Giant photorefractive and photoexpansion effects in a van der Waals semiconductor

Significance

We report a van der Waals semiconductor, crystalline As₂S₃, that exhibits an exceptionally large light-induced refractive index change (photorefraction Δn up to 0.3) together with controllable photoexpansion of up to 7%. These effects occur under continuous-wave illumination, enabling direct laser writing of nanoscale patterns at resolutions of up to 50,000 dpi without the use of pulsed lasers or lithography. At the same time, illumination can control the material thickness and modify its structure, providing a fully optical route for reconfiguring matter. This combination of high refractive index, optical transparency, and strong anisotropy establishes crystalline As₂S₃ as a versatile platform for adaptive photonics, holography, and other applications.

Abstract

Nanophotonics relies on precise nanoscale structuring, yet conventional fabrication techniques remain complex and costly. Layered van der Waals (vdW) materials, with their intrinsic anisotropy and high refractive indices, offer a promising route toward simplified nanostructuring and tunable optical functionality. However, no vdW material has previously been shown to exhibit a strong photorefractive effect—a key requirement for light-based modulation. Here, we report a giant photorefractive response (Δn up to 0.3) in crystalline arsenic trisulfide (As₂S₃), observed at low optical intensities. In addition to refractive-index modulation, light exposure enables controlled thickness tuning of As₂S₃. The material exhibits a giant photoexpansion of up to 7%, depending on the illumination intensity, which may originate from light-induced generation of point defects, consistent with molecular-dynamics modeling. Building on this photoexpansion effect, we introduce a maskless nanopatterning technique based on continuous-wave laser writing, achieving ~500 nm pitch (~50,000 dpi) without the need for ultrafast lasers. The combination of high photosensitivity, anisotropy, ease of exfoliation and transfer, and optical transparency positions vdW As₂S₃ as a practical platform for integrated photonics, adaptive optics, reconfigurable photonic elements, and dense optical encoding.

Precise nanostructuring is pivotal for enabling advanced functionalities in photonics and related fields (1–3). As a result, nanopatterning techniques have evolved from standard photo and electron-beam lithography (4–6) to self-assembling (7–10) and direct writing (11–13) techniques. Among them, laser nanostructuring continues to gain popularity due to its high-throughput and cost-efficient technology (14–18). However, traditionally, this method utilizes pulsed irradiation for efficient light–matter interaction (16, 19, 20), which requires high expertise and sophisticated equipment. In contrast, the use of continuous-wave lasers or even incoherent light sources could greatly simplify the process. Yet, direct writing under continuous illumination remains challenging due to thermal diffusion, which limits spatial resolution (21, 22).

In this regard, two-dimensional (2D) and van der Waals (vdW) materials serve as excellent platforms for light-driven nanostructuring thanks to their giant anisotropy of thermal (23), mechanical (24), and optical (25, 26) properties, arising from their layered crystal structure. This anisotropy, in particular, enables controlled laser thinning from bulk crystals down to atomically thin layers (27, 28). Although this approach affords atomic-layer control of vdW materials thickness, its lateral resolution is constrained by the high powers needed for material ablation. An alternative, non-ablative route is to pattern matter through light-induced refractive-index modulation; hence, vdW compounds that exhibit a giant photorefractive response (Δn) are critically sought for high-resolution, low-power optical nanostructuring.

A key challenge is the absence of photorefractive materials among the broad family of vdW compounds, which now includes over 1,000 reported structures (29, 30). This motivates the search for vdW analogs of traditional photorefractive materials. One promising candidate is recently emerging vdW As₂S₃ (24, 31), also known as orpiment (32), whose glassy form is a well-established photorefractive medium (33–35). The photorefractive effect in chalcogenide glasses enables recording of Bragg gratings, waveguides, and other photonic structures (33, 36–40). While these findings have motivated interest in As₂S₃ as a photosensitive material, the photorefractive behavior of its crystalline vdW counterpart has remained unexplored until now. Combined with its high refractive index n, exceeding 3, and record-high in-plane optical anisotropy of around 0.4 (31, 41, 42), the discovery of a strong photorefractive effect opens previously inaccessible opportunities for lithography-free optical nanostructuring in vdW As₂S₃.

