Interface-Engineered Highly Responsive ReS₂ Photodetector

Abstract

Trap states in 2D transition metal dichalcogenides significantly affect the responsivity and response time of photodetectors, and previous ReS₂/Si-based heterojunction photodetectors have struggled to simultaneously achieve high responsivity and fast response. To address this issue, we developed a n-type ReS₂/p-type Si heterojunction photodetector through interface engineering. Specifically, the silicon substrate with a silicon dioxide dielectric layer was treated with inductively coupled soft plasma to adjust the thickness and surface states of the dielectric layer. This treatment created a multilayered heterostructure, which increased carrier concentration, effectively passivated sulfur-vacancy-induced defects, and thereby improved responsivity. Experimental results showed that the silicon-based n-type ReS₂ photodetector achieved a responsivity of 0.88 A W⁻¹ with a rapid response rise time of 2.5 s, a significant improvement from the intrinsic values of 12 mA W⁻¹ responsivity and 6 s rise time. Additionally, due to the defect-tunable nature of this pretreatment technique, the device exhibited enhanced Raman peaks and intensified photoluminescence (PL) absorption features, confirming the effectiveness of the interface engineering in optimizing device performance.

1. Introduction

In the contemporary era, the development of high-performance photodetectors holds significant importance across multiple disciplines, including quantum communications, astronomical observation, energy monitoring, biomedical applications, security systems, and aerospace technologies [1,2,3,4,5,6]. Since the discovery of graphene decades ago, two-dimensional (2D) materials have attracted extensive research interest due to their exceptional mechanical, electronic, and optoelectronic properties [7,8]. Among them, transition metal dichalcogenides (TMDCs) have emerged as particularly promising candidates [9]. These atomically thin layers are coupled through van der Waals forces [10], demonstrating superior electronic characteristics such as improved integration density and mitigated short-channel effects [11,12,13]. In certain TMDCs, the indirect-to-direct bandgap transition between bulk and monolayer structures enables diverse optoelectronic applications ranging from photodetectors to light-emitting devices [14,15]. ReS₂ stands as a paradigm in the TMDC family, with its characteristic 2D layers comprising covalently bonded S-Re-S building blocks that weakly interact through van der Waals coupling along the c-axis [16]. ReS₂ exhibits a relatively wide electronic bandgap (~1.493 eV) and undergoes an indirect-to-direct transition as the number of stacking layers decreases [17,18]. Additionally, the SiO₂/Si substrate demonstrates high carrier mobility, a large current on/off ratio, and strong absorption across visible wavelengths [19,20]. Consequently, ReS₂ emerges as a strong candidate for semiconductor applications in phototransistors and photodetectors [21,22].

To date, numerous researchers have focused on developing ReS₂-based photodetectors and phototransistors. Kang et al. (2018) pioneered a novel optoelectronic device architecture based on a graphene-ReS₂ hybrid van der Waals heterostructure, demonstrating enhanced interfacial charge transport characteristics [23]. Remarkably, this device exhibits an ultrahigh photoresponsivity of 7 × 10⁵ A W⁻¹, which originates from the synergistic effects of direct bandgap transition in the multilayer structure, high quantum efficiency, and strong light absorption, collectively contributing to enhanced photocurrent generation in the heterostructure. Hyeran Cho et al. demonstrated enhanced electrical and optoelectronic performance through an In₂O₃-decorated heterostructure architecture [24]. The photoresponsivity significantly increased from 151 A W⁻¹ to 432 A W⁻¹, while the response time remained relatively unchanged at 5 s. In 2022, the research team led by Xing Xu at Shandong University fabricated an ReS₂-Ti₃C₂ photodetector via metal doping combined with liquid-phase exfoliation and electrophoretic deposition [25]. The device demonstrated a high photocurrent density of ~5.98 A cm⁻², broad detection capability across the UV–visible spectrum, and exceptional flexibility with robust folding endurance. However, these strategies typically involve the incorporation of additional dielectric materials, nanomaterials, or even other 2D materials, while interface engineering at the SiO₂/ReS₂ junction has never been reported for performance optimization in ReS₂ photodetectors.

