Introduction

Bismuth-based compounds are of significant interest1 due to their exceptional electronic properties, particularly their strong spin-orbit coupling, leading to intriguing quantum phenomena such as topological states. Compounds like e.g., Bi1-xSbx2, BiSbTe1.25Se1.753, Bi2Se3, and Bi2Te34 are well-known topological insulators with significant potential for advanced quantum devices, including those of spintronics5 topological superconductivity6, or quantum information processing7.

While the properties of group III-Bi compounds have been extensively explored, focusing on the tunability of band gaps8 and enhanced spin-orbit interaction9, the potential of low-dimensional Bi-group V materials remains largely uninvestigated. Theoretical studies have predicted that two-dimensional (2D) Bi-based binary compounds with group V elements10,11,12, such as BiAs, could exhibit novel electronic behaviors, including topological phase transitions when strained. However, experimental realization and study of such materials have been sparse. Recent experimental studies involving Bi on SiC13 and Bi on GaAs(111)B14 have demonstrated the formation of 2D Bi-based structures on top of substrates, respectively, as well as the presence of nontrivial band gaps and exotic quantum states within these gaps.

In this context, InAs is an interesting substrate material due to its direct band gap, high electron mobility15, and easy implementation into existing Si-based fabrication lines16,17. Its properties enable a versatile platform for various advanced applications, including infrared sensors18, low power electronics19, and quantum information processing20,21. Moreover, the interaction of Bi atoms with InAs(111)A22 and InAs(110)23 surfaces has been shown to induce quasi-one-dimensional surface states with large Rashba spin splitting, which is of great interest for spintronic devices that rely on spin-polarized transport.

In this study, we present the experimental realization of a 2D BiAs layer on an InAs(111)B substrate via molecular beam epitaxy (MBE). The BiAs layer is capped with an amorphous As overlayer to protect the interface during sample transfer. A gradual thinning of the overlayer is performed with low-energy sputtering and annealing, exposing the InAs:BiAs interface. A combination of experimental techniques, including synchrotron-based angle-resolved photoemission spectroscopy (ARPES) and X-ray photoelectron spectroscopy (XPS), low-energy electron diffraction (LEED), and scanning tunneling microscopy (STM), confirms, together with density functional theory (DFT) calculations, the formation of a stable (1×1) BiAs structure with significant Rashba splitting. Notably, the As overlayer plays a crucial role in maintaining a structural displacement between the InAs substrate and BiAs layer necessary for Rashba splitting, while preventing reconstruction. Therefore, it can be utilized to stabilize 2D compounds, especially group V-rich ones, which otherwise easily decompose.

Our findings establish BiAs as a promising 2D giant Rashba material, demonstrating its potential for applications in spintronic devices and field-effect transistors, where the ability to generate and control spin currents via spin-orbit coupling is key. The giant Rashba splitting observed in this system opens new avenues for exploring Bi-V compounds as highly tunable platforms for quantum technologies.

Results and discussion

Structural and chemical analysis

First, we will present the structural and chemical properties measured for the InAs(111)B sample after thinning the As overlayer to a fraction of a nanometer and with less than a monolayer (ML) of Bi at the interface. Probing the surface with LEED reveals a superposition of three hexagonal surface configurations, as shown in Fig. 1a. We observe the unreconstructed hexagonal pattern of the InAs(111)B substrate (green arrow) and a (2×2) reconstruction (purple arrows). A (1×1) layer with slightly smaller lattice constant is also present (red arrow). By analyzing the XPS peaks of the As 3 d and the Bi 5 d core level (see Fig. 1b, c), we can determine several chemical environments. For the As 3d core level peak, we relate the dominating component at higher binding energies to As-As bonds in the As overlayer24,25,26. Varying the As overlayer thickness induces changes within the As 3d core level spectrum (as well as the corresponding ARPES spectra) shown in Supplementary Fig. 1. The reduction of the pure As layer is correlated with the decrease of the component at higher binding energies as expected. Furthermore, two additional doublets are present: one at about 0.62 eV and the other at about 0.91eV lower binding energy than the As-As component. We attribute the first to the formation of Bi-As bonds on the surface of the InAs substrate below the As capping14,27. The second signal originates from the bulk As-In bonds in the substrate itself27. Analyzing the Bi 5d core level spectrum reveals two components with a significant peak at higher binding energies corresponding to Bi-As bonds, as observed in literature14,28. A small shoulder at about 0.65 eV lower binding energy is also seen, implying a presence of about 12% of Bi-Bi bonds on the surface27. The InAs substrate is originally As-terminated, and we do not expect In to leave the respective lattice position at the chosen annealing temperatures ( < 300 °C) to bond with Bi atoms. This is in agreement with the Bi 5d core level spectrum in Fig. 1c, where Bi-In bonds would be situated at lower binding energies compared to metallic Bi27,29 and are not detected. Additionally, the In 4d core level spectra do not display any bond formation with Bi (see Supplementary Fig. 2).

