Anisotropic Γ‑Valley Interlayer Excitonic Emission in Anisotropic-Isotropic 2D Heterostructures

Abstract

In-plane anisotropy of 2D materials and their heterostructures provides an essential degree of freedom for enabling direction-selective physical responses. However, realizing anisotropic optical responses in interlayer exciton (ILE) of type-II van der Waals heterostructures has remained elusive, as most heterostructures rely on high-symmetry group-VI transition metal dichalcogenides, and attempts to introduce low-symmetry layers are hindered by interlayer momentum mismatch leading to inefficient radiative recombination. Here, we demonstrate anisotropic ILE emission in ReS₂/WS₂ enabled by Γ-point momentum matching. Polarization-resolved photoluminescence (PL) reveals three distinct ILE peaks, appearing at 1.37–1.42 eV for 1L-ReS₂/2L-WS₂ and 1.32–1.37 eV for 2L-ReS₂/2L-WS₂. Time-resolved, power- and temperature-dependent PL measurements, together with first-principles many-body calculations, confirm strongly anisotropic interlayer excitons from a direct Γ-point interlayer bandgap. Furthermore, ILE intensity is independent of excitation polarization due to dominant WS₂ absorption, providing a stable exciton supply for exciton transport. These findings position momentum-matched anisotropic-isotropic heterostructures as a platform for polarization-selective optoelectronics and anisotropic excitonics.

Introduction

Anisotropic 2D materials, such as black phosphorus (BP) with a puckered structure and ReS₂ with a distorted 1T structure in group-VII transition metal dichalcogenides (TMDCs), have emerged as promising platforms for direction-selective physical responses owing to their reduced crystal symmetry and in-plane anisotropy. − Incorporating such low-symmetry 2D materials into van der Waals (vdW) heterostructures offers a viable route to impart anisotropy to heterostructure properties, particularly in type-II vdW heterostructures hosting interlayer exciton (ILE) with long lifetimes, aligned out-of-plane dipoles, and moiré tunability. − However, most previous studies have focused on heterobilayers of high-symmetry group-VI TMDCs, which exhibit in-plane polarization-isotropic emission. , While BP shows strong anisotropy and a thickness-independent direct band gap in the mid-infrared region, − its high chemical reactivity limits long-term stability. In contrast, ReS₂ is air-stable and has a strongly symmetry-broken triclinic lattice with pronounced anisotropic properties. − Its principal axis can also be reoriented under external stimuli, including e-beam irradiation and strain, , positioning ReS₂ as a compelling low-symmetry building block to realize anisotropic ILE when combined with high-symmetry group-VI TMDCs.

To date, most studies have focused on employing anisotropic 2D materials as templates to induce anisotropy in the intralayer excitonic emission of group-VI TMDCs, rather than realizing anisotropic ILE. − Theoretically, materials such as BP and GeSe have been proposed as potential hosts for anisotropic ILE. , Other than device-level photocurrent response based on interlayer absorption of type-II heterostructures involving anisotropic materials like BP and ReS₂, , to the best of our knowledge, direct observation of in-plane anisotropic ILE emission has not been reported.

The lack of optically active anisotropic ILE stems from momentum mismatch between anisotropic 2D materials and monolayer group-VI TMDCs. While BP and ReS₂ have band edges at the Γ-point, , group-VI monolayers exhibit their band edges at the K-point. This momentum mismatch renders the interlayer excitonic transitions momentum-forbidden, thereby suppressing ILE emission in such systems.

