AlN Passivation-Enhanced Mg-Doped β-Ga₂O₃ MISIM Photodetectors for Highly Responsive Solar-Blind UV Detection

Abstract

Mg-doped gallium oxide films were prepared on single crystal sapphire substrates through radio frequency magnetron sputtering technology, and then AlN films of different thicknesses were deposited on them as passivation layers. Finally, Pt interdigitated electrodes were prepared through mask plate and ion sputtering technology to make metal–insulator–semiconductor–insulator–metal (MISIM) photodetectors. The influence of the AlN passivation layer on the optical properties and photodetection performance of the device was investigated using UV-Vis (ultraviolet-visible absorption spectroscopy) spectrophotometer and a Keith 4200 semiconductor tester. The device’s performance was significantly enhanced. Among them, the MISIM-structured device achieves a responsivity of 2.17 A/W, an external quantum efficiency (EQE) of 1100%, a specific detectivity (D*) of 1.09 × 10¹² Jones, and a photo-to-dark current ratio (PDCR) of 2200. The results show that different thicknesses of AlN passivation layers have an effect on the detection performance of Mg-doped β-Ga₂O₃ films in the UV detection of the solar-blind UV region. The AlN’s thickness has little effect on the bandgap when it is 3 nm and 5 nm, and the bandgap increases at 10 nm. The transmittance of the film increases with the increase in AlN thickness and decreases when the AlN’s thickness increases to 10 nm. The photocurrent exhibits a non-monotonic dependence on AlN thickness at 10 V, and the dark current gradually decreases. The thickness of the AlN passivation layer also has a significant impact on the response characteristics of the detector, and the response characteristics of the device are best when the thickness of the AlN passivation layer is 5 nm. The responsiveness, detection rate, and external quantum efficiency of the device first increase and then decrease with the thickness of the AlN layer, and comprehensive performance is best when the thickness of the AlN passivation layer is 5 nm. The reason is that the AlN layer plays a passivating role on the surface of Ga₂O₃ films, reducing surface defects and inhibiting its capture of photogenerated carriers, while the appropriate thickness of the AlN layer increases the barrier height at the semiconductor interface, forming a built-in electric field and improving the response speed. Finally, the AlN layer inhibits the adsorption and desorption processes between the photogenerated electron–hole pair and O₂, thereby retaining more photogenerated non-equilibrium carriers, which also helps enhance photoelectric detection performance.

1. Introduction

Deep ultraviolet (DUV) light, with a wavelength of <280 nm, is known as “solar-blind” light because it cannot penetrate the Earth’s atmosphere, as it is strongly absorbed by the stratospheric ozone layer [1]. Due to their low environmental interference, solar-blind photodetectors (PDs) have shown great potential in many fields, such as ozone monitoring, fire warning, and missile tracking [2,3,4,5]. Among the many semiconductor materials, β-Ga₂O₃ has attracted considerable attention in the development of solar-blind photodetectors owing to its suitable bandgap (4.4–5.3 eV) and ability to respond to deep ultraviolet wavelengths [6,7,8]. Among the various crystal phases of Ga₂O₃, β-Ga₂O₃ exhibits excellent thermal and chemical stability, providing a guarantee for the stable operation of photodetectors in complex environments. However, there are still many challenges in improving the performance of β-Ga₂O₃ photodetectors, such as carrier recombination issues and interface state effects, which limit their further development in the field of high-performance detection [9]. To overcome these problems, elemental doping is an effective method [10,11,12]. Mg doping can adjust the electrical properties of β-Ga₂O₃, such as modifying the bandgap and regulating the carrier concentration, and is thus expected to enhance the detection performance of β-Ga₂O₃-based photodetectors [13,14,15,16]. β-Ga₂O₃ solar-blind photodetectors have expansive applications, from civil fields (ozone monitoring, flame early warning) to military use (missile tail flame detection) [16]. The synergy of interface engineering and doping modification is key to improving device performance, supporting this study’s design of Mg-doped β-Ga₂O₃ with an AlN passivation layer [17].

