betaFeedback

WorksPhysics and Astronomy

Shining light on short-range atomic ordering in semiconductors alloys

Abstract

The functional properties of semiconductors are typically controlled by tailoring their chemical composition and their state of strain, and by controlling their long-range structural order, including the presence of extended defects such as dislocations. In addition to these approaches, theoretical predictions suggest that short-range order (SRO) of atoms in group-IV semiconductor alloys can modify the bandgap, a defining property of any semiconductor. Herein, a new machine learning enabled, computation-guided methodology for extended X-ray absorption fine structure (EXAFS) analysis of SRO is used to quantify the effects of local atomic order on the bandgap of germanium-tin (GeSn) alloy single crystal nanostructures with well-controlled strain and composition. Correlative analysis of EXAFS and photoluminescence (PL) establishes the relationship between bandgap and the Warren-Cowley short-range order (WC-SRO) parameter of the GeSn alloys. It is further demonstrated that SRO can be tuned over a broad range by post-deposition annealing of the alloy crystals. This work establishes control of SRO as an important design parameter for semiconducting properties and suggests the potential for quantitative measurement and tuning of SRO in other semiconductor alloy systems.

For decades, the design of crystalline semiconductor materials has rested on two pillars: controlling which atoms are present (composition) and how the crystal is deformed (strain). Precise control of both composition and strain are now highly reproducible approaches for material design and manufacturing underpinning the global electronics and photonics industries. Theory predicts a third pillar; controlling how atoms are arranged locally for a given composition and strain state. Density functional theory (DFT) [1] and cluster expansion calculations [2] suggest that short-range ordering (SRO)—the preferential arrangement of chemical species within a few atomic nearest-neighbor distances—may be a fundamental and underexplored lever to tune the electronic structure of semiconductors. Similar phenomena examined theoretically in group-IV semiconductor alloys [1, 3], III-V semiconductors [4, 5], and high-entropy alloys [6] underscore the potential of SRO-driven electronic effects. However, controlled local order has remained largely inaccessible experimentally. Realizing this third pillar for practical application hinges on answering two fundamental questions. First, can SRO be manipulated deliberately and selectively to tailor semiconducting properties? Second, what approaches can enable statistically robust quantification of SRO? The absence of definitive answers to these questions has left SRO unexplored for materials design of semiconductors. A primary barrier to controlling SRO has been measurement: advanced microscopies and atom-probe based methods can map local atomic arrangements and chemical correlations [7, 8, 9, 10], but statistical SRO quantification remains unreliable. Similarly, EXAFS analysis has, to date, provided qualitative insights [11, 12, 13, 14, 15], but a robust quantitative framework has been lacking.

The challenges are addressed herein for GeSn alloys, a material system studied for application in integrated photonics [16], mid-infrared sensing [17, 18], and quantum information [19, 20]. A Ge/GeSn core/shell nanowire platform is employed, whereby a compliant Ge core relaxes strain in the shell and conformal ultrathin alumina (Al2_2O3_3) passivation enables thermal annealing high above the Ge-Sn eutectic temperature [21] without altering composition and strain. Thermal processing induces a pronounced, monotonic blueshift in photoluminescence (PL), suggesting a direct link between local atomic arrangement and optoelectronic response. To probe this link, an integrated methodology is developed, combining EXAFS with high-fidelity machine-learned atomistic models within a Bayesian inference framework. This approach quantitatively extracts the WC-SRO parameter, α\alpha, and reveals its systematic increase after annealing. Conventional mechanisms—compositional variation, strain relaxation, and defect formation—are systematically ruled out as plausible causes of the observed optical shift. The PL blueshift, a measure of increasing bandgap, is found to be a direct consequence of increasing SRO, thus revealing a powerful approach for semiconductor materials design.

Ab initio predictions of SRO-induced bandgap modulation

A quantitative baseline is first constructed by predicting how conventional factors (average composition, elastic strain) affect the bandgap of Ge1−x_{1-x}Snx_x alloys (Supplementary Information section 1). This framework is then extended to incorporate short-range chemical ordering, linking the WC-SRO parameter to bandgap variations. Figure 1 summarizes theoretical insights alongside representative experimental results. Central to this framework is the α\alpha parameter, defined as

Influence of SRO modulation on optoelectronic properties in Ge$_{0.9}$Sn$_{0.1}$ alloy

Figure 1. Influence of SRO modulation on optoelectronic properties in Ge0.9_{0.9}Sn0.1_{0.1} alloy. a, The radial distribution function (RDF) gSn−Sn(R)g_{\mathrm{Sn-Sn}}(R) for 1NN WC-SRO parameter α\alpha, illustrating the evolution from random alloy (blue) through partial order (α=0.5\alpha= 0.5, orange) to fully 1NN Sn-Sn depleted (α=1\alpha= 1, red) structure. The RDF is computed from ≈1780\approx1780 atoms, machine-learning potential (MLP)-generated supercells with specific SRO values. The inset shows the evolution of the ratio of the integrated intensity of the 3NN RDF peak to that of the 1NN peak, as a function of α\alpha. b, Unfolded electronic band structures back to the first Brillouin zone of the diamond cubic lattice (L, Γ\Gamma, and X) for special quasi-random structure (SQS) disordered (left) and short-range ordered (α=1\alpha= 1, right) configurations, with color scale denoting the corresponding spectral weight (Methods). The band structure is simulated at 0 K. c, Calculated bandgap variation ΔEg,αTheo.=(Eg(α)−Eg(α=−0.135))\Delta E_{\mathrm{g},\alpha}^{\mathrm{Theo.}} = (E_{\mathrm{g}}(\alpha) - E_{\mathrm{g}}(\alpha= -0.135)) as a function of the WC-SRO parameter, for a fixed strain (ε=0\varepsilon= 0) and Sn composition (x=10x = 10 at.%). d-e, Scanning electron microscopy (SEM) images of Ge/GeSn core/shell nanowires (NWs) in the as-grown state (blue) and after post-growth annealing at 450 ∘C450\,^{\circ}\mathrm{C} (red). Both samples were coated with a 3−43 - 4 nm layer of ALD-Al2_2O3_3; both scale bars: 500 nm. f, Normalized photoluminescence (PL) spectra at 80 K for the as-grown (blue) and annealed (red) NWs, showing a significant post-annealing PL blueshift (ΔEgExpt.\Delta E_{\mathrm{g}}^{\mathrm{Expt.}}). g, Schematic of the Bayesian inference approach for SRO quantification, where experimental EXAFS data are fit against a library of theoretical spectra generated from supercells with varying parameters α\alpha. Iterative optimization identifies the model with the minimum R-factor, yielding the most probable ordering parameter (α∗\alpha^*) for the sample.

αij1NN=1−P1NN(i∣j)cicj,i≠j,(1)\alpha_{ij}^{1\mathrm{NN}} = 1 - \frac{P^{1\mathrm{NN}}(i \mid j)}{c_i c_j}, \qquad i \ne j, \tag*{(1)}

where P1NNP^{1\mathrm{NN}} is the probability of finding ii-jj atomic neighbors in the 1NN shell, and cic_i and cjc_j are the corresponding

species concentrations [22, 23]. This parameter quantifies the extent to which atomic species ii and jj preferentially cluster or avoid each other, with α→1\alpha\rightarrow1 corresponding to suppression of Sn-Sn 1NN pairs compared to a random alloy (α=0\alpha= 0). Ab initio supercell calculations reveal evolution of the local RDF as α\alpha varies: attenuation of the 1NN peak intensity coupled with pronounced strengthening of the 3NN peak (Figure 1a).

