Heteroepitaxial growth of highly anisotropic Sb₂Se₃ films on GaAs†

Kelly Xiao ^a, Virat Tara ^b, Pooja D. Reddy ^a, Jarod E. Meyer ^a, Alec M. Skipper ^c, Rui Chen ^b, Leland J. Nordin ^de, Arka Majumdar ^bf and Kunal Mukherjee *^a
aDepartment of Materials Science and Engineering, Stanford University, Stanford, CA 94305, USA. E-mail: kunalm@stanford.edu
^bDepartment of Electrical and Computer Engineering, University of Washington, Seattle, WA 98195, USA
^cInstitute for Energy Efficiency, University of California Santa Barbara, Santa Barbara, CA 93106, USA
^dDepartment of Materials Science and Engineering, University of Centra Florida, Orlando, FL 32816, USA
^eCREOL, The College of Optics and Photonics, University of Central Florida, Orlando, FL 32816, USA
^fDepartment of Physics, University of Washington, Seattle, WA 98195, USA

First published on 12th May 2025

---

---

New concepts Anisotropic chalcogenides exhibit coveted properties but often require an understanding of synthesis routes that preserve their anisotropic properties in non-bulk forms. Insights into substrate engineering or other growth techniques are valuable in integrating low-dimensional chalcogenides with conventional cubic substrates. We demonstrate the concept of harnessing symmetry-breaking surface reconstructions to yield oriented growth of “quasi-1D” antimony selenide (Sb2Se3) on isotropic semiconductor templates. To date, Sb2Se3 growth is relatively immature, which is a potential reason the intrinsic anisotropy of Sb2Se3 single crystal is rarely reported in previous works on thin films. In contrast, we present findings on large optical anisotropy in the near-infrared among all three orthogonal directions in oriented Sb2Se3 thin films grown on reconstructed GaAs templates. This work aims to highlight fundamental synthesis–structure–property links in Sb2Se3. We turn the spotlight on Sb2Se3 thin films as a material with accessible strong anisotropy, opening further investigation for photonic applications that require phase or polarization control.

1. Introduction

Sb₂Se₃ is a layered material in the V₂–VI₃ chalcogenide family (e.g. Sb₂Te₃, Bi₂Se₃, Bi₂Te₃). Rather than crystallizing in the 2D rhombohedral structure as the prototypical members do, Sb₂Se₃ adopts the lower symmetry Pbnm orthorhombic structure.^1,2 Orthorhombic Sb₂Se₃ is isostructural with stibnite (Sb₂S₃), which exhibits covalently chained bonding along only a single “needle” or 1D axis to form parallel ribboned units. Along the other two crystallographic directions, the ribbons loosely pack into a van der Waals ensemble owing to steric effects imparted by the Sb 5s² lone pair.³ These asymmetric interactions along each primary axis in Sb₂Se₃ are the origin of its quasi-1D character. Unsurprisingly, a set of unique optoelectronic and structure-related properties converge in this low-dimensional system, among these favorable electronic and thermal transport along the needle axis,⁴ as well as high absorption of the visible wavelengths (E_g = 1.1–1.2 eV).^5,6 In light of these compelling qualities, Sb₂Se₃ has been proposed as a candidate material for thermoelectric alloys^7,8 and photovoltaic absorbers.^9–14 The inherent 1D nanostructure of this material has additionally inspired a technological focus on its “benign grain boundaries”, which in theory are intrinsically passivated and potentially self-healing, enabled by large reconstructive bonding displacements within the open structure.^15,16 Beyond optoelectronic applications, recent efforts are also revisiting the crystalline-amorphous transition of Sb₂Se₃ for phase change based reconfigurable photonic devices.^17–19 This reversible transition in the orthorhombic antimonides is rare in that it maintains near-zero losses in the near- to mid-infrared spectral ranges across both defective polycrystalline and amorphous states.¹⁹ Collectively, these applications underscore the broad impact of this material system, pointing to exciting avenues for future research and technological development.

From both fundamental characterization and device standpoints, the properties of Sb₂Se₃ are made more intriguing by the strong anisotropy of the crystal,^3,20 and as such, merit investigation of synthesis routes that reliably reproduce and realize such anisotropy to harness its full electronic or optical utility. Focusing on applications in photonics, the very low-loss, high-index, and birefringent character of Sb₂Se₃ makes it an ideal candidate for integrated photonics. While prior experimental work on the optical anisotropy of single crystal or bulk Sb₂Se₃ is limited, first-principles modeling suggests biaxial character with giant birefringence as high as 0.64 between the 1D axis direction (formed from covalently-bonded Sb and Se atoms) and the primary vdW-bonded direction.³ This is borne out experimentally in the isostructural bulk Sb₂S₃ crystals with birefringence of ∼0.9 at 800 nm.²⁰ Films of naturally anisotropic materials²¹ offer a complement to current approaches that use form-birefringence achieved by nanostructuring isotropic materials, which is lossy and challenging to fabricate. Single-crystalline or suitably textured Sb₂Se₃ films can offer much needed functionality where birefringence or dichroism is harnessed in integrated components for polarization-sensitive light detection, polarization rotating waveguides, and even higher index contrast in amorphous to single crystalline transitions. As a specific use case, we note integrated photonics are inherently polarization sensitive, which poses a serious limitation compared to free-space optics. Current efforts in polarization conversion in integrated photonics primarily employ sub-wavelength gratings that inevitably introduce additional loss.^22,23 Therefore, a low-loss anisotropic material integrated on photonic waveguides may help resolve this polarization discrimination.

