Ultrahigh Responsivity In₂O₃ UVA Photodetector through Modulation of Trimethylindium Flow Rate

Abstract

Oxygen vacancies (V_o) can significantly degrade the electrical properties of indium oxide (In₂O₃) thin films, thus limiting their application in the field of ultraviolet detection. In this work, the V_o is effectively suppressed by adjusting the Trimethylindium (TMIn) flow rate (f_TMIn). In addition, with the reduction of the f_TMIn, the background carrier concentration and the roughness of the film decrease gradually. And a smooth In₂O₃ thin film with roughness of 0.44 nm is obtained when the f_TMIn is 5 sccm. The MSM photodetectors (PDs) are constructed based on In₂O₃ thin films with different f_TMIn to investigate the opto-electric characteristics of the films. The dark current of the PDs is significantly reduced by five orders from 100 mA to 0.28 μA with the reduction of the f_TMIn from 50 sccm to 5 sccm. In addition, the photo response capacity of PDs is dramatically enhanced. The photo-to-dark current ratio (PDCR) increases from 0 to 2589. Finally, the PD with the f_TMIn of 5 sccm possesses a record-high responsivity of 2.53 × 10³ AW⁻¹, a high detectivity of 5.43 × 10⁷ Jones and a high EQE of 9383 × 100%. Our work provides an important reference for the fabrication of high-sensitivity UV PDs.

1. Introduction

Ultraviolet-A (UVA) photodetectors (PDs) have attracted wide attention in both civilian and military applications, such as flame detection, ultraviolet cameras, and missile early warning systems [1,2,3,4,5,6,7,8,9]. Indium compounds such as In₂Se₃, In₂S₃ and In₂O₃ are widely reported for efficient photoelectric detection [10,11,12,13,14,15,16,17,18]. Because metal oxides can be used as interface layers in metal/semiconductor contacts to change the optoelectronic properties of heterojunctions, they are widely used in optoelectronic devices [19]. Therefore, In₂O₃ has been considered a promising candidate for UVA PD fabrication among indium compounds, due to its suitable bandgap (3.7 eV), high mobility and stable chemical and optical properties [20,21,22]. Currently, UVA PDs based on In₂O₃ thin films mainly depends on the photovoltaic effect, because it can realize the efficient separation and collection of photogenerated carriers through the built-in electric field and reduce the dark current of the device [23,24,25]. The device structure based on photovoltaic effect mainly includes heterojunctions, transistors and metal–semiconductor–metal (MSM) structures. Among these, MSM PDs offer competitive advantages such as simple structure, ease of fabrication and convenient circuit integration [26].

In₂O₃ MSM PDs are typically integrated with circuits through flip-chip bonding, enabling their application in UV imaging via back illumination. Therefore, low-cost and UVA-transparent sapphire substrates are widely used for In₂O₃ thin film detectors. In addition, the epitaxy of high-crystal-quality In₂O₃ thin films has been reported through HVPE, MBE and MOCVD. Due to its ability to produce high-uniform and reproducible thin films, MOCVD offers advantages such as rapid growth rates that enhance production efficiency and reduce costs for the commercialization of high-quality In₂O₃ thin films and their detectors. However, defects such as oxygen vacancies (V_o) in the In₂O₃ thin film can deteriorate the performance of the PDs, which significantly limits their application [27]. It has been reported that the V_o has a significant effect on the metal/semiconductor interface and leads to a decrease in the Schottky barrier, resulting in a significant increase in the dark current of the detector [28]. In addition, the ionized oxygen vacancy can capture photogenerated electrons, and the release process is very slow after the light source is turned off, resulting in a persistent photoconductive effect and degradation of the response speed [29]. Ge et al. proposed hydrogen (H) doping in In₂O₃ through magnetron sputtering and proved H can occupy the V_o [30]. With the H doping concentration of 1%, the thin film showed a high mobility of 115.3 cm² V⁻¹s⁻¹. However, H exhibits low stability in In₂O₃ thin films and tends to escape at high temperatures.