Unfolded electronic structures (Figure 1b, Methods) compare special quasi-random structure (SQS) (left) and short-range ordered (α=1\alpha= 1, right) strain-free Ge0.9_{0.9}Sn0.1_{0.1} configurations. SRO increasing from a quasi-random to a fully 1NN Sn-Sn depleted alloy widens the direct bandgap, Eg(α)E_{\mathrm{g}}(\alpha), by >80> 80 meV. Figure 1c plots the theoretical bandgap difference versus α\alpha at a fixed Sn composition and zero strain, isolating the SRO effect. The predicted bandgap change rivals that induced by significant compositional or strain variations [24]. For example, achieving 10 meV bandgap shift via composition change requires ≈ 0.75 at.% Sn reduction. For intermediate SRO values, supercells are generated using machine-learning potential (MLP) (Methods), enabling efficient statistical estimation of the bandgap across a wide range of WC-SRO values (0<α≤10 < \alpha\le1). Random alloys exhibit lattice disorder-induced electronic state localization, narrowing the bandgap. Conversely, greater SRO fosters a more regular atomic motif, reduces local strain, and

promotes electronic delocalization, collectively increasing the bandgap [25].

Experimental evidence supporting SRO-driven bandgap modulation is highlighted in Figure 1d-Figure 1f. Panels d and e display SEM images of as-grown and annealed Ge/GeSn nanowire ensembles, demonstrating morphological stability. Panel f shows normalized infrared PL spectra at 80 K: annealed samples exhibit significant blueshift (ΔEg,αExpt.\Delta E_{g,\alpha}^{\mathrm{Expt.}}). To link optoelectronic changes to atomic-scale structure, an adaptive Bayesian inference algorithm resolves the α\alpha

SamplesTT (± 5°C)dshelld_{\mathrm{shell}} (nm)Sn content (± 0.5 at. %)ᵃEgRT→EgTE_g^{\mathrm{RT}} \rightarrow E_g^{T} (± 5 meV)ΔEgExpt.\Delta E_g^{\mathrm{Expt.}} (± 5 meV)ΔEg,xTheo.\Delta E_{g,x}^{\mathrm{Theo.}} (meV) (DFT)SRO (α\alpha) (±0.05)
AN.A.140±7140 \pm79.5481→481 \rightarrow N.A.0N.A.0.20
B300145±7145 \pm79.5481→490481 \rightarrow490840.30
C350146±10146 \pm109.5489→502489 \rightarrow5021250.36
D400141±7141 \pm79.5486→509486 \rightarrow5092170.43
E450139±12139 \pm129.6489→515489 \rightarrow5152580.52

Table 1. Structural, compositional, and electronic properties of post-annealed Ge/GeSn core/shell nanowires. Sn content is measured by HRXRD and TEM-EDX, thickness by SEM, and bandgap shift ΔEgExpt.=Eg(T)−Eg(25∘C)\Delta E_g^{\mathrm{Expt.}} = E_g(T) - E_g(25^\circ\mathrm{C}) in meV by infrared PL spectroscopy at 80 K. TT denotes the post-growth annealing temperature. Theoretical bandgap shifts ΔEg,xTheo.\Delta E_{g,x}^{\mathrm{Theo.}} (meV) is computed via DFT using SQS supercells that represent the measured alloy disorder at the measured Sn content. ΔEg,xTheo.\Delta E_{g,x}^{\mathrm{Theo.}} isolates composition-driven changes at fixed strain (ε=0\varepsilon= 0) by varying Sn content (xx) in SQS cells. Finally, for each entry, the WC-SRO parameter is reported as α\alpha (± σα\sigma_\alpha), inferred via the adaptive Bayesian EXAFS fitting framework that integrates DFT-generated supercells. The average WC-SRO uncertainty (σα\sigma_\alpha) is estimated to be 0.05. Reported uncertainties reflect the measurements or computation associated with each technique/estimate.

parameter directly from EXAFS spectra. The workflow, illustrated in Figure 1g (Supplementary Information Figure 4–5), links experimental data with first-principles structural models. This model-based approach moves beyond conventional EXAFS analysis (Supplementary Information section 2) by mapping experimental spectra into a physically meaningful order parameter, enabling statistically robust SRO quantification.

Bayesian inference short-range order quantification and bandgap-SRO correlation

Figure 4. Bayesian inference short-range order quantification and bandgap-SRO correlation. a, Temperature-dependent single-scattering path analysis for first three coordination shells (1NN, 2NN, 3NN) Sn∗^{*}–Sn (top) and Ge∗^{*}–Sn (bottom) pairs. Asterisk denotes absorber atom. Aggregate EXAFS contributions from these paths, isolate element-specific local coordination environments. b, The MSRD (σ2\sigma^{2}) for Sn∗^{*}–Sn pairs across coordination shells. Colored bars: theoretical MSRD from machine-learning potential (MLP)-supercells (25 °C to 450 °C); bar heights reflect ensemble variance at fixed α\alpha. Circles: experimental MSRD extracted from EXAFS Bayesian inference; error bars indicate posterior uncertainties (Methods). Insets: Warren-Cowley short-range order (WC-SRO) parameter α\alpha versus annealing temperature (circles, right). Integrated intensity ratio of 3NN to 1NN (I3NN/I1NNI_{\mathrm{3NN}}/I_{\mathrm{1NN}}) of Sn∗^{*}–Sn peaks in panel a (stars, left). c, Bayesian inference validation through MSRD: experimental (σExpt.2\sigma_{\mathrm{Expt.}}^{2}) versus theoretical (σTheo.2\sigma_{\mathrm{Theo.}}^{2}) for Sn∗^{*}–Sn scattering pairs across 1NN and 3NN coordination shells (25 °C to 450 °C). Colored ellipses represent 3σ\sigma (95 % confidence) centroid estimation of the data distribution for each shell. d, Bandgap shifts vs. the WC-SRO parameter α\alpha: experimental (ΔEg,Expt.\Delta E_{\mathrm{g,Expt.}}, from PL) and theoretical (ΔEg,αTheo.\Delta E_{\mathrm{g,\alpha}^{\mathrm{Theo.}}}, from DFT) relative to the as-grown sample WC-SRO parameter (α25∘C=0.20±0.05\alpha_{25^\circ\mathrm{C}} = 0.20 \pm0.05) quantified using the current Bayesian inference. Error bars: vertical (from structural and optical measurements); horizontal (from statistical uncertainty in α\alpha from Bayesian analysis). Red curve and band show DFT ensemble mean and variance.

Sample preparation and multimodal characterization

Single-crystalline Ge/GeSn core/shell nanowires were epitaxially grown on Ge (111) substrates (Methods). SEM verified vertical alignment, uniform diameter, and morphological stability (Figure 2a–Figure 2d). HRXRD validated phase purity, ⟨111⟩\langle111 \rangle crystallographic orientation, and yielded an ensemble-averaged nominal Sn content of (9.50 ± 0.25) at.% (Figure 2e). A reference plan-view TEM image of the as-grown structure is shown in Figure 2f. EDX in the TEM mapped elemental distributions (Figure 2g–Figure 2h), confirming uniform Sn content of (9.40 ± 0.75) at.% (Methods, Supplementary Information Figure 1 and Table I).

