1. Introduction

UV PDs are important devices for detecting and measuring UV photons. They have a wide range of applications, including environmental and biochemical sensing, fire warning, medical imaging and optical communications^[1−6]. Generally, UV PDs can be classified into two main types according to their operational principles: those with built-in electric field and those without. The first type, devoid of a built-in electric field, is typically categorized within a metal−semiconductor−metal (MSM) structure. In this structure, both metal−semiconductor interfaces have Ohmic contact. This type of UV PD follows a photoconductive mechanism, and the device is known as a photoconductor^[3−5]. The UV PD of this type typically necessitates external power sources like batteries or alternating current, resulting in escalated operating costs and heightened system complexity due to power dependency, cost implications, and complexity enhancements in the system’s design^{[7, 8]}. Therefore, researchers propose using a self-powered UV PD, specifically the second type, which operates through the photovoltaic effect. This type of UV photodetector generates current in response to absorbed UV radiation without needing an external power source^{[7, 8]}. The built-in electric field, formed at the junctions between two semiconductors or between a metal and a semiconductor, enables self-sustaining operation. Remarkably, these self-powered PDs exhibit extensive benefits like manufacturing simplicity‌, cost-effectiveness, high compatibility and portability.

In recent years, a variety of wide bandgap semiconductors have been explored and developed for the fabrication of self-powered UV PDs, including ZnO, TiO₂, MgZnO, Ga₂O₃, GaN, SnO₂ and so on^[5−9]. Among them, SnO₂, with a bandgap of 3.6 eV, is a particularly promising candidate for self-powered UV PDs due to its outstanding properties. As a result, various fabrication techniques have been developed to harness its potential in self-powered UV PDs^[10−18]. For instance, Cai et al.^[10] employed chemical vapor deposition and chemical bath deposition method to obtain the microwire based n-SnO₂/p-CuZnS, showcasing notable enhancements of on−off ratio and response speed compared to the plain SnO₂ microwire PD. Wu et al.^[11] developed a self-powered UV PD based on SnO₂/Cs₂AgBiBr₆ heterojunction prepared by a two-step spin coating method, exhibiting a high responsivity and fast response time. Hao et al.^[14] reported a SnO₂/CsBi₃I₁₀ based self-powered UV PDs by a two steps spin coating approach, showing a good long-tern stability, broad spectral and rapid μs-scale response time. Besides, other inorganic materials such as NiO^[15], CuI^[16], ZnO^[17], Bi₂O₃^[18], and p-Si^[19−23] wafer or nanowires were also utilized to form heterojunction with SnO₂ and functioned as self-powered UV PDs. However, most of these studies require an additional growth step to obtain a heterojunction, increasing process complexity and cost, making widespread application potentially less feasible.

In this study, we introduce a straightforward method for fabricating of a p-Si/n-SnO₂ junction self-powered UV PD. This method eliminates extra growth steps and complicated processing procedures. It is demonstrated to be facile and effective in fabricating self-powered UV PDs. This approach takes the advantage of the manipulability of millimeter-length SnO₂ microbelts, which can be easily patterned and assembled into self-powered UV PDs. The p-Si/n-SnO₂ junction photodetector exhibited self-powered properties with high responsivity and rapid response speed, this suggests that SnO₂ microbelts are great for making low-cost, high-performance UV PDs. This work contributes to the ongoing effort to enhance the fabrication of self-powered UV PDs, making them more accessible and easier to produce for practical applications.

2. Experiment

2.1 Materials and synthesis of SnO₂ microbelts

All materials and reagents utilized in the experimentation were employed without additional purification. For synthesizing the SnO₂ microbelts, a mixture of graphite (Aladdin, 99.95%) and SnO₂ nanoparticles (Aladdin, 99.99%, 50−70 mm) with a weight ratio of 1 : 9 was used as the raw material. This mixture was ground and transferred into an alumina boat, which was then positioned at the center of a horizontal tube furnace. Then, a piece of Si substrate with 300 nm thermally oxidized layer was positioned over the source material for collecting the SnO₂ microbelts.

The furnace (Hefei kejing Co., GSL−1500X−OTF) was heated to 1200 ^oC (15 ^oC/min). Then the temperature was sustained at this level for 60 min under a steady flow of argon (100 sccm, standard cubic centimeters) and oxygen (2 sccm). Since no catalyst is used, the growth of SnO₂ follows a vapor−solid (VS) mechanism. This process needs graphite as a reducing agent, as reported in the references^[16]. Upon cooling to room temperature, the final products were formed on the substrate’ surface (Fig. 1(b)). These materials were the SnO₂ microbelts which were used to fabricate the self-powered UV photodetectors.