While dissimilar growth interfaces in heteroepitaxy are energetically unaccommodating, the van der Waals epitaxy growth mode adopted by low-dimensional materials partially relaxes structural and bonding constraints,²⁴ opening exciting pathways towards integration with conventional single crystal cubic platforms. In this respect, healthy initial progress in heteroepitaxy of layered Bi₂Se₃, Bi₂Te₃, and Sb₂Te₃ on GaAs(001) and Si(111) substrates has been made,^25–29 yet ultra-high vacuum (UHV) film deposition methods for the orthorhombic V–VIs (Sb₂Se₃ and Sb₂S₃) remain underdeveloped. With molecular beam epitaxy (MBE), capabilities to monitor surface preparation and produce controlled growth rates may advance tuning of substrate–film interactions. We note demonstrations of heteroepitaxial growth of Sb₂Se₃ on muscovite mica substrates using vapor transport deposition³⁰ and on Bi₂Se₃ epilayers via MBE.^31,32 Both reports are encouraging examples of oriented Sb₂Se₃ growth, and they target distinct applications for Sb₂Se₃ in flexible electronics and topological phenomena, respectively. In this study, we aim to address practical challenges involving Sb₂Se₃ films on surfaces not of hexagonal or trigonal arrangement as has been demonstrated, but on technologically prevalent (001)-type cubic surfaces utilized in photonics. Here, we focus on the commercially accessible GaAs(001) substrate with the objective of exploring the Sb₂Se₃ synthesis space through growth temperature, Se and Sb beam fluxes, and substrate symmetry parameters. These constitute an initial investigation of whether Sb₂Se₃ growth is amenable to atomically clean III–V surfaces, and if the film structure can be systematically controlled despite the unorthodox stibnite structure. Sb₂Se₃ growth on conventional substrates may facilitate device integration beyond photonics; more recently, MBE growth of Sb₂Se₃ also on GaAs yielded nanostructures potentially useful for electronic applications.³³

It is important to state that the Sb₂Se₃ literature has mainly cited two space group conventions (Pnma and an equivalent setting Pbnm). As evidenced by numerous works dedicated to characterizing and optimizing the orientation of Sb₂Se₃,^9,13,34–37 explicit and consistent crystallographic indexing of the low symmetry Sb₂Se₃ material is critical to accurate cross-interpretation.³⁸ To avoid ambiguity, we reiterate that the Pbnm convention (a = 11.63 Å, b = 11.78 Å, c = 3.98 Å in the bulk, with [001] covalently-bonded chains) has been adopted for conceptual convenience throughout this work.

2. Results and discussion

2.1. Nucleation, growth, and morphology

Sb₂Se₃ films were prepared using molecular beam epitaxy (MBE) and separate Sb and Se solid sources. The high vapor pressure of Sb and Se require lower growth temperatures to enable sufficient adsorption and subsequent incorporation into the Sb₂Se₃ compound. With Sb and Se shutters open, we did not observe changes in reflection high energy electron diffraction (RHEED) until the substrate temperature approached ∼280 °C and below, suggesting desorption of the impinging Sb and Se at temperatures above 280 °C. We find successful growth temperatures (T_g) for crystalline films were between 180–265 °C. At 150 °C or below, we observe disordered phases of Sb₂Se₃ (Fig. S1 and S2, ESI†). The quality of the crystalline films was found to depend on substrate preparation. The first preparation is termed Se-treated GaAs, an epi-ready wafer which undergoes 600 °C thermal deoxidation under a Se overpressure prior to growth; the second is pristine regrown (homoepitaxial) GaAs preserved under an arsenic cap. This surface cap is also thermally desorbed prior to growth, but at much lower temperatures near 400 °C and without an intentional Se overpressure. We refer the reader to the film synthesis methodology for further details on III–V substrate preparation. Other growth parameters are also summarized for the films in this study in Table SI (ESI†).

Several film morphologies measured by atomic force microscopy (AFM) and corresponding RHEED patterns are compared in Fig. 1. At substrate temperatures of 265 °C or below, we observe that growth of crystalline Sb₂Se₃ directly on Se-treated GaAs forms a rough microstructure. As shown in Fig. 1a, protruding and mixed orientation crystallites are plentiful in the film grown on Se-treated GaAs and compromise the overall film quality. In Fig. 1b, the chevron pattern RHEED, typical of a rough 3D growth mode, further supports that a Se-treated surface is suboptimal for nucleation and growth. Over the growth duration, the intensity of RHEED slightly dims, suggesting that film quality struggles to improve even after the interface has been overgrown. Misoriented grains are particularly detrimental for Sb₂Se₃ film morphology because the crystal habits in stibnites may embody prismatic or bladed forms. These high aspect ratio crystals are clearly seen in an additional film grown at 265 °C with an increased Sb BEP of 2 × 10⁻⁷ Torr. (This Sb BEP is 4× greater than the 5 × 10⁻⁸ Torr BEP otherwise used throughout this work.) These growth conditions promoted inclined shards with poor coalescence across the substrate (Fig. S3, ESI†).

The structural quality of the films improves significantly on arsenic-capped regrown GaAs templates. We prepared films under the same growth rate of ∼0.4 Å s⁻¹ on regrown GaAs at T_g = 265 °C and 200 °C. Inclined crystallites are significantly reduced, resulting in an overall smoother and continuous film (Fig. 1c and e). AFM surface topography indicates an order of magnitude reduction in root mean square (RMS) surface roughness from 14 nm to 1.5 nm by switching to regrown GaAs templates. As captured by SEM and AFM, the primary surface feature of the smoother films is parallel faceting, which creates a distinct ribbon- or rod-like surface structure. At T_g = 200 °C, these parallel structures become less clearly defined, we suspect due to decreased adatom diffusion at the lower growth temperature. Geometric “rod” surface features were also observed for films grown on mica and Bi₂Se₃.^30,32 In these reports, domains manifested in parallelogram shapes with 60° or 120° angles, as opposed to the 0°/180° or parallel features we observe.