This work reveals a pronounced photorefractive response in vdW As₂S₃ and establishes its broadband photosensitivity from the ultraviolet (UV) to near-infrared (NIR). We also show that even incoherent illumination induces a photorefractive effect in As₂S₃, eliminating the need for lasers in structural nanopatterning. In addition, this response is accompanied by controllable thickness changes in both directions. The material’s high photosensitivity further enables submicron-resolution photonic nanopatterning using continuous-wave laser irradiation. To gain insight into the underlying mechanisms, we also discuss possible microscopic origins of the observed effects, supported by molecular dynamics (MD) simulations. By bringing together refractive, structural, and morphological responses within a single vdW crystal, this work establishes a platform for both fundamental studies of light–matter interactions and practical applications in next-generation vdW nanophotonics.

Results

We begin by investigating the anisotropic optical response of crystalline As₂S₃ before and after 1-h UV illumination to evaluate its photorefractive behavior. We used a mechanically exfoliated flake with a lateral size of ~20 × 10 µm² and a thickness of ~128 nm (≈256 layers). The sample was exposed to 408 nm radiation from a light-emitting diode (LED) at a power density of ~2.25 × 10⁻¹⁰ W/µm² for 1 h, as illustrated in Fig. 1A. Ellipsometry spectra recorded before and after exposure reveal pronounced changes in both the refractive index (Fig. 1B) and extinction coefficient (Fig. 1C), indicating a strong anisotropic photorefractive effect (see SI Appendix, Supplementary Note 1 and Fig. S1 for the ellipsometry data fitting details). Interestingly, prior to illumination, we observed distinct peaks in the extinction coefficient at 395 nm (a-axis) and 430 nm (c-axis) (Fig. 1B), most likely originating from excitons, which is a typical signature of vdW materials (43). Its excitonic origin is further supported by ab initio Bethe–Salpeter equation (BSE) calculations (Materials and Methods), shown in Fig. 1 B and C. Hence, the anisotropic photorefractive change (Fig. 1 D–F) results from excitonic quenching (Fig. 1C) under UV-light exposure.

Fig. 1.

Photorefractive effect in vdW As₂S₃. (A) Schematic of the UV exposure process; the Inset shows the crystal structure of As₂S₃, viewed along the crystallographic b-axis. (B) Anisotropic refractive index n before and after 1 h UV exposure. (C) Anisotropic extinction coefficient k under the same conditions; the Inset shows Raman spectra confirming structural stability. (D–F) Change in refractive index along the (D) a-, (E) b-, and (F) c-axes after 1-h UV exposure; Insets show magnified views of the 550 to 950 nm region. (G) Comparison of the photorefractive response in As₂S₃ (solid lines) with other materials. (H and I) Optical microscopy images of an As₂S₃ flake (H) before and (I) after UV exposure.

Of particular interest is the high magnitude of the photorefractive effect Δn = n^before – n^after in vdW As₂S₃, especially its pronounced anisotropy, as shown in Fig. 1G. Along the crystallographic a-axis, Δn_a exceeds values reported for conventional photorefractive materials such as BaTiO₃, LiNbO₃, and ZnTe:V (44). Along the c-axis, Δn_c is comparable to that of photopolymer films (45). Most notably, along the b-axis, the photorefractive change Δn_b reaches a giant value of ~0.1 at 600 nm, which already exceeds the maximum Δn reported for any previously studied photorefractive material, including glassy As₂S₃ (Fig. 1G) (44–46). Strikingly, at even shorter wavelengths (e.g., 350 nm), Δn_b increases further—up to ∼0.3—highlighting the exceptional optical tunability of crystalline As₂S₃. This demonstrates that pronounced photorefractive responses are not limited to isotropic chalcogenide glass systems but can also arise in vdW crystalline materials with markedly different structure and symmetry.

In addition, the crystalline structure of vdW As₂S₃ remains intact after UV exposure, with no signs of amorphization, as confirmed by Raman spectroscopy (Fig. 1 C, Inset). All characteristic vibrational modes of vdW As₂S₃ are preserved (24, 47, 48), with only a slight intensity reduction observed in the 310 cm⁻¹ A_g mode (SI Appendix, Fig. S2). This subtle change likely reflects a mild redistribution of the local polarizability tensor, consistent with UV‐induced exciton quenching. Simultaneously, the flake undergoes a pronounced color change (Fig. 1 H and I). Given the preserved crystallinity, this change can be attributed to a variation in thickness. Such light-induced thickness modulation in a vdW crystalline system suggests a previously unrecognized photoresponse mechanism, distinct from conventional photorefractive or ablative effects.