Fig. 1: Structural and chemical analysis of the sample.
Fig. 1: Structural and chemical analysis of the sample.The alternative text for this image may have been generated using AI.
Full size image

a LEED pattern of a sample exhibiting the same band structure as the sample shown in Fig. 2b (taken at 102 eV). Three hexagonal configurations are visible: green arrow – (1×1), purple – (2×2) reconstruction, red – (1×1) with smaller lattice vector. XPS spectra of b the As 3 d, and c the Bi 5 d core level. Both spectra were taken with a photon energy of 70 eV. d Cross-sectional model of the sample suggested based on the experimental data and related to the ARPES spectrum in Fig. 2b.

By combining LEED and XPS results, we propose a model for the atomic arrangement on the InAs surface after most of the As overgrowth has left the surface shown in Fig. 1d. Starting with the deepest layer, we do not expect reconstruction of the InAs surface corresponding to the strong (1×1) LEED pattern in Fig. 1a. Based on the Bi 5 d core level spectrum, we conclude that the next layer is dominated by BiAs. As we will discuss in more detail below, DFT calculations exclude the presence of a (2×2) reconstructed BiAs layer when compared to the experimental ARPES data at this stage and suggest a (1×1) arrangement. The detectable amount of metallic Bi indicates that the original Bi deposition resulted in cluster/island formation rather than a well-defined monolayer, in agreement with literature29. This local inhomogeneity, combined with the fact that less than a monolayer of Bi was deposited, suggests that the (1×1) BiAs compound is not expanding over the whole surface. We expect the rest of the InAs surface to be covered with As atoms from the capping. Therefore, we attribute the (1×1) diffraction spots with slightly smaller lattice constant to either the BiAs layer or the As atoms right at the InAs surface. During the annealing of the substrate for several hours at temperatures <300 °C, enough energy was provided for excess As atoms to desorb and move into gas phase. This illustrates that the rearrangement of the As atoms is possible during this process step. Therefore, we conclude that the (2×2) reconstruction observed by LEED corresponds to a rearrangement of the remaining As atoms at the interface to the BiAs layer induced by the underlying crystalline ordering. Interestingly, this ordering does not carry through the whole overlayer. STM images indicate an amorphous structure on top of the As capping (see Supplementary Fig. 3).

Electronic properties of the BiAs layer

To understand the possible changes in the electronic band structure induced by the BiAs layer, ARPES spectra are recorded. We first consider a pristine InAs(111)B surface after oxide removal and without Bi deposition, shown in Fig. 2a. The ARPES spectrum agrees with experimental and theoretical data found in the literature30 and our DFT calculations along the Γ-M direction (see Supplementary Fig. 4). Only parabolic bulk valence bands are observed for the unreconstructed surface with the Fermi level positioned inside the band gap (see the STM image in Supplementary Fig. 5). Next, we analyze the InAs(111)B sample with Bi atoms and less than a nanometer of As capping left on the surface (as determined by XPS). The corresponding ARPES spectrum along the Γ-M direction is significantly different from the pristine InAs(111)B surface (see Fig. 2b–d). The bulk bands of the InAs(111)B surface appear to be shifted downwards by approximately 0.8 eV. Two additional features can be identified as shown in Fig. 2b: (i) a characteristic M-shaped band consisting of two hole pockets moving around Γ with maxima at about ± 0.2 Å−1 (white arrows), and (ii) an electron pocket centered at the Γ-point (red arrow).

Fig. 2: Results of the ARPES measurements.
Fig. 2: Results of the ARPES measurements.The alternative text for this image may have been generated using AI.
Full size image

a ARPES spectrum of pristine InAs(111)B surface with the Fermi level (EF = 0) positioned inside the band gap. b ARPES spectrum of an InAs(111)B substrate with less than a ML of Bi and an overlayer of As of less than a nanometer. The states below 1 eV dominate the contrast. Both spectra were taken at 24 eV around normal emission. The induced band structure changes are highlighted with differently colored arrows. The ARPES spectrum corresponds to LEED and XPS data shown in Fig. 1a–c. c Normalized momentum dispersion curves of (b) highlighting the electron pocket and newly formed electronic states close to the Γ-point. d Constant energy map at the Fermi level taken at 25 eV corresponding to the sample conditions in (b).