Improved nucleation can also be immediately observed in the RHEED patterns for films synthesized on regrown GaAs (Fig. 1d and f). Instead of a chevron pattern previously seen in Fig. 1b, Sb₂Se₃ RHEED remains streaky throughout the growth period along the GaAs [10] direction. Differences emerge in the RHEED pattern following nucleation for these two growth temperatures. At T_g = 265 °C, a faint streaky pattern forms by 60 seconds only along GaAs [110] and further degrades as the film grows thicker. At T_g = 200 °C, upon opening shutters, highly disordered growth first produces a temporary hazy pattern. The hazy pattern lasts across the “10–60 s” frame in Fig. 1f. Following the haze, a two-fold streaky pattern along two primary axes of the orthorhombic cell emerges, indicating a transition to longer range order crystalline growth. By the 30 minute mark in Fig. 1f, the [110]-aligned streaks are still weak. Extending the growth period, we see by the 90 minute mark that the streak intensity has further increased and shows a clearer periodicity. This likely corresponds with gradual improvement in crystallinity along this second axis in Sb₂Se₃ as the film increases in thickness. We note that different streak periodicities along the [10] and [110] azimuths suggest that diffraction arises from real space gratings of shorter and longer periodicity, consistent with the highly dissimilar lattice constants of orthorhombic Sb₂Se₃.

The stark contrast in morphology and RHEED between growths on Se-treated and regrown GaAs suggests that a smoother regrown zinc blende surface benefits growth quality significantly. Se-treated GaAs may include defects such as Se-terminated dimer rearrangement or Ga–Se byproducts.^39,40 Additionally, Se-treated GaAs develops surface pits due to the high temperatures required for in situ oxide removal and related outgassing.⁴¹ Pit formation roughening is quite visible in Fig. S3 (ESI†) and degrades the seed quality. Lastly, but perhaps most importantly, Se-treated GaAs does not include a so-called “buffer layer”, which is used in traditional III–V MBE to smooth out and restore the surface before subsequent epitaxial material is grown. Sb₂Se₃ appears to be structurally sensitive to surface defects on GaAs(001) and thus all other films discussed in this study are deposited on pristine regrown GaAs templates to preserve smoothness and crystallinity.

2.2. Epitaxial relationship and structural quality

As visually suggested in AFM and SEM, the ribbon surface features of Sb₂Se₃ films are oriented in-plane (IP) to the GaAs substrate, prompting us to investigate the structural properties further with X-ray diffraction (XRD). We report two types of crystalline structure in two growth temperature regimes of 230–265 °C (in-plane/IP textured) and 180–200 °C (epitaxial). Symmetric out-of-plane (OP) and phi scans, in addition to several structure schematic diagrams, are shown in Fig. 2 to characterize the grain orientations of Sb₂Se₃ on GaAs.

First, we characterize the out-of-plane orientations indicated by 2θ − ω scans in Fig. 2a. At higher growth temperatures between 230–265 °C, the film contains mostly mixed (hk0)-oriented OP domains, indicated by the presence of both low- and high-index X-ray reflections in the Pbnm space group (upper panel of Fig. 2a). Several (hk0) planes are illustrated in Fig. 2f. The abundance of both high and low symmetry orientations in Sb₂Se₃ is an unusual observation; we hypothesize that the predominantly vdW character common to (hk0)-plane terminations is responsible for this phenomenon. In general Sb₂Se₃ bonding can be decomposed into two components: (i) intrachain interactions (forming Sb–Se covalent bonds of length 2.59–3.21 Å) and (ii) interchain interactions (vdW-based with longer interatomic distances of 3.24–3.74 Å).² The (hk0) planes cut through the ribbons at an oblique angle and therefore primarily terminate the weaker interchain interactions within the Sb₂Se₃ unit cell.

Intermediate growth temperatures of 180–200 °C produce films favoring a primary (010)-orientation, as indicated by the first allowed (020) and higher order (040) and (060) reflections, seen in the lower panel of Fig. 2a. We believe it is most accurate to categorize this intermediate regime not as entirely independent from the high temperature regime (230–265 °C), but rather as a subtle transition away. Viewed on a logarithmic scale, symmetric 2θ − ω scans reveal that several (hk0) OP orientations present in the T_g = 230–265 °C films persist down to 180–200 °C, but their intensities have been dramatically suppressed. This (010)-orientation dominant XRD pattern may point to a more conventional case of vdW materials growth where exposure on a single low energy basal plane is preferred. Several reports have indicated that Sb₂Se₃ may exhibit such hallmarks of 2D vdW systems despite the quasi-1D nature.³ The weakest bonding exists along one axis in particular, namely [010], which has the furthest atom-to-atom distance in Sb₂Se₃.² There are several experimental reports on single crystal Sb₂Se₃ that support this primary (010)-cleavage or (010)-layered behavior.^38,42,43

As for the in-plane (IP) orientation of Sb₂Se₃, we find the Sb₂Se₃ needle axis (c-axis) is aligned to the GaAs [110] directions in both films. This crystallographic alignment is consistent with work by Wojnar et al.³³ Phi scans of GaAs {224} and two specific asymmetric planes in Sb₂Se₃ were probed depending on the grain OP orientation. For mixed (hk0)-oriented films, we probed the (120)-plane to understand the IP relationship of the (020) OP grains. Likewise, we used the (020)-plane to understand the orientation of (120) and (130) OP grains. These phi scans are shown in relation to the substrate {224} peaks in Fig. 2b. We see that two 180°-separated film peaks share the same azimuth as the GaAs [110] IP directions, indicating the IP relationship, Sb₂Se₃ [001]‖ GaAs [110], holds for multiple (hk0) OP-oriented grains. An illustrative schematic of these textured films is presented in Fig. 2c. In this work, we refer to these films as IP-textured or (hk0)-oriented Sb₂Se₃. The phi scans also reveal that 180°-rotated twin domains constitute the film. Single crystal orthorhombic Sb₂Se₃ has a single (020) pole; we have instead observed two 180°-separated peaks arising from the asymmetric (020)-plane in Fig. 2b.