2.2 Characterizations of SnO₂ microbelts and device

In the experimentation, we used a variety of techniques to characterize the as prepared samples. The morphology of the sample was investigated using scanning electron microscopy (SEM). The chemical composition of the SnO₂ microbelts was ascertained using X-ray photoelectron spectroscopy (XPS). The crystal structure analysis of SnO₂ microbelts was performed through X-ray diffraction (XRD) with Cu K_α radiation (λ = 0.15406 nm) over a 2θ range of 15° to 65° on a single SnO₂ microbelt. The optical properties analysis of SnO₂ microbelts, encompassing their absorption spectrum and optical bandgap, was performed using UV−vis spectroscopy. The effective absorption areas of the device were measured directly using an Olympus optical microscope.

2.3 Device configuration and characterization

To fabricate the self-powered UV photodetectors, we used a simple and cost-effective process. A p-type heavy doping Si wafer (purchased from Zhejiang lijing Electronic Optical Co., (100), 0.01−0.012 Ω·cm) was parallel bonded to a thermally oxidized Si substrate with a 300 nm thick SiO₂ overlayer, and these two substrates were adhered to a glass slide using tape. A millimeter-length SnO₂ microbelt was gently lifted with tweezers and positioned on the substrate, ensuring that one end rested on the thermally oxidized silicon substrate, while the other touched the p-type silicon wafer. An indium (In) particle was employed to secure the microbelt onto the oxidized silicon substrate, simultaneously functioning as an electrode for the device. An additional In particle was applied onto the p-type Si wafer to act as the other electrode for the self-powered UV PD. To facilitate comparison, in-situ measurements were conducted on the identical SnO₂ microbelt by pressing additional In particles onto the microbelt's side on the thermally oxidized Si substrate to act as an electrode. Fig. 1 displays the schematic and optical images of the prepared SnO₂ microbelt and the UV PD structure.

For assessing the optoelectronic characteristics of the devices (Fig. 1(c)), including their current−time (I−t) response and current−voltage (I−V) features, a programmable semiconductor parameter analyzer equipped with a four−probe station (Keithley 4200, USA) was utilized. A 450 W Xenon lamp, coupled with a monochromator for wavelength-specific light irradiation, was employed to illuminate the devices. The intensity of the light was verified using a NOVA II power meter (OPHIR photonics). The spectral response of the PDs was captured across a range from 250 to 700 nm, with immediate responses being documented using a digital oscilloscope (Tektronix DPO5140B) and a Nd:YAG 355 nm pulsed laser system.

3. Results and discussion

3.1 Characterizations of SnO₂ microbelts

Fig. 2 presents the morphological and phase structural characterization results of the SnO₂ microbelts prepared via the CVD process. The SEM images depicted in Figs. 2(a)−2(c) reveal that the sample predominantly consists of freestanding microbelts, which ranging 15−25 µm in width and 1−1.5 µm in thickness. The microbelts extend to several millimeters in length, as evidenced by the inset optical image in Fig. 2(a). Fig. 2(d) shows the XRD pattern of a single SnO₂ microbelt. The pattern shows a pronounced diffraction peak at 34°, which corresponds to the (101) plane of the tetragonal SnO₂ (JCPDS NO. 41−1445, space group P₄₂/mnm), signifying the high quality and purity of the samples. The SEM and XRD analysis corroborate the belt-like morphology and crystallographic structure of the SnO₂ microbelts.

To confirm the surface elemental composition and oxidation states of the obtained samples, XPS measurements was conducted on the SnO₂ microbelts. As shown in Fig. 3(a), the XPS survey spectrum clearly shows only Sn and O core levels. There is a minor C 1s peak around 284.8 eV, which comes from adventitious carbon contamination. Fig. 3(b) display the high-resolution XPS spectra for Sn 3d core levels, exhibiting two distinct peaks at 487.7 and 493.5 eV. These peaks, with a spin−orbit splitting of 8.4 eV, are indicative of the 3d_5/2 and 3d_3/2 peaks of Sn⁴⁺ in SnO₂. Fig. 3(c) depicts the XPS O 1s line for the synthesized SnO₂ microbelts, revealing a broad, asymmetrical peak with a noticeable shoulder on the high binding energy side. The peak deconvolution revealed that it is composed of two distinct components. The first component, with a binding energy of 530.2 eV, is assigned to O−Sn⁴⁺ ion, while the second component, at a higher binding energy of 531.5 eV, is likely associated with the oxygen atoms chemisorbed on the surface of the SnO₂ microbelts, such as OH⁻ and CO₃²⁻ species^[24−26]. These results indicate that the SnO₂ microbelts are composed of Sn⁴⁺ and O²⁻ ions, consistent with the expected composition of SnO₂.

