Monolithic integration of optically anisotropic GeSe-based films on GaAs by templated solid-phase epitaxy

Aaron M. Lindenberg; Ashlee M. Garc\'ia; Autumn Y. Lee; Kelly Xiao; Kira J. Martin; Kunal Mukherjee; Pooja D. Reddy; Pranav Mahaadev; Tri Nguyen

REVIEW 2 major objections 5 minor 2 references

Monolithic integration of optically anisotropic GeSe-based films on GaAs by templated solid-phase epitaxy

T0 review · 2 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read On offcut GaAs with a SnSe buffer, solid-phase crystallized GeSe films grow in a single in-plane orientation and show twice the polarized reflection contrast of mixed-orientation films.

desk verdict A solid, incremental integration paper: single-orientation GeSe films on GaAs via SnSe-buffer templated SPE, with a plausible 2× anisotropy enhancement that needs a control to separate GeSe from the SnSe buffer contribution. read the letter →

arxiv 2608.04418 v1 pith:E3SSSKQR submitted 2026-08-05 cond-mat.mtrl-sci

Kira J. Martin , Autumn Y. Lee , Pranav Mahaadev , Pooja D. Reddy , Kelly Xiao , Tri Nguyen , Ashlee M. García , Aaron M. Lindenberg

show 1 more author

Kunal Mukherjee

This is my paper · ORCID

classification cond-mat.mtrl-sci PACS 81.15.Hi

keywords GeSesolid-phaseepitaxyin-planeopticalanisotropySnbufferoffcutGaAssubstratealloyspolarization-sensitivereflectionmolecularbeam

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that GeSe, a layered semiconductor whose low-symmetry crystal gives it strong in-plane optical anisotropy, can be integrated on a GaAs substrate as a single-orientation crystalline film, and that this orientation control directly improves the optical response. Because GeSe forms a glass at low temperatures and re-evaporates at high temperatures, it cannot be grown epitaxially the usual way. The authors deposit amorphous GeSe and crystallize it against a crystalline SnSe buffer, which templates both the out-of-plane stacking and the in-plane orientation (the armchair and zigzag axes of the puckered structure); on a substrate miscut $4^\circ$ toward $\langle 111\rangle$B, the buffer locks in one in-plane orientation. They measure a 2× larger polarization-dependent reflectivity ratio for the single-orientation film (0.08) than for a film with two competing in-plane variants (0.04) at 800 nm. If correct, this gives a scalable, monolithic route to polarization-sensitive photonic devices from an air-stable material, without the exfoliation and transfer steps used for black phosphorus.

What carries the argument

The mechanism is ex-situ solid-phase epitaxy of amorphous GeSe against a crystalline SnSe buffer, combined with step-edge orientation control on an offcut substrate. GeSe deposited at 160°C is smooth, amorphous, and optically isotropic; a 5-minute anneal at 400°C under a silicon proximity cap crystallizes it. The SnSe buffer, grown at 300°C where GeSe will not stick, shares GeSe's orthorhombic Pnma structure, so it templates both the out-of-plane van der Waals axis and the in-plane armchair and zigzag directions. On GaAs (001) miscut $4^\circ$ toward $\langle 111\rangle$B, SnSe nucleates with its zigzag edge aligned to the step edges, giving a single in-plane variant, and the overlying GeSe keeps that orientation during crystallization. The quantitative observable carrying the argument is the polarization-dependent reflectivity ratio $(R_{\max}-R_{\min})/R_{\max}$, measured as the incident polarization is rotated through 180° at 800 nm and 1030 nm.

What would settle it

Measure the same 800 nm polarization-dependent reflection on a bare single-variant SnSe buffer on offcut GaAs and a bare 80:20 double-variant SnSe buffer on on-axis GaAs; if their anisotropy ratios already differ by about a factor of 2, the claim that the GeSe layer itself exhibits the 2× enhancement is not established. A complementary check is to measure the films at a wavelength where SnSe's anisotropic response is negligible, or to remove the GeSe layer and re-measure the remaining stack.