Here, we demonstrate anisotropic ILE emission in heterostructures composed of monolayer or bilayer ReS₂ and bilayer WS₂, enabled by Γ-point momentum matching. Polarization-resolved photoluminescence (PL) reveals three distinct ILE peaks whose polarization directions correlate with those of ReS₂ intralayer excitons in both monolayer- and bilayer-ReS₂-based heterostructures. , The ILE peak energies appear at 1.37–1.42 eV for 1L-ReS₂/2L-WS₂ and 1.32–1.37 eV for 2L-ReS₂/2L-WS₂. Time-resolved, excitation-power-dependent, and temperature-dependent PL measurements confirm that these peaks exhibit key signatures of direct Γ-matched ILE. Based on first-principles GW and Bethe-Salpeter equation (BSE) calculations, we show that the ReS₂/WS₂ heterostructures have a direct Γ-point interlayer bandgap and find strongly anisotropic interlayer excitons consistent with experimental observations. In addition, the overall ILE emission intensity remains independent of excitation polarization due to dominant WS₂ absorption at 2.33 eV excitation, whereas ReS₂ intralayer excitons exhibit anisotropy. , These results establish momentum-matched anisotropic-isotropic heterostructures, such as ReS₂/WS₂ heterostructures as a hybrid anisotropic-isotropic platform enabling both polarization-selective optoelectronics and anisotropic excitonics.

Results and Discussion

The schematic drawing of the ReS₂/WS₂ heterostructure and its anisotropic ILE emission is shown in Figure a. The ILE is illustrated to exhibit pronounced anisotropy with respect to the Re-chain direction of ReS₂ (crystallographic b-axis), consistent with the polarization behavior of intralayer excitons in ReS₂. , All samples were encapsulated with hexagonal boron nitride (hBN) to enhance the spectral resolution and emission intensity of the ILE. PL measurements were performed at 3 K in a cryostat using a continuous-wave 532 nm laser that was linearly polarized in-plane, unless otherwise noted. Time-resolved PL (TRPL) measurements were conducted using a pulsed 405 nm laser (see Methods). For convenience, we denote n-layer ReS₂ and WS₂ as nR and nW, respectively.

1.

Formation of ILE in ReS₂/WS₂ heterostructures. (a) Schematic of ILE formation and their anisotropic optical properties. (b) PL spectra measured from different layer configurations (nR: n-layer ReS₂ and nW: n-layer WS₂). (c) Band structures of ReS₂/WS₂ heterostructures calculated from the GW method. The blue and red dots indicate ReS₂ and WS₂ layer states, respectively. (d) Optical microscopy image of the ReS₂/WS₂ heterostructure. The highlighted regions within the black box indicate areas with different layer configurations. (e) PL mapping result corresponding to the boxed region in (d), with intensity representing the integrated signal over the magenta-shaded energy range (1.35–1.44 eV) in (b).

To investigate whether momentum matching induces ILE emission, we measured PL spectra of ReS₂/WS₂ heterostructures with different layer number configurations. Figure b compares the PL spectra of individual layers and nR/2W (n = 1, 2) heterostructures (PL spectra of 1W and nR/1W are shown in Supporting Figure S6). Distinct low-energy emission peaks in the range of 1.2–1.4 eV are observed exclusively in nR/2W, appearing at energies well below those of the constituent intralayer excitons, allowing unambiguous identification of interlayer excitonic transitions. Detailed spectral features of the individual layers are provided in Supplementary Note 1.

Figure c shows the GW quasiparticle band structure of both the 1R/2W and 2R/2W systems. Both systems have direct Γ-point bandgaps of 1.669 and 1.574 eV, respectively. The conduction band is dominated by ReS₂ states and the valence band by WS₂ states, indicating a type-II band alignment.

Figure d shows an optical microscopy (OM) image of a representative fabricated heterostructure. PL mapping performed over the boxed region reveals uniform low-energy emission across the heterostructure area (Figure e), confirming the interlayer origin of the observed ILE emission. Fabrication details and sample information are provided in Supplementary Note 1 and Supporting Table S1.

To extract the characteristics of direct ILE, we performed temperature-dependent PL, TRPL, and power-dependent PL measurements. Figure a shows temperature-dependent PL spectra from 3 to 293 K. Because direct excitonic emission does not require phonon-assisted recombination, the ILE intensity becomes stronger at lower temperatures, , which is clearly illustrated in Figure b where the integrated intensity within the shaded region of Figure a is plotted as a function of temperature. The slight decrease between 3 and 10 K is likely associated with lower-lying dark states, as previously observed in ReS₂ and WS₂. ,

2.

