln \mathrm{A}$$

(2)

where A is a constant. Accordingly, E_U can be extracted from the inverse of the slope based on a linear fit for a plot of lnα versus hν. As shown in Figure 3c, the E_U of Ga₂O₃ films without RF-bias is 299 meV. When the RF-bias power adds to 20 W, E_U is slightly reduced to 290 meV, and then increased with the RF-bias power, reaching 447 meV at 60 W. A larger E_U indicates a higher density of structural disorder and defects [20]. Thus, the application of 20 W plasma power slightly reduces the E_U, suggesting that an appropriate plasma power can improve the structural disturbances. During the plasma-assisted deposition of Ga₂O₃, the high-energy plasma could provide a specific energy for the sputtered particles, and it would be helpful for the atoms move to a stable position, thereby promoting a more orderly structure of the films. However, the elevated ion’s bombardment energy can disrupt the chemical bonds within the film, resulting in the formation of additional defects and consequently inducing increased disorder in the Ga₂O₃ film.

The photoelectric performance was also measured both in the dark and under 254 nm DUV radiation, as shown in Figure 4a,b, respectively. It is obvious that the photoelectric characteristics of the a-Ga₂O₃ PDs are critically sensitive to the RF-bias power. The a-Ga₂O₃ PDs without RF-bias exhibited the lowest photocurrent (I_photo), which was only 1.12 × 10⁻⁸ A at a bias voltage of 20 V. Then, the I_photo continues to rise with the augmentation of RF-bias power, reaching values from 1.20 × 10⁻⁶ A to 8.64 × 10⁻⁶ A. The variation trend in the dark current (I_dark) is similar to that of I_photo, with I_dark experiencing a significant boost following the application of substrate RF-bias, and the current continues to escalate with the increasing RF-bias power [21]. The responsivity (R) of the device was calculated using Equation (3) [22] and is shown in Figure 4c:

$$\mathrm{R}={(\mathrm{I}}_{\mathrm{p}\mathrm{h}\mathrm{o}\mathrm{t}\mathrm{o}}-{\mathrm{I}}_{\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{k}})/{\mathrm{P}}_{\mathsf{\lambda }}\mathrm{S}$$

(3)

where P_λ is the irradiation power density, and S is the effective illuminated area. Furthermore, EQE is defined as [21]

$$\mathrm{E}\mathrm{Q}\mathrm{E}=\mathrm{R}\mathrm{h}\mathrm{c}/\mathrm{q}\mathsf{\lambda }$$

(4)

where h is Planck’s constant, c is the velocity of light, q is the elementary charge, and λ is the wavelength of the incident light. In addition, detectivity (D*) is another key figure of merit, which was employed to evaluate the ability of the photodetector to sense light signals from noise. It is usually calculated using the following formula [23]:

$${\mathrm{D}}^{\mathrm{*}}=\mathrm{R}\sqrt{\mathrm{S}/2\mathrm{q}{\mathrm{I}}_{\mathrm{dark}\mathrm{a}\mathrm{r}\mathrm{k}}}$$

(5)

The highest D* appears at an RF-bias power of 20 W, which is attributed to high responsivity, and at a I_photo/I_dark of S20. The performances of all a-Ga₂O₃ photodetectors with and without substrate RF-bias during sputtering are summarized in

Table 1. The performance parameters of amorphous Ga₂O₃ photodetectors fabricated under different RF-bias powers ^a.

Device	Power	R	D*	EQE	τr	τd
	(W)	(A/W)	(Jones)	(%)	(s)	(s)
S0	0	0.34	8.31 × 1013	1.66 × 102	1.49	0.24
S20	20	30.59	6.20 × 1014	1.49 × 104	1.15	0.12
S40	40	120.04	9.32 × 1013	5.87 × 104	3.13	6.05
S60	60	160.06	6.11 × 1013	7.83 × 104	4.99	>10

^a The bias voltage was 20 V.

. Although the S40 and S60 have a larger I_photo compared with S20, the similarly larger I_dark implies that because of the larger background noise in these two PDs, the D* of the PDs deteriorates after the RF-bias power exceeds 20 W. The response properties of S0–S60 are shown in Figure 4d. The peak response of S0–S60 slightly red-shift from 248 nm (S0) to 254 nm (S20) and 260 nm (S40 and S60), all in the solar-blind region, indicating the good wavelength selectivity of the Ga₂O₃ PDs.