Furthermore, the optical band gaps were ascertained from the UV−vis spectrum to gain a deeper understanding of the band structure of SnO₂ microbelts, as shown in Fig. 3(d). SnO₂, recognized as a direct bandgap semiconductor, has its optical bandgap (E_g) and absorption coefficient (α) characterized by the equation that follows^{[16, 27]}:

$$( {\alpha h\nu } )^{ {2}} ={{A}}( {h\nu } - {E} \mathrm{g).}$$

(1)

In this equation, h represents Planck’s constant, ν is frequency, hν denotes the photon energy, α is the absorption coefficient, and A is the absorption constant. The bandgap of the SnO₂ microbelts, as indicated by the inset image of Fig. 3(d), is determined to be 3.6 eV by extrapolation a tangent line to the photon energy within the 3.5−4.0 eV range. This suggests that the SnO₂ microbelts have a highly efficient in absorbing ultraviolet light, rendering them ideal for applications in UV PDs.

3.2 Photoelectrical characteristics of the single SnO₂ microbelt based UV PD

A significant benefit of our method for creating self-powered UV PDs is the incorporation of ultra-long SnO₂ microbelts. The millimeter-scale length of these microbelts endows them with macroscopic manipulability, facilitating their easy handling and enabling the efficient, cost-effective production of photodetectors. An additional benefit of our methodology is the superior crystalline quality and belt-like shape of the SnO₂ microbelts. These characteristics promote the establishment of robust electrical contacts with the p-type Si substrate, which in turn enhances the detection of UV photons with heightened sensitivity and precision.

To evaluate the photoelectric performance of the single SnO₂ microbelt and p-Si/n-SnO₂−based UV PDs, we conducted in-situ measurements of the I−V and I−t transient response characteristics of the SnO₂ microbelt under ambient conditions at room temperature, with the findings depicted in Fig. 4. The I−V curves depicted in Fig. 4(a) indicate that the single SnO₂ microbelt based device demonstrated a minimal dark current of 0.26 pA at 3 V. Upon exposure to a visible light at a wavelength of 500 nm, the device's photocurrent exhibited a negligible increase (~1 pA at 3 V). Nevertheless, there is a significant surge in photocurrent when the microbelt is irradiated with UV light at a wavelength of 300 nm, reaching 289.6 pA at 3 V, which represents an increase of more than three orders of magnitude in photocurrent. It is worth to mention that the symmetric nature of all I−V curves indicates the formation of high-quality Ohmic contacts between the SnO₂ microbelt and the indium (In) electrodes. The presence of these Ohmic contacts enables the photodetector to function as a photoconductor within a MSM configuration of In−SnO₂−In. As observed from the I−t curve in Fig. 4(b), the photocurrent remains stable and reproducible under a 3 V bias and 300 nm irradiation at an intensity of 0.311 mW∙cm⁻². Furthermore, the current signals exhibit a swift response to the toggling of the light source, demonstrating the device's capability for rapid detection of fluctuations in incident light.

Response times are crucial for photodetectors to effectively follow and react to light variations. To precisely measure the response time of the single SnO₂ microbelt-based device, a pulsed Nd:YAG laser at 355 nm, in conjunction with a digital oscilloscope, was employed. As seen from Fig. 4(c), the device's rise time (τ_r), which is the duration required for the current to increase from 10% to 90% of its maximum steady-state value, was determined to be 50 µs. Conversely, the decay time (τ_d), the period needed for the current to decrease from 90% to 10% of its maximum steady-state value, was measured at 900 µs. These response times are comparable to those of other photodetectors fabricated from individual SnO₂ nanostructures or pure SnO₂ films^[28−30]. In addition to response speed, responsivity is a critical metric for assessing the effectiveness of a photodetector, and it can be determined using the following formula:^[30−32]

$${R} _{ {\lambda } }=\Delta {I} / {P} _{ \mathrm{in}} =( {I} _{ \mathrm{ph}}-{I}_{ \mathrm{d}} )/{SP} _{ {\lambda } } .$$

(2)

In this formula, R_λ signifies the responsivity, ΔI represents the variation in current from the dark current I_d to the photocurrent I_ph, and P_in indicates the effective power of the incident light, which is derived from multiplying the illuminated area S (0.018 mm², Fig 5(a)) by the incident light's intensity P_λ. Fig. 4(d) illustrates the variation of responsivity with wavelength at a 3 V bias. The UV/vis rejection ratio^[33], which is the ratio of responsivity at 270 nm to that at 400 nm, exceeds three orders of magnitude, highlighting the "visible−blind" characteristic of the single SnO₂ device (as shown in the inset of Fig. 4(d)).