Watch

Extended reading notes

Core claim

The central claim is that templated solid-phase epitaxy can stabilize a single in-plane crystallographic orientation of GeSe on GaAs(001), and that removing the degenerate second orientation roughly doubles the measured in-plane optical anisotropy. On exact (001) GaAs, both the SnSe buffer and the GeSe crystallized on top of it form two orthogonal in-plane variants, with the armchair and zigzag directions interchanged, in an 80:20 volume ratio; the minority grains average down the anisotropic response. On a $4^\circ$ offcut substrate, the SnSe buffer grows as a single variant and the GeSe inherits that orientation, confirmed by reciprocal space maps showing only (820) peaks at $\varphi = 0^\circ$ and only (802) peaks at $\varphi = 90^\circ$. Polarized 800 nm reflection gives an anisotropy ratio (maximum minus minimum reflected power over maximum) of 0.08 for the single-orientation film versus 0.04 for the double-variant film, a 2× enhancement close to the 1.7× expected from removing the 20% minority grains. The paper states explicitly that these measurements probe the multilayer GeSe/SnSe/GaAs stack, and it reports a smaller 1.4× gain at 1030 nm, so the size of the effect is wavelength-dependent.

Load-bearing premise

The paper assumes the 2× enhancement comes from the GeSe layer's orientation, even though the polarization-dependent reflection measurement probes the whole GeSe/SnSe/GaAs stack, and the single-orientation sample also has a single-orientation SnSe buffer while the double-variant sample has an 80:20 buffer, so a substantial buffer contribution to the measured ratio is not excluded by the data presented.

Editorial extensions

If this is right

Single-orientation GeSe on GaAs delivers the material's full in-plane anisotropy, giving a polarization-dependent reflectivity contrast roughly twice that of double-variant films at 800 nm and 1.4× at 1030 nm.
The SnSe-buffer-plus-SPE route spans the full SnSe–GeSe composition range, so alloy composition becomes a handle for tuning lattice constants, bandgap, and the spectral position of the anisotropic response.
Offcut substrates do not by themselves guarantee single-orientation films at low growth temperatures: SnGeSe alloys grown at 160°C form two in-plane orientations even on offcut GaAs, so step-edge selection must be recovered at low temperature for alloys.
Because structural in-plane anisotropy increases with Ge content while SnSe and GeSe have opposite absorption trends along the two axes, some intermediate alloy composition is expected to show weak or vanishing optical anisotropy despite strong structural anisotropy.
Devices built from single-orientation films, such as polarization-sensitive photodetectors, polarizers, and on-chip polarimetry components, would gain the full contrast of the material without exfoliation or transfer.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A decisive control experiment the paper leaves implicit is measuring bare SnSe buffers (single-variant on offcut GaAs and 80:20 double-variant on on-axis GaAs) under the same reflection setup; if their anisotropy ratios already differ by about 2×, the enhancement cannot be assigned to the GeSe layer from the stack measurement alone.
The same approach should transfer to other glass-forming layered chalcogenides, such as GeS or Sb2Se3, whenever a lattice-matched crystalline template of the same orthorhombic structure can be grown first, since the templating mechanism is structural rather than chemical.
The 1.4× versus 2× difference between 1030 nm and 800 nm suggests the orientation gain itself is wavelength-dependent; measuring the full visible-to-NIR spectrum of single- and double-variant samples would map where orientation control matters most for devices.
Within the paper's linear model, the contrast gain grows as the minority variant fraction shrinks, so tuning nucleation density, cap quality, or anneal profile to push the observed 80:20 ratio toward single-variant is a direct path to test the predicted scaling.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is a competent, incremental integration paper. The genuinely new result is that GeSe—a glass former that resists direct epitaxy—can be crystallized in single in-plane orientation on GaAs via an SnSe buffer and offcut substrate, and that the oriented film shows roughly twice the polarization-dependent reflectivity contrast of the double-variant film. The structural evidence is solid: RHEED/XRR establish the growth window, RSMs show single-variant GeSe on offcut, and the 80:20 variant ratio on on-axis is quantified. The alloy sweep is exploratory but honest.

What it does well: clear writing, careful characterization, and the authors explicitly acknowledge the measurement probes a multilayer stack. The 1.7× expected vs 2× observed consistency check at 800 nm and 1.4× at 1030 nm are reasonable back-of-envelope numbers, not fitted targets.

Soft spots, in proportion: the headline 2× enhancement is not cleanly attributable to the GeSe layer. The comparison is between two samples that differ in both the GeSe and the SnSe buffer orientation (80:20 vs single-variant). Since SnSe is also anisotropic and the POM of the as-grown stack shows the buffer alone gives optical contrast, the reflection ratio could be partly or largely from the buffer. The authors' own note about probing the multilayer does not resolve this. A control (buffer without GeSe, or a multilayer model) is needed before the 2× can be pinned on GeSe. Also, there are no error bars or repeated measurements on the central ratios; the numbers are single-spot comparisons.

The 1030 nm data, where the measured 1.4× falls short of the 1.7× expectation, is honestly reported and attributed to misalignment, but that same discrepancy is consistent with the buffer-confound worry. The alloy offcut results are presented as a limitation rather than overclaimed, which I credit.