Figure 2. Structural and compositional characterization of Ge/GeSn core/shell nanowires (NWs) after rapid thermal annealing. a–d, SEM micrographs of NW arrays post-growth annealing (300 °C to 450 °C). e, High-resolution X-ray diffraction 2θ2\theta–ω\omega scans around the (333) reflection. The intense peak at 90.05° corresponds to the Ge (333) substrate; the broader peak at 88.5° arises from the GeSn shell. Sn content determination is described in Methods. Solid lines: experimental data from annealed samples. Dotted lines: experimental data from as-grown samples. All samples were coated with a ≈ 3 nm alumina layer. f, Side-view TEM of as-grown reference Ge/GeSn nanowire structure. g, Side-view TEM of representative annealed nanowire (450 °C), with corresponding EDX elemental maps displaying the spatial distributions of Ge and Sn. h, Cross-sectional EDX color overlay map. i, HAADF-STEM image of a single NW annealed at 450 °C, with indexed {112} and {110} crystal facets. j, HRTEM image of the Ge/GeSn interface. Inset (upper left): diffraction pattern image indexed to the [111] zone axis. Inset (lower right): aberration-corrected lattice image at the heterointerface. k, HRTEM image at the GeSn/Al₂O₃ interface with an inset showing an aberration-corrected lattice image at the interface.

A central challenge in tuning SRO in GeSn alloys is their intrinsic metastability and strong tendency for Sn segregation above the eutectic point [26]. Rapid thermal annealing (RTA) is employed on nanowire assemblies encapsulated by an ultrathin (≈ 3–4 nm), conformal alumina layer deposited via atomic layer deposition (ALD) (Methods). This capping inhibits Sn surface segregation, a primary kinetic pathway for GeSn partial melting [21], while preserving crystal quality and optical transparency [27] during annealing. Studied annealing temperatures spanning 300 °C to 450 °C may permit atomic diffusion to reorganize local bonding within the GeSn alloy shells.

Post-annealing nanowire luminescence

Low-temperature infrared PL spectroscopy elucidates the influence of RTA on the optical response of GeSn NWs (Methods). The core/shell heterostructure shapes emission characteristics through three mechanisms. First, at compositions exceeding ≈ 8 at.% Sn, the shell undergoes an indirect-to-direct bandgap transition [28], rendering Γ\Gamma-valley emission dominant. The resulting type-I band alignment confines both electrons and holes to the GeSn shell. Second, when shell thickness (≈ 150 nm) exceeds optical penetration depth (≈ 100 nm at 808 nm laser excitation [29]), photoexcitation is largely shell-confined, improving radiative efficiency and enabling direct measurement of bandgap modulation. Third, strain partitioning between core and shell enables wavelength engineering [30, 31]. Consequently, PL spectra reflect shell-dominated, direct-gap recombination, establishing a baseline for examining RTA-induced modifications.

Figure 3a displays normalized PL spectra acquired at 80 K for samples annealed at selected temperatures (Table I), benchmarked against unannealed alumina-coated controls. Alumina encapsulation preserves spectral characteristics while providing modest (≈ 10%) intensity gain, possibly due to suppressed non-radiative recombination via surface passivation of dangling bonds [32]. Three trends emerge across all annealed samples: (i) a monotonic blueshift commensurate with rising annealing temperature, indicating progressive bandgap widening; (ii) pronounced PL intensity enhancement, reaching twenty-fold at 450 °C, consistent with reduced non-radiative recombination; and (iii) stable full-width at half maximum (FWHM), suggesting uniform average composition. Spectra exhibit single symmetric peaks (Supplementary Information Fig. 7, Supplementary Information section 3) confirming that the Γ\Gamma-minimum lies sufficiently below the

Spectroscopic and structural evolution of Ge/GeSn with thermal annealing

Figure 3. Spectroscopic and structural evolution of Ge/GeSn with thermal annealing. a, 80 K PL spectra for nanowires annealed across a range of temperatures (300 °C to 450 °C) compared to as-grown, unannealed (25 °C), alumina-coated control samples (yellow-green curves); all are normalized to their peak maximum. The graded colored band illustrates the bandgap blueshift with TT. Dashed red lines denote the JDOS model line shape fits (Equation (2) in Methods). The excitation power density was fixed to 1.9 kW cm−2^{-2} for all samples. b, Extracted bandgap shift ΔEgExpt.\Delta E_{\mathrm{g}}^{\mathrm{Expt.}} plotted against annealing temperature (dashed line is guide to the eye). Error bars denote three standard deviations. c, k2k^{2}-weighted EXAFS spectra, k2⋅χ(k)k^{2} \cdot\chi(k), acquired at the Sn-K edge. The line plotted below each spectrum displays the normalized residual. For all temperatures, the residuals remain below 10−210^{-2}. d, Corresponding Fourier transforms, ∣FT(k2⋅χ(k))∣|\mathrm{FT}(k^{2} \cdot\chi(k))| as a function of radial distance RR. Each trace is segmented into two regions: the unshaded area (left) highlights the dominant 1NN peak with its associated y-axis, and the shaded gray region (right) isolates the typically weaker 2NN and 3NN peaks, for which a secondary (right-side) y-axis with expanded scale is provided. Dashed gray lines serve as guides to the eye for each peak.

LL-minimum to suppress intervalley tunneling [33]. The as-grown reference exhibits a narrow average PL bandgap of 488 ± 4 meV, consistent with TEM-EDX and HRXRD evidence of shell compositional stability and closely aligned with prior literature [31] for nanowires of similar geometry and composition (Supplementary Information section 1).

Temperature-dependent PL emission from the annealed samples (Supplementary Information Fig. 7) shows systematic shifts and broadening from 80 K to 300 K, consistent with band-to-band recombination and intrinsic thermal effects. The pronounced intensity rise upon cooling is a defining characteristic of direct bandgap semiconductors [34], corroborated by fits to the combined JDOS/Viña model [35] (Supplementary Information section 3, Methods). fig:3b summarizes the experimental bandgap shift versus annealing temperature. The measured bandgap closely mirrors theoretical predictions (Figure 1c), suggesting that enhanced chemical SRO drives the observed PL blueshift. However, identifying the blueshift origin requires a critical evaluation of plausible contributing factors other than SRO.

Plausible annealing-induced blueshift mechanisms

Four plausible mechanisms for the observed spectral shift during annealing are evaluated: Sn composition reduction in the GeSn shell, a change in its elastic strain state, crystalline defect formation, and increased SRO.

Sn composition: Bandgap shifts may arise from changes in average Sn content, as the GeSn bandgap increases with decreasing Sn concentration [36]. To ascertain whether post-growth annealing modifies shell composition, HRXRD and TEM-EDX measurements were compared across the annealed sample series. Both techniques confirm stable Sn content of (9.5±0.5)(9.5 \pm0.5) at.% independent of thermal history (Figure 2e, Figure 2g-Figure 2h, Table I) with no systematic trends correlating with annealing temperature. EDX maps (Figure 2e,Figure 2f; Supplementary Information Fig. S1) show uniform Sn distribution with no evidence of clustering, segregation, or secondary phase formation. The characteristic faceted “sunburst” GeSn shell morphology remains unaltered [31].