Through the adoption of inductively coupled plasma (ICP) treatment technology to pretreat the surface of P-type SiO₂/Si substrates, the thickness and surface state of the SiO₂ dielectric layer can be artificially regulated. The optimized surface treatment leads to a substantial enhancement in the photoelectric performance of the device, demonstrating a maximum responsivity of 0.88 A W⁻¹ and a response rise time of 2.5 s. These values represent significant improvements over the intrinsic ReS₂-based device, which exhibited a responsivity of merely 0.012 A W⁻¹ and a considerably slower response rise time of 6 s. The underlying physical mechanisms can be attributed to two key aspects. First, the treated SiO₂ surface possesses reactive oxygen bonds that form connections with the sulfur atoms in the ReS₂ channel layer. This interaction effectively reduces the optical bandgap of ReS₂, thereby enabling the detection of mid-infrared light. The second mechanism involves the accumulation of electrons at the ReS₂ interface induced by the high concentration of holes at the treated SiO₂-Si interface. This phenomenon creates a gating effect that effectively enhances the device’s photoconductive gain.

The capacitively coupled plasma (CCP) is primarily generated by applying a radio-frequency (RF) electric field between parallel electrodes, which accelerates electrons in the gas phase. These energized electrons then collide with gas molecules, leading to ionization and subsequent plasma formation. However, the collision process between accelerated high-energy electrons and gas molecules may generate secondary electrons, thereby significantly increasing the electron density in the plasma. This phenomenon leads to elevated plasma density while making precise control of ion energy particularly challenging, which could adversely affect process uniformity and device performance [26,27]. Unlike capacitively coupled plasma (CCP), inductively coupled plasma (ICP) predominantly transfers energy to electrons through an inductive coupling coil, exhibiting fundamentally distinct plasma characteristics. Electrons exhibit orbital motion around magnetic field lines under the influence of a magnetic field, demonstrating a high collision ionization efficiency with gas molecules. Additionally, inductively coupled plasma (ICP) allows for the control of plasma density and electron energy by adjusting the radio-frequency (RF) power of the coil. Therefore, the plasma density is low, and the ion energy is easy to control (which is why we refer to ICP as a soft plasma) [28]. By treating the substrate with inductively coupled plasma (ICP), we can generate precisely controlled SiO₂ layers and interface states, enabling charge accumulation at the interface gate and achieving high gain in n-type 2D material photodetectors.

2. Materials and Methods

The study involved the fabrication of a broadband n-type ReS₂/p-type Si heterojunction photodetector using interface engineering techniques. A degenerate p-type silicon substrate with a 300 nm SiO₂ dielectric layer was pretreated with inductively coupled soft plasma (ICSP) gases, including nitrogen (N₂) and sulfur hexafluoride (SF₆), with a flow rate ratio of 2:1 (N₂:SF₆) and a total gas flow rate of 40 sccm. The plasma treatment was conducted in a vacuum chamber maintained at a pressure of 3 Pa, operating in the transition region between capacitive (E-mode) and inductive (H-mode) discharge modes of the ICP system. An RF power of 80 W (frequency: 13.56 MHz) was applied to the inductive coil, with a bias power of 20 W applied to the substrate stage, and the treatment duration was 45 s, modulating the SiO₂ layer thickness from 300 nm to ~100 nm (verified by spectroscopic ellipsometry). Multilayer ReS₂ flakes (~50 nm thick) were mechanically exfoliated from bulk crystals and transferred onto both pristine and plasma-treated SiO₂/Si substrates for comparative analysis. The samples were characterized using Raman spectroscopy (excitation wavelength: 532 nm, laser power: 1 mW) and photoluminescence (PL) measurements (excitation wavelength: 488 nm, laser power: 500 μW) to evaluate the effects of defect passivation and carrier dynamics.

The photodetector’s optoelectronic performance was assessed through current–voltage (I–V) measurements under dark and illuminated conditions (637 nm laser). Responsivity and response time were calculated using the photocurrent, dark current, and incident laser power data, with the rise time (τ_r) and fall time (τ_f) determined by analyzing the photocurrent transition between 10% and 90% of its maximum amplitude. The plasma-treated substrate demonstrated enhanced responsivity (0.88 A W⁻¹) and faster response (2.5 s rise time) compared to the intrinsic substrate (0.012 A W⁻¹, 6 s rise time). The study confirmed that the ICSP treatment effectively passivated sulfur vacancies, reduced deep-level traps, and improved carrier mobility, as evidenced by intensified Raman and PL peaks.