Citations: mostly the group's own prior work—that is normal for a series and not a problem here. No missing competitor jumps out. Methods are described in enough detail to reproduce.

Who is this for: people working on anisotropic IV-VI integration on III-Vs, polarization-sensitive photonics, or SPE templating. It is a useful data point, not a paradigm shift.

Verdict: deserves a serious referee. I would send it to review with the expectation that the authors add a control or model to separate the GeSe and buffer contributions, and report uncertainties. The integration result and structural data are worth publishing regardless.

Referee Report

2 major / 5 minor

Summary. The paper reports a method for heteroepitaxial integration of GeSe and SnGeSe films on GaAs substrates by molecular beam epitaxy followed by ex-situ solid-phase epitaxy, using a crystalline SnSe buffer layer to template crystallization. The authors show that on-axis GaAs leads to a double-variant in-plane orientation (about 80:20 volume fraction), while a 4-degree offcut substrate yields a single in-plane orientation. Using polarization-dependent reflectivity at 800 nm, they report an anisotropy ratio of 0.04 for the double-variant film and 0.08 for the single-variant film, i.e., a 2x enhancement; at 1030 nm the enhancement is 1.4x (0.04 to 0.06). The paper also extends the growth method to SnGeSe alloys and characterizes their structural and optical anisotropy, noting that low-temperature alloy growth does not yet preserve single-variant orientation.

Significance. The templated solid-phase epitaxy route addresses a real integration challenge: GeSe is a promising anisotropic semiconductor but is difficult to grow directly because of glass formation and high vapor pressure. The structural characterization is thorough—reciprocal space maps, X-ray diffraction, AFM, and polarized optical microscopy—and the transfer of the offcut-substrate templating concept from prior SnSe work to GeSe is a valuable extension. The central optical claim, that single-orientation GeSe provides a 2x increase in anisotropic reflectivity, is plausible and important for polarization-sensitive device applications, but it currently rests on a comparison in which the SnSe buffer orientation also changes. If the buffer contribution is controlled, the result would be a significant demonstration. The paper contains no machine-checked proofs or code, but the experimental method is clearly described and the supporting information provides useful detail.

major comments (2)

[Section II.b, Figure 4] The central claim that single-IP-orientation GeSe gives a 2× increase in anisotropic reflectivity rests on comparing two samples that differ in more than the GeSe layer. The on-axis sample has an 80:20 double-variant SnSe buffer while the offcut sample has a single-variant SnSe buffer, and the authors explicitly note that the polarization-dependent reflection probes the multilayer GeSe/SnSe/GaAs stack. Figure S2 shows that the SnSe buffer alone produces crossed-polarizer optical contrast in the as-grown sample, and no evidence is given that the 100 nm GeSe layer dominates the reflection at 800 nm. Before the 2× enhancement can be attributed to the GeSe film, the buffer's own orientation-dependent reflectivity must be quantified (e.g., by measuring buffer-only samples on on-axis and offcut substrates) or a multilayer optical model must be used to deconvolve the contributions.
[Section II.b and Supporting Information S3] The agreement with the expected 1.7× increase is presented as support for the attribution, but the expectation is computed from the same multilayer measurements: the single-variant reflected intensities I_ac and I_zz are taken from the offcut sample, which includes the single-variant SnSe buffer, and the double-variant combination assumes an 80:20 mixture of those same intensities. This consistency check therefore inherits the confounding buffer contribution and cannot by itself establish that GeSe is responsible for the enhancement. In addition, the mixing model treats the reflected power as a linear weighted sum of the two variant intensities; for a multilayer with polarization-dependent reflections at each interface, an intensity-only superposition may not be valid. Please justify the linear mixing approximation or provide a control that isolates the GeSe contribution.

minor comments (5)

[Section II.b, Figure 4] The reported ratios (0.04, 0.08) appear to be from single measurements with no stated uncertainty or spot-to-spot variation; please add error estimates or state how many locations were measured and whether the values are representative across the sample.
[Abstract and Section II.b] The '2× increase' is only measured at 800 nm; at 1030 nm the increase is 1.4×. Please specify the wavelength in the abstract to avoid overgeneralization.
[Section II.b] There is a grammatical error: 'two distinct family of grains' should be 'two distinct families of grains.'
[Section II.c and Figure S6] The statement in the main text that 'Post-growth annealing had no effect on the IP orientation of the alloy' is supported by the supplementary text but no annealing data are shown; please either include the relevant data or reference a figure that displays them.
[Supporting Information S1] The 80:20 volume fraction estimate is based on integrating RSM peak intensities and correcting for structure factors. Please state the estimated uncertainty in this ratio and whether the integration was repeated on multiple azimuths or samples to confirm the reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the 2× anisotropy claim is a direct experimental comparison, and the S3 1.7× estimate is an independent consistency check, not a fitted target.