Temperature and power dependence of PL and TRPL results. (a) PL spectra of 2L-ReS₂/2L-WS₂ at different temperatures. (b) Integrated PL intensity over the yellow-shaded energy range in (a) as a function of temperature. (c) TRPL results for direct excitons in 1L-WS₂ (intralayer exciton), for ILE in 1L-ReS₂/2L-WS₂, and for ILE in 2L-ReS₂/2L-WS₂, as shown in the inset. (d) Power-dependent PL intensity maps of 2L-ReS₂/2L-WS₂ in the spectral ranges containing ILE and the intralayer excitons of 2L-ReS₂. The dashed lines are guides to the eye showing the evolution of the peak positions. (e) Peak energies of IX₃ and X₂ as a function of excitation power, extracted from (d). (f) Integrated PL intensities of IX₃ and X₂ as a function of excitation power on log–log scales. The solid lines are fits to I ∝ P ^α , yielding ${\alpha }_{I{X}_3}=0.81$ and ${\alpha }_{X_2}=0.93$ .

Since an ILE features spatially separated electron and hole, it is expected to have a longer lifetime than an intralayer exciton. To verify this, we compared the TRPL signals of the 1W intralayer exciton and the ILEs in 1R/2W and 2R/2W (Figure c). The decay curves were fitted with a biexponential function (solid lines), yielding average lifetimes of 257, 489, and 650 ps for 1W, 1R/2W, and 2R/2W, respectively (inset of Figure c). Although reported K-valley ILE population lifetimes can range from approximately 1 ns to over 100 ns, these long decay times can arise because excitons outside the light cone must first relax into radiative states. In contrast, an intrinsic radiative lifetime of 0.40 ns has been measured for a momentum-direct ILE. Our ILE lifetimes of 489–650 ps are comparable to this intrinsic time scale, consistent with a reduced momentum-space relaxation bottleneck in the proposed momentum-matched Γ-point transition. The longer lifetime of 2R/2W compared with 1R/2W is consistent with the increased electron–hole separation in the thicker ReS₂ layer, as illustrated schematically in the inset (all fitted parameters are summarized in Supporting Table S2).

Figure d presents the power-dependent PL spectra of 2R/2W as intensity maps. The extracted peak energies in Figure e show a clear blueshift of the peak labeled IX₃ (details discussed later) with increasing excitation power, whereas the intralayer exciton X₂ shows no systematic blueshift. This behavior is consistent with the interaction-induced blueshift commonly observed for ILEs at increasing exciton density. , In addition, the integrated PL intensities follow I ∝ P ^α , with ${\alpha }_{I{X}_3}=0.81$ and ${\alpha }_{X_2}=0.93$ (Figure f). The more sublinear power dependence of IX₃, compared with the nearly linear dependence of X₂, is consistent with previous observations of interlayer and intralayer excitons, respectively.

To investigate the anisotropic optical response of ILE in 2R/2W, polarization-resolved PL measurements were performed by systematically varying both the analyzer angle Φ and the excitation polarization angle Θ relative to the b-axis of ReS₂ (Figure a). We first examined the polarization anisotropy of the ILE emission by fixing Θ = 0° and rotating Φ counterclockwise in 10° increments. The analyzer angle-dependent color map and stacked spectra in Figure b reveal a periodic modulation of the ILE intensities with Φ, indicating pronounced polarization anisotropy inherited from the in-plane anisotropy of ReS₂.

3.