The 200 °C epitaxial film, with dominant (010) OP grains, is similarly aligned in-plane to GaAs, as evidenced by a phi scan of the asymmetric (120)-Sb₂Se₃ plane (Fig. 2d). A schematic of the epitaxial grain structure is depicted in Fig. 2e. The epitaxial films are slightly higher crystalline quality than the IP-textured films. Rocking curves (RCs) are large for the IP-textured films, demonstrated by a 1.4° FWHM of the (020) RC and 1.2° FWHM of the (120) RC. The epitaxial films exhibit an out-of-plane (020) RC FWHM = 0.87°, a slight improvement but nonetheless with broadening contributions from low lateral coherence within the film. While these films can be improved structurally, they are distinguished from previous reports of epitaxial Sb₂Se₃ because 90°-rotated orthorhombic domains have been largely suppressed on the cubic substrate—resulting in a fully in-plane anisotropic structure not observed for hexagonally-twinned Sb₂Se₃ grown on more compatible Bi₂Se₃ templates.³²

Lattice mismatch is a critical parameter that determines the quality of traditional crystalline interfaces and subsequent growth, leading us to consider its potential effects on the orthorhombic-cubic Sb₂Se₃/GaAs interface. The alignment of orthorhombic Sb₂Se₃ (a = 11.63 Å, b = 11.78 Å, c = 3.98 Å) on GaAs (a = 5.65 Å) is depicted in Fig. 2g. Epitaxial Sb₂Se₃ crystallizes with the a- and c-axes in-plane, where the highly rectangular a–c face is 45°-rotated with respect to the GaAs (001) face. The a- and c-directions experience different lattice mismatch to the GaAs cubic diagonals. The longer a-axis exhibits a +3% mismatch to GaAs [110], derived from a d_SbSe(100): 3d_GaAs(110) ratio where d_SbSe(100) = 11.63 Å and d_GaAs(110) = 4.00 Å. There exists a smaller and thus more favorable +0.5% lattice mismatch between the shorter c-axis and GaAs [110] (d_SbSe(001): d_GaAs(110), where d_SbSe(001) = 3.98 Å). However, the role of favorable lattice mismatch in mediating the crystallographic relationship between Sb₂Se₃ [001] and GaAs [110] is unclear. We also observe evidence that substrate symmetry anisotropy plays a large role in influencing the needle direction, as we discuss at the end of this section.

We aim to further understand the transition in growth at the interface for epitaxial films through cross-sectional transmission electron microscopy (XTEM) shown in Fig. 3. The narrow growth regime in which (010)-oriented epitaxial Sb₂Se₃ forms is identifiable by the evolution of the RHEED pattern from hazy to streaky, shown previously in Fig. 1f. High-resolution (HR) or phase-contrast XTEM images in Fig. 3a show that Sb₂Se₃ closer to the interface (corresponding to the initial hazy RHEED) contains regions of misoriented lattice but is otherwise crystalline. Although hazy RHEED patterns are typically suggestive of amorphous character, in this case, it is possible that the initial growth suffers from such short-range lateral coherency relative to that of the electron beam that coherent diffraction does not occur. RHEED detects coherence on lengths of at least 100 nm.^44,45 We speculate that local clusters which had fortuitously nucleated in the primary vdW (010) orientation were energetically favorable sites for subsequent growth over other grain orientations. An example of such a (010)-oriented region is shown in Fig. 3b.

An alternative explanation for a disorder-to-order RHEED transition was proposed by Shayduk and Braun in MBE-grown cubic Ge₂Sb₂Te₅ on GaSb(001).⁴⁶ They hypothesized that the initial disordered phase crystallized epitaxially after sufficient incubation time to seed further epitaxial growth. This growth mechanism is not active in our experiment. Had the epitaxial seed of Sb₂Se₃ come from a transformed interface layer, we would expect to resolve lattice fringes signaling improved registry between the interface and the rest of the film, without signatures of overgrowth. In Fig. 3a(i)–(iii), we find that fast Fourier transforms (FFTs) taken along three lateral regions gradually further from the interface increasingly sharpen, indicating one contributor to improved structural quality is lower misorientation within the film as the (010)-orientation successfully overgrows. Inspecting the XTEM images further away from the Sb₂Se₃/GaAs interface, we find that some regions of the (010)-Sb₂Se₃ film produce clearer lattice fringes than others. This additionally points to a wide range of in-plane Sb₂Se₃ grain orientations about the GaAs [110] zone axis. The skew-symmetric (120)-Sb₂Se₃ RC exhibits a large FWHM of ∼1.5°, further supporting the presence of severe twist distortion in the 200 °C film. These structural characteristics point to the soft bonding of the Sb₂Se₃ crystal and the relatively low growth temperatures used.

The Sb₂Se₃ films otherwise growing epitaxially or with IP parallel domains requires factors which break the high symmetry of the substrate. We find that the (2 × 1) GaAs surface reconstruction, which likely modulates the interface energy as well as in-plane adatom diffusivity, plays a critical role. We hypothesize anisotropic cation mobility and diffusion across GaAs⁴⁷ has promoted the in-plane ordering of Sb₂Se₃. Directions with longer migration lengths would preferentially promote Sb₂Se₃ growth along its c-axis, as [001] sites are more active and readily available for adatom incorporation. When (2 × 1) GaAs was exchanged with a higher symmetry rocksalt (1 × 1) PbSe(001) template, the film consisted of four 90°-rotated populations instead (Fig. S4, ESI†). There is an even greater lattice mismatch of +8% between Sb₂Se₃ [001] and PbSe [110], as the PbSe lattice constant is 6.12 Å.