To quantify how subtle compositional variations would alter the bandgap, first-principles DFT sensitivity simulations were conducted (zero strain, α=0\alpha= 0), spanning Sn compositions from 9 − 11 at.% (Supplementary Information section 1). The calculations predict a maximum sensitivity of ≈ 35 meV/at.% Sn, in agreement with experimental reports of ≈ 30 − 40 meV at.% Sn for GeSn alloys [37]. Assuming maximum compositional uncertainty (±1.0 at.%, combining HRXRD and EDX errors), composition-driven bandgap shifts should not exceed 35 meV, nor exhibit a systematic monotonic increase with annealing temperature (Figure 3b). The absence of detectable Sn redistribution, which is likely required for a notable change in average shell composition, also indicates this mechanism is not the cause of the measured PL blueshift.

Strain state: Lattice mismatch between the Ge core and Ge0.905Sn0.095\mathrm{Ge}_{0.905}\mathrm{Sn}_{0.095} shell may induce residual strain, thereby altering the bandgap [28, 38]. However, elasticity theory [39] predicts that coherent core/shell nanowires achieve > 95 % strain relaxation in the shell when the shell thickness-to-core radius ratio exceeds 3. For the geometry studied herein (ratio ≈ 5.6), the GeSn shell is predicted to be nearly fully strain relaxed. Moreover, HRXRD ω−2θ\omega- 2\theta scans of the symmetric (333) Bragg reflection (Figure 2e) confirm this: the GeSn shell peak position remains invariant, indicating negligible lattice parameter change during annealing. If residual compressive strain in GeSn were relaxing, the bandgap should decrease [38]. The monotonic blueshift (Figure 3b) is, therefore, inconsistent with strain-driven bandgap modulation.

Aberration-corrected HAADF-STEM and HRTEM (Figure 2i–Figure 2k, Supplementary Information Fig. S1) corroborate this, revealing defect-free Ge/GeSn epitaxy and an abrupt GeSn/Al2_2O3_3 interface. Diffraction patterns from the core region (Figure 2j, top inset) indicate single-crystalline structure is maintained after annealing. The agreement between elasticity theory predictions, HRXRD lattice parameter stability, and atomic-resolution imaging confirms both composition and strain state remain effectively constant, eliminating these as plausible mechanisms for the GeSn bandgap change.

Shell crystalline imperfections: Annealing could, in principle, introduce crystalline defects (e.g., dislocations, stacking faults, impurity clusters) that degrade optical performance. TEM performed across the annealed series consistently reveal no evidence of such defects (Figure 2i-Figure 2k, Supplementary Information Fig. S1). The power-law dependence of PL intensity on excitation power density (IPL∝PmI_{\mathrm{PL}} \propto P^m, Supplementary Information section 3) provides insight into defect-related carrier recombination mechanisms. For the as-grown sample, linear scaling (m=1.04±0.02m = 1.04 \pm0.02) reflects low-injection operation, where background carrier concentration (associated with unintentional doping [40]) exceeds photo-generated carriers across the measured power range. Annealing reduces background carriers via defect passivation, enabling partial access to the high-injection regime at higher powers. The higher exponent (m=1.76±0.05m = 1.76 \pm0.05), approaching the m≈2m \approx2 limit characteristic of Shockley-Read-Hall (SRH) recombination, is consistent with the observed twenty-fold PL intensity amplification (Figure 3a): defect passivation and/or annihilation reduces the non-radiative loss rate, improving the internal quantum efficiency [41]. These observations indicate that, although defect populations are modified during annealing, they do not account for the observed blueshift.

SRO evolution: We now consider whether RTA measurably modifies SRO. EXAFS measurements at the Sn and Ge K-edges (Figure 3c–Figure 3d; Supplementary Information Fig. S9) are combined with first principles–based supercell modeling and adaptive Bayesian inference to extract values of the WC-SRO parameter α\alpha across the annealed sample series. Adaptive Bayesian inference identifies, for each dataset, the most probable SRO state from a library of DFT-based MLP generated supercells and returns credible intervals of EXAFS-related parameters (Supplementary Information section 4).

The Fourier transform magnitudes, ∣F(kw⋅χ(k))∣|F(k^w \cdot\chi(k))|, exhibit peak positions tracking mean neighbor distances and widths reflecting the mean-square relative displacement (MSRD) [42]. Nonetheless, phase shifts, windowing/kk-weight effects, and multiple scattering limit conventional analysis to qualitative interpretation (Supplementary Information section 2). The measured EXAFS spectra evolve systematically with annealing (Figure 3d). Chemically selective single-scattering path decomposition from both K-edges (Figure 4a, Supplementary Information Fig. S9; Supplementary Information section 5) decouples Sn*-Sn from Ge*-Sn coordination. The Ge*-Sn scattering path intensities increase consistently across coordination shells with rising annealing temperature, indicating enhanced Ge-Sn bonding (Figure 4a). This trend is mirrored in the Sn*-Ge scattering paths (Supplementary Information Fig. S9), consistent with depletion of 1NN Sn-Sn pairs and with RDF calculations (Supplementary Information Fig. S2).

The Sn*-Sn paths reveal a more complex evolution. While the 1NN peak intensity decreases monotonically with annealing, the 3NN peak enhancement saturates or weakly reverses at the highest annealing temperature. This non-monotonic 3NN behavior is consistent with multi-scattering (MS) interference effects in partially ordered crystalline systems, where changes in the chemical identity of intermediate scattering atoms shift MS path phases [43]. The Bayesian framework, by fitting the entire experimental EXAFS spectrum, implicitly accounts for MS interference to quantify the SRO order parameter, α\alpha, versus annealing temperature (Figure 4b).

Within the Bayesian inference framework, each best-fit supercell carries a distinct WC-SRO value, enabling mapping between EXAFS features and α\alpha. The MSRD is computed independently for each scattering path and coordination shell (Supplementary Information section 2) [44]. The bond lengths are also evaluated (Supplementary Information Table 1). The WC-SRO parameter α\alpha increases monotonically from 0.20 (±\pm0.05) for the as-grown material to 0.52 (±\pm0.05) after annealing at 450 °C (Figure 4b, top inset, left). Progressive redistribution of Sn atoms from first- to third-neighbor shells is tracked by the 3NN-to-1NN peak integrated intensity ratio (Figure 4b, top inset, right), mirroring the predicted evolution of the RDF (Figure 1a, Extended Data Fig. 6d) and providing a self-consistent link between theory and experiment. Shell-resolved MSRDs (Figure 4b) quantify the atomistic reordering. While the extracted values are consistent with prior reports [13, 15], the adaptive Bayesian framework greatly reduces statistical uncertainty (Supplementary Information Table 1, Supplementary Information section 4).

Discussion and Outlook

An increase in GeSn bandgap after annealing is attributed to increasing SRO resulting from local atomic redistribution. This conclusion is reinforced by considering three independent observables: (1) local structural order inferred from EXAFS, (2) bandgap shifts from PL, and (3) first-principles ensemble predictions. Figure 4c compares experimental and theoretical MSRD for Sn∗^{*}–Sn scattering pairs at the inferred α\alpha for each sample (Table I). Strong correlation around the y=xy=x line for the 1NN and 3NN coordination shells across the annealing series demonstrates internal consistency, indicating that the model-based inference successfully captures the local order probed by EXAFS. The overlaid first-principles predictions from 128-atom supercell ensembles in Figure 4d are not fitted data; rather, α\alpha values are independently extracted from EXAFS, and the supercells are constructed at those α\alpha values. The red curve traces the ensemble mean of calculated bandgap shifts, and the shaded band reports configurational variability. For partially ordered alloys, different local atomic arrangements can yield identical mean α\alpha but distinct bandgaps due to motif-level heterogeneity (Supplementary Information section 6).