Anisotropic optical properties of ILEs in 2L-ReS₂/2L-WS₂. (a) Experimental setup for tuning analyzer and excitation polarization angles relative to the b-axis of ReS₂. (b) PL spectra with fixed excitation polarization (0°) and varying analyzer angles (10° steps). Top: color map, bottom: stacked spectra (0–90°). (c) Enlarged view of the ILE region in (b). Top: no analyzer, middle: analyzer at 0°, bottom: analyzer at 90°. Three ILE peaks (IX₁–IX₃) are identified. (d) Polar plots of normalized fitted amplitudes. Left: ILE peaks, right: intralayer excitons (X₁, X₂) of ReS₂. ILE shows similar anisotropy to intralayer excitons. (e) PL spectra with varying excitation polarization angle (10° steps) without analyzer. Top: color map, bottom: stacked spectra (0–90°), showing isotropic ILE behavior. (f) Polar plots of normalized fitted amplitudes for IX₃ and X₁.

To resolve the spectral components contributing to this angular dependence, Figure c displays the magnified ILE energy range (1.25–1.44 eV) under three analyzer conditions: without analyzer (top), Φ = 0° (middle), and Φ = 90° (bottom). Without an analyzer, IX₂ is largely obscured by IX₃, whereas polarization-resolved spectra at 0° and 90° resolve three ILE peaks at 1.329 eV (IX₁), 1.351 eV (IX₂), and 1.363 eV (IX₃). Their microscopic origins and corresponding interlayer transitions are further supported by the BSE calculations discussed in Figure . Two intralayer excitonic peaks from 2R, X₁ (1.603 eV) and X₂ (1.652 eV), are also observed, exhibiting the same polarization dependence as previously reported. , While IX₁ intensity remains almost unchanged with Φ, IX₂ and IX₃ exhibit strong analyzer angle dependence (Figure b).

4.

Ab initio calculations of anisotropic ILEs in 2L-ReS₂/2L-WS₂. (a) Band-to-band composition of the three ILEs at Γ. The valence band v₁ lies on the WS₂ layers and the two lowest conduction bands c₁ and c₂ on the ReS₂ layers, and all three transitions are momentum-direct at Γ. At Γ, v₁ is W d _z2, and c₁ and c₂ share the same Re d _z2 character. Arrow widths are proportional to the weight of each exciton carried by that channel, and the weight is printed above each arrow. The weight is that of the BSE eigenvector, summed over the Brillouin zone. IX₁ carries 79.5% into c₂ and 20.3% into c₁, IX₂ 17.6% and 82.2%, and IX₃ 5.1% and 94.8%. (b, c) Exciton wave function |χ _S (r _e ,r _h )|² of IX₁ in real space as a function of (b) electron positions with a hole fixed in WS₂, and (c) hole positions with an electron fixed in ReS₂, respectively. (d) Optical absorption spectra from the GW-BSE calculations. The solid line corresponds to polarization parallel to the Re-chain direction (0°), while the dotted line indicates polarization perpendicular to it (90°). The ILEs show an anisotropic optical response depending on the polarization angle. The energy of the first exciton is below the experimental measurement range. (e) Polarization-dependent oscillator strengths of IX₁ (red), IX₂ (green), and IX₃ (blue), respectively.

To compare anisotropy, the fitted peak amplitudes of the ILE peaks (IX₁, IX₂, IX₃) and intralayer excitons (X₁, X₂) were plotted independently in polar coordinates, as shown in Figure d (left: ILEs, right: intralayer excitons). The detailed spectral features of ILE peaks and the principle used to quantify their anisotropy are discussed in Supplementary Note 2. The peak amplitudes as a function of Φ were fitted using I(Φ) = I _min + I _p cos²(Φ – Φ _max ) where I _min , I _p , and Φ _max are the fitting parameters corresponding to the offset, polarization strength, and the angle of maximum intensity, respectively. The extracted values of Φ _max are 87.4° and 0.5° for IX₂ and IX₃, respectively, while those for X₁ and X₂ are 3.9° and 77.5°. Similar ILE energies and emission polarization anisotropy are observed across samples with substantially different twist angles (Supporting Figures S7 and S8), demonstrating the relative twist-angle insensitivity of the Γ-matched ILE, consistent with prior reports on other Γ-matched systems. , Details of the twist-angle determination and sample-to-sample comparison are provided in Supplementary Notes 1 and 2. The extracted fitting parameters are summarized in Supporting Table S1.