To explore this, we investigated the thickness evolution of vdW As₂S₃ under illumination using atomic force microscopy (AFM). Several flakes of different initial thicknesses were exposed to the same UV conditions as in Fig. 1. In all cases (Fig. 2A), the thickness decreased by 3 to 4 nm, independent of the starting value. This reduction also correlates with a visible color shift, as seen in Fig. 2A. These results demonstrate that the thickness of vdW As₂S₃ can be tuned with nanometric precision by adjusting illumination parameters.

Fig. 2.

Irradiation-controlled thinning of vdW As₂S₃. (A) Optical images of five As₂S₃ flakes before and after UV exposure (1 h, 0.81 µJ/µm² radiant exposure), showing color changes corresponding to decreasing thickness (AFM-measured thickness values are indicated). (B) AFM topography scans illustrating the gradual thinning of As₂S₃ flakes under 408-nm UV light, with irradiation times and radiant exposures specified in each panel. (C) Thickness (left axis) and estimated number of layers (right axis) as a function of radiant exposure for the three flakes shown in (B).

We next address the extent of thickness change in As₂S₃ under UV exposure. Thinning proceeds down to a few layers (~2 nm), as shown in Fig. 2 B and C. In this experiment, flakes were sequentially irradiated and analyzed by AFM at 1-h intervals over a 10-h period (see SI Appendix, Fig. S3 for details). Over time, a progressive decrease in thickness was observed, with one of the thinnest flakes reducing from ~12 to ~2 nm. Notably, the interlayer thickness in vdW As₂S₃ is ≈0.5 nm per layer (31), which allows us to convert the AFM thickness values in Fig. 2C into an estimated number of layers. The thinning rate gradually declined and eventually plateaued. This saturation is likely driven by two factors: i) the optical absorption of vdW As₂S₃ decreases upon UV exposure (Fig. 1C); and ii) the shrinking thickness modifies the band structure, increasing the bandgap (49). A similar thinning phenomenon was previously described for vdW MoS₂, but in that case, it was attributed to thermal sublimation, whereas here thinning occurs at much lower illumination intensities ($I_{{\mathrm{MoS}}_2}/I_{{\text{As}}_2{\mathrm{S}}_3}$ ~ 10⁴) (27). The irradiance of our UV LED (~2.25 × 10⁻¹⁰ W/µm²) is unlikely to heat the flake by more than a few degrees. Moreover, the bottom layer is thermally anchored to the substrate, enabling efficient cooling.

In general, irradiation-induced thickness changes in layered materials can arise from several mechanisms: i) photothermal heating leading to sublimation/evaporation; ii) photoexfoliation—both thermal (50) and athermal (51)—resulting in layer-by-layer separation; and iii) photoactivated decomposition of the lattice (52, 53).

To rule out a purely photothermal mechanism, we conducted heating experiments: mechanically exfoliated As₂S₃ flakes were placed in a heating chamber (70 to 100 °C) for several hours in the dark, with a set of Raman–AFM measurements before and after. No measurable change in thickness, flake color, or sharp Raman features was detected, indicating that thermal exposure alone does not induce spontaneous layer removal (SI Appendix, Fig. S4). This conclusion is further supported by MD simulations (SI Appendix, Supplementary Note 2 and Movie S1), which show that the melting temperature of vdW As₂S₃ is approximately 590 ± 20 K, and that spontaneous thermal degradation occurs only above ~590 K—well above the temperatures reached in our experiments. Therefore, even prolonged heating cannot account for the observed 3 to 4 nm thickness reduction after 1 h of UV irradiation, and a purely thermal mechanism can be ruled out.

Photoexfoliation (both thermal and athermal) is typically reported under pulsed irradiation with high peak intensities, far above the irradiance of our UV LED. For example, reported graphene photoexfoliation thresholds correspond to very high peak irradiances under nanosecond pulses (50). Our MD and DFT simulations further show that the energy required to exfoliate a single layer of As₂S₃ is E_exfol ≈ 29 to 35 meV/Ų (SI Appendix, Fig. S5), comparable to graphene and MoS₂ (15 to 20 meV/Ų) (54). This makes photoexfoliation under our low-irradiance UV-LED conditions unlikely.