In this work, we significantly suppress the defects of V_o in In₂O₃ thin films by adjusting the Trimethylindium (TMIn) flow rate (f_TMIn) during film growth. With a reduction in the f_TMIn, not only the concentration of V_o is suppressed, but also the performance of the In₂O₃ PDs is improved. Consequently, the dark current of the PDs based on the In₂O₃ thin film decreased by five magnitudes. In addition, the UV PD exhibits a high responsivity (2.53 × 10³ AW⁻¹) and a high detectivity (5.43 × 10⁷ Jones). This work provides an important reference for the fabrication of high-performance UVA PDs based on In₂O₃ thin film.

2. Materials and Methods

The unintentionally doped (UID) In₂O₃ thin films are grown on sapphire substrates using MOCVD. The growth is performed at 600 °C for 120 min. TMIn is used as the In precursor and the flow rate of O₂ (99.999%) is fixed at 1000 sccm. High-purity nitrogen (N₂) serves as the carrier gas. To construct a UVA PD, the Pt/Au (50/100 nm) bilayer metal is deposited on the In₂O₃ thin film by electron beam evaporation. Finally, the electrodes are patterned into intercalated fingers with a length of 480 μm, a width of 10 μm and a spacing of 10 μm. The electrodes contain 12 pairs of fingers, and the area of the active region is 1.3 × 10⁻³ cm².

The thickness of the In₂O₃ thin films is measured to be approximately 50 nm using SEM as shown in Figure S1 in Supplementary Materials. The crystal structure and crystallinity of In₂O₃ thin films are characterized by X-ray diffraction (XRD) with a test voltage of 40 kV, a current of 40 mA and a step size of 0.02°/s The chemical compositions of the In₂O₃ thin films are analyzed by X-ray photoelectron spectroscopy (XPS). An XPS instrument with a base pressure better than 4.4 × 10⁻⁸ Pa is used to acquire core-level XPS spectra from the sample. Monochromatic Al Kα radiation source (hv = 1486.6 eV) is employed. The size of analyzed area is 100 × 100 μm and the optoelectronic detection angle was 45°. Argon ion etching is not carried out to avoid the influence of etching damage on the atomic ratio [31]. The surface morphologies are obtained by atomic force microscopy (AFM) with am amplitude of 200~300 mV and a scanning speed of 1 Hz. The carrier concentration and mobility are tested by Hall measurements. The optoelectronic performance of the devices is measured via a probe station equipped with Keysight B1505A (Colorado Springs, United States), and a commercial monochromatic tunable light sources system (Omno150300, Abuja, NigeriaNBET), including Xenon lamp, Monochromator, chopper and optical fibers.

3. Results

The In₂O₃ thin films grown at the f_TMIn of 50, 25, 10 and 5 sccm are labeled as S₁, S₂, S₃ and S₄, respectively. In order to study the effect of f_TMIn on the crystal structure and crystallinity of the In₂O₃ thin films, XRD measurements are carried out. As shown in Figure 1a, the diffraction peak observed near 41.68° corresponds to the (0006) plane of the sapphire substrate. In addition, two diffraction peaks located at 30.65° and 64° are observed, which correspond to the (222) and (444) planes of In₂O₃. The results indicate that the heteroepitaxial In₂O₃ thin films grown on c-plane sapphire have a preferred orientation along the (222) direction. The coherent volume of specific diffraction peaks in thin films is related to the size of crystallite [32]. According to Scherrer’s relation, the size of the crystallites in S₁ to S₄ can be calculated using the following equation:

$$\mathrm{Crystallite}\mathrm{size}(\mathrm{D})=\frac{0.9\lambda }{\beta \mathrm{c}\mathrm{o}\mathrm{s}\theta }$$

(1)

where λ = 1.54 Å is the wavelength of the Cu Kα line, θ is the Bragg’s angle and β is the full width at half maximum (FWHM). Therefore, the D of S₁ to S₄ can be calculated to be 28.0 nm, 15.9 nm, 7.4 nm and 4.3 nm, respectively. In addition, the dislocation density can be expressed as

$$\mathrm{Dislocation}\mathrm{density}(\delta )=\frac{1}{D^2}$$

(2)