To directly compare the anisotropy of ILE peaks with X₁ and X₂, we analyzed all peaks extracted from the same 2R/2W spectrum. The polarization axes of the two ILE peaks can be assigned: IX₂ shares the polarization axis of X₂ (approximately perpendicular to the b-axis), whereas IX₃ aligns with X₁ (approximately parallel to the b-axis). We also note that IX₂ is completely quenched near Φ = 0° whereas X₂ is not (Figure c). X₁ and X₂ exhibit nearly identical anisotropy in both 2R/2W and isolated 2R (Supporting Figure S10), consistent with previous reports.

To quantify the anisotropic behavior, we calculated the degree of linear dichroism $\rho =\frac{I_{max}-I_{min}}{I_{max}+I_{min}}$ for each peak, where I _max and I _min denote the fitted extreme amplitudes. IX₂ and IX₃ exhibit substantially larger ρ values than their corresponding intralayer excitons, X₂ and X₁, with enhancement factors of 1.88 and 1.28, respectively. This enhancement may additionally reflect interface-induced symmetry breaking, consistent with reports that interlayer-coupling-induced lattice relaxation can enhance excitonic polarization anisotropy in ReS₂-based heterostructures. All extracted values are summarized in Supporting Table S1.

Since ReS₂ exhibits anisotropic absorption , whereas WS₂ is isotropic, it is important to clarify whether ILE formation depends on the excitation polarization. Using 2.33 eV excitation, which lies well above the ReS₂ excitonic resonances and also efficiently excites WS₂, we removed the analyzer and rotated the excitation polarization angle in 10° increments (Figure a). The resulting spectra in Figure e show that the ILE intensities remain nearly constant regardless of Θ, whereas X₁ and X₂ retain clear angular dependence, consistent with the anisotropic absorption of ReS₂. , This contrast is highlighted in Figure f, where the same polar-fit analysis described above reveals that IX₃ exhibits negligible excitation angle dependence, whereas X₁ remains anisotropic. The anisotropic ratio $\eta =\frac{I_{max}}{I_{min}}$ of IX₃ and X₁ is 1.02 and 1.60, respectively, confirming isotropic and anisotropic formation of the ILE and intralayer exciton, respectively. Results for the remaining peaks are provided in Supporting Figure S11. The ILE formation remains excitation-polarization-insensitive in the two samples with the same stacking sequence despite their substantially different twist angles (7.6° and 27.9°), whereas the sample with reversed stacking at 3.1° shows excitation-polarization dependence (Supporting Figure S9). These results suggest that both stacking sequence and low-angle structural reconstruction may influence optical anisotropy, which requires further investigation. Anisotropic ILE emission is also observed in the 1R/2W, and the same set of analyses applied to 2R/2W is provided in Supplementary Note 3.

We further performed first-principles GW-BSE calculations , to investigate excitons in the 2R/2W heterostructure. Figure b and c show the modulus squared of the exciton wave functions ${|{\chi }_S(r_e,r_h)|}^2$ , where we find that electrons are strongly confined within the ReS₂ layers while holes are confined to the WS₂ layers, confirming their interlayer nature. Figure d shows optical absorption spectra obtained from the BSE calculations. We identify three ILE peaks IX₁ (1.186 eV), IX₂ (1.273 eV), and IX₃ (1.293 eV) corresponding to the experimental measurements. These ILE peaks are predominantly composed of transitions involving the highest valence band (v₁) and the lowest two conduction bands (c₁, c₂) at the Γ-point. IX₁ is built mainly on c₂ while IX₂ and IX₃ are built mainly on c₁ (Figure a). Figure e shows polarization-dependent oscillator strengths. IX₁ and IX₂ are polarized perpendicular and IX₃ is parallel to the Re-chain direction. The three ILEs share the same dipole matrix elements and differ only in their envelopes (Supplementary Note 4). The optical anisotropy of the ILEs does not depend on the stacking registry of the interface (Supplementary Note 5). We note an additional exciton peak at 1.135 eV, which is below the experimental measurement range.