To gain a deeper understanding, we explore the effects of Bi deposition and subsequent As capping by performing DFT calculations in two stages: (i) the initial adsorption of Bi on the pristine InAs(111)B surface without any As capping, and (ii) Bi adsorption followed by an As overlayer with varying thickness. In the first stage, we evaluate three possible adsorption sites for the Bi atom: type-I (directly on top of As atoms in the first layer), type-II (above In atoms in the second layer), and type-III (above In atoms in the first layer, see Supplementary Fig. 6 for details).

After conducting force minimization via DFT, we find that one Bi atom per unit cell is most energetically stable in the type-I configuration, with its total energy per unit cell approximately 85 meV lower than the type-II configuration, and 254 meV lower than type-III. However, adding one ML of As on top shifts the relative stability between types I and II. Now, the type-II configuration becomes energetically more favorable by 65 meV. This suggests that the deposition of an As overlayer stabilizes the shifted BiAs monolayer in the type-II structure. However, the actual configuration adopted by the system may also depend on the energy barriers between configurations. Given the consistently higher energy of type-III (by over 250 meV), our subsequent discussion focuses exclusively on the type-I and type-II structures.

In Fig. 3, we present the calculated band structures for the type-I and type-II configurations. In the absence of the As capping, the states at the Fermi level are predominantly derived from Bi atoms, and their dispersions are highly sensitive to the Bi adsorption site, as shown in Fig. 3a, b. However, with the introduction of the As capping layer, the As atoms saturate the dangling bonds of Bi, pushing the Bi-related states down to approximately 1.3 eV below the Fermi level. Consequently, the states at the Fermi level are now dominated by the As atoms from the 2D BiAs layer, exhibiting the characteristic M-shaped dispersion for both type-I and type-II configurations.

Fig. 3: DFT calculation of band structures for different placement of the Bi atom.
Fig. 3: DFT calculation of band structures for different placement of the Bi atom.The alternative text for this image may have been generated using AI.
Full size image

Calculated band structure of the InAs(111)B slab unit cells exhibiting Bi adatoms at different positions on the surface before and after capping with a ML of As atoms. Panels (a) and (c) refer to the type-I slab in which the Bi atom is directly placed on top of an As atom in the first layer. b,d depict the projected band structures for the type-II slab in which the Bi atom is placed above the In atom in the second layer of the InAs substrate. The band wavefunction are projected onto all As atoms in the InAs slab for (a) and (b), and only onto the As atoms within the BiAs and first layer of the InAs plane for (c) and (d). The Fermi level is shifted and set to zero in all cases. After the deposition, the type-II structure becomes stable and recovers the pronounced M-shape, which was evident in the type-I band structure before As capping. The colored dots represent contributions from different elements: In (magenta), As (blue), and Bi (gold).

Overall, an important distinction between the two bonding configurations can be made after the formation of BiAs: the higher-energy type-I configuration follows the overall crystal structure of the InAs(111)B substrate and represents a substitution of In atoms by Bi in the top surface layer. This structural continuation limits any significant Rashba splitting in the electronic structure, see Fig. 4a, c. In contrast, the type-II configuration exhibits a shift in the crystal structure at the interface of the InAs(111)B and the BiAs layer. This displacement relative to the substrate is key to the emergence of Rashba-type splitting (Fig. 4b, d) within the BiAs layer after the deposition of As on top of Bi.

Fig. 4: DFT analysis of the Rashba splitting.
Fig. 4: DFT analysis of the Rashba splitting.The alternative text for this image may have been generated using AI.
Full size image

Structural and electronic properties of the type-I and type-II configuration on InAs(111)B slab unit cells with BiAs monolayers including a single layer As on top. a,b show the structural configurations, with Bi (gold), As (blue), and In (gray) atoms. In type-I, the BiAs monolayer is unshifted, in contrast to type-II. The energy difference between the two structures is EI − EII = 65 meV. c,d display the spin-resolved band structures for type-I and type-II along the M − Γ−M direction. Red and blue dots represent spin-up (Sy ↑ ) and spin-down (Sy ↓ ) states, respectively. Type-I exhibits k-independent atomic spin-orbit splitting between the M and Γ points, while type-II shows Rashba splitting around Γ, combined with both Rashba and atomic spin-orbit effects between M and Γ.