According to Equation (2), the δ of S₁ to S₄ can be estimated to be 1.27 × 10¹¹ cm⁻², 3.95 × 10¹¹ cm⁻², 1.83 × 10¹² cm⁻² and 5.53 × 10¹² cm⁻², respectively. The size of the crystallites has a significant influence on the crystallite number (N_C). The N_C can be estimated using the following equation:

$${\mathrm{N}}_{\mathrm{C}}=\frac{d}{D^3}$$

(3)

where d is the thickness of the thin film. Therefore, the N_C of S₁ to S₄ is 2.29 × 10¹¹ cm⁻², 1.23 × 10¹² cm⁻², 1.24 × 10¹³ cm⁻² and 6.49 × 10¹³ cm⁻², respectively. With the reduction of the f_TMIn, the film crystallite size decreases significantly and the dislocation density increases. This is due to the fact that a large number of In atoms are deposited on the substrate surface with a high f_TMIn, which promotes the growth of crystallites through the diffusion and migration of the deposited atoms [33]. Therefore, larger crystallite size can be obtained at high f_TMIn in this study. The decrease of crystallite size leads to the increase of dislocation density, so the decrease of f_TMIn will reduce the crystal quality of thin films to some extent. As shown in Figure 1b, high-resolution XRD rocking curve of (222) plane is performed to identify the crystallinity of the In₂O₃ thin films. The FWHM for (222) plane of S₁ to S₄ are extracted to be 0.26°, 0.31°, 0.44° and 0.48°, respectively. The results indicate that the crystallinity of In₂O₃ thin films slightly decreased with the decrease of the f_TMIn due to the decrease of the grain size.

The AFM measurements with the scanning area of 5 × 5 μm² are carried out to analyze the surface morphology of the In₂O₃ thin films. As shown in Figure 2a, the large island structure can be observed on the surface of the film when the f_TMIn is 50 sccm. With the reduction of the f_TMIn, the size of the island structure gradually shrinks, and the density increases gradually. The results are consisted of that of XRD. Finally, the surface of the In₂O₃ thin films gradually becomes smooth and the roughness of S₁, S₂, S₃ and S₄ are 2.41 nm, 1.08 nm, 0.56 nm and 0.44 nm, respectively. It is worth noting that compared with the scattering of carriers through crystallite boundaries, the capture and release of carriers by point defects (e.g., V_o) have a more significant effect on the optical response of the PDs [34,35].

In order to study the effect of the f_TMIn on the defects in In₂O₃ thin films, the XPS measurements are carried out. Figure 3a shows the XPS spectra of the In₂O₃ thin films grown on sapphire substrates. The binding energy of C 1s peak plus the work function of In₂O₃ is a constant value of 289.58 eV, which can be used for the alignment of XPS spectra [36]. The work function of In₂O₃ is 5.0 eV, and the C 1s peak of In₂O₃ can be set at 284.58 eV [37,38]. As shown in Figure 3a, peaks of C 1s, O 1s, In 3d, In 3d3/2, In 3p, In 3p1/2 and In 4p signals are observed in the general scan range of 0 to 1200 eV. The result indicates that the chemical composition of the In₂O₃ thin films grown on sapphire mainly consists of In and O. The position of the O 1s peak is located at 529.58 eV. The O 1s spectrum can be deconvoluted into two components. The component located at 529.58 eV (O_I) corresponds to binding energy of lattice oxygen in the In₂O₃ thin film, while the component at 531.2 eV (O_II) corresponds to binding energy of defects (V_o) [39]. The V_o in the In₂O₃ thin film can lead to a deterioration in the photo response performance of the PDs and the oxygen-rich atmosphere is beneficial for suppression of the V_o. As shown in Figure 3b–e, the intensity ratios of O_II/(O_I + O_II) for S₁ to S₄ are 27.23%, 25.78%, 25.43% and 24.78%, respectively. The result indicates that the V_o is reduced with the reduction of the f_TMIn. The suppression of V_o in In₂O₃ thin films is due to the fact that with the decrease of the f_TMIn, the V_o sates are compensated by oxygen in oxygen-rich environment [40].