A bandgap–SRO sensitivity metric S=dEg/dαS = \mathrm{d}E_g/\mathrm{d}\alpha with an experimental slope 79 ±\pm 10 meV/α\alpha agrees with the DFT–ensemble trend (65 ±\pm 6 meV/α\alpha) within uncertainty. This agreement is consistent with SRO being the principal driver of observed bandgap changes across the explored annealing range. Atomic-level analysis pinpoints local Sn–Sn motif heterogeneity as the source of residual variation: larger bandgaps correlate with suppressed nearest-neighbor Sn–Sn pairs and an augmented third-neighbor population. The mean SRO is, therefore, an effective descriptor of the trend, but the distribution of local motifs contributes to the bandgap variance at fixed SRO [45].

The ability to both quantify the extent of SRO and predict its effect on the bandgap enables a significant advance in controlling this overlooked degree of freedom for semiconductor alloy design. Intentional tuning of SRO may be broadly useful across multiple classes of such materials. Specific to the GeSn system, this work establishes thermal processing as a viable approach for manipulating atomic short-range order, thus engineering the bandgap independent of changes in composition or strain of the crystal.

METHODS

First-Principles Calculations. Structural models of GeSn alloys containing 1728 atoms with 10 at.% Sn

with varying degrees of SRO were generated via Monte Carlo sampling at a range of temperatures (300 K to 1500 K with 300 K step) using state-of-the-art and highly accurate MLP for GeSn alloys [3]. For band structure calculations (Figure 1b), all 128-atom generated structures with variable SRO were further optimized with DFT using local local density approximation (LDA) for the exchange-correlation functional, known for yielding excellent agreement with experimental results for geometry optimization in pure Ge and Sn [1, 3, 45, 46]. A Monkhorst-Pack 2×2×22 \times2 \times2 kk-point mesh [47] and a plane-wave cutoff energy of 300 eV were applied. Structural relaxations during total energy minimization were performed with the conjugate gradient algorithm, imposing convergence criteria of 1×10−41 \times10^{-4} eV and 1×10−31 \times10^{-3} eV for electronic and ionic steps, respectively. To improve the fidelity of bandgap estimations, the modified Becke-Johnson (mBJ) exchange potential [48] combined with LDA correlation was employed, as implemented in the Vienna Ab initio Simulation Package (VASP) software package [49], using a cc-mBJ parameter of 1.2. This computational scheme demonstrates excellent agreement with experimental bandgap data for Si, Ge, and α\alpha-Sn, while offering greater efficiency compared to hybrid functional or GW approximation methods [48, 50, 51]. Band structures were unfolded into the primitive Brillouin Zone of the diamond cubic lattice using the spectral weight method [52], as implemented in the fold2bloch code [53]. Relativistic effects, including spin-orbit coupling, were incorporated throughout the calculations, as these are essential to reproduce the experimental band structures of Ge and α\alpha-Sn [50, 51]. While a perfectly random alloy is characterized by a WC-SRO parameter α=0\alpha= 0, the actual supercell configurations employed in the bandgap calculations yield a distribution of α\alpha values centered near zero. For the 128-atom SQS ensemble used in this particular calculation, the reference point is α=−0.125\alpha= -0.125, reflecting the practical limitations in generating fully random atomic arrangements in finite-size cells. Computational tractability limits band structure calculations to small supercells (128 atoms), in contrast to the significantly larger supercells (≈1780\approx1780 atoms) employed in the adaptive Bayesian inference approach for SRO quantification from EXAFS measurements described below. Multiple independent SQS configurations (typically 5) are necessary to statistically sample the configurational space and estimate both the mean and standard deviation of the bandgap for the random alloy baseline (error bars in Figure 1c at α=−0.125\alpha= -0.125).

Epitaxial Growth of Ge/GeSn Core/Shell NWs.

The vapor-liquid-solid (VLS) growth of Au-catalyzed Ge/GeSn core/shell NWs was performed in a reduced-pressure chemical vapor deposition (RPCVD) system utilizing ultra-high purity H2_2 as the carrier gas, with 10 % germane (GeH4_4) and tin(IV) chloride (SnCl4_4) as precursor gases. The full experimental procedure has been comprehensively described elsewhere [54, 55]. In brief, NWs were synthesized on single-crystal Ge (111) substrates by drop-casting Au colloids (diameter 40 nm, Sigma-Aldrich) onto the substrate surface, followed by immediate transfer to the RPCVD reactor. The fabrication of the Ge NW cores was carried out using a carefully optimized two-step thermal process under a constant pressure of 30 Torr: (i) an initial nucleation phase at 370 °C for 2 min to initiate axial nanowire growth at the base, and (ii) subsequent core growth at a reduced temperature of 300 °C to promote morphological uniformity by minimizing tapering and kinking. After core synthesis, the SnCl₄ precursor was introduced at 270 °C to enable radial overgrowth, yielding coherent and epitaxial GeSn shells around the Ge cores. For the present study, five samples were fabricated, each featuring a nominal Sn incorporation of (9.5 ± 0.5) at.%, an average shell thickness of (142 ± 12) nm, and an average wire length of (4.1 ± 0.3) μm. Significantly, this NW platform offers the benefit that the GeSn shell remains dislocation-free and strain relaxed as a result of growth on the elastically compliant (much smaller radius) Ge core, as evidenced by previous studies [30, 39, 56].

ALD Growth of Al₂O₃ Protective Layer. A conformal Al₂O₃ layer was deposited onto the entire NW growth chip via ALD following a standardized protocol. Immediately prior to ALD processing, the chip was immersed for 40 s in a 1:1 (v/v) mixture of concentrated 37 % HCl and deionized (DI) water to remove any Ge- and Sn-induced native surface oxides. This etching step was succeeded by thorough rinsing with DI water and then drying in a nitrogen stream. The cleaned sample was then loaded into the ALD reactor, which was maintained at a base pressure of 7.5 Pa (5.6 × 10⁻² Torr). Alumina deposition was performed at a substrate temperature of 200 °C, using trimethylaluminum (TMA) as the aluminum precursor and ozone as the oxidant. Each ALD cycle consisted sequentially of a 1 s TMA pulse, a 25 s purge, a 10 s ozone pulse, and a final 30 s purge. A total of 24 cycles were executed, resulting in a (3 ± 1) nm Al₂O₃ film, determined by spectroscopic ellipsometry on a concurrently processed silicon reference substrate.

Rapid Thermal Annealing. RTA was performed using an RTA-610 system (AllWin21 Corp.) to thermally process the NWs under a nitrogen (N₂) environment. Each sample was annealed for 12 min at each temperature from 300 °C to 450 °C, in 50 °C increments. The heating rate was 100 °C s⁻¹, followed by a ramp-down at 50 °C s⁻¹.

