Sputtered WS2/CuO Heterojunction Photodetector Runs on Nothing but Light

Imagine a light sensor that needs no battery, no external voltage, and no power supply at all — just the photons it is built to detect. Researchers Monireh Jafari and Nafiseh Memarian have reported exactly that: a self-powered, broadband photodetector built from a carefully stacked sandwich of tungsten disulfide, copper oxide, and a whisper-thin titanium dioxide buffer layer, all grown by magnetron sputtering on ordinary fluorine-doped tin oxide glass. Published in Results in Physics, the work demonstrates stable photoresponse across the ultraviolet and visible spectrum, from 368 to 622 nanometers, while drawing zero bias from an external source. In an era when the Internet of Things, wearables, and autonomous sensor networks are hungry for low-power electronics, a photodetector that generates its own operating voltage is more than a laboratory curiosity — it is a template for how smart interface engineering can replace brute-force biasing.

The heart of the device is a p-n heterojunction formed between two very different semiconductors. Copper oxide, or CuO, is a p-type material with a band gap ranging roughly from 1.2 to 2.1 electronvolts, prized for its structural stability and good conductivity. Tungsten disulfide, WS2, is an n-type transition metal dichalcogenide with a direct band gap of about 2.1 electronvolts and exceptional optical properties that make it a favorite for sensitive photodetection. When these two are layered together, a built-in electric field forms at their interface. Under illumination, that field sweeps photogenerated electrons and holes in opposite directions, separating charge carriers before they can recombine and driving a measurable current without any applied voltage. It is the same fundamental physics that powers a solar cell, here repurposed for light sensing.

What elevates this design from a simple two-layer diode is the insertion of an approximately 20-nanometer titanium dioxide buffer layer between the fluorine-doped tin oxide substrate and the WS2 film. TiO2 is a wide-band-gap, transparent n-type semiconductor, with a band gap in the range of 3 to 3.8 electronvolts, high optical transmittance across the visible region, and strong chemical stability. In this architecture it plays several roles at once: it optimizes the electron transport path, reduces charge recombination, and physically protects the WS2 layer from direct contact with the substrate, mitigating surface mismatch and instability. In effect, the buffer layer is the quiet architect of the device’s energy-band landscape, aligning the conduction and valence bands of the three semiconductors so that electrons flow toward the TiO2 side while holes migrate toward the CuO side.

The fabrication itself is a showcase of industrial-friendly vacuum deposition. Using a three-cathode magnetron sputtering system, the team deposited the layers sequentially: TiO2 from its own target at 200 watts at room temperature, WS2 at 75 watts with the substrate heated to 200 degrees Celsius, and CuO reactively sputtered from a copper target in a mixed argon-oxygen atmosphere at 300 degrees Celsius. The chamber was pumped down to a base pressure of 8.5 times ten to the minus five torr before each run, a detail that matters, because residual water vapor in a sputtering chamber is a notorious source of hydroxyl defects that degrade film quality. A gold top electrode completed the Glass/FTO/TiO2/WS2/CuO/Au stack. The deposition rates — 0.75, 8.56, and 6.93 nanometers per minute for TiO2, WS2, and CuO respectively — were tightly controlled, and cross-sectional electron microscopy confirmed a WS2 layer of about 250 nanometers and a CuO layer of about 100 nanometers with uniform, continuous coverage.

Structural characterization backed up the claim that the films grew as intended. X-ray diffraction revealed the hexagonal phase of WS2 with reflections indexed to its (104), (009), and (116) planes, and the monoclinic phase of CuO with peaks matching its standard reference card, alongside contributions from the substrate. Crystallite sizes calculated with the Scherrer equation came out at roughly 11.6 nanometers for WS2 and 10.2 nanometers for CuO, with correspondingly modest dislocation densities, indicating relatively clean lattices with few structural defects. Raman spectroscopy sealed the case: the characteristic WS2 vibrational signatures at 408 and 318 inverse centimeters and the CuO modes near 282 and 210 inverse centimeters all appeared, with no evidence of unwanted secondary crystalline phases. Notably, no TiO2 peaks were seen in the diffraction pattern — expected, given that the buffer layer is thinner than the instrument’s effective resolution limit.