It is reported that the V_o in In₂O₃ thin films can provide electrons as shallow donors, thus UID-In₂O₃ thin films often contain very high background carrier concentration [41,42,43]. To evaluate the electrical properties of the In₂O₃ crystal film, Hall measurements are carried out and the measurement structure has been illustrated in Figure 4a. As shown in Figure 4c, the carrier concentration deceases significantly with the decrease of the f_TMIn, which is beneficial to suppress the dark current of the PDs. The carrier concentrations of S₁ to S₄ are 1.53 × 10¹⁹ cm⁻³, 8.31 × 10¹⁸ cm⁻³, 4.94 × 10¹⁸ cm⁻³ and 4.31 × 10¹⁶ cm⁻³, respectively. It worth noting that the mobility also gradually decreases with the decrease of the f_TMIn. Since the dislocation density and number of crystallites of the In₂O₃ thin films increases with the decrease of the f_TMIn, the decrease of mobility is speculated to be attributed to the scattering of the grain boundary [30]. The optimization of crystallinity and mobility will be further studied in future work. In order to investigate the effect of the f_TMIn on the performance of the PDs, the MSM PDs based on S₁ to S₄ are constructed and tested. Figure 4b illustrate the Schematic diagram device structure. The PDs based on S₁ to S₄ are labeled as PD₁ to PD₄, respectively. To prevent the PDs from being damaged by too large a current, the current is limited to 100 mA in the test. Figure 4d shows the dark current (I_d) of the PDs and the inset shows the microscope image of the device structure. As shown in Figure 4d, the PD₁, PD₂ and PD₃ exhibit extremely high current due to the high carrier concentration and reach the current limits at 0.75 V, 0.95 V and 2.95 V, respectively. However, when the f_TMIn is reduced to 5 sccm, the I_d of the PDs decreases significantly to the microampere level. Compared with PD₁, the I_d of PD₄ is reduced by 4.5 × 10⁵ times at 0.75 V. It is worth mentioning that with the decrease of the f_TMIn, the devices gradually show Schottky behavior. As shown in Figure S2a–c in Supplementary Materials, there is a linear relation between the I_d and voltage for PD₁ to PD₃. The results indicate that the Pt/Au bilayer metal forms Ohmic contact with In₂O₃ thin films when the f_TMIn is 50 sccm, 25 sccm and 10 sccm. And the resistance of PD₁ to PD₃ can be calculated to be 7.08 Ω, 8.68 Ω and 15.06 Ω. Due to the low resistance between Pt/Au bilayer metal and In₂O₃ thin films, the I_d of the PDs are extremely huge. However, the relationship between voltage and I_d deviates from linearity when the f_TMIn is 5 sccm as shown in Figure S2d in Supplementary Materials. This is because with the decrease of the f_TMIn, the concentration of V_o in In₂O₃ thin films decreases and Schottky behavior recovers gradually [35]. The I_d of the devices with Schottky behavior will be significantly reduced, and the photo response ability will be improved. The results are consistent with that of Hall and XPS measurements, which indicates that the reduction of the f_TMIn can effectively suppress the V_o thus reduce the I_d of the In₂O₃ PDs.