Instead, we attribute the gradual thinning to a photoactivated desorption (52, 53) of S-rich clusters—triggered by UV-induced bond-weakening—rather than by any thermal sublimation, which MD indicates would require temperatures >500 K. In addition, AFM reveals that after 1 h of UV irradiation, the RMS roughness increases from ≈0.3 nm to ≈0.5 nm, which is consistent with a stochastic cluster-desorption mechanism, as such events are expected to occur nonuniformly across the surface.

Interestingly, the thickness of vdW As₂S₃ also changes under NIR irradiation (SI Appendix, Fig. S7). However, the required irradiation dose is approximately six orders of magnitude higher, with fluences approaching ~1 J/µm². This markedly reduced efficiency is attributed to the strong detuning of the 785 nm laser from the material’s absorption peaks (Fig. 1C). As a result, fine control over the geometry of vdW As₂S₃ can be achieved by adjusting both the incident wavelength and power.

Notably, the illumination-induced modifications reported here are not reversible under our experimental conditions. UV exposure results in a measurable thickness reduction (Fig. 2), i.e., material removal, and therefore the initial state cannot be restored once illumination is stopped.

Given the pronounced changes observed under low-intensity LED illumination, we next investigate the effects of high-power laser radiation on the structure of vdW As₂S₃ to unlock additional functionalities. For this purpose, we utilize a standard continuous-wave (CW) 532-nm laser integrated into a Raman microscope. This simple configuration allows us to pattern a grating with a 600 nm period and ~100 µm lateral size on the surface of a vdW As₂S₃ flake. The fabrication process is schematically illustrated in Fig. 3A, with additional details provided in Materials and Methods and SI Appendix, Supplementary Note 3.

Fig. 3.

CW-laser-written grating on the surface of an As₂S₃ flake. (A) Schematic of the fabrication process: a 532 nm laser integrated into a Raman microscope performs a line-by-line area scan, producing a surface grating. (B) Optical image of the flake with the grating visible in the bottom-right corner. (C) AFM topography image of the area indicated by the yellow square in (B). The Inset shows the 2D Fourier transform (FFT) of the topography in (C), with distinct peaks corresponding to the grating period. (D) Higher-magnification AFM image of the area outlined by the red rectangle in (C); the Inset displays a height profile across the dashed line. (E and F) Transmission spectra acquired (E) outside and (F) inside the grating area. (G) Calculated transmission spectra. Peaks corresponding to waveguided modes are indicated by vertical gray lines in (F) and (G).

The resulting grating is observable via both optical microscopy and AFM, demonstrating the height modulation exceeding 40 nm (Fig. 3 B–D). Moreover, transmittance spectra indicate the presence of waveguided modes in the grating, confirming its optical functionality (see Fig. 3 E–G and SI Appendix, Supplementary Note 4 for calculation details). Additionally, illumination with a 785 nm laser—well above the grating period—produces edge emission, consistent with grating-assisted coupling into waveguided modes (SI Appendix, Fig. S8).

In contrast to low-power illumination, this intense irradiation induces significant structural transformations in the vdW As₂S₃ flake. Raman spectroscopy (SI Appendix, Fig. S9) reveals a crystalline-to-amorphous transition within the grating region, evidenced by a suppression of sharp Raman peaks characteristic of vdW As₂S₃ and the emergence of a broad band near 340 cm⁻¹, typically associated with vibrations of AsS₃ pyramids in the amorphous state (52). This transformation is attributed to the high density of structural defects inherent to the mechanically sensitive As₂S₃ crystal (52). These defects may arise during exfoliation, and subsequent laser exposure can provide sufficient energy to overcome the activation barrier for partial amorphization. Moreover, in contrast to the thinning observed under low-intensity exposure, high-power irradiation leads to local expansion. The grating shown in Fig. 3 was fabricated at a radiant exposure of ~0.83 mJ/µm², nearly three orders of magnitude higher than the low-dose regime used for UV-induced thinning (Figs. 1 and 2).