In addition to elemental doping for Ga₂O₃ films, a passivation layer can be deposited on Ga₂O₃ films to form a MISIM structure photodetector, which can improve photodetection performance by reducing the defect state on the surface of the film [18,19]. Jing et al. prepared and compared a β-Ga₂O₃ metal–insulator–semiconductor–insulator–metal (MISIM) photodetector passivated with an ultra-thin Al₂O₃ layer and explored the physical mechanism behind the passivation technology. The results show that the 3 nm Al₂O₃ layer passivated device decreases the dark current and increases the photocurrent and responsivity, and the responsivity and specific detection rate reach 83.3 A/W and 1.35 × 10¹⁵ Jones, respectively, due to the change in the carrier transport mechanism and effective physical isolation at the metal/semiconductor interface. Studies have shown that precise control of the thickness of the Al₂O₃ passivation layer can comprehensively improve the performance of the β-Ga₂O₃ solar-blind photodetector, which provides an effective technology for reducing energy consumption and noise levels and improving performance [20]. Zhang et al. studied the fabrication of β-Ga₂O₃ MSM and MISIM solar-blind photodetector photodetectors, controlling their performance by inserting AlN layers of different thicknesses into the MISIM structure. The results show that the responsivity of the MISIM detector with a 3 nm AlN layer reaches 482 A/W, the EQE is 2.36 × 10⁵%, the response time is 0.10 s, and the dark current is 0.17 nA. XPS analysis shows that the AlN/β-Ga₂O₃ interface is arranged in type I bands, and the migration of carriers at the interface explains the performance change. It has been shown that the AlN layer can improve the performance of β-Ga₂O₃ photodetectors through interface engineering [21]. Moreover, Dong et al. introduced a high-κ HfO₂ interlayer into β-Ga₂O₃ MIS solar-blind photodetectors via atomic layer deposition, effectively reducing the dark current by nearly two orders of magnitude (from 2.53 × 10⁻¹² A to 4.29 × 10⁻¹⁴ A) and enhancing the detectivity to 2.39 × 10¹² Jones. The improvement was attributed to the high dielectric constant and insulating nature of HfO₂, which increased the Schottky barrier height, suppressed surface and interface defects, and accelerated carrier transport. The study demonstrates that HfO₂ passivation is an efficient strategy to suppress leakage current and boost the performance of β-Ga₂O₃ solar-blind photodetectors [22]. Lv et al., from our group, prepared β-Ga₂O₃ MISIM solar-blind photodetectors with AlN passivation layers of varying thicknesses. The device with a 10 nm AlN layer showed the best performance, with responsivity of 8.4 A W⁻¹, EQE of 4108%, detectivity of 3.6 × 10¹² Jones, and PDCR of 8.4 × 10⁴ at 10 V. Optical measurements indicate strong absorption in the 200–300 nm band and a blue shift of the absorption edge with increasing AlN thickness. Electrical analysis shows that the performance improvement mainly arises from photo-assisted Fowler–Nordheim tunneling and interface passivation (reduced traps and enhanced carrier separation); AlN’s polarization also contributes. The study demonstrates that optimizing AlN thickness via interface engineering can significantly enhance β-Ga₂O₃ solar-blind photodetector performance [23].

Although some results have been achieved in researching β-Ga₂O₃ photodetectors, the research on AlN passivation layers combined with Mg-doped β-Ga₂O₃ MISIM-type photodetectors is still quite scarce, and how the performance of the AlN passivation layer affects Mg-doped β-Ga₂O₃ MISIM photodetectors and their internal physical mechanisms is not fully understood. Therefore, this work focuses on the AlN passivation layer Mg-doped β-Ga₂O₃ MISIM photodetector, deeply explores the mechanism of the AlN passivation layer, and systematically studies the influence of key factors in the preparation process on the performance of the detector, aiming to provide a theoretical basis and experimental support for the development of high-performance solar-blind photodetectors.

2. Experimental

In this experiment, a JCP-350M2 high-vacuum multi-target magnetron sputtering coating machine (Technol Co., Ltd., Beijing, China) was used to sputter and deposit Mg-doped gallium oxide films on c-(001) sapphire and (100) monocrystalline silicon substrates, and monocrystalline silicon was used as the substrate, mainly for EDS testing, to obtain the composition of the thin films [11]. A mosaic gallium oxide ceramic target (99.99% purity) with different amounts of Mg metal blocks was used to control the Mg doping concentration. The sputtering power was 100 W, the argon flow rate was 25 sccm, the working air pressure was 0.2 Pa, and the sputtering time was 90 min. AlN targets were used to deposit passivating layers of different thicknesses (0 nm, 3 nm, 5 nm, and 10 nm), and the samples were labeled as S0, S1, S2, and S3, respectively. Finally, four samples, including S0–S3, were annealed at 800 °C for 1 h in an argon atmosphere. In order to realize current collection, Pt interdigitated electrodes were fabricated via ion sputtering combined with a metal mask and the electrode’s dimensions were listed as follows: electrode thickness: approximately 30 nm; finger width: 200 $\mathsf{\mu }$m; finger length: 2800 $\mathsf{\mu }$m; finger spacing: 200 $\mathsf{\mu }$m. These parameters guaranteed good contact with test probes; moreover, all samples were prepared using the same mask, resulting in an electrode structure consistency error of <2%. Thus, the MSM photodetector and MISIM sunblind photodetector were obtained, and their photoelectric detection performance was evaluated and analyzed.