High-Resolution X-Ray Diffraction of the Annealed Series. Structural and compositional characterization of the Ge/GeSn core-shell nanowire arrays was performed using an HRXRD on a PANalytical Empyrean system (45 kV, 40 mA, 1/4° incident slit, 2 mm mask). Symmetric ω–2θ scans of the (333) direction (Figure 2e) exhibit well-resolved diffraction peaks from the single-crystalline Ge(111) substrate (∼ 90.05°) and the Ge₁₋ₓSnₓ shell (∼ 88.48°). Quantitative Sn content in the GeSn shell was extracted using Vegard’s law, with a bowing parameter of 0.041 Å [57]. Gaussian fitting of the HRXRD peaks was performed to extract the peak positions and quantify the associated uncertainty. These GeSn shell composition estimates, when compared to the TEM-EDX composition maps, show excellent agreement. Analyses across multiple fields of view and NW batches confirmed uniform composition and structure for the analyzed samples (Supplementary Information Fig. S1). Relaxed shell lattice parameters enable a direct correlation of the shell Bragg angle to the Sn alloy fraction. For the nanowire geometries and shell thicknesses considered here, prior combined diffraction and modeling studies have shown that GeSn shells grown on elastically compliant ∼ 50 nm diameter Ge cores are effectively strain-relaxed [58]; under these conditions, the measured out-of-plane lattice parameter can be related to the average Sn alloy fraction without requiring additional reciprocal-space mapping. The HRXRD-derived compositions reported in Table I therefore provide a consistent and reliable estimate of the Sn content across the annealing series. HRXRD composition uncertainty arises from two sources. First, peak-fitting precision is limited by instrumental resolution and counting statistics: σα≈±0.003\sigma_{\alpha} \approx\pm0.003 Å. Second, the Vegard-linear calibration carries uncertainty of approximately ±0.3–0.5 at.% from literature scatter. Error propagation in quadrature yields total composition uncertainty of Δ(Sn)≈±0.5\Delta(Sn) \approx\pm0.5 at.%.

HRTEM Imaging. TEM investigations were performed using a ThemIS 60-300 instrument (Thermo Fisher Scientific) operated at an accelerating voltage of 300 kV. The microscope was equipped with an image aberration corrector and a Bruker SuperX EDX detector for spectroscopic analysis. HRTEM images and selected area electron diffraction patterns were acquired with a Ceta2 CMOS camera, while scanning transmission electron microscopy (STEM) images were collected using the high-angle annular dark field (HAADF) detector. For plan-view analysis, NWs were mechanically transferred onto TEM grids, and cross-sectional specimens were prepared via focus ion beam (FIB) milling. Insets in Figure 2j display fast-Fourier transform (FFT) filtered HRTEM images. EDX quantification employed 2 × 2 binning. The parameters for EDX acquisition were a probe current of 615 pA, a spot size of 2 nm, a step size of 1.5 nm, and a dwell time of 8 μs. EDX uncertainty (± 0.75 at.%) is determined from counting statistics at the measured photon intensity and is consistent with repeated measurements on standard reference samples.

EXAFS Spectroscopy & Bayesian SRO Analysis. Measurement: Fluorescence-yield EXAFS measurements were performed at the Stanford Synchrotron Radiation Lightsource (SSRL) beamline 7-3. To eliminate potential substrate background contributions, Ge/GeSn core/shell NWs were mechanically exfoliated from their Ge (111) growth substrates and transferred onto Kapton tape. Owing to the small volume fraction of Ge cores (∼2%) relative to the GeSn shells in the core/shell geometry, the measured EXAFS signal originates almost exclusively from the shell region. Nanowire samples were maintained at liquid helium temperatures (7 K) to suppress thermal disorder. Ge K-edge and Sn K-edge spectra were collected using a passivated implanted planar silicon (PIPS) detector (with Ga filter) and a 30-element Ge detector (with Cd filter), respectively. Data reduction, including dead-time correction, energy calibration, and background subtraction, was performed using SIXPACK and ATHENA software [59]. The isolated fine-structure oscillations χ(k)\chi(k) extraction involved: (i) fitting and subtracting a linear polynomial in the pre-edge region to eliminate instrumental background and non-resonant absorption, and (ii) applying cubic spline functions to model the atomic background in the post-edge region. All data processing steps were performed with the ATHENA software package. Sub-1NN features (Figure 3d) result from the presence of SnOx\mathrm{SnO}_{x} from surface oxidation of the shells.

Bayesian Inference: Local structural ordering was quantified using q-SRO, a custom Bayesian inference framework that extracts short-range order directly from EXAFS spectra. To capture the full coordination environment, Ge and Sn K-edge spectra were simultaneously fitted in RR-space (1.3 Å to 6.0 Å) using theoretical scattering paths generated from 1728-atom SRO supercells via FEFF7 [43]. Critically, scattering paths were not averaged: FEFF7 calculated individual backscattering amplitudes f(k)f(k) and phase shift δ(k)\delta(k) functions for each unique absorber-scatterer configuration from the DFT-relaxed supercells. Paths within the same coordination shell and atomic pair type share scattering functions but have occupation weights Neff(α)N_{\mathrm{eff}}(\alpha) determined by the WC-SRO parameter α\alpha (Eq. S1, Supplementary Information section 2). This preserves full chemical sensitivity without imposing random-alloy assumptions. The fitting model included nine single-scattering paths per edge, accounting for the first three coordination shells (Ge∗^{*}–Ge/Sn, Sn∗^{*}–Ge/Sn), plus dominant multiple-scattering paths (e.g., triangular MS Ge∗^{*}–Sn1–Sn2 and Sn∗^{*}–Ge–Ge [60]). Refined parameters (11 total): (1) α\alpha, (2) ΔRGe−Sn\Delta R_{\mathrm{Ge-Sn}}, (3) σGe−Sn2\sigma_{\mathrm{Ge-Sn}}^{2}, (4) shell-specific coordination numbers (constrained by stoichiometry and supercell SRO), and (5) ΔE0\Delta E_{0} per edge, with ΔRGe−Sn\Delta R_{\mathrm{Ge-Sn}} and σGe−Sn\sigma_{\mathrm{Ge-Sn}} constrained to be identical across both edges for physical consistency. To resolve spectral degeneracies, we implemented Bayesian inference with DFT-calculated bandgap priors (Supplementary Information section 6):

P(α∣DEXAFS)∝L(DEXAFS∣α)⋅PDFT(α),P(\alpha\mid D_{\mathrm{EXAFS}}) \propto\mathcal{L}(D_{\mathrm{EXAFS}} \mid\alpha) \cdot P_{\mathrm{DFT}}(\alpha),

where L(DEXAFS∣α)\mathcal{L}(D_{\mathrm{EXAFS}} \mid\alpha) represents the EXAFS likelihood function derived from spectral goodness-of-fit, and PDFT(α)P_{\mathrm{DFT}}(\alpha) is the informative prior defined by the consistency between the theoretical bandgap of the short-range order structure and experimental optical absorption data. Final spectra were ensemble averaged over 180 Sn and 50 Ge absorber sites per supercell to account for local structural heterogeneity. This DFT-informed prior approach reduces SRO parameter uncertainty by approximately 50% compared to uninformed fitting (uniform prior) while maintaining unbiased central estimates (Supplementary Information Fig. S10 and Table I), demonstrating that the framework enhances statistical precision without introducing systematic bias. The MSRD for each scattering path and coordination shell was computed independently using the DFT-relaxed supercell geometries (Supplementary Information Eq. S2). Complete details of the Bayesian inference algorithm, clustering methodology, and validation against uninformed priors are provided in Supplementary Information sections 2 and 6.