The correlation between structural transformations and illumination intensity naturally raises a key question: Does a threshold radiant exposure exist at which photoexpansion transitions into rapid evaporation? To address this, we irradiated vdW As₂S₃ flakes using CW 532-nm laser beams of varying power, producing a range of point-like features from small surface bumps to deep crater-shaped holes (Fig. 4 A and B and SI Appendix, Fig. S10). The transition from expansion to rapid evaporation occurred at a radiant exposure of ~3 mJ/µm², corresponding to a power density of ~1.5 × 10⁶ W/cm², which should be considered the ablation threshold for vdW As₂S₃. This value is approximately 50% higher than the ablation threshold reported for glassy As₂S₃ films (55).

Fig. 4.

Laser nanopatterning of As₂S₃ flakes. All patterns were obtained using the Sample Raster point-by-point mode of a Raman microscope equipped with a CW 532-nm laser. (A and B) Effect of laser power on surface morphology: (A) 3D rendering of (B) corresponding AFM topography maps. The Z-axis in (A) is enhanced for clarity. (C–F) Optical images of laser-written patterns: (C) a 25 × 25 bump array with 700-nm-period, (D) a diamond pattern with 500-nm spacing between points, (E) a portrait imprinted with a 700-nm point spacing, (F) a QR code with a point spacing of 600 nm, encoding the phrase “Let’s connect.”

In contrast, exposures below 0.6 mJ/µm² at this wavelength produced no visible surface changes, and we consider this value as the threshold between photoexpansion and slow thinning. Further increases in laser power led to rapid enlargement of the holes, eventually resulting in full penetration of the flakes (SI Appendix, Fig. S11). Importantly, the dimensions of individual features can be precisely controlled by adjusting either the laser power or irradiation time. For clarity, the characteristic ranges of illumination dose and the corresponding material response are summarized in Table 1.

Table 1.

Illumination regimes and resulting photoinduced responses in crystalline As₂S₃

Regime	Instrument	λ, nm	Power, W/cm2	Duration, s	Exposure, mJ/µm2	Observed effects	Structure after exposure
Low	LED	408	2.25 × 10−2	3,600	0.0008	Δn ≈ 0.1, gradual thinning	Crystalline: Raman peaks of vdW As2S3
High	CW laser	532	0.3 to 1.5 × 106	0.2	0.6 to 3	Expansion up to 7%, amorphization	Amorphous: Raman peaks of glassy As2S3
Ablation	CW laser	532	>1.5 × 106	0.2	>3	Ablation	Local evaporation

Low-intensity UV LED irradiation produces a purely photorefractive response (Δn ≈ 0.1) with gradual thinning, whereas high-intensity 532 nm CW laser leads to local photoexpansion (up to 7 %) and amorphization; still higher fluence produces ablation/evaporation.

Notably, the tallest observed bumps reached ~200 nm, corresponding to a giant photoexpansion of about 6 to 7%—exceeding the expansion magnitudes previously reported for chalcogenide glass films (55–57). While such effects have been attributed to light-induced charge redistribution on sulfur atoms leading to network expansion (58), the models proposed in those works were developed for isotropic glassy phases and do not directly apply to our highly anisotropic, layered crystalline system. In particular, the computational approaches previously employed (58) relied on simplified anharmonic potentials and Monte Carlo displacement of small clusters, whose relevance to real material behavior remains uncertain.

In contrast, our work presents a different scenario: we attribute photoexpansion in vdW As₂S₃ to the displacement of atoms and generation of point defects under illumination, resulting in a measurable volume increase within the crystal. This mechanism is supported by our MD simulations (SI Appendix, Supplementary Note 2 and Fig. S6), which use state-of-the-art machine-learning interatomic potentials and explicitly account for the layered, anisotropic structure of the material. As shown in SI Appendix, Fig. S6B, a lattice expansion of approximately 10% can occur at defect concentrations (displaced atoms) around 5%. Notably, at these defect levels, the lattice retains its crystal-like character (SI Appendix, Fig. S6A) and does not undergo a complete transition to the amorphous state. We anticipate that the structural transformation associated with the photoexpansion regime is depth-dependent, due to the finite optical penetration depth and the strongly focused irradiation conditions and is therefore not necessarily uniform through the full flake thickness.