Figure 5 shows the IV curves of PD₁ to PD₄ in the dark and under 335 nm illumination. Due to the large I_d, PD₁ and PD₂ exhibit almost no photo response as shown in Figure 5a,b. The photo-to-dark current ratio (PDCR) is an important figure-of-merit to evaluate the capacity of the photo response, which can be calculated by the following equation:

$$\mathrm{PDCR}=\frac{I_p-I_d}{I_d}$$

(4)

where I_p refers to photo current. Therefore, the PDCR of PD₁ and PD₂ can be calculated to be 0 in the test range. With the decrease of the carrier concentration, the photo response capacity of the PDs is obviously enhanced. The PDCR of PD₃ is calculated to be 0.94 at 1.5 V before the I_p reaches the limitation of the current. When the carrier concentration is 4.31 × 10¹⁶ cm⁻³ with the f_TMIn of 5 sccm, the PDCR of the PDs is dramatically improved. As shown in Figure 5d, PD₄ exhibits an obvious photo response with an extremely high PDCR of 2588 at 5 V. Furthermore, the responsivity (R) and external quantum efficiency (EQE) of the PD₄ are extracted using the following equation:

$$\mathrm{R}=\frac{I_p-I_d}{PS}$$

(5)

$$\mathrm{EQE}=\frac{Rhc}{e\lambda }$$

(6)

where P, S, h, c, e and λ refer to the intensity of the illumination (2000 μW/cm²), the effective area, the Planck constant, the speed of the light, the electric charge and the wavelength, respectively. The R and EQE of PD₄ are 2.53 × 10³ AW⁻¹ and 9383 × 100%, respectively. According to the results of XPS and Hall measurements, when the indium flow rate is 50 sccm, 25 sccm and 10 sccm, the concentration of V_o in the In₂O₃ film is higher, and the background carrier concentration is also higher due to the donor action of the V_o. Under this condition, the change of the device conductance caused by photogenerated minority carriers will not be obvious, so the device has almost no optical response. However, with the decrease of f_TMIn, the concentration of V_o decreases, the background carrier concentration of In₂O₃ thin film decreases significantly and the Schottky behavior is restored. The existence of the built-in electric field increases the resistance of the device, thus reducing the I_d of the device and improving the light response ability of the device. The above results indicate that reducing the background carrier concentration of the UID-In₂O₃ thin films by reducing the f_TMIn can effectively improve the photo response of the PDs.

In order to explain the effect of V_o on the performance of the PDs in detail, the energy band diagram of the In₂O₃ MSM PDs is shown in Figure 6. When the f_TMIn is high, the concentration of V_o in In₂O₃ thin films is also higher. On the one hand, the V_o as a donor state can increase the carrier concentration in In₂O₃ thin films, thus reducing the height of Schottky barrier. On the other hand, the V_o at the metal/In₂O₃ interface can cause Fermi pinning, which promotes the tunneling of electrons at the reverse biased Schottky junction, as the process 1 shown in Figure 6a [44,45]. What’s more, some unoccupied states, such as ionized V_o states, can capture electrons, making it easier for electrons to tunnel through the barrier. This process is marked as process 2 and is depicted in the energy band structure of Figure 6a. It is worth mentioning that the V_o defects within the tunneling distance at and near the interface can promote the occurrence of tunneling. Finally, the electrons of the devices with high V_o concentration will mainly conform to the field emission or thermal field emission mechanism, rather than the thermal emission mechanism (process 3) [46,47]. This is why the PDs have Ohmic behavior rather than Schottky behavior and the I_d is extremely high. Under 335 nm illumination, the electrons in the In₂O₃ film will transition from the valence band to the conduction band, resulting in photogenerated electron-hole pairs, as shown in process 4 in Figure 6b. Photogenerated electron-hole pairs will be separated by external electric field and built-in electric field and collected by electrodes. However, in the process of drift to the electrode, the photogenerated electrons will be trapped on the surface of the Metal/In₂O₃ thin film, due to the existence of a large number of oxygen vacancies (process 5) [48]. When the light source is turned off, the trapped photogenerated carriers will be released. However, this process is very slow, resulting in the PDs cannot quickly return to the dark state, thus deteriorating the response speed of the PDs. Therefore, in order to obtain highly sensitive and high-speed In₂O₃ MSM UV PDs, the defects such as V_o in the film should be further suppressed, which will be further studied in the future.