The relative group VI and V fluxes may also modulate diffusivity anisotropy on GaAs. We have demonstrated that under Sb and Se BEPs of 5 × 10⁻⁸ Torr and 1 × 10⁻⁶ Torr, respectively, that the surface is lightly corrugated from parallel-faceted grains. However, additional films prepared at 200 °C and lower Se BEPs suggest that the film morphology and in-plane needle alignment suffer (Fig. S5, ESI†) without sufficient Se overpressure at 1 × 10⁻⁶ Torr, highlighting the role of flux tuning to additionally set the epitaxial relationship. This Se BEP is indeed quite large, and we additionally notice that accumulated Se in the chamber negatively impacts the (010) OP orientation upon repeating consecutive 200 °C growths (Fig. S6, ESI†). With sufficient time to pump away the volatile Se vapor, we see recovery of epitaxial growth and repeatability of this procedure. Therefore, (i) exploring other low symmetry templates with exaggerated surface anisotropy, and (ii) ensuring a low Se environment prior to growth (to preserve a pristine surface) may improve monocrystallinity of Sb₂Se₃ thin films. This potentially has positive implications for UHV growth on CMOS-compatible cubic substrates that also develop a (2 × 1) reconstruction, such as Si and Ge (001).⁴⁸

2.3. Optical anisotropy

The relationship between the structure and optical properties in the Sb₂Se₃ films are highlighted in this section. Orthorhombic crystals are optically biaxial, an intriguing property which fundamentally originates from anisotropic atomic arrangements and their differences in polarizability. Conventional layered materials, with strong covalent bonding within layers and weak vdW bonding between layers, exhibit strong out-of-plane birefringence.^49,50 In orthorhombic layered materials, the refractive properties are different along all three crystal axes, uniquely enabling both out-of-plane and in-plane anisotropy. The films in this study preserve structural anisotropy and may serve as a platform to investigate the expectedly biaxial properties of Sb₂Se₃.

The film anisotropy is indeed reflected in the non-zero Mueller matrix off-diagonal terms (Fig. S7 and S8, ESI†). The complex refractive index (n + ik) across 210–2500 nm obtained from generalized ellipsometry for the IP-textured and epitaxial (010)-Sb₂Se₃ films are shown in Fig. 4. We present the in-plane n and k optical constants of the films approximately along the distinct GaAs [110] and [10] directions. Additionally, for epitaxial Sb₂Se₃, the out-of-plane index is presented to look for giant birefringence.

Fig. 4a shows the in-plane indices for the IP-textured Sb₂Se₃ film. The first set of constants along GaAs [110] (n_c, k_c) corresponds to polarization along the Sb₂Se₃c-axis, and the second set of constants along GaAs [10] (n_a*b, k_a*b) corresponds to polarization along the Sb₂Se₃a- and b-directions. (The a*b subscript denotes weighted average contributions from the a- and b-axes because of the rotated Sb₂Se₃ grains about their c-axis, as depicted previously in Fig. 2c).

Fig. 4b shows the full indices for epitaxial (010)-Sb₂Se₃. Here, the first set of constants are also n_c, k_c. However, the second set of constants corresponds to polarization primarily along the a-axis (n_a, k_a). The final third set of constants corresponds to polarization in the out-of-plane direction or primarily along the b-axis (n_b, k_b). We caution the reader against ascribing certainty to n_b and k_b in the absorbing region approaching the visible and UV wavelengths. We note that reflection-mode ellipsometry loses sensitivity to out-of-plane orientation when the refracted p-polarization is strongly absorbed in the film, therefore leaving s-polarization, which probes only the in-plane properties.⁵¹ Also, the out-of-plane absorption towards the near-infrared range is assumed to be zero (k_b = 0) in our model. We make this assumption based on other experimental reports suggesting Sb₂Se₃ absorbs minimally below its bandgap.^6,19 In the transparent region, the out-of-plane refractive index fit can be reported with greater confidence.

We compare the refractive indices between IP-textured and epitaxial films (from Fig. 4a and b, respectively) more closely in Fig. 4c. Looking at the transparent C-band telecom wavelength (λ = 1550 nm), IP-textured (hk0)-Sb₂Se₃ exhibits in-plane indices of n_c = 4.57 and n_a*b = 4.28, with a remarkable in-plane birefringence of 0.29. We have not measured the out-of-plane index. Epitaxial (010)-Sb₂Se₃ exhibits n_c = 4.59, n_a = 4.41, and n_b < 3.5—a giant birefringence exceeding 1 in the b–c plane that is among the largest values reported for an epitaxially integrated thin film in the near-infrared. This hierarchy of optical properties arises from its underlying bonding—the needle direction exhibits the highest refractive index (Pnma: n_b, Pbnm: n_c), ribbon sheet direction second (Pnma: n_c, Pbnm: n_a), and primary vdW direction lowest (Pnma: n_a, Pbnm: n_b), in agreement with studies on single crystal Sb₂S₃ (reported in Schubert et al. under the Pnma convention).²⁰ This hierarchy in the orthorhombic Sb₂X₃ crystals reflects that the refractive indices at sub-bandgap wavelengths tend to be highest along the covalently-bonded direction (here, the c-axis) relative to more weakly-bonded directions. The lowest index is indeed along the b-axis (Pbnm convention) which bears the greatest van der Waals character.

In Fig. 4c, we note the n_c indices are comparable across these samples as expected, suggesting refractive properties are similar in both IP-textured and epitaxial structures for polarization maintained parallel to the needle direction. This considerably expands the temperature growth window for Sb₂Se₃ on GaAs for photonic applications from epitaxial films to also include the IP-textured films. We also observe that n_a*b of the IP-textured film is lower than n_a of the epitaxial film. This is due to the index along the primary vdW direction (n_b) being lowest in the hierarchy and decreasing the weighted n_a*b index, yielding enhanced IP anisotropy in these samples compared to the epitaxial films. The in-plane and out-of-plane birefringence at the O-band (λ = 1310 nm) is similar in value to that at 1550 nm.