3.3 Photoelectrical performances of the self-powered p-Si/n-SnO₂ UV PD

Despite the single SnO₂ devices exhibit high photoelectric performance in response to UV light, they suffer from a major limitation for practical use: the photoconductor device with a MSM structure (In−SnO₂−In), necessitates an external power supply for its operation. This implies that the photodetector operates only when an external power is connected, leading to increased energy usage. To overcome this limitation, self-powered p-Si/n-SnO₂ device was configured on the same SnO₂ microbelts by simply integrating a parallel−bonded p-type Si wafer to this single SnO₂ device, as described in the experimental Section of Fig. 1.

The photoelectrical characteristics of the self-powered p-Si/n-SnO₂ UV PD are shown in Fig. 5. Fig. 5(b) presents the I−V curves of the device under dark conditions and when exposed to 500 and 300 nm light, plotted on a semi-logarithmic scale. The I−V curve of the p-Si/n-SnO₂ PD in the dark displays an asymmetrical pattern, suggesting rectification properties. Upon exposure to UV light with a wavelength of 300 nm, the device produces a substantial photocurrent, the photocurrent can peak at 673 pA at a bias of 3 V, which is marginally greater than the photocurrent of 529 pA at −3 V. Additionally, the photocurrents at both polarities are markedly higher than the dark current, being 448 and 440 times larger than the respective dark currents of 1.5 pA at 3 V and 1.2 pA at −3 V. It's somewhat odd that the current in the dark state doesn't appear to cross the zero point. This phenomenon of the dark current not crossing the zero point has also been observed in other materials in previous studies, such as Cs₂AgBiBr₆/SnO₂ heterojunction^[11], n-SnO₂ modified p-Si nanowires heterojunction^[20], LaNb₂O₇^[34], (BA)₂MAPb₂Br₇−MAPbBr₃ lateral heterostructures^[35], TiO₂/Cs₃Cu₂I₅ heterojunction^[36], GaN−CsCu₂I₃ heterostructures^[37]. This behavior may be attributed to the different carrier dynamics under varying injection conditions, distinct conduction mechanisms at the metal/semiconductor, and particularly the n-type semiconductor/p-type semiconductor interfaces, as well as the inhomogeneity of the contacts^[37].

Moreover, the I−V curve of the p-Si/n-SnO₂ photodetector undergoes a change when illuminated with UV light at 300 nm, leading to the creation of an open-circuit voltage of −0.2 V. This change signifies the self-powered capability of the photodetector, as it can produce a substantial electrical current in response to light without requiring an external power supply. In essence, when exposed to 300 nm light, a substantial number of electron−hole pairs are generated and separated, facilitating the self-powered operation. This built-in electric field inherent to the heterojunction serves as the driving force, facilitating the transfer of holes from the n-type SnO₂ to the p-type Si, while electrons move in the opposite direction. The I−t curve depicted in Fig. 5(c) demonstrates that the photocurrent produced by the device remains stable and reproducible under a bias of 3 V and an irradiance of 300 nm at 0.311 mW∙cm⁻². The current signals also exhibit a swift response when the light is switched on and off, signifying the device's rapid detection capability for changes in light intensity. The device's τ_r and τ_d were determined to be 150 µs and 3.3 ms, respectively, as shown in Fig. 5(d). The rise time and fall time of the device are very different, the potential cause of this phenomenon may be explained as follows: Generally, oxygen molecules undergo chemisorption on the side surfaces of metal oxides, capturing free electrons from the conduction band of n-type metal oxides, as represented by the reaction: O_2(g) + e⁻ →${\mathrm{O}}_{2({\mathrm{ad}})}^- $. As a result, a low-conductivity depletion layer is formed on the side surfaces of metal oxides^{[38, 39]}. When metal oxides are exposed to UV irradiation with photon energies exceeding the semiconductor band gap, electron−hole pairs are generated through the process described by the equation: hν → e⁻ + h⁺. These photogenerated holes react with adsorbed oxygen ions to release oxygen molecules, as shown by the reaction: ${\mathrm{O}}_{2({\mathrm{ad}})}^- $ + h⁺ → O_2(g). As a result, the high-resistance depletion layer is diminished, and the unpaired electrons remaining after the recombination of holes with oxygen ions enhance the conductivity under an applied electric field^{[38, 39]}. The photodecay process in Fig. 5(d) can be modeled using the following exponential relaxation equation:

$$I=I_0+A\rm{e}^{-\tfrac{t}{\tau_1}}+B\rm{e}^{-\tfrac{t}{\tau_2}},$$

(3)

with two relaxation time constants denoted as τ₁ and τ₂. These constants reflect the existence of two distinct mechanisms involved in the decay process, as documented in the literature^{[28, 29, 37]}. The time constant τ₁ is associated with band-to-band recombination within the bulk material, whereas τ₂ is influenced by the presence of chemisorbed oxygen and oxygen vacancies, which are responsible for the persistent photoconductivity in metal oxides^[38]. The time constants under positive voltage are τ₁ = 0.77 ms and τ₂ = 0.84 ms (Fig. 5(d)).

In addition, the wavelength response spectrum of the p-Si/n-SnO₂ heterojunction device closely matches that shown in Fig. 5(e), with the exception of a slight decrease in the maximum response rate (from 12 to 9 mA/W), attributed to variances in the effective absorption area of the device (0.0135 mm² for p-Si/n-SnO₂ PD, Fig. 5(a)). Table 1 presents a comparison of the characteristic parameters of SnO₂ based self-powered photodetectors with those of other nanomaterials. Given similar self-powered performance found in the literature, our method of fabrication is notably simpler and more convenient. Moreover, the UV detection capabilities of our designed architecture outperform those of structures that rely on SnO₂/p-Si configurations.

The self-powered functionality of the p-Si/n-SnO₂ UV PD is depicted in Fig. 6. The I−t curve in presented in Fig. 6(a) illustrates that the device maintains a consistent and repeatable light response signal under zero bias and 300 nm ultraviolet illumination (0.311 mW∙cm⁻²). Furthermore, the device's current signal exhibits a light-to-dark current ratio exceeding three orders of magnitude, which fully underscores its excellent self-powered capabilities. It is observed that the decay curve of the I−t curve shows a tailing effect, a characteristic shared with the I−t curves of both the single SnO₂ microbelt and the p-Si/n-SnO₂ UV photodetector. This phenomenon is attributed to the sluggish dynamics of oxygen adsorption on the SnO₂ microbelt's surface, given its relatively large specific surface area^{[29, 37, 38]}.

The response spectrum curve depicted in Fig. 6(b) reveals that the device exhibits no response within the 350−700 nm wavelength range, signifying that the p-type Si, being a semiconductor itself, does not react to light within this spectrum. The reason for this lack of response is that the p-type Si employed in the experiment is heavily doped, and its function within the p-Si/n-SnO₂ heterojunction device is to serve as the p-type material that facilitates the creation of the junction^[40]. Consequently, leveraging the energy band characteristics of the materials, a plausible response mechanism for the p-Si/n-SnO₂ heterojunction device has been proposed, as illustrated in the inset of Fig. 6(b). The electron affinities of Si and SnO₂ are approximated to be 4.05 and 4.5 eV, respectively, and their respective energy band gaps are considered to be 1.12 and 3.6 eV, based on literature values^{[15, 16, 41]}. The energy band diagram clearly illustrates the formation of a typical type-II heterojunction at the interface between p-Si and SnO₂. Upon exposure to ultraviolet light with energy exceeding that of the SnO₂ bandgap, electron−hole pairs are generated within SnO₂ and migrate towards the depletion region. Here, they are swiftly separated by the built-in electric field, with electrons moving to the conduction band of SnO₂ and holes to the valence band of p-Si. Ultimately, at a zero bias, the photo-generated current is collected at the indium electrodes on either end, enabling the device to function effectively as a self-powered UV PD.

4. Conclusion

In summary, a p-Si/n-SnO₂ junction based self-powered UV PD was constructed using a straightforward and economical method. This self-powered UV PD was realized by simply integrating a p-type Si wafer with an n-type SnO₂ microbelt, which was produced via CVD approach. The high quality and belt-like form of the SnO₂ microbelt facilitate a robust contact with the p-type silicon, establishing a built-in electric field at the junction and thus enabling the device's self-powered capabilities. The p-Si/n-SnO₂ junction photodetector demonstrated self-powered characteristics, boasting a high responsivity of 0.12 mA/W and a rapid response time under zero bias conditions. Hence, the utilization of ultra-long SnO₂ microbelts provides opportunities for the creation of cost-effective, high-performance self-powered UV PDs.

Acknowledgments

Financial support for this research was provided by the High−Level Scientific Research Cultivation Project at Hubei Minzu University, with the grant identifier PY22001, and also by the Guiding Projects from the Department of Education in Hubei Province, identified by the grant number B2018088.