full rationale

The paper's claimed derivation chain is experimental rather than mathematical. The central claim—that stabilizing a single in-plane orientation increases the polarization-dependent reflectivity ratio from 0.04 to 0.08—is a direct measurement on two samples (Section II.b, Figure 4), not an output of a model that already encodes the 2× result. The S3 expectation of 1.7× is a consistency check: it is computed from the independently measured 80:20 volume fractions (XRD) and from the armchair/zigzag intensities of the single-orientation sample, then compared to the separately measured double-variant ratio of 0.04. Thus the expectation uses measured inputs but does not fit or presuppose the 2× claim. The offcut templating is supported within this paper by reciprocal space maps (Figure 3c) showing only one in-plane GeSe orientation at each azimuth, so the prior SnSe offcut result (ref 7) is background/method rather than a load-bearing imported uniqueness theorem. The only notable caveat is the authors' own statement that 'these measurements probe a multilayer structure of GeSe, SnSe, and GaAs'; because the SnSe buffer also changes from 80:20 to single-variant between the compared samples, the 2× enhancement cannot be unambiguously isolated to the GeSe layer by this measurement alone. That is a potential attribution confound or correctness risk, but it is not circularity: no equation, fitted parameter, or cited theorem in the paper reduces to the quantity it is used to predict. No self-definitional steps, fitted inputs renamed as predictions, or renamed known results were identified.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

No invented entities. The method relies on material-specific assumptions (glass formation, templating, step-edge control) that are standard for this materials system and supported by prior work from the same group. The only fitted numerical input is the assumed constant strain state used to convert XRD peak shifts into alloy compositions.

free parameters (1)

Assumed constant strain for SnGeSe alloys = 0.3% compressive (from pure SnSe)
Used to convert the (800) XRD peak position into alloy composition. This value is measured on pure SnSe and assumed unchanged across the alloy series; no per-sample strain measurement is provided.

assumptions (5)

domain assumption GeSe is a glass former at low temperature and has high vapor pressure at high temperature, preventing direct crystalline epitaxy on GaAs.
Invoked in the Introduction and Section II.a to motivate the solid-phase epitaxy approach; supported by cited references on GeSe glass structure (ref 20) and sublimation kinetics (ref 34).
domain assumption SnSe, being isostructural with GeSe and capable of direct epitaxy on GaAs, templates the crystallization of amorphous GeSe.
The templating mechanism is central to the method (Section II.b). It relies on the structural similarity of SnSe and GeSe and the authors' prior demonstration of SnSe heteroepitaxy on GaAs (refs 7, 38).
domain assumption A 4-degree offcut GaAs(001) substrate in the <111>B direction stabilizes a single in-plane orientation by aligning the zigzag edge to the step edge.
This mechanism was established for SnSe in the authors' prior work (ref 7) and is assumed to hold for solid-phase crystallized GeSe without direct verification of the nucleation mechanism in this paper.
domain assumption The ratio of maximum to minimum reflected power in normal-incidence polarization-dependent reflectivity is a valid metric for the in-plane optical anisotropy of the film.
The measurements probe the GeSe/SnSe/GaAs stack, not the GeSe layer alone, as acknowledged in Section II.b. The metric is used directly for the 2x enhancement claim.
domain assumption SnGeSe alloy compositions are derived from XRD peak positions assuming a strain state equal to that of pure SnSe (0.3% compressive).
Described in Section II.c and Supplementary S5. If the strain state varies with composition, the reported compositions would be systematically shifted.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Monolithic integration of optically anisotropic GeSe-based films on GaAs by templated solid-phase epitaxy." pith.science (2026). https://pith.science/paper/E3SSSKQR

@misc{pith2026260804418,
  author       = {Pith},
  title        = {Pith review of: Monolithic integration of optically anisotropic GeSe-based films on GaAs by templated solid-phase epitaxy},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/E3SSSKQR}},
  note         = {Machine review of arXiv:2608.04418}
}

read the original abstract

Layered IV-VI semiconductors such as GeSe exhibit strong in-plane optical anisotropy, making them promising candidates for polarization-sensitive photonic devices. However, realizing these properties in scalable platforms requires heteroepitaxial integration on technologically relevant substrates like GaAs. Direct growth of GeSe is complicated by its glass formation at low temperatures and high vapor pressure at elevated temperatures. To overcome this, we develop a method for ex-situ solid-phase epitaxy utilizing a SnSe buffer and offcut GaAs substrate to enable single-orientation crystalline GeSe films. Using polarized reflection measurements, we find that stabilizing a single-in-plane-orientation results in a 2x increase in anisotropic response between the armchair and zigzag directions. This work provides a new integration route to harness the anisotropic optical properties of GeSe and its alloys for polarization-sensitive technologies.