3. Results and Discussion

Figure 1a schematically illustrates the complete process of interface engineering, involving the treatment of a degenerate p-type silicon substrate covered with a 300 nm SiO₂ dielectric layer using soft plasma gases (nitrogen and sulfur hexafluoride). Nitrogen and sulfur hexafluoride (SF₆) were excited in the transition region between the capacitive (E-mode) and inductive (H-mode) discharge modes of the ICP system. As a control experiment, no treatment was applied to the intrinsic substrate sample. Using mechanical exfoliation, multilayer ReS₂ flakes (~50 nm in thickness) were prepared from bulk crystals and then transferred onto two types of substrates: untreated intrinsic substrates and plasma-treated SiO₂/Si substrates. For better comparative analysis, the selected samples had nearly identical thicknesses and were all strip-shaped, as shown in Figure 1c. The representative Raman spectra are presented in Figure 1b. The two characteristic Raman peaks at 211.8 cm⁻¹ and 521.4 cm⁻¹ correspond to the vibrational modes of Re and S atoms, respectively. Previous studies have confirmed that, unlike MoS₂ or WSe₂, the optical and electrical properties of ReS₂ can be modulated through carrier doping due to its reduced crystal symmetry, enabling effective band structure engineering [29]. After treatment with soft plasma gases, the highly reactive SF₆ and N₂ disrupt the Si-O-Si bonds on the SiO₂ surface, generating Si-O⁻, Si radicals, and free electrons. These liberated electrons adsorb onto ReS₂ molecules, enhancing in-plane anisotropy and intensifying the Raman absorption peaks. The photoluminescence (PL) spectra of ReS₂ samples on both pristine and plasma-treated SiO₂/Si substrates are compared in Figure 1d. The ReS₂ crystal contains abundant sulfur vacancies, which generate substantial free carriers [30]. These excess charge carriers promote the formation of charged excitons (trions), consequently quenching the photoluminescence (PL) intensity of ReS₂. After soft plasma treatment, the highly reactive SF₆/N₂-generated oxygen anions effectively passivate sulfur vacancies in ReS₂ through molecular modification, achieving significant defect remediation. The passivation of sulfur vacancies substantially reduces deep-level traps, leading to enhanced carrier mobility and consequently a significant increase in photoluminescence (PL) intensity.

Figure 2a presents the output characteristics of ReS₂ photodetectors on both substrate types under zero gate bias and dark conditions. During the bias sweep from −2 V to 2 V, the current through ReS₂ on the intrinsic substrate consistently exceeded that on the treated substrate by approximately 25-fold. Figure 2c presents the output characteristics (I_DS-V_DS) of ReS₂-based photodetectors fabricated on both pristine and plasma-treated substrates under 637 nm illumination at zero gate bias (V_GS = 0 V). Consistent with dark measurements, the photodetector on the intrinsic substrate exhibited significantly higher current (I_DS) than the plasma-treated counterpart under 637 nm illumination, though the disparity reduced to approximately 4-fold. This phenomenon may originate from plasma-induced substrate modifications: (1) the treatment generates surface wrinkles that alter dielectric properties, enhancing light scattering/reflection losses and reducing effective photon absorption [31]; (2) simultaneously, etching-induced interfacial stress (e.g., amorphization) coupled with the thermal expansion coefficient mismatch between the substrate and ReS₂ promotes defect formation (dislocations/cracks), thereby increasing non-radiative recombination [32]. Figure 2b reveals a linear increase in output current (I_DS) with drain voltage (V_DS), demonstrating the formation of excellent Ohmic contacts at the source/drain electrodes. Figure 2d presents the photoresponse characteristics under nine precisely controlled optical power levels ranging from 770 nW to 123.4 μW (0.77 μW to 123.4 μW/cm², assuming a standard illumination spot size). The photocurrent ranges from −1.21 × 10⁻⁵ A to 8.36 × 10⁻⁶ A, while the resistance varies between 1.96 × 10⁵ Ω and 8.30 × 10⁵ Ω. This inverse correlation between resistance and optical power clearly demonstrates enhanced carrier mobility under stronger illumination. The observed slight current–voltage (I–V) asymmetry under positive/negative biases suggests near-Ohmic contact formation at the metal electrodes.

Figure 3a,b present the photoresponse characteristics of ReS₂ photodetectors on both substrate types under 637 nm illumination with varying optical power densities. Responsivity (R) can be quantitatively determined using the following fundamental photodetector equation:

$$R={\displaystyle \frac{I_{light}-I_{dark}}{P_{opt}}}$$

(1)