Optical Characterization. Quasi-continuous wave (qCW) PL measurements were performed with samples mounted in a liquid-nitrogen-cooled cryostat featuring a CaF2_{2} optical window. Sample excitation was provided by a 1 W, 808 nm diode laser, which was electrically modulated as a square wave at 10 kHz with a 50% duty cycle. The emitted PL was collected by a reflective parabolic mirror and directed into a Bruker Invenio-R Fourier-transform infrared (FTIR) spectrometer for spectrally resolved analysis. The spectrometer was operated in step-scan mode, and the signal was detected using a liquid-nitrogen-cooled mercury cadmium telluride (MCT) detector coupled to a lock-in amplifier to reject thermal background. For power-dependent measurements, the incident excitation density was varied using a series of neutral density filters. The laser spot area on the sample surface was determined to be 3670 μ\mum2^{2} via the knife-edge method, corresponding to an excitation density range of approximately 1 kW cm−2^{-2} to 22 kW cm−2^{-2}. Spectral line shapes were fit using a model based on the convolution of a Gaussian function with a JDOS term [29, 35], described by:

IPL=A[E−Eg×exp⁡(−EkBT)]∗[1γ2πexp⁡(−E22γ2)],(2)I_{\mathrm{PL}}=A\left[\sqrt{E-E_{g}}\times\exp\left(\frac{-E}{k_{\mathrm{B}}T}\right)\right]*\left[\frac{1}{\gamma\sqrt{2\pi}}\exp\left(\frac{-E^{2}}{2\gamma^{2}}\right)\right], \tag*{(2)}

where AA is a scaling constant with units of (energy)1/2^{1/2}, EgE_{g} is the bandgap energy, TT is the absolute temperature, and γ\gamma is the spectral broadening parameter. The first term, representing the JDOS multiplied by the Boltzmann distribution, models the emission profile for parabolic bands under low-excitation conditions. The second term accounts for broadening effects—including alloy composition fluctuations and strain inhomogeneities—through convolution with a Gaussian function [61, 62]. More details related to the optical analysis are presented in Supplementary Information section 3.

Code Availability

All codes that support the findings of this study are available from the corresponding authors upon reasonable request.

Acknowledgements

This work was supported by μ\mu-ATOMS, an Energy Frontier Research Center funded by the U.S. Department of Energy (DOE), Office of Science, Basic Energy Sciences, under award DE-SC0023412 and SubAward No. UA2023-351. This study includes work supported by the DOE Office of Science, Office of Workforce Development for Teachers and Scientists, through the Office of Graduate Student Research (SCGSR) program, administered by the Oak Ridge Institute for Science and Education under contract DE-SC0014664. Additional support was provided by the U.S. National Science Foundation under grant DMR-2003266 and by the National Institute of Standards and Technology (NIST). We acknowledge the use of the Stanford Synchrotron Radiation Lightsource (SSRL), SLAC National Accelerator Laboratory, supported by the DOE Office of Science, Office of Basic Energy Sciences, under contract DE-AC02-76SF00515. This research used resources of the National Energy Research Scientific Computing Center (NERSC), a DOE Office of Science User Facility, under contract DE-AC02-05CH11231 and NERSC Award BES-ERCAP0027056. Additional computing support was provided by the Sherlock cluster at Stanford University; we thank Stanford University and the Stanford Research Computing Center for computational resources and support. The authors also acknowledge the George Washington University High Performance Computing facility for computing support. Part of this work was performed at the Stanford Nano Shared Facilities (SNSF), supported by the National Science Foundation under award ECCS-2026822. Reference to specific software packages, equipment, and materials is only for completeness and does not represent an endorsement by NIST or the Federal Government of the USA.

AUTHORS INFORMATION

A.A. carried out thermal annealing and EXAFS-related theoretical analysis and measurements. J.Z.L. carried out the epitaxial growth of the Ge/GeSn NWs. J.M. and K.M. provided the PL measurements. J.C.W supported the EXAFS analysis. L.V. and A.M.M. performed all the TEM characterizations of the annealed NWs. S.C. and T.L. performed the DFT calculations and conducted the large-scale EXAFS simulations using supercells generated from machine-learning atomistic simulations. P.C.M. conceptualized the research, acquired funding for it, and led the research. All authors contributed to the manuscript.

Corresponding Authors:

*[email protected]

† [email protected]

NOTES

The authors declare no competing financial interests.