In Fig. 4c, d, we show the spin-resolved DFT band structures for the type-I and type-II configurations, respectively. These calculations reveal that the origin of the Bi-induced spin-orbit interaction depends on the relative positioning of the Bi atoms to the substrate. Only the type-II configuration exhibits the Rashba splitting and spin-resolved band structures for the Sx, Sy and Sz component can be found in the Supplementary Fig. 7, respectively. The resulting Rashba coefficient is determined as \({\alpha }_{R}\) = 2.141 eV·Å by including higher k-terms. An in-depth description of the fitting procedure is provided in the Supplementary Methods. According to literature, materials with \({\alpha }_{R}\) larger than 1.0 eV·Å are categorized as materials with a giant Rashba splitting31. Therefore, we conclude that a 2D BiAs layer on an InAs(111)B substrate sheltered by an amorphous As capping belongs to the group of giant Rashba materials. We believe that the observed effect is in good agreement with the experimental data reported here without resolving the direct band splitting at the Γ-point. However, the momentum dispersive curves of the ARPES spectrum presented in Fig. 2b reveal the presence of two bands within the M-shaped electronic states, with their respective maxima position highlighted in Supplementary Fig. 8, analog to the states in Fig. 4d for the energetically more favorable type-II configuration.

In the DFT calculations, we extend the investigation to include the deposition of crystalline As beyond the first monolayer. The resulting band structures (see Supplementary Fig. 10 and 11) exhibit significant deviations from the experimentally measured ARPES spectra, confirming that the As overlayer beyond the initial 2D BiAs layer is not crystalline but amorphous. Additionally, we calculate the band structure for varying thicknesses of As overlayers without Bi atoms (see Supplementary Fig. 12). While the excess As modifies the band structure of the InAs slab, the resulting bands do not match with our ARPES data, further confirming that the presence of Bi on the InAs(111)B substrate is crucial in determining the correct hopping interactions between As atoms and inducing the spin-orbit interaction necessary for Rashba splitting. Up to this point, the calculations focused on modeling an unreconstructed (1×1) BiAs layer. As a (2×2) reconstruction is also detected via LEED, we construct an InAs(111)B slab in the (2×2) supercell with a BiAs layer. Here, the Bi coverage is ¼ of the (2×2) supercell with four As atoms. The calculations show that the resulting band structure does not reproduce the M-shaped electronic states at the Γ-point, as shown in Supplementary Fig. 13. This confirms that the excess As atoms surrounding the BiAs layer effectively prevent a significant diffusion of Bi atoms across the InAs surface, and highlights the critical role of the As capping layer in maintaining the unreconstructed (1×1) structure for BiAs necessary for the emergence of Rashba-split states.

Chemical, structural, and electronic changes induced via additional annealing

When the sample is annealed further, all excess As atoms are removed. The corresponding As 3d and Bi 5d core level spectra are shown in Fig. 5. Here, no metallic As is detected. Instead, a shoulder at lower binding energies suggests that parts of the InAs(111)B substrate are not covered any longer, displaying an As-In surface component32. In general, Bi is still present on the surface and strongly bound to As atoms, as shown in Fig. 5b, c. Concerning the Bi 5d core level peak, we observe that the ratio between metallic Bi and Bi-As bonds is slightly shifted, promoting Bi-Bi bonds. It should be noted that the intensities for both spectra shown in Figs. 1c and 5c are similar, indicating that the majority of Bi atoms remained on the surface. As atoms are expected to be more volatile compared to Bi when heating the sample below 300 °C. As a result, they will evaporate at a higher rate, allowing for more Bi-Bi bonds to be formed on the surface, corresponding to the detected shift in the Bi 5d core level spectrum.