The optical properties of the samples were characterized using ultraviolet-visible absorption spectroscopy (UV-Vis) (Agilent Technologies, Santa Clara, CA, USA), and the photoelectric detection performance of the prepared MSM-type photodetectors (PDs) was evaluated on the Keithley 4200-SCS semiconductor test platform (Keithley Co., Ltd., Solon, OH, USA) using dual-wavelength (254 nm and 365 nm) ultraviolet light illumination.

To verify the phase purity of Mg-doped β-Ga₂O₃ films and the crystallinity of AlN passivation layers, Raman spectroscopy with 532 nm laser excitation was collected, as shown in Figure 1. We can see that the AlN-free sample (S0) shows β-Ga₂O₃ characteristic peaks at 197 cm⁻¹, 344 cm⁻¹, and 414 cm⁻¹ without impurity peaks, confirming no phase transition from Mg doping. While the 10 nm AlN sample shows sharp peaks at 613 and 654 cm⁻¹ (A1(LO) mode of hexagonal wurtzite AlN, indicating good AlN crystallinity and a stable interface) and no significant attenuation of β-Ga₂O₃ peak intensity (proving no damage to the underlyingβ-Ga₂O₃ film’s crystalline integrity from AlN deposition and annealing).

3. Results and Discussions

3.1. Light Absorption and Bandgap

As we all know, optical absorption measurements are of crucial significance in determining optical parameters, such as the absorption coefficient and the optical bandgap of thin films. Figure 2a shows the UV-Vis (UV-Vis) absorption spectra of Mg-doped β-Ga₂O₃ films with different AlN thicknesses. The absorption coefficient of Mg-doped β-Ga₂O₃ films was calculated from the transmittance (T) data using the formula [24]

$$\begin{array}{c}\alpha ={\displaystyle \frac{-\ln \left(\mathrm{T}\right)}{\mathrm{d}}} \end{array}$$

(1)

where T denotes the transmittance of the film–substrate system and d is the thickness of the Mg-doped β-Ga₂O₃ film. This calculation accounts for the influence of light reflection and film thickness on transmittance, ensuring the accuracy of the obtained absorption coefficient for subsequent bandgap analysis.

It can be seen that the Mg-doped films with different AlN thicknesses β-Ga₂O₃ exhibit significant UV absorption in the sunblind wavelength band of 200–300 nm, and a steep optical absorption edge appears. In view of the fact that β-Ga₂O₃ is a typical direct bandgap semiconductor material, the following relationship between the optical bandgap and the absorption coefficient is met experimentally [25]:

$$\begin{array}{c} \alpha \left(\mathrm{h}\mathsf{\nu }\right)= {\mathrm{B}\left(\mathrm{h}\mathsf{\nu }-{\mathrm{E}}_{\mathrm{g}}\right)}^{{\displaystyle \frac{1}{2}}} \end{array}$$

(2)

where hν is the photon energy, B is the constant, Eg is the optical bandgap, and α is the absorption coefficient; hν and (αhν)² are used to fit the curves in the transverse and ordinate coordinates. When the ordinate abscissa value is 0, the intercept between the tangent line of the curve and the transverse axis is the value of the bandgap of the thin film, as shown in the inset (a), and it can be found that the thickness of the AlN layer has little effect on the Mg-doped β-Ga₂O₃ bandgap. When the AlN layer’s thickness is 3 nm and 5 nm, the bandgap values are around 5.17 eV. This may be due to the fact that the AlN layer is thinner and therefore has less of an effect on the light absorption of the film. When the thickness of the AlN layer reaches 10 nm, the bandgap value increases to about 5.24 eV, as shown in the inset of Figure 2b, because the bandgap value of AlN is larger than that of β-Ga₂O₃ [26,27], which increases the overall bandgap width.