Previous nanoscale patterning in chalcogenide glasses required high-fluence femtosecond pulsed lasers (59), whereas our approach achieves similar results in a crystalline vdW material under low-power CW illumination. Importantly, unlike conventional chalcogenide glass films, thin vdW As₂S₃ flakes can be easily exfoliated and transferred onto a variety of substrates—including other 2D materials—enabling the fabrication of complex heterostructures. Combined with the intrinsic anisotropy of vdW As₂S₃, this flexibility opens a broad design space for nanoscale optical engineering, allowing direct integration of laser-patterned elements with tailored orientation and stacking. The approach demonstrated here is therefore not only energy-efficient and accessible with standard laboratory equipment but also uniquely suited for modern photonic architectures requiring both material anisotropy and integration versatility.

The observed photoexpansion effect thus enables direct fabrication of arbitrary structures on vdW As₂S₃ flakes using a 532 nm CW laser with a standard Raman microscope equipped with a motorized stage (or any other system that provides precise laser beam focusing and sample movement). A laser power of 0.3 to 0.4 mW (corresponding to radiant exposure of ~1 mJ/µm²) was found to be optimal based on the dimensions of the resulting surface features (Fig. 4 A and B). We determined the minimal separation distance between two distinct points to be ~500 nm, corresponding to a dot density of ~50,000 dots per inch (DPI). As a result, we imprint a variety of structures ranging from periodic arrays to arbitrary monochromatic images (Fig. 4 C–F; see also SI Appendix, Fig. S12 A–C for AFM data). The QR-code in Fig. 4F was patterned on a flake preliminarily transferred onto a flexible PDMS substrate, demonstrating proof of concept for information encoding in this transparent vdW material, provided appropriate encapsulation.

Outlook.

vdW materials exhibit exceptional optical, electronic, and magnetic properties, enabled by their weak interlayer bonding, which facilitates isolation into atomically thin layers. However, practical methods for actively controlling these properties remain limited, impeding their broader technological adoption. At the same time, nanophotonic applications inherently require efficient and precise structuring at the nanoscale—a task particularly challenging for layered systems. In this context, light-based sculpting techniques have emerged as powerful and versatile methods to achieve controlled nanostructuring, potentially addressing the specific fabrication challenges posed by vdW materials. Here, we demonstrate that light not only enables precise structural patterning of crystalline As₂S₃ but also actively modulates its optical properties, significantly broadening its applicability in advanced nanophotonics. Fig. 5 schematically illustrates the light-induced multifunctionality of this material.

Fig. 5.

Light-driven multifunctionality of crystalline vdW As₂S₃. The combination of a high refractive index, strong anisotropy, and extreme light sensitivity enables optical control in As₂S₃. The material exhibits tunable excitonic properties (Top Left), structural transformations (Top Right), and thickness modification (Center Right) upon illumination. Its high refractive index and strong anisotropy position it among the most optically responsive materials (Center Left). Light-induced refractive-index tuning allows for controlled change of optical properties (Bottom Left), while direct CW laser writing enables high-resolution nanopatterning with feature sizes down to 0.5 µm, even when the material is placed onto a flexible substrate (Bottom Right).

First, the discovery of a giant photorefractive effect in crystalline As₂S₃, reaching values as high as Δn = 0.1 (and even reaching 0.3 at 350 nm), significantly extends the scope of photonic modulation achievable in anisotropic vdW crystals. Such optically induced changes in refractive index surpass those typically observed in conventional photorefractive materials, positioning crystalline As₂S₃ as a compelling platform for next-generation adaptive photonic elements. The combination of strong anisotropy and exceptionally large refractive-index tunability makes As₂S₃ highly suitable for dynamic waveguiding, optically reconfigurable integrated photonic circuits, and high-precision optical sensors (60–62). This large photorefractive response may also benefit holographic storage solutions, adaptive microlenses, and tunable polarization optics (9, 63, 64), highlighting the broad technological potential of crystalline As₂S₃ in future optical systems.

Second, it was surprising to discover that the thickness of crystalline As₂S₃ flakes can be precisely controlled by illumination, even without lasers: we observed a gradual evaporation and refractive-index modification at low illumination intensities. With increasing intensity, the behavior transitions first to photoexpansion and then to rapid ablation. This intensity-dependent tuning unlocks multiple operational modes: nanometer-scale thickness tuning, photorefractive effects for tailored light propagation and absorption, structural and geometric patterning by imprinting features such as gratings or bumps, and even rapid material ablation for laser cutting and drilling. Strikingly, precise nanostructuring of crystalline As₂S₃ is achievable with a simple CW laser. Such extensive control through a single parameter—illumination intensity—opens exciting and far-reaching possibilities for advanced applications.