Furthermore, near-infrared attenuation is found to be low in these films (Fig. 4a and b), as the extinction coefficients k are on the order of 10⁻² and below when the polarization lies within the plane of the film. The confluence of birefringence and high transmission in Sb₂Se₃ in the near-infrared may be beneficial in devices that require polarization tuning or phase control. Towards the visible range, the in-plane extinction coefficients k_cvs. k_a*b/k_a are no longer so similar and become of notable magnitude, suggesting dichroism could be an additional functionality for shorter wavelengths, especially around 600–1000 nm.

3. Conclusion

We have developed thin films of orthorhombic Sb₂Se₃ on cubic GaAs(001) using molecular beam epitaxy. Structurally, these films exhibit preferred alignment of the Sb₂Se₃ 1D axis to GaAs [110] at low growth temperatures of 180–265 °C. We demonstrate that these films are low-loss and are optically biaxial. They preserve out-of-plane and in-plane optical anisotropy with giant birefringence exceeding 1 at telecom bands, and uniquely also an in-plane anisotropy along orthogonal directions approaching 0.3. Our work marks a development towards fully anisotropic Sb₂Se₃ films synthesized on technologically relevant cubic templates, a significant departure from previous reports achieving three or six rotational domains that yield in-plane isotropic behavior. We anticipate two important avenues for further study in electronics and photonics. Improvement in the mosaicity and in-plane twist in the epitaxial films will lead to application in a range of minority carrier devices, while transfer of the growth principles derived from our work on GaAs to CMOS-compatible substrates such as Si and Ge would further facilitate integration of Sb₂Se₃ and bring exciting orientation-dependent functionalities to existing photonic schemes.

4. Experimental

4.1. Sb₂Se₃ thin film synthesis

A Riber Compact 21 (C21) chalcogenide MBE equipped with elemental valved Sb and Se cracker sources was used to grow Sb₂Se₃ thin films of thicknesses 72–228 nm on single crystal GaAs(001) substrates. Growth temperatures were read from an optical pyrometer. The beam equivalent pressures (BEPs) of the Sb and Se sources were maintained at 5 × 10⁻⁸ Torr and 1 × 10⁻⁶ Torr, respectively, for a growth rate of ∼0.4 Å s⁻¹. Film growth rates were determined using thicknesses values obtained from X-ray reflectivity measurements (XRR).

To prepare pristine regrown (homoepitaxial) GaAs(001) templates prior to Sb₂Se₃ growth, commercial epi-ready GaAs wafers were first loaded into a separate Veeco Gen III MBE. The surface oxide was removed by thermal desorption at 600 °C, followed by growth of ∼100 nm homoepitaxial GaAs under an arsenic overpressure. The homoepitaxial GaAs templates were then capped in situ with an amorphous arsenic layer. These capped homoepitaxial substrates were taken out-of-vacuum, cleaved into 1 × 1 cm² sections, and indium-bonded to molybdenum platens. For cold growths below 150 °C, cleaved quarter-wafer pieces were secured to a separate molybdenum sample holder designed for indium-free mounting. (An indium-free mount avoids heating effects that would have otherwise been introduced in the indium-debonding process following growth).

The substrates were subsequently introduced into the Riber C21 chamber, where the amorphous arsenic cap was thermally desorbed and the regrown GaAs template was held to stabilize a (2 × 4) reconstruction. Residual chalcogenide fluxes from beam pressure calibrations or preceding growths within a single day commonly resulted in an immediate conversion to a (2 × 1) surface, as indicated by RHEED. After stabilization of the GaAs surface reconstruction, Sb and Se source shutters were opened to initiate film growth at Sb₂Se₃ growth temperatures.

A Se-treated GaAs surface preparation in the C21 MBE is also presented for comparison, in which the substrate native oxide is thermally desorbed under a Se overpressure of 2 × 10⁻⁷ Torr prior to initiating growth. The Se flux is supplied due to the absence of an As source in the C21 MBE.

4.2. Structural characterization

Film structure was characterized with scanning electron microscopy (SEM), atomic force microscopy (AFM), X-ray diffraction (XRD), and transmission electron microscopy (TEM). Standard plan-view and 45°-mounted scanning electron micrographs of films were acquired using a ThermoFisher Apreo SEM (5 kV, 50 pA) and JEOL JSM-IT500HR SEM (10 kV). In the TF Apreo instrument, the electron signal was collected using in-lens detectors. Surface topographical features were analyzed with a Park NX-10 AFM and a NCS15 probe in tapping mode. XRD scans were acquired on a PANalytical X’Pert Pro MRD instrument with mirror and parallel-plate collimator optics and Ni-filtered Cu-Kα radiation. Cross-sectional lamella of Sb₂Se₃ films were milled in a FEI Helios NanoLab 600i DualBeam. High resolution TEM (HR-TEM) imaging was conducted using a FEI Titan E-TEM (300 kV) and images recorded on a Gatan OneView camera.

4.3. Optical characterization

Variable angle reflection-mode Mueller matrix ellipsometry measurements between wavelengths of 210–2500 nm were acquired on a J. A. Woollam RC2. To ascertain the anisotropy of the crystalline thin films, three datasets were recorded at multiple azimuthal orientations: the samples were rotated in-plane 0°, 45°, and 90° counterclockwise relative to the Sb₂Se₃ 1D axis. The three datasets collected at multiple incidence angles between 55°–75° were used to fit the complex anisotropic refractive index within J. A. Woollam's CompleteEASE analysis software. For crystalline-Sb₂Se₃, multiple Gaussian oscillators were employed for fitting of in-plane optical constants. For epitaxial films, films of separate thicknesses were used to help fit the out-of-plane index under a separate Tauc–Lorentz model.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors thank N. Hong from J. A. Woollam for helpful discussions on ellipsometry measurements and fitting. We gratefully acknowledge support for materials synthesis and structural characterization via the NSF CAREER Award under grant no. DMR-2036520. Part of this work was performed at the Stanford Nanofabrication Facilities (SNF) and Stanford Nano Shared Facilities (SNSF), supported by the National Science Foundation under award ECCS-2026822. A. M., V. T., and R. C. are supported by the DARPA-ATOM program. J. E. M. gratefully acknowledges support from the Tomkat Center for Sustainable Energy's Tomkat Center Graduate Fellow for Translational Research Fellowship.