Figure 2b presents the light transmittance curves of Mg-doped β-Ga₂O₃ films with different AlN thicknesses. It can be observed that all four types of films possess relatively high transmittance. Among them, the transmittance of S0, S1, and S2 is greater than 80%, and with the increase in AlN thickness, transmittance shows an upward trend. The possible reason is that when light travels from AlN to β-Ga₂O₃, reflection and refraction phenomena will occur at the interface between the two media. The refractive index of the AlN transparent film is between that of air and β-Ga₂O₃. When light irradiates onto the AlN film, reflections will be generated on the upper and lower surfaces of the film, thus forming two reflected light beams. These two reflected light beams can interfere destructively, thereby reducing the intensity of the reflected light and increasing the intensity of the transmitted light [28,29]. In addition, the AlN film can make the film’s surface smoother and flatter, which helps to reduce the diffuse reflection of light and further improve the transmittance. As for S3, its transmittance is somewhat decreased, which may be due to the excessive thickness of the film, which is not conducive to the occurrence of the interference phenomenon.

3.2. I-V Curve and I-T Response Curve

The I-V curves of Mg-doped β-Ga₂O₃ metal–semiconductor–metal (MSM) and metal–insulator–semiconductor–insulator–metal (MISIM) Schottky barrier photodetectors (SBPDs) were measured under dark conditions and upon irradiation with 254 nm ultraviolet light (2 μW/cm²), respectively. The results are presented in Figure 3a. It can be observed that the dark current of the MISIM photodetector is relatively less affected by bias voltage, with weak symmetry between the photocurrent and the dark current; both currents exhibit strong Schottky contact characteristics. Additionally, at an applied bias of 10 V, the photocurrent first increases and then decreases with increasing AlN thickness, while the dark current decreases gradually. Figure 3b shows the current values for different AlN thicknesses at a 10 V bias. The dark current of the β-Ga₂O₃ MSM-type SBPD is 0.16 nA, whereas the dark currents of the MISIM-type SBPDs with 3 nm, 5 nm, and 10 nm AlN layers are 0.12 nA, 0.09 nA, and 0.08 nA, respectively. However, the introduction of the AlN layer only slightly increases the photocurrent of the MISIM-type SBPDs, and a relatively higher photocurrent (218 nA) is observed in the MISIM-type SBPD with a 5 nm AlN layer. Compared with β-Ga₂O₃, AlN has greater resistance and a wider bandgap; the larger conduction band offset between them is highly conducive to the establishment of a strong built-in electric field, which can enhance carrier transport at the AlN/β-Ga₂O₃ interface. Moreover, AlN forms a higher Schottky barrier height with platinum electrodes [30]. Nevertheless, when the AlN layer is ultra-thin, electrons tend to tunnel directly through the layer and then drift at the AlN/β-Ga₂O₃ interface under the action of the built-in electric field; this causes the dark current of the MISIM SBPD to be higher than that of the MSM-type SBPD at low bias voltages. As the bias voltage increases, the electron emission via field emission (FE) and thermionic field emission (TFE) in the MISIM-type detector decreases [31], leading to a reduction in dark current.

The I-T plot shows the light response characteristics of the device when the UV lamp is turned on and off at different time points, as shown in Figure 4c. The test conditions enable observation of the instantaneous response characteristics of the device to ultraviolet light irradiation with a wavelength of 254 nm (optical power density of 2 μW/cm²) and three bias voltages of 5 V, 10 V, and 15 V at both ends of the device. As you can see, all devices exhibit good repeatability and stability, as well as a remarkable light response. Under the same bias condition, with the increase in the thickness of the AlN interlayer, the photocurrent first increases and then decreases.

3.3. Transient Response Characteristics and Photoconductive Gain

The transient response curve is an important means to characterize the dynamic characteristics of the detector, which can show the current variation characteristics of the detector over time after being stimulated by the change in external light intensity, which is helpful for in-depth understanding and evaluation of the detector performance. Figure 5a–c show the transient response curves of the S0, S2, and S3 devices under 254 nm UV irradiation, respectively, and the results of fitting these curves using the double exponential relaxation equation. The equation is as follows:

$$\begin{array}{c}\mathrm{I}={\mathrm{I}}_0+\mathrm{A}{\mathrm{e}}^{-{\displaystyle \frac{\mathrm{t}}{{\mathsf{\tau }}_1}}}+\mathrm{B}{\mathrm{e}}^{-{\displaystyle \frac{\mathrm{t}}{{\mathsf{\tau }}_2}}}\end{array}$$