Fig. 5: Structural, chemical and electronic analysis of the sample after annealing.
Fig. 5: Structural, chemical and electronic analysis of the sample after annealing.The alternative text for this image may have been generated using AI.
Full size image

Measurements taken after the full As overlayer is removed: a LEED image displaying two main surface configurations: green arrow- (1×1), and purple – (2×2) reconstruction (taken at 102 eV). b XPS spectra of the As 3d, and c the Bi 5d core level. Both spectra were taken at 70 eV. d STM image of the surface exhibiting several areas with seemingly different symmetry (exemplary indicated by the red arrows). The brighter protrusions and clusters are interpreted as leftovers of the As overlayer (I = 130 pA, V = 2 V). e ARPES spectrum taken at 24 eV in normal emission. The new, M-shaped electronic states around the Γ-point are barely visible. Only the bulk bands of InAs(111)B and the electron pocket remain dominant.

Additionally, the LEED signal corresponding to the (1×1) layer with smaller lattice constant has disappeared when comparing Figs. 1a and 5a. STM images reveal a variety of different surface domains with hexagonal and seemingly square symmetry (red arrows), as observed in Fig. 5d and in the Supplementary Fig. 5 for a larger overview. Measuring the atomic distance at several locations in the STM image yields an average lattice constant of 8.8 ± 0.8 Å, which is in good agreement with a (2×2) reconstruction based on an unreconstructed InAs(111)B surface (lattice parameter 4.3 Å)33. The overall appearance of the surface is ordered giving rise to the (2×2) diffraction points in LEED superimposed with the (1×1) InAs substrate signal. However, variations within the position of the surface atoms (see Fig. 5d) give rise to streaks between the (2×2) diffraction spots in the LEED pattern. We conclude that the additional annealing facilitated the evaporation of most of As atoms involved in the overlayer. This allowed for easier diffusion of Bi atoms over the InAs surface leading to a (2×2) BiAs reconstruction.

These structural changes eliminate the newly formed electronic states around the Γ-point almost completely as displayed in Fig. 5e. The electron pocket is still present, which we attribute to band bending at the InAs(111)B surface induced by Bi atoms, accompanied by a downward shift of the InAs valence band of about 0.4 eV in our study. This effect has been observed before on InAs(111)B substrates after depositing other adatom species30,34.

Conclusion

In this study, we successfully demonstrated the formation of a 2D BiAs layer on an InAs(111)B substrate, revealing significant insights into its structural and electronic properties. By combining surface-sensitive experimental techniques with DFT calculations, we confirm the emergence of electronic M-shaped Rashba-split states near the Fermi level, centered at the Γ-point. This study demonstrates that the formed 2D BiAs layer belongs to the group of giant Rashba materials. Furthermore, we found that the presence of an amorphous As overlayer is essential for stabilizing the (1×1) BiAs layer, preventing unwanted (2×2) reconstructions. This acts as a capping lock-in the structural asymmetry required for Rashba effects, preserving the giant Rashba splitting. Our findings highlight the importance of employing a group V overlayer and controlled sample annealing to ensure a robust and reproducible growth process of novel 2D materials. The concept of using elemental capping layers as a route to stabilize or alter structural and electronic properties of 2D films is not confined to the present system, but could be widely expanded. This work paves the way for further exploration of Bi-V compounds as 2D Rashba materials and expands the potential for integrating BiAs into many quantum technologies, e.g., as spintronic devices, magnetic memory applications, devices utilizing spin-orbit torque, and spin-polarized LEDs.

Methods

Substrate preparation

A commercially available InAs(111)B wafer (n ~ 1018 cm−3) from Wafer Technology Ltd. was deoxidized in a Riber MBE 412 system in an ultra-high vacuum (UHV) environment and under an As flux of 1.2 ×10−5 Torr. The temperature was slowly ramped up to 450 °C, while checking the surface quality with reflection high-energy electron diffraction. After deoxidation, the sample temperature was set to 100 °C for the deposition of less than a monolayer of Bi. Finally, an amorphous As overlayer of 20 nm was grown at around 5 °C to prevent surface oxidation.

To investigate the sample, single pieces of the substrate were sputtered for 10-15 min with an angle of 45° between surface normal and sputter gun with an Ar pressure of 5×10−6 mbar. Keeping the acceleration voltage to 0.5 keV allowed for a reduction of the As-oxide thickness without implementing a larger number of impurities. Subsequently, the sample was annealed step-by-step for several hours to temperatures around 200 °C–300 °C to desorb the excess As but prevent the desorption of Bi atoms from the substrate surface. The annealing temperature was measured with a thermocouple mounted on each sample plate.