In general, the concentration of oxygen vacancies in Ga₂O₃ films is a key factor affecting the performance of Ga₂O₃-based photodetectors [8,16]; therefore, the X-ray photoelectron spectroscopy (XPS) measurement was performed on the samples of S0–S60. As shown in Figure 5a, the O1s core-level spectrums were deconvoluted into three components based on Gaussian fitting analysis: (O_I) 530.6 eV, which corresponds to the Ga-O bond in Ga₂O₃ film; (O_II) 531.2 eV, attributed to the O²⁻ ions in an oxygen-deficient region, which is generally considered to be the oxygen vacancies in Ga₂O₃; and (O_III) 532.2 eV, usually related to chemisorbed species on the surface of the film [15,24]. Generally, the intensity ratios of O_II/(O_I + O_II) are used to reflect the density of oxygen vacancies in the film, and the oxygen vacancies of S0–S60 were determined to be 33.35%, 34.47%, 40.18%, and 43.44%, respectively. The spectral analysis of Ga2p_3/2 reveals a similar trend. It can be seen from Figure 5b that the intensity area ratio of Ga³⁺/(Ga⁺ + Ga³⁺) of S0–S60 exhibits a continuous decrease from 63.52% to 49.21%, indicating a significant increase in the concentration of Vo. Clearly, the introduction of substrate plasma and the application of a higher RF-bias power result in a substantial increase in the concentration of Vo within the Ga₂O₃ film. According to a previous report [25], the deep-level neutral oxygen vacancies (Vo) can be excited by illumination, thus generating ionized oxygen vacancies (Vo²⁺) and the photogenerated electron that excites the conduction band. Due to the existence of energy barriers, Vo²⁺ does not immediately recombine with photogenerated electrons, resulting in a long lifetime of Vo²⁺ and photogenerated electrons. The longer the lifetime of free photogenerated electrons, the larger the gain and responsivity, as well as the longer photocurrent rise and decay times. Consequently, the introduction of plasma could enhance the device performance by effectively modifying the concentration of oxygen vacancies in Ga₂O₃. During the sputtering process, ion bombardment can peel off the molecules with weak bonding, resulting in the destruction of some Ga-O bonds and the increase in Vo within the film.

Response time is another important parameter for photodetectors, so a time-dependent response test of a-Ga₂O₃ photodetectors was performed at 20 V bias to investigate the PDs’ response speed and repeatability. The normalized transient responses of S0–S60 are shown in Figure 6a, for which 254 nm UV light was periodically on/off during the test. The rise time (τ_r) is defined as the time required for the current to transform from 10% of the peak value to 90%, and the decay time (τ_d) is defined as the time required for the current to transform from 90% of the peak value to 10% [15]. The τ_r/τ_d of these four samples were 1.49 s/0.24 s, 1.15 s/0.12 s, 3.13 s/6.05 s, and 4.99 s/>10 s, respectively. Notably, an appropriate RF-bias power contributes to the improvement in response speed, potentially linked to the earlier-discussed enhancement in the Ga₂O₃ structure. However, the τ_r/τ_d experiences a significant increase with the further elevation of RF-bias power. In addition, it can be seen that S0 has no significant persistent photoconductivity (PPC) effect after introducing the substrate RF-bias in the magnetron sputtering process; only S20 still shows the sharp falling edge, and S40 and S60 both exhibit severe PPC phenomenon. Overall, in comparison with other reported Ga₂O₃ PDs, as listed in

Table 2. Performance comparison of the Ga₂O₃-based solar-blind UV photodetectors prepared using different deposition techniques to date.

Method of Deposition	Bias (V)	R (A/W)	D* (Jones)	τd (s)	Ref.
sputtering	20	30.59	6.20 × 1014	0.12	This work
MOCVD	10	38.82	9.0 × 1015	0.50	[16]
exfoliation	30	1.68	3.73 × 1010	0.24	[29]
PA-MBE	20	170.2	1.3 × 1014	2.1	[30]
Sol–gel	15	0.028	5.41 × 1011	0.04	[31]
PECVD	0	2.6 × 10−4	6.67 × 1010	0.013	[14]
MOCVD	5	7.9 × 10−2	-	0.80	[32]
sputtering	8	0.78	6.22 × 1010	0.154	[33]

, both PDs in this work exhibited faster response and recovery. According to the XPS results, there are higher concentrations of oxygen vacancy in S40 and S60. Generally, the PPC phenomenon in oxide semiconductors is attributed to the increase in oxygen vacancy [26]. Upon turning off the light, conduction band electrons and valence band holes undergo rapid recombination [27,28]. However, photogenerated carriers captured by deep-level Vo will gradually release, causing a gradual decline in the device current, resulting in the phenomena of PPC [26]. A repetition test of the four samples was conducted, and the results are shown in Figure 6b. All Ga₂O₃ photodetectors exhibited good cyclicity during the measurement of their time-dependent response, demonstrating excellent reliability.

4. Conclusions

In summary, amorphous Ga₂O₃ solar-blind photodetectors were fabricated with CCP-assistant magnetron sputtering at room temperature. The effects of the RF-bias power on the morphology of Ga₂O₃ films and the performances of photodetectors were investigated. Through increasing the RF power from 0 W to 60 W in the sputtering process, the responsivity exhibited a robust improvement, while the response speed continued to slow down. Appropriate RF power contributes to the overall enhancement of photodetector performance. According to the XPS results, the increase in oxygen vacancy concentration caused by plasma is the reason for the change in device performance. The best overall performance of a-Ga₂O₃ photodetectors fabricated at 20 W exhibits a high responsivity of 30.59 A/W, a low dark current of 4.18 × 10⁻¹¹ A, and a decay time of 0.12 s. This research prepared a high-performance a-Ga₂O₃ based photodetector at room temperature and provided a promising way to balance the device performance.