(3)

where I₀ represents the initial steady-state current, A and B are the parameters used for fitting, t represents time, and τ₁ and τ₂ are the relaxation time constants for the fast and slow ascent phases, respectively. The on–on light and light-off response constants are denoted as τ_r and τ_d, respectively. The fast rise time and the fast fall time are compared in Figure 5d, and it can be seen that the rise and fall times both decrease and then increase with the thickness of the AlN interlayer. This may be due to the fact that the surface passivation of the β-Ga₂O₃ film by the AlN layer effectively reduces the capture of photogenerated holes by surface traps, thereby improving the response speed of MISIM SBPDs. However, as the thickness of the AlN layer increases, the probability of carrier tunneling decreases, and it takes more time for the carriers to pass through the AlN layer, so the response speed decreases. In summary, when the AlN layer’s thickness is 5 nm, the device has a short response time of 0.16 s for a fast rise and 0.01 s for a fast fall.

Photoconductive gain refers to the amplification of the current generated by photogenerated carriers under the action of an electric field relative to the number of original carriers generated by photoexcitation in photoconductive materials. It reflects the combined effect of multiple physical processes, such as carrier generation, transport, and recombination, in the photoconductive process, and the photoconductive gain (G) is defined as follows [32]:

$$\begin{array}{c}\mathrm{G}={\displaystyle \frac{\frac{{\mathrm{I}}_{\mathrm{p}} - {\mathrm{I}}_{\mathrm{d}}}{\mathrm{e}}}{{\mathrm{P}}_{\mathrm{l}}\cdot \frac{\mathrm{S}}{\mathrm{h}\mathsf{\nu }}}}\end{array}$$

(4)

where I_p is the photocurrent, I_d is the dark current, e is the amount of electron charge, P_l is the optical power, S is the illumination area, h is the Planck constant, ν is the frequency of the light, and the G_S0 and G_S2 of the S0 and S2 data are 12.12 and 14.66, respectively, under the 15 V bias, which indicates that the AlN layer improves the photoconductance gain of the detector. On the one hand, the AlN layer, as a sandwich, has the effect of passivating the interface, which can reduce the interfacial density of states on the surface of β-Ga₂O₃. In MSM structures, due to the lack of an effective passivation layer, the surface state may trap photogenerated carriers, resulting in a decrease in the number of carriers involved in conduction, which in turn reduces the photoconductance gain. In the MISIM structure, the lower interface density of states makes it easier for photogenerated carriers to be transported inside of the material, reduces the scattering and recombination of carriers at the interface, and helps to improve the photoconductance gain. On the other hand, in the MISIM structure, the internal electric field formed between the AlN layer and the β-Ga₂O₃ and electrodes is more conducive to the separation and transport of the carriers, which accelerates the drift motion of the photogenerated carriers and allows them to reach the electrodes more quickly to form currents [21]. In contrast, the internal electric field of the MSM structure is relatively weak and cannot effectively drive the carriers to move quickly, which limits the improvement of the photoconductance gain.

3.4. Key Response Parameters of Optoelectronic Devices

In order to study the photoelectric response characteristics of different optoelectronic devices in greater depth, we can calculate and compare the following key parameters. The light-to-dark current ratio [33] (PDCR) can be calculated as

$$\begin{array}{c}\mathrm{P}\mathrm{D}\mathrm{C}\mathrm{R}={\displaystyle \frac{{\mathrm{I}}_{\mathrm{p}} - {\mathrm{I}}_{\mathrm{d}}}{{\mathrm{I}}_{\mathrm{d}}}}\end{array}$$

(5)

where I_p is the photocurrent and I_d is the dark current. Responsivity [34] (R) can be expressed as

$$\begin{array}{c}\mathrm{R}={\displaystyle \frac{{\mathrm{I}}_{\mathrm{p}} - {\mathrm{I}}_{\mathrm{d}}}{{\mathrm{P}}_{\mathsf{\lambda }}\mathrm{S}}} ,\end{array}$$

(6)

where I_p is the photocurrent, I_d is the dark current, S is the effective illumination area (S ≈ 0.05 cm² in this experiment), and P_λ is the optical power density, which describes the photoelectric conversion efficiency of the device, which is defined as the ratio of the photogenerated current to incident optical power, which is usually expressed in units A/W. Detectivity (D*) can be expressed as

$$\begin{array}{c}{\mathrm{D}}^{\mathrm{*}}={\displaystyle \frac{{\mathrm{R}}_{\mathsf{\lambda }}}{\sqrt{2\mathrm{q}{\mathrm{J}}_{\mathrm{d}}}}}\end{array}$$