In this context, I_light, I_dark, and P_opt denote the photocurrent, dark current, and incident laser power, respectively. Through systematic calculations, the ReS₂-based photodetector exhibits responsivities of 0.012 A W⁻¹, 0.015 A W⁻¹, 0.014 A W⁻¹, and 0.010 A W⁻¹ on its intrinsic substrate under incident laser powers of 770 nW, 1.25 μW, 4.65 μW, and 10.4 μW, respectively. The ReS₂ photodetector on the treated substrate demonstrates responsivities of 0.88 A W⁻¹, 0.64 A W⁻¹, 0.26 A W⁻¹, and 0.14 A W⁻¹ under incident laser powers of 770 nW, 1.25 μW, 4.65 μW, and 10.4 μW, respectively. These results clearly demonstrate that the post-treatment of the substrate leads to a significant enhancement in the photoresponsivity of the device. This improvement indicates that the modified substrate substantially enhances the detector’s photosensitivity, enabling more efficient conversion of incident optical signals into electrical signals. Notably, we observe that, at an incident power of 770 nW, the dark current and photocurrent curves nearly overlap on the intrinsic substrate, whereas the etched substrate still exhibits a distinguishable photoresponse signal. This finding confirms that the substrate treatment effectively enhances the detector’s sensitivity to weak optical signals, demonstrating its improved capability for low-power photodetection. Furthermore, we extracted the photoresponse time to evaluate the device’s response speed to optical signals. The photoresponse rise time (τ_r) and fall time (τ_f) are defined as the time intervals required for the photocurrent to transition between 10% and 90% of its maximum amplitude during the switching curve’s rising and falling edges, respectively. As shown in Figure 3c,d, through systematic measurements and calculations, the photoresponse rise time (τ_r) and fall time (τ_f) were determined to be 6 s and 6.4 s for the intrinsic substrate, while the etched substrate exhibited values of 2.5 s and 5.4 s, respectively. The plasma-treated substrate demonstrates significantly reduced photoresponse rise time (τ_r) and slightly decreased fall time (τ_f), indicating enhanced switching speed and improved temporal response to optical signals. The deep trap states induced by intrinsic defects are identified as the primary factor prolonging the response time [33]. Through preemptive substrate treatment for defect engineering, we demonstrate controllable modulation of response times in the fabricated devices. The substrate treated with SF₆/N₂ soft plasma demonstrates effective defect passivation, leading to the enhanced response speed of the device. The device demonstrates stable and reproducible photoresponse characteristics, with both dark current and photocurrent returning to their baseline levels between consecutive illumination cycles. This confirms the ReS₂ photodetector exhibits reliable cyclic operation with consistent signal recovery capability. As shown in Figure 3e, the ratio of the photocurrent (I_on) generated under illumination to the dark current (I_off) in the dark for the treated rhenium disulfide (ReS₂) photodetector is significantly higher than that of the intrinsic device. This indicates that the plasma-treated substrate can more efficiently convert optical signals into electrical signals while suppressing unwanted dark current. The intrinsic ReS₂ photodetector, due to the presence of a large number of sulfur vacancies and deep-level traps, exacerbates non-radiative recombination and carrier trapping, resulting in a relatively high dark current and limited photocurrent gain, thus leading to a low I_on/I_off ratio. In contrast, after inductively coupled soft plasma (ICSP) treatment, sulfur vacancies are effectively passivated, deep-level traps are reduced, and carrier mobility is improved. As a result, the photocurrent is significantly enhanced while the dark current is suppressed, ultimately increasing the on/off ratio. This confirms the optimizing effect of defect engineering on device performance.

The detectivity of a photodiode (D) represents the signal-to-noise ratio output by the detector when 1 W of optical power is input. The D can be calculated using the following formulas:

$$D={\displaystyle \frac{R}{NEP}}$$

(2)

Here, R is the photoresponsivity and NEP is the noise equivalent power. It can be seen from Figure 3f that the detectivity of the photodetector after substrate etching is higher than that of the intrinsic device. This is because etching reduces defects or impurities in the substrate, optimizes the material interface properties, and makes carrier transport and separation more efficient. Thus, under the same optical power, the device can generate a stronger effective signal while reducing the noise level, ultimately improving the detectivity.

In semiconductor materials, defects and impurities introduce additional localized energy levels within the bandgap. These defect-induced energy levels exhibit amphoteric trapping behavior for charge carriers: unoccupied states serve as electron traps by capturing conduction band electrons, while occupied states subsequently function as hole traps through valence band hole capture. It should be noted that such carrier trapping processes are typically transient in nature, as the trapped carriers may be thermally re-emitted back to their respective bands upon acquiring sufficient energy. Moreover, certain deep-level defects exhibit another critical characteristic: the trapped carriers may recombine with oppositely charged carriers before being released. These defects are conventionally termed recombination centers and play a pivotal role in carrier recombination processes in semiconductor devices. From the preceding discussion, it becomes evident that the fundamental distinction between carrier traps and recombination centers lies in the competition between two processes: whether trapped carriers undergo recombination first or are thermally re-emitted back to the band. This dichotomy can be quantitatively characterized by comparing their recombination probability (P_rec) and emission probability (P_emit). Figure 4a. Schematic illustration of a typical semiconductor band structure, depicting an n-type non-degenerate semiconductor with a single recombination center, shallow-level traps, and deep-level traps, as considered in this study. The fall time (τ_decay) of the photodetector can be expressed as follows:

$${\tau }_{decay}={\tau }_r+{\tau }_t(1+p)$$

(3)

where τ_r represents the recombination lifetime of excess carriers (essentially corresponding to carrier lifetime), which is primarily determined by the properties of recombination centers; τ_t denotes the thermal excitation time required for trapped carriers to return to either the valence or conduction band, mainly governed by the characteristics of carrier traps; and p is the recapture probability. The equation reveals that, in addition to the carrier lifetime, an additional response time component emerges. When recapture effects are neglected, this extra delay term reduces to τ_t, which is inversely proportional to the carrier thermal emission rate. For n-type semiconductors, this relationship can be expressed as follows [34,35,36]:

$${\tau }_t^{-1}=s_p{N}_v{V}_{th}\mathrm{e}\mathrm{x}\mathrm{p}\left({\displaystyle \frac{-∆E}{KT}}\right)$$

(4)

where S_p is the capture cross-section of the trap, N_v represents the effective density of states in the valence band, V_th denotes the carrier thermal velocity, and ΔE corresponds to the energy difference between the trap level and valence band maximum. The equation reveals that even minor energy differences (0.1–0.2 eV) between trap states can induce orders-of-magnitude variations in response time. Consequently, deep-level traps (with larger ΔE) significantly prolong the device’s temporal response. The mechanically exfoliated pristine ReS₂ samples exhibit slow response speeds due to abundant sulfur vacancies (V_s), which introduce additional localized electronic states within the bandgap (i.e., both deep-level and shallow-level traps). Moreover, when both deep-level and shallow-level traps coexist, photo-generated carriers preferentially fill the deep-level traps [37]. This conclusion can be rigorously demonstrated by the following formulation [38]:

$$p_t=p\left({\displaystyle \frac{P_t}{N_v}}\right)\exp \left({\displaystyle \frac{E_{tp}-E_v}{kT}}\right)=P_t\mathrm{e}\mathrm{x}\mathrm{p}\left({\displaystyle \frac{E_{tp}-E_{Fp}}{kT}}\right)$$

(5)

In the equation, p_t represents the density of trapped holes, P_t denotes the trap state density, E_tp corresponds to the energy level of hole traps, and E_Fp is the hole quasi-Fermi level. The formula explicitly demonstrates that traps near the hole quasi-Fermi level exhibit the highest capture efficiency. Furthermore, the presence of high-density deep-level traps can induce Fermi-level pinning near these trap states even under strong illumination conditions, as illustrated in Figure 4c. Consequently, although both deep-level and shallow-level traps coexist, the device’s temporal response and photoconductive gain are predominantly governed by deep-level traps, with shallow-level traps contributing less than 10% to the net photocurrent [39]. As shown in Figure 4b, the substrate after interface engineering treatment demonstrates effective passivation of sulfur vacancies through free oxygen anions (O²⁻). As the carrier traps are effectively passivated, the majority of deep-level defects are repaired, while shallow-level traps begin to dominate the carrier dynamics. Consequently, the response time of the ReS₂ photodetector is dramatically reduced from 6 s to 2.5 s, as clearly demonstrated in Figure 4d.

4. Conclusions

In summary, through innovative interface engineering employing inductively coupled soft plasma (ICSP) pretreatment of SiO₂/Si substrates, we have successfully fabricated high-performance ReS₂/Si heterojunction photodetectors with exceptional optoelectronic characteristics. This technology achieves significant remediation of sulfur-vacancy-induced deep-level defects in ReS₂ through the precise modulation of dielectric layer thickness and surface states, enabling breakthrough optoelectronic performance. The device demonstrates remarkable performance enhancement, with the responsivity surging from the intrinsic value of 0.012 A W⁻¹ to 0.88 A W⁻¹, while the response time is dramatically reduced from 6 s to 2.5 s. Simultaneously, the detector’s weak-light detection capability is significantly enhanced, enabling high-sensitivity optoelectronic performance. The enhancement in Raman and photoluminescence (PL) spectra provides direct evidence of deep-level trap reduction and suppression of non-radiative recombination, demonstrating that interface engineering effectively optimizes the optoelectronic performance of ReS₂ through defect remediation mechanisms. This establishes a new paradigm for performance optimization in 2D material-based photonic devices.