Optical and electrical measurements of the individual layers explained why the stack works as a cohesive photodetector. Tauc analysis of the absorption spectra yielded direct band gaps of 3.24 electronvolts for TiO2, 2.43 for WS2, and 1.60 for CuO — a staggered ladder of energy levels that is critical for steering charge carriers in the right directions at each interface. Hall effect measurements in a four-probe configuration confirmed the intended doping characters: TiO2 and WS2 are n-type, while CuO is p-type with a high carrier concentration of about ten to the seventeenth per cubic centimeter and moderate mobility, well suited to collecting holes. Photoluminescence under 320-nanometer excitation revealed multiple emission bands tied to excitonic transitions and defect states in the layered structure, underscoring that both surface and defect-related electronic states shape the material’s optical response.

Under monochromatic illumination from ultraviolet to red LEDs at a fixed 1-watt output, the device behaved like a genuine photovoltaic diode. The dark current-voltage curve showed a rectification ratio of about 195, and analysis with the standard diode equation gave an ideality factor of 3.02 and a reverse saturation current density of 1.52 microamperes per square centimeter. In the fourth quadrant of the illuminated curves, the team extracted an open-circuit voltage near 0.008 volts and a short-circuit current density of roughly 23 microamperes per square centimeter under ultraviolet light — the telltale fingerprint of an internal built-in electric field capable of powering photodetection at zero bias. Quantitative band-alignment estimates, built from the measured gaps and reported electron affinities, support this picture: the offsets at the TiO2/WS2 and WS2/CuO interfaces channel electrons and holes along separate, energetically downhill paths while suppressing recombination.

The transient photoresponse told an equally interesting story. Switched on and off in ten-second cycles under zero bias, the device produced reproducible photocurrent over repeated cycles, with rise and decay times of 50 and 100 milliseconds under ultraviolet illumination — fast for an unbiased, multi-layer structure. Applying reverse bias did increase the raw photocurrent, from about 8 microamperes at zero volts to 34 microamperes at 1 volt, but it came at a cost: the photoresponse ratio collapsed from over 100 to around 1.1, and the switching times stretched to 130 and 170 milliseconds as trap states and defect-assisted charge transfer slowed the current dynamics. The lesson is counterintuitive but important — for this device, self-powered operation is not just the frugal option, it is the better-performing one, delivering the highest relative sensitivity and the fastest response. Under ultraviolet light the device achieved a responsivity of 0.25 milliamperes per watt, a specific detectivity of 3.9 times ten to the seventh Jones, and an external quantum efficiency of about 0.084 percent, with these figures declining toward the red end of the spectrum as longer-wavelength photons generate carriers deeper in the film, farther from the junction.

The authors are candid that the responsivity and detectivity trail those of some recently reported heterojunction photodetectors, many of which rely on aggressive plasmonic interface engineering or heavy external biasing to reach spectacular sensitivities. They attribute the shortfall to interfacial defects, trap-assisted recombination, carrier scattering across the multilayer stack, series resistance — including the relatively high resistivity of the TiO2 buffer — and the 250-nanometer thickness of the WS2 layer, which lengthens the journey carriers must make before extraction. Yet the comparison table in the paper makes the trade-off plain: among self-powered devices, this sputtered triple stack holds its own, and it does so with a fabrication route — scalable magnetron sputtering on inexpensive FTO glass — that is far closer to mass production than exfoliated van der Waals assemblies. As sensor networks proliferate and energy harvesting becomes a design requirement rather than an afterthought, this work suggests that sometimes the smartest way to boost a photodetector is not to push harder with power, but to arrange the energy bands so skillfully that the light does all the work.

Subject of Research: A self-powered broadband TiO2/WS2/CuO heterojunction photodetector fabricated by magnetron sputtering for zero-bias UV-to-visible light detection.

Article Title: Self-powered broadband WS 2 /CuO heterojunction photodetector with a TiO 2 buffer layer

Article References: Jafari, M., & Memarian, N. (2026). Self-powered broadband WS2/CuO heterojunction photodetector with a TiO2 buffer layer. Results in Physics, Article 108758. https://doi.org/10.1016/j.rinp.2026.108758

Image Credits: AI Generated

DOI: 10.1016/j.rinp.2026.108758

Keywords: photodetector, self-powered, heterojunction, tungsten disulfide, copper oxide, titanium dioxide, magnetron sputtering, band alignment, broadband photodetection, p-n junction, thin films, optoelectronics