(7)

where R_λ is the responsivity, J_d is the dark current density, and q is 1.6 × 10⁻¹⁹ C [35], which is a comprehensive performance index that represents the detection ability of the device to the optical signal and is defined as the ratio of the optical responsivity of the device to the background noise, which is used to evaluate the sensitivity of the device. External quantum efficiency (EQE) [36] can be expressed as

$$\begin{array}{c}\mathrm{E}\mathrm{Q}\mathrm{E}={\displaystyle \frac{\mathrm{h}\mathrm{c}\mathrm{R}}{\mathrm{q}\mathsf{\lambda }}}\end{array}$$

(8)

where h is Planck’s constant, c is the speed of light, q is 1.6 × 10⁻¹⁹ C, and λ is the wavelength of incident light (254 nm), which measures the part of the optical quantum efficiency of the device that generates electron–hole pairs, which reflects the efficiency of the device in converting light energy into electrical energy.

Figure 6 shows the responsivity (R), light-to-dark current ratio (PDCR), detection rate (D*), and external quantum efficiency (EQE) curves of S0, S1, S2, and S3 devices under ultraviolet light irradiation at a wavelength of 254 nm (optical power density of 2 μW/cm²). The S2 device exhibits responsivity, a light–dark current ratio, detectability, and external quantum efficiency of 2.17 A/W, 2200, 1.09 × 10¹² Jones, and 1100%, respectively. These parameters indicate that the device has good sensitivity and detection ability for ultraviolet light signals, which is very beneficial for the development of efficient optoelectronic devices and ultraviolet detection applications.

The improvement of responsivity, detectability, and external quantum efficiency lies in the fact that the nanoscale AlN layer may produce a quantum confinement effect, and the movement of electrons and holes is confined to a small space, resulting in band splitting and energy level discretization. This effect causes a change in the effective mass of electron–hole pairs, which affects their transition probabilities and recombination probabilities [37,38]. Under illumination, it is more conducive to the generation and separation of electron–hole pairs, increasing the photocurrent and improving the photoresponsivity. In addition, the built-in electric field helps to separate the photogenerated carriers so that the electrons and holes move in different directions, reducing recombination, increasing the photocurrent, and thus improving the photoresponsivity. For MISIM Schottky barrier photodetectors, the elevation of EQE stems from the different tunneling probabilities between electrons and holes. The electrons with smaller effective mass in the β-Ga₂O₃ film are more likely to overcome the AlN barrier, resulting in an excess of photogenerated holes in the semiconductor. To maintain charge neutrality, the metal electrode supplies an excess of electrons, thereby generating an internal photoelectric gain [18]. When the thickness of the AlN layer is too great, the quantum confinement effect and electron tunneling are weakened, causing the responsivity, detectivity, and external quantum efficiency to decrease.

Figure 7a and b show the I-T response curves of S0 and S2 at different optical power densities. It can be seen that the photocurrent increases with the increase in light intensity, and the increase in light intensity will increase the absorption of photons, resulting in the generation of more photoexcited charge carriers inside of the gallium oxide material. It can be seen that the photodetector can switch well between the open and closed states.

To explore the photoelectric response mechanism, Figure 7c,d show the dependence of S0 and S2 on the photocurrent to the excitation power. To determine the photogenerated current (I_ph), the total current measured under illumination (I_p) is subtracted from the dark current (I_d), which can be expressed as (I_ph = I_p − I_d). The photocurrent (I_p) increases linearly with the increase in light intensity (P). The correlation between the photocurrent and light intensity can be mathematically expressed by the power law as I_ph = P_α, where P and α correspond to light intensity and empirical coefficients, respectively [39]. The index provides information about the charge trap of the photodetector. Under ideal trap-free conditions, α is one, and the closer it is to one, the fewer traps there are [40,41]. As shown in Figure 7c,d, the nonlinear fitting of S0 and S2 shows that α_S2 (0.94) is very close to one and greater than α_S0 (0.68), which may be attributed to the presence of the AlN layer, which reduces the defect states on the surface, increases the contribution of photogenerated carriers to the photocurrent, and reduces carrier trapping and recombination near the interface. In addition, the linear relationship between the photocurrent and light intensity is stronger (the correlation coefficient R² increases from 0.94 to 0.97), confirming that the photocurrent is proportional to the number of photogenerated carriers and that the generation of the photocurrent is determined by the number of photogenerated carriers [42].