Discussion

While previous studies on ReS₂/WS₂ heterobilayers have reported ultrafast charge transfer arising from their type-II band alignment, ILE emission has not been observed, which has been primarily attributed to momentum mismatch between monolayer ReS₂ with its conduction band minimum (CBM) at the Γ-point, , and monolayer WS₂, whose valence band maximum (VBM) lies at the K-point. Momentum matching of the interlayer band edges is crucial for optically active ILE emission. ,, In this context, Γ-point ILE emission has been demonstrated in type-II heterostructures composed of isotropic 2D materials such as InSe, or PbI₂, whose CBM lies at Γ, combined with bilayer WS₂ with its VBM also at Γ. , However, the 3-fold in-plane symmetry of the isotropic constituents limits anisotropic optical responses. In contrast, we observed anisotropic ILE peaks in the range of 1.2–1.4 eV by combining bilayer WS₂ with ReS₂, demonstrating that the anisotropy of ReS₂ is imparted to the resulting ILE. These ILE peaks appear well below the energies of the constituent intralayer excitons, unlike prior reports where spectral overlap led to ambiguity. , Moreover, the observed anisotropy arises intrinsically from ReS₂ rather than from extrinsically induced symmetry breaking, such as strain, metastructure, or magnetic field. −

Conclusions

In conclusion, direct interlayer excitonic transitions enabled by Γ-point momentum matching between monolayer or bilayer ReS₂ and bilayer WS₂ give rise to robust anisotropic interlayer excitonic emission consisting of three distinct ILE peaks, whose polarization orientations originate from the intrinsic optical anisotropy of ReS₂ and remain similar across the investigated twist angles. The ILE peak centers appear in the range of 1.37–1.42 eV for 1L-ReS₂/2L-WS₂ and 1.32–1.37 eV for 2L-ReS₂/2L-WS₂ heterostructures. Polarization analysis further reveals that ILE exhibits markedly stronger anisotropy than the intralayer excitons of ReS₂. In contrast, the excitation-insensitive formation of anisotropic ILE provides a stable exciton supply for reliable exciton transport. Our GW–BSE calculations further corroborate the experimental observations by revealing a direct Γ-point interlayer bandgap and strongly anisotropic ILE. These findings position momentum-matched anisotropic-isotropic heterostructures as a platform for polarization-selective optoelectronics and anisotropic excitonics.

Methods

Sample Preparation

To assemble the ReS₂/WS₂ heterostructure, we used mechanical exfoliation and dry transfer. Multilayer ReS₂ and WS₂ were exfoliated from their bulk crystals (2D Semiconductors and HQ Graphene) with mechanical exfoliation (the Scotch tape method) onto SiO₂/Si substrate and PDMS (Gel-Pak X4), respectively. We prepared WS₂ with a cleaved edge along the isotropic crystallographic direction, exhibiting 60° angles at the corners, and ReS₂ with a clearly defined b-axis along the anisotropic crystallographic direction. We sequentially stacked WS₂ and ReS₂ onto SiO₂/Si while aligning the crystal axes, utilizing a rotation stage (Zaber XLSM025A) having an angular accuracy of up to 80 millidegrees (0.08°). We used a transfer stage (Newport XYZ-PPP) inside a humidity-controlled glovebox to reduce the effects of adsorbates, contaminants, and oxidation of sample during transfer.