Third, we explore possible microscopic mechanisms underlying the observed light-induced thickness changes—both thinning and photoexpansion. In particular, MD simulations reveal that illumination-induced generation of point defects can result in a local volume increase of up to 10%, in close agreement with the experimentally observed photoexpansion of 7%. At the same time, the combination of MD and controlled heating experiments rules out purely photothermal origins for the gradual thinning observed under low-intensity UV exposure. While a full theoretical model is beyond the scope of this study, we provide initial insight into the physical origins of these effects and highlight the potential of vdW As₂S₃ as a model system for further mechanistic investigations into light–matter interaction in layered semiconductors.

Fourth, we demonstrate the capacity of vdW As₂S₃ to serve as an efficient optical memory medium. Arbitrary patterns can be written with submicron precision using a low-power CW laser integrated into a standard Raman microscope. This technique achieves resolutions up to ~50,000 DPI (point spacing 500 nm), currently limited by the diffraction limit but with potential for further optimization. Crucially, the patterning is not based on ablative damage but on local photoexpansion, enabling high-fidelity structure formation.

The formed nanostructures offer wide opportunities across multiple domains. Periodic patterns may find direct use in optics (65), biomedicine (65, 66), tribology (65, 67), sensing (68, 69), and encryption (70). Furthermore, individual photoinduced bumps can serve as microlenses, making them attractive components for integrated 2D optoelectronic circuits (57, 71). Complex images can encode information, including QR codes (SI Appendix, Fig. S12D), user-specific tags, or metadata. Moreover, the intrinsic optical transparency of vdW As₂S₃ ensures visibility of these features under transmission illumination (SI Appendix, Fig. S13), further expanding their utility.

In summary, this accessible and straightforward nanostructuring approach circumvents the complexity and cost of femtosecond laser techniques while offering strong photorefractive functionality, high refractive index, and anisotropy. These attributes position crystalline As₂S₃ as a promising vdW material for next-generation nanophotonics and related emerging technologies.

Materials and Methods

Sample Preparation.

Bulk As₂S₃ crystals were purchased from 2D Semiconductors (Scottsdale, USA) and exfoliated onto the desired substrates. Adhesive tape from Nitto Denko Corporation (Osaka, Japan) carrying the As₂S₃ crystal was brought into contact with the substrates. The flakes were then transferred onto silicon substrates or Schott glass. Subsequently, the tape was removed, completing the exfoliation procedure. The choice of substrate was based on the intended characterization and use of the samples. An optical microscope (Nikon ECLIPSE LV150NA, 100× objective, Nikon CFI TU Plan Fluor BD) was employed to visually identify exfoliated flakes of suitable size.

To place flakes onto a flexible substrate, a PDMS-based pick-up technique was used. Flakes were first mechanically exfoliated onto PDMS (41), and the exfoliated flakes were then transferred onto a flexible PDMS substrate at room temperature using a manual transfer system (HQ2D MAN).

UV and NIR Exposure.

As₂S₃ flakes on substrates were exposed to UV light for 66 min (that equals to ~1 h, which was used in the article text) using an Omicron_LedHUB (408.4 nm, Power 100%) light source of Accurion Nanofilm EP4 Imaging Ellipsometer (the corresponding parameters of polarizer, compensator, and analyzer were set to 50°, 45°, and 30°), utilized in reflection mode. The angle of incidence and angle of view of the ellipsometer were set to an equal angle of 50°. For the described ellipsometer configuration, the light source power on the surface of the substrates (265 µW) was estimated using Power Meter Detector “ThorLabs TH-084 (CAL 05-09-2023) S120VC 200-1100 nm 50mW”. The spot size on the surface of the substrate had an elliptical shape with a semimajor axis and a semiminor axis of ~750 µm and ~500 µm, respectively. The corresponding power density on the surface of the substrate was ~2.25 × 10⁻¹⁰ W/µm².