Figures

Figures reproduced from arXiv: 2608.04418 by the authors.

**Figure 1.** (a) RHEED pattern of [110] azimuth at the start, 30 seconds, 5 minutes, and 10 minutes into GeSe growth at three different growth temperatures. A schematic illustrates experimental setup. (b) XRR data and best fit of GaAs (001) substrate with native oxide and amorphous GeSe grown at a growth temperature of 200˚C and 160˚C. The curves are vertically offset for clarity. The vertical line marks the critical angle of th… view at source ↗

**Figure 2.** (a) Schematic of annealed GeSe on GaAs (001). (b) Schematic of annealed GeSe with crystalline SnSe buffer on GaAs (001). (c) Open detector symmetric 2θ-ω scans of both films post annealing. The scans are vertically offset for clarity. The asterisk is used to denote GaAs substrate peaks. (d) AFM scan of as-grown amorphous GeSe with inset of RHEED pattern from the end of the growth and (e) post annealing. (f) Polarize… view at source ↗

**Figure 3.** (a) Schematic of amorphous GeSe with crystalline SnSe buffer on 4˚offcut GaAs (001). (b) AFM scan of asgrown amorphous GeSe and post annealing. (c) RSMs of GeSe film with SnSe buffer grown on offcut GaAs (001) post annealing at azimuths φ=0˚ and 90˚. Schematics highlight how the GeSe layer maintains the IP orientation of the SnSe buffer after crystallization [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Normalized reflected power vs polarization angle for GeSe with single IP orientation (offcut) and double IP orientation (on-axis) [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

**Figure 5.** Figure 5: (a) RSMs of as grown crystalline SnGeSe alloys with SnSe buffer on GaAs (001). Two peaks for the SnSe buffer and SnGeSe film corresponding to two in-plane orientations. (b) Lattice constant as a function of Ge composition compared to bulk alloy trends. Lattice constant…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

2 extracted references · 2 canonical work pages

[1]

Über Struktur Und Eigenschaften Der Halbmetalle

(1) Krebs, H.; Langner, D. Über Struktur Und Eigenschaften Der Halbmetalle. XVI. Mischkristallsysteme Zwischen Halbleitenden Chalkogeniden Der Vierten Hauptgruppe. II. Zeitschrift anorg allge chemie 1964 , 334 (1–2), 37–49. https://doi.org/10.1002/zaac.1964334

work page doi:10.1002/zaac.1964334 1964
[107]

-Y .; Xiang, P

(2) Zhao, Y .-F.; Guan, Z.; Zhong, N.; Yue, F. -Y .; Xiang, P. -H.; Duan, C. -G. Raman Spectra of Bulk and Few-Layer GeSe From First-Principles Calculations. Front. Mater. 2021 , 8, 736057. https://doi.org/10.3389/fmats.2021.736057. (3) Chandrasekhar, H. R.; Zwick, U. Raman Scattering and Infrared Reflectivity in GeSe. Solid State Communications 1976 , 18...

work page arXiv 2021

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Über Struktur Und Eigenschaften Der Halbmetalle

(1) Krebs, H.; Langner, D. Über Struktur Und Eigenschaften Der Halbmetalle. XVI. Mischkristallsysteme Zwischen Halbleitenden Chalkogeniden Der Vierten Hauptgruppe. II. Zeitschrift anorg allge chemie 1964 , 334 (1–2), 37–49. https://doi.org/10.1002/zaac.1964334

work page doi:10.1002/zaac.1964334 1964

[2] [107]

-Y .; Xiang, P

(2) Zhao, Y .-F.; Guan, Z.; Zhong, N.; Yue, F. -Y .; Xiang, P. -H.; Duan, C. -G. Raman Spectra of Bulk and Few-Layer GeSe From First-Principles Calculations. Front. Mater. 2021 , 8, 736057. https://doi.org/10.3389/fmats.2021.736057. (3) Chandrasekhar, H. R.; Zwick, U. Raman Scattering and Infrared Reflectivity in GeSe. Solid State Communications 1976 , 18...

work page arXiv 2021