To obtain the specific detectivity (D*) of the PD₄, the noise characteristics are evaluated. By performing the Fourier transform of the dark current in the time domain, the noise power spectrum is obtained [49,50]. As shown in Figure 7a, the noise power density of PD4 is frequency dependent. The result indicates that the noise of PD₄ is mainly dominated by 1/f noise and g-r noise, which are charge-traps-related [51]. Therefore, although the V_o has been suppressed by reducing the f_TMIn, the defects in the In₂O₃ thin films need to be further studied. The spectra density of the noise power can be fitted with a Hooge-type equation:

$$S_n(f)=S_0(\frac{I_d^{\beta }}{f^{\xi }})$$

(7)

where S₀ is a constant, β and ξ are two fitting parameters. The calculated ξ of PD₄ is 1.91. The noise current (in) can be estimated by integrating the Hooge-type equation for a given bandwidth, B [52]:

$$i_n^2={\int }_0^B{S}_n\left(f\right)df={\int }_0^1{S}_n\left(f\right)df+{\int }_1^B{S}_n\left(f\right)df=S_0[\ln (B)+1]$$

(8)

Here, S_n(f) is assumed to be S_n(1) when f < 1 Hz. Thus, the i_n of S₄ can be estimated to be 1.68 × 10⁻⁶ A H^1/2. The D* can be expressed as

$$\mathrm{D}*=\frac{{(S·B)}^{1/2}}{NEP}$$

(9)

$$\mathrm{NEP}=\frac{i_n}{R}$$

(10)

where NEP is the noise equivalent power. Therefore, the D* of PD₄ can be calculated to be 5.43 × 10⁷ cm·Hz^1/2 W⁻¹ (Jones). Figure 7b shows the time-dependent response of PD₄ under period illumination. As shown in Figure 7b, the PD₄ exhibits excellent electric stability after multiple illumination cycles. In addition, the response time of PD₄ is extracted from the time-dependent response curve. The rise time, τ_r, (decay time, τ_d) is defined as the time during which the current rises (decays) from 10% to 90% (90% to 10%). The values of τ_r/τ_d are 0.23 s/0.24 s. Figure 7c shows the time-dependent photo response of S₄ within 30 s, and excellent periodicity can be observed.

Figure 7. (a) Noise power spectra density and (b) time-dependent response of PD₄. (c) Time-dependent response of S₄ in a 30-s period.

Figure 7. (a) Noise power spectra density and (b) time-dependent response of PD₄. (c) Time-dependent response of S₄ in a 30-s period.

Figure 8a shows the spectra response of the PD₄. As shown in Figure 8a, PD₄ exhibits a peak response in the UVA band. The UV/visible (R_peak/R₄₀₀) rejection ratio can be calculated to be 1241%. The above results indicate that the PD based on In₂O₃ thin film with the f_TMIn of 5 sccm mainly works in UVA band. Finally, compared with most of the reported UV PDs, our devices have impressive responsivity as shown in Figure 8b.

4. Conclusions

In summary, UID-In₂O₃ thin films are grown on sapphire substrates by MOCVD. And the defects such as V_o and background carrier concentration are effectively regulated by changing the f_TMIn. With the decrease of the f_TMIn, the growth environment gradually becomes oxygen-rich, so that the V_o sates are compensated by O atoms. Consequently, the background carrier concentration decreases from 1.53 × 10¹⁹ cm⁻³ to 4.31 × 10¹⁶ cm⁻³ with the reduction of the f_TMIn from 50 sccm to 5 sccm. And the I_d of the PDs is reduced by more than five orders due to the decrease of the background carrier concentration. In addition, the photo response capacity of the PDs enhances dramatically. The PDCR increases from 0 to 2588 when the f_TMIn decreases from 50 sccm to 5 sccm. Finally, a high-performance MSM PD with a high R (2.53 × 10³ A W⁻¹), a large EQE (9383 × 100%), an impressive D* (5.43 × 10⁷ Jones) and a fast response speed (τ_r/τ_d = 0.23 s/0.24 s) is obtained. This work provides an important reference for the realization of high-performance UVA PDs and has great application potential in the field of high-sensitivity UV imaging and communication.