References

1. G. P. Voutsas, A. G. Papazoglou, P. J. Rentzeperis and D. Siapkas, Z. Kristallogr. - Cryst. Mater., 1985, 171, 261–268 CrossRef CAS .
2. N. W. Tideswell, F. H. Kruse and J. D. McCullough, Acta Crystallogr., 1957, 10, 99–102 CrossRef CAS .
3. X. Wang, Z. Li, S. R. Kavanagh, A. M. Ganose and A. Walsh, Phys. Chem. Chem. Phys., 2022, 24, 7195–7202 RSC .
4. C. Chen, D. C. Bobela, Y. Yang, S. Lu, K. Zeng, C. Ge, B. Yang, L. Gao, Y. Zhao, M. C. Beard and J. Tang, Front. Optoelectron., 2017, 10, 18–30 CrossRef .
5. L. R. Gilbert, B. Van Pelt and C. Wood, J. Phys. Chem. Solids, 1974, 35, 1629–1632 CrossRef CAS .
6. C. Wood, Z. Hurych and J. C. Shaffer, J. Non-Cryst. Solids, 1972, 8–10, 209–214 CrossRef CAS .
7. W. M. Yim, E. V. Fitzke and F. D. Rosi, J. Mater. Sci., 1966, 1, 52–65 CrossRef CAS .
8. H. L. Uphoff and J. H. Healy, J. Appl. Phys., 1963, 34, 390–395 CrossRef CAS .
9. X. Wen, C. Chen, S. Lu, K. Li, R. Kondrotas, Y. Zhao, W. Chen, L. Gao, C. Wang, J. Zhang, G. Niu and J. Tang, Nat. Commun., 2018, 9, 2179 CrossRef PubMed .
10. C. Chen, L. Wang, L. Gao, D. Nam, D. Li, K. Li, Y. Zhao, C. Ge, H. Cheong, H. Liu, H. Song and J. Tang, ACS Energy Lett., 2017, 2, 2125–2132 CrossRef CAS .
11. Y. C. Choi, T. N. Mandal, W. S. Yang, Y. H. Lee, S. H. Im, J. H. Noh and S. I. Seok, Angew. Chem., Int. Ed., 2014, 53, 1329–1333 CrossRef CAS PubMed .
12. M. Leng, M. Luo, C. Chen, S. Qin, J. Chen, J. Zhong and J. Tang, Appl. Phys. Lett., 2014, 105, 083905 CrossRef .
13. Z. Li, X. Liang, G. Li, H. Liu, H. Zhang, J. Guo, J. Chen, K. Shen, X. San, W. Yu, R. E. I. Schropp and Y. Mai, Nat. Commun., 2019, 10, 125 CrossRef PubMed .
14. G.-X. Liang, Y.-D. Luo, S. Chen, R. Tang, Z.-H. Zheng, X.-J. Li, X.-S. Liu, Y.-K. Liu, Y.-F. Li, X.-Y. Chen, Z.-H. Su, X.-H. Zhang, H.-L. Ma and P. Fan, Nano Energy, 2020, 73, 104806 CrossRef CAS .
15. I. Mosquera-Lois, S. R. Kavanagh, A. Walsh and D. O. Scanlon, npj Comput. Mater., 2023, 9, 25 CrossRef CAS .
16. K. P. McKenna, Adv. Electron. Mater., 2021, 7, 2000908 CrossRef CAS .
17. M. Delaney, I. Zeimpekis, D. Lawson, D. W. Hewak and O. L. Muskens, Adv. Funct. Mater., 2020, 30, 2002447 CrossRef CAS .
18. D. Lawson, D. W. Hewak, O. L. Muskens and I. Zeimpekis, J. Opt., 2022, 24, 064013 CrossRef CAS .
19. K. Aryana, H. J. Kim, Md. R. Islam, N. Hong, C.-C. Popescu, S. Makarem, T. Gu, J. Hu and P. E. Hopkins, Opt. Mater. Express, 2023, 13, 3277 CrossRef CAS .
20. M. Schubert, T. Hofmann, C. M. Herzinger and W. Dollase, Thin Solid Films, 2004, 455–456, 619–623 CrossRef CAS .
21. S. Niu, G. Joe, H. Zhao, Y. Zhou, T. Orvis, H. Huyan, J. Salman, K. Mahalingam, B. Urwin, J. Wu, Y. Liu, T. E. Tiwald, S. B. Cronin, B. M. Howe, M. Mecklenburg, R. Haiges, D. J. Singh, H. Wang, M. A. Kats and J. Ravichandran, Nat. Photonics, 2018, 12, 392–396 CrossRef CAS .
22. F. Zhang, J. Zheng, Y. Song, W. Liu, P. Xu and A. Majumdar, OSA Contin., 2020, 3, 560 CrossRef .
23. H. Jin, L. Niu, J. Zheng, P. Xu and A. Majumdar, Opt. Express, 2023, 31, 10684 CrossRef CAS PubMed .
24. A. Koma, Thin Solid Films, 1992, 216, 72–76 CrossRef CAS .
25. X. Liu, D. J. Smith, H. Cao, Y. P. Chen, J. Fan, Y.-H. Zhang, R. E. Pimpinella, M. Dobrowolska and J. K. Furdyna, J. Vac. Sci. Technol., B: Nanotechnol. Microelectron.: Mater., Process., Meas., Phenom., 2012, 30, 02B103 Search PubMed .
26. Z. Zeng, T. A. Morgan, D. Fan, C. Li, Y. Hirono, X. Hu, Y. Zhao, J. S. Lee, J. Wang, Z. M. Wang, S. Yu, M. E. Hawkridge, M. Benamara and G. J. Salamo, AIP Adv., 2013, 3, 072112 CrossRef .