3.5. Mechanism Analysis of Photoelectric Detection Enhancement

Figure 8 illustrates the schematic of the Mg doped β-Ga₂O₃ MSM and MISIM-type photodetector (PD), where the back-biased Schottky barrier between the electrode and the β-Ga₂O₃ plays a key role in determining the carrier transport and thus the dark current [31]. As shown in Figure 8a, when the carriers have sufficient thermal energy, they can cross the barrier through the thermal electron emission (TE) mechanism or by means of field emission (FE) and thermal field emission (TFE) mechanisms for direct and trap-assisted tunneling [31,43]. The transport of carriers is influenced by the barrier height (Ф) and the barrier width (λ) between the electrode and β-Ga₂O₃ [44]. As can be seen in Figure 8b, the inserted AlN dielectric layer has a wider barrier (Ф′) as well as additional physical thickness (w). This results in an increase in both the height and width of the barrier, which has a significant effect on the electrons emitted in the field emission (FE) or thermal field emission (TFE), resulting in a decrease in the number of emitted electrons and a significant decrease in the dark current. As can be seen from Figure 8c,d, in the case of deep ultraviolet (DUV) illumination, light excitation increases the carrier concentration, which in turn leads to a narrowing of the depletion layer (λ′) in the β-Ga₂O₃ body and a consequent decrease in the effective barrier width, although the barrier height remains basically unchanged. The reduction of the effective barrier width increases the probability of carrier tunneling on the one hand and increases the probability of thermal field emission (TFE) and field emission (FE) electrons on the other hand. In addition, the ultra-thin AlN layer strengthens the carrier tunneling effect, which is more conducive to carrier transport, which in turn promotes the increase in photocurrent.

The performance of Mg-doped β-Ga₂O₃ thin-film-based solar-blind ultraviolet detectors is related to the generation of electron–hole pairs and the adsorption and desorption processes of O₂ on the surface of the β-Ga₂O₃ thin film [20]. As shown in Figure 8e, in a dark environment, when the Mg-doped β-Ga₂O₃ thin film comes into contact with air, oxygen will spontaneously adsorb on the film’s surface and combine with the majority carriers (holes) in the film to form ${\mathrm{O}}_2^{+}$. As depicted in Figure 8f, when the Mg-doped β-Ga₂O₃ thin film is irradiated by an ultraviolet lamp with a wavelength of 254 nm, a large number of photogenerated electron–hole pairs will be generated within the film. When the electrons migrate to the film’s surface and combine with ${\mathrm{O}}_2^{+}$ to generate O₂, the desorption process of ${\mathrm{O}}_2^{+}$ occurs. The chemisorption and desorption processes of oxygen will consume carriers both in the dark and under deep ultraviolet illumination. As shown in Figure 8g,h, the chemisorption/desorption process can be effectively inhibited due to the physical isolation of the active region from the air environment by the AlN layer, thus retaining more photogenerated non-equilibrium carriers, a mechanism that may also help explain the monotonic reduction in decay time observed with increasing AlN thickness.

From the perspective of basic physicochemical properties, Ga₂O₃:Mg is a typical p-type wide-bandgap semiconductor—Mg, as a shallow acceptor impurity, replaces Ga³⁺ in the β-Ga₂O₃ lattice to form acceptor levels, and Luchechko et al. [17] identified two trap levels in it via DLTS: shallow ones (≈0.18 eV, dominating carrier transport, with a hole concentration ≈1.2 × 10¹⁶ cm⁻³ measured by the Hall effect in this study) and deep ones (≈0.65 eV, slightly capturing photogenerated carriers, partly causing the longer response time (0.32 s) of unpassivated S0 (0 nm AlN)). In terms of optical properties, Mg doping mildly modulates β-Ga₂O₃’s bandgap (~5.15–5.24 eV in this study, 5.15 eV without AlN passivation), consistent with Luchechko et al.’s conclusion that the “β-Ga₂O₃ bandgap shifts only 0.05–0.1 eV at Mg concentration ≤2 at.%”, which means Mg doping optimizes electrical properties mainly by regulating carrier concentration rather than significantly altering the optical absorption range, ensuring the device maintains an intrinsic response to the 200–280 nm solar-blind UV region (absorption coefficient of Ga₂O₃:Mg at 254 nm ≈ 1.2 × 10⁴ cm⁻¹ here). For interfacial properties and stability, Ga₂O₃:Mg’s surface tends to form oxygen vacancy defects due to Mg segregation (increasing surface state density, which Luchechko et al. [17] noted intensifies O₂ adsorption/desorption to raise the dark current and reduce photoresponse stability), while the AlN passivation layer in this study effectively suppresses this issue; via physical isolation and interface charge regulation, it reduces Ga₂O₃:Mg’s surface state density from ≈1.5 × 10¹² cm⁻² (S0) to ≈8 × 10¹¹ cm⁻² (S2 with 5 nm AlN), decreasing the dark current from 0.16 nA to 0.09 nA and improving photocurrent stability (I-T curve fluctuation < 5%).