Characterization

PL/Raman spectroscopy, and time-resolved PL measurements were conducted using a customized confocal Raman microscope (NanoBase XPER Raman system). For PL and Raman measurements, gratings with 300 lines per mm (lpmm) and 1800 lpmm were used, respectively, providing spectral resolutions of 0.9 nm and 4.3 cm^–1. For excitation angle-dependent microscopic investigations, a linear polarizer and a half-wave plate were incorporated in the incident laser path to adjust the laser polarization direction relative to the b-axis of ReS₂. An analyzer was added and rotated for polarization-resolved PL measurements. Area mapping of optical responses was performed with 1 μm spatial resolution using a galvo-mirror scanner, and area-averaged spectra were extracted to minimize discrepancies caused by surface roughness in single-point spectra.

For PL/Raman spectroscopy, a continuous-wave 532 nm green laser (Cobolt Tor) was used as the excitation source and emitted and scattered light was collected by the detector. Laser power was controlled using a continuous neutral density filter. Charge-coupled device camera (iVac 316, Andor) was employed for detection of PL/Raman signals.

TRPL measurements were performed using a time-correlated single photon counting (TCSPC) system, consisting of a single-photon avalanche detector (PDM series, PicoQuant) and time-tagging electronics (TimeHarp 260, PicoQuant) with a timing resolution of 30 and 25 ps, respectively. A 405 nm picosecond pulsed diode laser (PDL 800D, PicoQuant) operating at an 80 MHz repetition rate was used for excitation. To isolate the 405 nm excitation light and selectively detect either neutral 1W excitons or the ILE emission from 1R/2W and 2R/2W, a 405 ± 10 nm bandpass filter was inserted in the excitation path, and a 610 ± 10 nm bandpass filter or an 875 nm long-pass filter was placed before the detector.

Atomic force microscopy (AFM) topography images and data were obtained using a Park Systems XE-7 under ambient conditions.

Calculations

Density functional theory (DFT) calculations were performed using the Quantum ESPRESSO package. We used the Perdew–Burke–Ernzerhof (PBE) functional and the optimized Vanderbilt norm-conserving pseudopotentials. , The van der Waals interactions were included within the Grimme-D2 scheme. Electron wave functions are expanded in plane wave basis with an energy cutoff of 80 Ry, and the Brillouin zone was sampled by an 8 × 8 × 1 k-point grid.

We used energetically the most favorable stacking structures between each ReS₂ and WS₂ layer. The determination of these structures and the resulting stacking at each interface are given in Supplementary Note 5. We construct a commensurate supercell based on 1 × 1 ReS₂ unit cell and 2 × 2 WS₂ unit cells. This configuration results in strains of +0.06% and −0.76% (+0.03% and +0.60%) for ReS₂ (WS₂) along a- and b-axes, respectively.

GW-BSE calculations were performed by using the BerkeleyGW package. In the GW calculations, the dielectric matrix was calculated with a cutoff of 20 Ry, and 5000 bands were included and we used the generalized plasmon pole model. In self-energy calculations, we employed the nonuniform neck subsampling (NNS) method with 10 subsampled points, corresponding to an effective q-grid of 1154 × 1154 × 1. The Bethe-Salpeter equation (BSE) was solved on a fine k-point grid of 48 × 48 × 1, considering 10 valence and 10 conduction bands. The Coulomb truncation scheme for slab geometry was applied.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.6c11574.

• Sample characterization and structural information, including layer-number and crystal-orientation determination, heterostructure fabrication, AFM measurements, and SAED-based twist-angle identification; additional ILE spectra, twist-angle and stacking-sequence statistics, polarization-resolved and excitation-polarization-dependent PL analyses, spectral fitting, and anisotropy quantification; anisotropy and GW-BSE analysis of the 1R/2W heterostructure; analysis of ILE polarization axes, interference of electron–hole pair contributions, band anisotropy, and interface stacking configurations; sample parameters and TRPL fitting results (PDF)

#.

T.K.Y. and B.L. contributed equally to this work.

The authors declare no competing financial interest.