Photovoltaic Performance of GaSe/MoSe₂ Vertical Heterojunctions: Insights from 2‑D Semiconductors

Abstract

Two‑dimensional (2‑D) materials, with atomically thin layers, offer transformative potential for electronics and optoelectronics. Transition‑metal monochalcogenides and dichalcogenides—specifically gallium selenide (GaSe) and molybdenum diselenide (MoSe₂)—feature intrinsic bandgaps and high carrier mobilities even as monolayers, making them attractive for flexible photovoltaic devices. Here, we fabricate a vertical GaSe/MoSe₂ heterojunction by mechanically exfoliating and transferring natural flakes onto titanium electrodes patterned on Si/SiO₂ substrates. The device, comprising ~118 nm p‑type GaSe and ~79 nm n‑type MoSe₂, is tested in the dark and under simulated sunlight ranging from 0.5 to 1.5 sun. Both short‑circuit current (I_sc) and open‑circuit voltage (V_oc) increase with illumination intensity, reaching 0.41 V and 0.46 % efficiency at 1.5 sun. These findings confirm the viability of 2‑D heterojunctions for next‑generation photovoltaic technology.

Introduction

Two‑dimensional (2‑D) materials exhibit unique optical and electronic properties that set them apart from bulk counterparts. Their strong absorption across the solar spectrum, high internal radiative efficiency, and tunable bandgaps make them prime candidates for both single‑junction and tandem solar cells. While in‑plane heterostructures offer atomically sharp interfaces through continuous growth, out‑of‑plane (vertical) heterojunctions allow larger active areas and stackable architectures—an approach we exploit with a GaSe/MoSe₂ vertical device.

GaSe, long sought for photodetectors and nonlinear optics, has historically been limited by crystal growth challenges. Recent advances in 2‑D material synthesis have revived interest, revealing a quasi‑direct bandgap near 2 eV and high optical absorption. MoSe₂, a prototypical transition‑metal dichalcogenide, displays a direct bandgap of ~1.6 eV in monolayer form and carrier mobilities in the hundreds of cm² V⁻¹ s⁻¹ range, positioning it as both an optoelectronic active layer and a transistor channel material.

Despite theoretical predictions of high conversion efficiencies for 2‑D heterojunctions, experimental reports lag behind due to material and interface quality limitations. Moreover, the carrier separation dynamics in vertical 2‑D stacks remain poorly understood. This study addresses these gaps by measuring the current‑voltage (I‑V) characteristics of a GaSe/MoSe₂ heterojunction across a range of light intensities, providing insights into the underlying device physics.

Methods

Four‑terminal Ti (50 nm) electrodes were deposited on p‑type Si wafers with 300 nm thermally grown SiO₂ via electron‑beam evaporation. Natural GaSe and MoSe₂ flakes were mechanically exfoliated onto PDMS and sequentially transferred onto the Ti electrodes, following established protocols. The completed Ti/GaSe/MoSe₂ stack was annealed at 400 °C in N₂ for two hours to improve interlayer coupling.

Optical characterization involved micro‑UV–Vis spectroscopy (JASCO MSV‑5300) to obtain transmittance (T) and reflectance (R) over 200–1600 nm. Absorbance (A) was calculated as A = 1 – T – R, and the absorption coefficient (α) derived via α = [ln(1 – R) – ln T]/d, where d is the flake thickness measured by AFM (HITACHI Nano Navi). Raman and micro‑photoluminescence (PL) spectra were recorded with a 532 nm laser (1.5 mW for Raman, 0.3 mW for PL) using a 100× objective at 25 °C.

Photovoltaic performance was measured at 25 °C under a solar simulator (0.5–1.5 sun). The active area—defined from optical microscopy—was ~490 µm². Spectral response was obtained by combining a monochromatic light source with a pico‑ammeter.

Results and Discussion

Figure 1 shows the transmittance, reflectance, and derived absorbance of a GaSe flake (~638 nm thick). The absorption coefficient rises sharply near the bandgap (~2 eV), confirming its quasi‑direct nature. In contrast, Figure 2 reveals that a ~99 nm MoSe₂ flake exhibits an absorption coefficient exceeding that of GaSe by more than an order of magnitude, with pronounced excitonic peaks at 1.5 eV.

Raman spectra of the GaSe/MoSe₂ device display clear peaks at 133, 214, and 309 cm⁻¹, corresponding to the A_1g, E_1g, and A_2g modes, respectively—evidence of high crystallinity. PL from GaSe shows peaks at 626 nm and 655 nm, representing direct and indirect bandgaps, while MoSe₂ Raman exhibits A_1g peaks at ~236 and 243 cm⁻¹, confirming the material quality.

Figure 5 illustrates the device layout: GaSe contacts the left and bottom Ti electrodes, MoSe₂ contacts the right and top electrodes, with a ~490 µm² active area. AFM confirms flake thicknesses of 118 nm (GaSe) and 79 nm (MoSe₂), corresponding to ~120–130 layers.

Under illumination, the I‑V curves (Figure 6a) demonstrate clear rectification and a photovoltaic response. I_sc scales linearly with light intensity, while V_oc follows the ideal diode relation V_oc = (nk_BT/q) ln(I_L/I_dark + 1), yielding an ideality factor n ≈ 1.11—indicative of an efficient heterojunction with strong internal electric field. At 1 sun, J_sc is 3.11 mA cm⁻², fill factor 0.44, and efficiency 0.54 %. Although series resistance limits the fill factor at higher irradiances, the overall performance remains comparable.

Optical simulations (e‑ARC) predict a maximum J_sc of 19.29 mA cm⁻² across 300–950 nm if all generated carriers are collected. The discrepancy between simulated and experimental currents suggests sub‑optimal built‑in potential and highlights opportunities to optimize absorber thickness, contact work functions, and light‑trapping strategies—such as plasmonic enhancement—to boost efficiency.

Conclusions

We successfully fabricated and characterized a vertical GaSe/MoSe₂ heterojunction via mechanical peeling. MoSe₂ exhibits a markedly higher absorption coefficient than GaSe, and both materials retain high crystallinity after device assembly. Increasing illumination from 0.5 to 1.5 sun elevates both I_sc and V_oc, achieving 0.41 V Voc and 0.46 % efficiency at 1.5 sun. Optical modeling indicates that further optimization of absorber thickness and interface engineering could substantially enhance J_sc and overall efficiency. Light‑trapping techniques on both device surfaces also present a promising route for future performance gains.