In short, the performance enhancement of MISIM structured photodetectors mainly stems from three aspects: first, the AlN layer has the effect of passivating the interface, which can effectively reduce the density of the β-Ga₂O₃ surface interface state. Secondly, the band alignment effect leads to the formation of an internal electric field between the AlN layer and β-Ga₂O₃, which is conducive to the separation and transmission of carriers and can accelerate the drift movement of photogenerated carriers and make them reach the electrode faster to form higher photogenerated currents. Finally, the AlN layer inhibits the adsorption and desorption processes between the photogenerated electron–hole pair and O₂, thereby retaining more photogenerated non-equilibrium carriers.

Table 1. Comparison of the main performance indicators of Ga₂O₃ thin-film-based photodetectors at home and abroad.

Material	Type	R(A/W)	EQE	D* (Jones)	PDCR	Ref.
β-Ga2O3	MSM	0.74	3.603%	0.34 × 1011	103	[45]
β-Ga2O3	MSM	1.48	727%	—	340	[46]
β-Ga2O3	MISIM	4.12	4000%	—	—	[47]
Al2O3-β-Ga2O3	MISIM	83.3	—	1.35 × 1015	—	[20]
HfO2-β-Ga2O3	MIS	1.2	600%	2.39 × 1012	1.2 × 103	[22]
Si-β-Ga2O3	MSM	5	2500%	—	9	[48]
Cu-β-Ga2O3	MSM	1.73	841%	5.56 × 1012	372	[49]
Mg-β-Ga2O3	MSM	0.14	—	—	338	[50]
Mg-β-Ga2O3	MSM	1.3	750%	0.4 × 1012	800	This work
Mg-β-Ga2O3	MISIM	2.17	1100%	1.09 × 1012	2200	This work

summarizes and compares the responsivity (R), external quantum efficiency (EQE), detectability (D*), and light-to-dark current ratio (PDCR) in this study and others. Compared to intrinsically doped β-Ga₂O₃ with other elements, it is found that the photodetector based on Mg-doped β-Ga₂O₃ thin film obtained in this study has greater responsivity, external quantum efficiency, and detectability and a higher light–dark current ratio. The MISIM type is better than the MSM type in all aspects, so it can be seen that the preparation of the MISIM ultraviolet photodetector through Mg doping is an extremely effective way to achieve a faster response.

4. Conclusions

In this study, we investigated Mg-doped β-Ga₂O₃ MISIM photodetectors with AlN passivation layers, and we draw the following key conclusions:

• Mg-doped β-Ga₂O₃ films with AlN layers exhibit strong UV absorption in the 200–300 nm solar-blind region. A thinner AlN layer has little effect on the bandgap, whereas increasing the thickness to 10 nm causes a noticeable increase. The light transmittance for samples S0, S1, and S2 appears to be fairly high and increases with thickness, while for sample S3 the transmittance shows a decrease due to excessive thickness.
• The dark current of MISIM photodetectors is relatively less affected by bias voltage. At 10 V voltage, with the increase in AlN thickness, the photocurrent first increases and then decreases, and the dark current gradually decreases. The thickness of the AlN passivation layer also has a significant impact on the response characteristics of the detector, and the response characteristics of the device are best when the AlN passivation layer is 5 nm. The AlN passivation layer enhances the photoconductive gain of the detector, which is attributed to its modulation of interface state density, carrier transport, and the internal electric field.
• The photocurrent increases with light intensity, and the presence of the AlN layer strengthens this linear relationship while reducing surface defect states, thereby improving photocarrier dynamics. The AlN layer inhibits the adsorption and desorption processes between the photogenerated electron–hole pair and O₂, thereby retaining more photogenerated non-equilibrium carriers, which is also helpful in enhancing the photoelectric detection performance.