Characterization techniques

Measurements involving ARPES, XPS, LEED (a ScientaOmicron 3-grid SPECTRALEED), and STM (a ScientaOmicron VT-STM XA) were carried out at the Bloch beamline35 (Max IV, Lund). Additional measurements were carried out at the SGM336,37 beamline (ASTRID2, Aarhus). ARPES spectra were recorded with photon energies ranging between 18 and 150 eV. As 3d, Bi 5d and In 4d core level spectra were taken with a photon energy of 70 eV and 130 eV in conjunction with the respective ARPES spectra at the same sample position. To analyze the As 3d and Bi 5d core level peaks, a Shirley background was removed and fitting of the individual doublets (branching ratio: 1.54 ± 0.02 and 1.37 ± 0.01; spin-orbit splitting: 0.70 eV and 3.02 eV for As and Bi, respectively) was performed with a pseudo-Voigt function. The Lorentzian full width half maximum was kept at 0.18 eV and 0.25 eV for the As and Bi spectra, respectively. The thickness of the As overgrowth was estimated by analyzing the change in intensity of the In 4d core level spectra originating from the InAs substrate as described and demonstrated elsewhere27,38. The energy resolution was below 40 meV for XPS and 20 meV for ARPES measurements, with the sample temperature being around 17 K (Bloch) and 35 K (SGM3). Where necessary, the Fermi Energy was calibrated on the sample holder in electrical contact with the sample. The binding energy was determined via Ekin-EF considering the kinetic energy of the detected photoelectrons (Ekin) and the of the Fermi level (EF). STM and LEED measurements were done at room temperature.

As a reference, a similar InAs(111)B substrate was analyzed after removing the native InAs oxide in UHV via hydrogen treatment39. Here, the sample was annealed to 400 °C for about 40 min while exposing to 5×10−6 mbar of hydrogen with a temperature of the Hydrogen cracker of 1700 °C.

Theoretical calculations

Density Functional Theory (DFT) calculations were performed using the Vienna Ab-initio Simulation Package, employing the projector-augmented wave method40,41,42. The Perdew-Burke-Ernzerhof generalized gradient approximation was used to model the exchange-correlation interactions. For the plane-wave expansion of the wavefunctions, a kinetic energy cutoff of 400 eV was set. All geometric relaxations were carried out until the total energy and ionic forces met a threshold convergence criterion of 10⁻⁸ eV and 10⁻⁷ eV/Å, respectively.

During the relaxation, the effects of spin-orbit coupling (SOC) were excluded to reduce computational cost. However, the electronic structure calculations were performed both with and without the inclusion of SOC for comparative analysis. Unlike previous studies30, no Hubbard-U corrections were applied in our calculations. While Hubbard-U corrections are generally required to reproduce the correct band gap at the Γ-point for a bulk 3D-periodic DFT calculation of zinc-blende (ZB) InAs, we opted not to apply these corrections for our slab model since the primary goal was not to achieve the precise band gap value but rather to qualitatively investigate the surface electronic states. Our slab model produced a gap at the Γ-point without such adjustments, which was sufficient for the scope of our study.To model the InAs(111)B surface, we employed a (1×1) periodic slab supercell, with the As-terminated InAs(111)B surface defined as the upper surface of the slab. The Brillouin zone was sampled using an 8×8×1 k-point grid for accurate integration during DFT calculations. A vacuum layer of 15 Å was introduced along the z-axis to minimize interactions between periodic images of the slab.

The creation of the top and bottom surfaces in the slab model inherently breaks the symmetry in the out-of-plane direction, leading to dangling bonds at both surfaces. To mitigate the impact of these dangling bonds and better mimic the bulk-like chemical environment, the bottom surface (In-terminated) was saturated with pseudo-hydrogen atoms carrying a nuclear charge of 1.25e, while the top surface (As-terminated) was left unsaturated to preserve the surface characteristics under study. Atomic relaxation was performed by allowing all atoms to move freely, with both forces and energy minimized until convergence thresholds were met. This process resulted in relaxed lattice parameters of a = b = 4.39 Å. To investigate the influence of Bi on the electronic structure of the InAs(111)B surface, we modeled three distinct (1×1) surface configurations, each corresponding to a different adsorption site for the Bi atom on the InAs(111)B substrate (see the experimental part for more details). After depositing Bi, both the Bi atom and the topmost InAs surface atoms were allowed to relax fully through energy and force minimization. Finally, we introduced an As capping on top of the InAs(111)B:Bi system to explore its effects on the BiAs formation. This is modeled by constructing up to six layers of metallic crystalline As above the Bi atom and subsequently relaxing the entire system.