27. X. Liu, Y. P. Chen, D. J. Smith, Y.-H. Zhang, C. Liu, M. Z. Hasan, M. Dobrowolska, J. K. Furdyna, J. Fan, H. Cao, T.-L. Wu and R. E. Pimpinella, in Bismuth-Containing Compounds, ed. H. Li and Z. M. Wang, Springer, New York, NY, 2013, vol. 186, pp. 263–279 Search PubMed .
28. Z. Wang and S. Law, Cryst. Growth Des., 2021, 21, 6752–6765 CrossRef CAS .
29. J. Momand, J. E. Boschker, R. Wang, R. Calarco and B. J. Kooi, CrystEngComm, 2018, 20, 340–347 RSC .
30. X. Wen, Z. Lu, L. Valdman, G.-C. Wang, M. Washington and T.-M. Lu, ACS Appl. Mater. Interfaces, 2020, 12, 35222–35231 CrossRef CAS PubMed .
31. A. V. Matetskiy, V. V. Mararov, I. A. Kibirev, A. V. Zotov and A. A. Saranin, J. Condens. Matter Phys., 2020, 32, 165001 CrossRef CAS PubMed .
32. Y.-J. Chen, Y.-C. Jhong, P.-Y. Chuang, C.-W. Chong, J.-C.-A. Huang, V. Marchenkov and H.-C. Han, Chin. J. Phys., 2019, 62, 65–71 CrossRef CAS .
33. P. Wojnar, S. Chusnutdinow, A. Kaleta, M. Aleszkiewicz, S. Kret, J. Z. Domagala, P. Ciepielewski, R. Yatskiv, S. Tiagulskyi, J. Suffczyński, A. Suchocki and T. Wojtowicz, Nanoscale, 2024, 16, 19477–19484 RSC .
34. S.-W. Jung, S.-M. Yoon, I.-K. You, B.-G. Yu and K.-W. Koo, Nanosci. Nanotechnol., 2011, 11, 1569–1572 CAS .
35. X. Wen, Z. Lu, B. Li, G.-C. Wang, M. A. Washington, Q. Zhao and T.-M. Lu, J. Chem. Eng., 2023, 462, 142026 CrossRef CAS .
36. J. Otavio Mendes, A. Merenda, K. Wilson, A. Fraser Lee, E. Della Gaspera and J. Van Embden, Small, 2023, 2302721 Search PubMed .
37. R. Kondrotas, J. Zhang, C. Wang and J. Tang, Sol. Energy Mater. Sol. Cells, 2019, 199, 16–23 CrossRef CAS .
38. T. D. C. Hobson and K. Durose, Mater. Sci. Semicond. Process., 2021, 127, 105691 CrossRef CAS .
39. H. Abe, K. Ueno, K. S. Koichiro Saiki and A. K. Atsushi Koma, Jpn. J. Appl. Phys., 1993, 32, L1444 CrossRef CAS .
40. H. Cheng, J. M. DePuydt, M. A. Haase and J. E. Potts, J. Vac. Sci. Technol., B: Nanotechnol. Microelectron.: Mater., Process., Meas., Phenom., 2018, 8, 181–186 Search PubMed .
41. A. Guillén-Cervantes, Z. Rivera-Alvarez, M. López-López, E. López-Luna and I. Hernández-Calderón, Thin Solid Films, 2000, 373, 159–163 CrossRef .
42. N. Fleck, T. D. C. Hobson, C. N. Savory, J. Buckeridge, T. D. Veal, M. R. Correia, D. O. Scanlon, K. Durose and F. Jäckel, J. Mater. Chem. A, 2020, 8, 8337–8344 RSC .
43. T. D. C. Hobson, O. S. Hutter, M. Birkett, T. D. Veal and K. Durose, in 2018 IEEE 7th World Conference on Photovoltaic Energy Conversion (WCPEC) (A Joint Conference of 45th IEEE PVSC, 28th PVSEC & 34th EU PVSEC), IEEE, Waikoloa Village, HI, 2018, pp. 0818–0822.
44. L. Däweritz, Handbook of Surfaces and Interfaces of Materials, 2001, pp. 351–386 Search PubMed .
45. A. Ohtake, Surf. Sci. Rep., 2008, 63, 295–327 CrossRef CAS .
46. R. Shayduk and W. Braun, J. Cryst. Growth, 2009, 311, 2215–2219 CrossRef CAS .
47. Y. Watanabe, T. Scimeca, F. M. Fumihiko Maeda and M. O. Masaharu Oshima, Jpn. J. Appl. Phys., 1994, 33, 698 CrossRef CAS .
48. R. J. Hamers, R. M. Tromp and J. E. Demuth, Phys. Rev. B: Condens. Matter Mater. Phys., 1986, 34, 5343–5357 CrossRef CAS PubMed .
49. G. A. Ermolaev, D. V. Grudinin, Y. V. Stebunov, K. V. Voronin, V. G. Kravets, J. Duan, A. B. Mazitov, G. I. Tselikov, A. Bylinkin, D. I. Yakubovsky, S. M. Novikov, D. G. Baranov, A. Y. Nikitin, I. A. Kruglov, T. Shegai, P. Alonso-González, A. N. Grigorenko, A. V. Arsenin, K. S. Novoselov and V. S. Volkov, Nat. Commun., 2021, 12, 854 CrossRef CAS PubMed .
50. A. Segura, L. Artús, R. Cuscó, T. Taniguchi, G. Cassabois and B. Gil, Phys. Rev. Mater., 2018, 2, 024001 CrossRef CAS .
51. N. Hong, R. A. Synowicki and J. N. Hilfiker, Appl. Surf. Sci., 2017, 421, 518–528 CrossRef CAS .

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5mh00225g

This journal is © The Royal Society of Chemistry 2025