For NIR exposure, a 785 nm laser was employed in continuous Raman Area Scan mode using an alpha300 RA confocal Raman–AFM microscope (WITec, Ulm, Germany). The 4 × 4 µm areas were exposed with laser power increasing from 0 to 80 mW with the total exposure time per area of 400 s (resulting in a radiant exposure from 0 to 2 J/µm²).

Ellipsometry.

To analyze the anisotropic optical constants of As₂S₃ before and after UV irradiation, we implement the Accurion EP4 imaging spectroscopic ellipsometer in the rotating compensator mode. Two measurements were performed with the plane of incidence aligned with the crystallographic a-axis and c-axis, respectively, which allows for quasi-isotropic optical recording of Ψ and Δ instead of Mueller matrix measurements. Ellipsometry spectra of Ψ and Δ were recorded in the spectral range from 360 to 950 nm for two incident angles 45° and 50°.

Laser Nanostructuring of As₂S₃ Flakes.

All precise surface nanostructuring of crystalline As₂S₃ flakes described in this work was conducted using a 532 nm continuous wave laser and the alpha300 RA confocal Raman–AFM microscope. Parameter control was managed via WITec Control SIX software. A 100× objective (Zeiss EC Epiplan-Neofluar, NA 0.9 DIC) with a spot diameter of approximately 0.3 µm for the 532 nm laser was employed for focusing. Sample positioning was achieved using a motorized stage with a step precision of 25 nm.

Raman and AFM Analysis.

Raman analysis was performed using the WITec alpha300 RA confocal Raman–AFM microscope. Spectral processing was executed with WITec Project SIX software. Raman spectra were acquired with the Zeiss 100× objective, producing a spot diameter of approximately 0.45 µm for the used 785 nm laser (5 mW, equivalent to a laser density of ~13 mW/µm²). A 1,200 lines/mm grating was utilized, and the backscattered light was detected with a backilluminated deep depletion CCD detector cooled to −60 °C, achieving a spectral resolution of ~0.5 cm⁻¹. To eliminate any influence of crystal anisotropy, each spectrum was recorded by rotating the polarizer‐analyzer combination in 5° steps from 0° to 360°, then averaging all acquired scans. Each spectrum acquisition lasted 10 s and was repeated five times.

AFM imaging was conducted in tapping mode using the WITec alpha300 RA microscope equipped with a NanoWorld ARROW-FMR probe (75 kHz, 2.8 N/m). The scanning speed, scan size, and resolution were tailored to each study area to ensure optimal visualization quality. Surface morphology and thickness of the As₂S₃ flakes (Fig. 2 and SI Appendix, Fig. S3) were estimated via Cypher S microscope (Oxford Instruments) operated in tapping mode, using AC160TSA-R3 tip type. AFM image processing was conducted using Gwyddion software.

First Principles Calculations.

The optical response of bulk vdW As₂S₃ (a = 4.255 Å, b = 9.578 Å, c = 11.415 Å, α = 90°, β = 90.44°, γ = 90°) was computed using the BSE@GW approach as implemented in the VASP package (72). Single-particle ground-state wavefunctions were first obtained from a self-consistent density functional theory (DFT) calculation. These were then used to initialize the GW step and compute the screened Coulomb interaction kernels. The resulting quasiparticle energies served as input for a Bethe–Salpeter Equation (BSE) calculation to evaluate the frequency-dependent dielectric function including electron–hole interactions. A plane-wave cutoff energy of 400 eV and 512 bands were used. The exchange correlation effects in the DFT run were described with the generalized gradient approximation (Perdew–Burke–Ernzerhof functional) (73). The behavior of the core electrons and their interaction with the valence electrons was described using the projector augmented wave pseudopotentials (74). The BSE step included 24 valence and 24 conduction bands.

MD Calculations.

All MD simulations were performed using the LAMMPS software package (75), in combination with the modern machine-learning interatomic potential PET-MAD, which demonstrates superior accuracy in modeling the energies of a wide range of atomic structures (76). For the simulation of As₂S₃ melting, a time step of 0.25 fs was employed. Temperature was regulated using a Nosé–Hoover thermostat with a damping constant of 25 fs. Further methodological details are provided in SI Appendix, Supplementary Note 2.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

A snapshot at 00:07 is shown here. The movie presents the molecular dynamics simulation (LAMMPS) showing the time evolution of the melting front propagation in a single-layer As2S3 crystal at 670 K.

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.