References

  1. [1]Cao, B., Chen, S., Jin, X., Liu, J. & Li, T. Short-Range Order in GeSn Alloy. ACS Applied Materials & Interfaces 12, 57245–57253 (2020). URL https://doi.org/10.1021/acsami.0c18483.
  2. [2]Xu, X. & Jiang, H. Cluster expansion based configurational averaging approach to bandgaps of semiconductor alloys. The Journal of Chemical Physics 150, 034102 (2019). URL https://doi.org/10.1063/1.5078399.
  3. [3]Chen, S., Jin, X., Zhao, W. & Li, T. Intricate short-range order in GeSn alloys revealed by atomistic simulations with highly accurate and efficient machine-learning potentials. Physical Review Materials 8, 043805 (2024). URL https://link.aps.org/doi/10.1103/PhysRevMaterials.8.043805.DOI
  4. [4]Bellaiche, L. & Zunger, A. Effects of atomic short-range order on the electronic and optical properties of GaAsN, GaInN, and GaInAs alloys. Physical Review B 57, 4425–4431 (1998). URL https://link.aps.org/doi/10.1103/PhysRevB.57.4425.DOI
  5. [5]Liu, J., Fernandez-Serra, M. V. & Allen, P. B. Special quasiordered structures: Role of short-range order in the semiconductor alloy (GaN)₁−ₓ(ZnO)ₓ. Physical Review B 93, 054207 (2016). URL https://link.aps.org/doi/10.1103/PhysRevB.93.054207.DOI
  6. [6]Singh, P., Smirnov, A. V. & Johnson, D. D. Atomic short-range order and incipient long-range order in high-entropy alloys. Physical Review B 91, 224204 (2015). URL https://link.aps.org/doi/10.1103/PhysRevB.91.224204.DOI
  7. [7]Pekin, T. C., Gammer, C., Ciston, J., Ophus, C. & Minor, A. M. In situ nanobeam electron diffraction strain mapping of planar slip in stainless steel. Scripta Materialia 146, 87–90 (2018). URL https://www.sciencedirect.com/science/article/pii/S1359646217306462.DOI
  8. [8]Li, Y. et al. Quantitative three-dimensional imaging of chemical short-range order via machine learning enhanced atom probe tomography. Nature Communications 14, 7410 (2023). URL https://www.nature.com/articles/s41467-023-43314-y.DOI
  9. [9]He, M., Davids, W. J., Breen, A. J. & Ringer, S. P. Quantifying short-range order using atom probe tomography. Nature Materials 23, 1200–1207 (2024). URL https://www.nature.com/articles/s41563-024-01912-1.DOI
  10. [10]Vogl, L. M. et al. Identification of short-range ordering motifs in semiconductors. Science 389, 1342–1346 (2025). URL https://www.science.org/doi/10.1126/science.adu0719.DOI
  11. [11]Shimura, Y. et al. EXAFS study of local structure contributing to ss stability in SiᵧGe₁−ᵧ−zSnz. Materials Science in Semiconductor Processing 70, 133–138 (2017). URL https://www.sciencedirect.com/science/article/pii/S1369800116305182.
  12. [12]Robouch, B. V., Valeev, R. G., Kisiel, A. & Marcelli, A. Atomic distributions observed in group IV-IV binary tetrahedron alloys: A revised analysis of SiGe and GeSn compounds. Journal of Alloys and Compounds 831, 154743 (2020). URL https://www.sciencedirect.com/science/article/pii/S0925838820311063.DOI
  13. [13]Lentz, J. Z. et al. Local ordering in ge/GeSn semiconductor alloy core/shell nanowires revealed by extended x-ray absorption fine structure (EXAFS). Ap- plied Physics Letters 122, 062103 (2023). URL https://doi.org/10.1063/5.0136746.
  14. [14]Lentz, J. Z. et al. Composition dependence of atomic order in strain-relaxed, metastable GeSn alloys. Journal of Applied Physics 137, 145701 (2025). URL https://doi.org/10.1063/5.0245736.
  15. [15]Gougam, S. et al. EXAFS sheds light on short-range ordering in Ge1−xSnx heteroepitaxial layers grown by MBE and CVD. Physical Review Materials 9, 064601 (2025). URL https://link.aps.org/doi/10.1103/PhysRevMaterials.9.064601.DOI
  16. [16]Kelzenberg, M. D. et al. Enhanced absorption and carrier collection in Si wire arrays for photovoltaic applications. Nature Materials 9, 239–244 (2010). URL https://www.nature.com/articles/nmat2635.DOI
  17. [17]Soref, R. Enabling 2 μm communications. Nature Photonics 9, 358–359 (2015). URL https://www.nature.com/articles/nphoton.2015.87.DOI
  18. [18]Moutanabbir, O. et al. Monolithic infrared silicon photonics: The rise of (Si)GeSn semiconductors. Applied Physics Letters 118, 110502 (2021). URL https://doi.org/10.1063/5.0043511.
  19. [19]Assali, S. et al. A Light-Hole Germanium Quantum Well on Silicon. Advanced Materials 34, 2201192 (2022). URL https://onlinelibrary.wiley.com/doi/abs/10.1002/adma.202201192.DOI
  20. [20]Kaul, P. et al. GeSn quantum wells as a platform for spin-resolved hole transport. Communications Materials 6, 216 (2025). URL https://www.nature.com/articles/s43246-025-00934-9.
  21. [21]Braun, M. R. et al. Oxide Decomposition and Sn Surface Segregation on Core/Shell Ge/GeSn Nanowires. ACS Applied Electronic Materials 4, 5406–5412 (2022). URL https://doi.org/10.1021/acsaelm.2c01061.
  22. [22]Cowley, J. M. An Approximate Theory of Order in Alloys. Physical Review 77, 669–675 (1950). URL https://link.aps.org/doi/10.1103/PhysRev.77.669.DOI
  23. [23]Sheriff, K., Cao, Y., Smidt, T. & Freitas, R. Quantifying chemical short-range order in metallic alloys. Proceedings of the National Academy of Sciences 121, e2322962121 (2024). URL https://www.pnas.org/doi/10.1073/pnas.2322962121.DOI
  24. [24]Gupta, S., Magyari-Köpe, B., Nishi, Y. & Saraswat, K. C. Achieving direct band gap in germanium through integration of sn alloying and external strain. Journal of Applied Physics 113, 073707 (2013). URL https://aip.scitation.org/doi/full/10.1063/1.4792649.DOI
  25. [25]Liang, Y. et al. Group IV topological quantum alloy and the role of short-range order: The case of Ge-rich Ge1−xPbx npj Computational Materials 10, 82 (2024). URL https://www.nature.com/articles/s41524-024-01271-0.DOI
  26. [26]Chen, R. et al. Material characterization of high Sn-content, compressively-strained GeSn epitaxial films after rapid thermal processing. Journal of Crystal Growth 365, 29–34 (2013). URL https://www.sciencedirect.com/science/article/pii/S0022024812008913.DOI
  27. [27]Berghuis, W. J. H. W.-J. et al. Excellent surface passivation of germanium by a-si:h/al2o3 stacks. Journal of Applied Physics 130, 135303 (2021). URL https://doi.org/10.1063/5.0064808.
  28. [28]Attiaoui, A. & Moutanabbir, O. Indirect-to-direct band gap transition in relaxed and strained Ge1−xSixSny ternary alloys. Journal of Applied Physics 116, 063712 (2014). URL https://doi.org/10.1063/1.4889926.
  29. [29]Assali, S. et al. Midinfrared Emission and Absorption in Strained and Relaxed Direct-Band-Gap Ge1−xSnx- Semiconductors. Physical Review Applied 15, 024031 (2021). URL https://link.aps.org/doi/10.1103/PhysRevApplied.15.024031.DOI
  30. [30]Meng, A. C. et al. Coupling of coherent misfit strain and composition distributions in core–shell Ge/Ge1−xSnx nanowire light emitters. Materials Today Nano 5, 100026 (2019). URL https://www.sciencedirect.com/science/article/pii/S258884201830155X.DOI
  31. [31]Assali, S. et al. Growth and Optical Properties of Direct Band Gap Ge/Ge0.87Sn0.13 Core/Shell Nanowire Arrays. Nano Letters 17, 1538–1544 (2017). URL https://doi.org/10.1021/acs.nanolett.6b04627.
  32. [32]Gupta, S., Chen, R., Harris, J. S. & Saraswat, K. C. Atomic layer deposition of Al2O3 on germanium-tin (GeSn) and impact of wet chemical surface pre-treatment. Applied Physics Letters 103, 241601 (2013). URL https://doi.org/10.1063/1.4850518.
  33. [33]Pezzoli, F. et al. Ge Crystals on Si Show Their Light. Physical Review Applied 1, 044005 (2014). URL https://link.aps.org/doi/10.1103/PhysRevApplied.1.044005.DOI
  34. [34]Pavesi, L. & Guzzi, M. Photoluminescence of AlxGa1-xAs alloys. Journal of Applied Physics 75, 4779–4842 (1994). URL https://doi.org/10.1063/1.355769.
  35. [35]Viña, L., Logothetidis, S. & Cardona, M. Temperature dependence of the dielectric function of germanium. Physical Review B 30, 1979–1991 (1984). URL https://link.aps.org/doi/10.1103/PhysRevB.30.1979.DOI
  36. [36]Von Den Driesch, N. et al. Direct Bandgap Group IV Epitaxy on Si for Laser Applications. Chemistry of Materials 27, 4693–4702 (2015).DOI
  37. [37]Gallagher, J. D., Senaratne, C. L., Kouvetákis, J. & Menéndez, J. Compositional dependence of the bowing parameter for the direct and indirect band gaps in Ge1−ySny alloys. Applied Physics Letters 105, 142102 (2014). URL https://doi.org/10.1063/1.4897272.
  38. [38]Wirths, S. et al. Lasing in direct-bandgap GeSn alloy grown on si. Nature Photonics 9, 88–92 (2015). URL https://www.nature.com/articles/nphoton.2014.321.DOI
  39. [39]Wang, Y., Meng, A. C., McIntyre, P. C. & Cai, W. Phase-field investigation of the stages in radial growth of core–shell ge/Ge1−xSnx nanowires. Nanoscale 11, 21974–21980 (2019). URL https://pubs.rsc.org/en/content/articlelanding/2019/nr/c9nr07587a.
  40. [40]Seifner, M. S. et al. Epitaxial Ge0.81Sn0.19 Nanowires for Nanoscale Mid-Infrared Emitters. ACS Nano 13, 8047–8054 (2019). URL https://doi.org/10.1021/acsnano.9b02843.
  41. [41]Gerber, M. W. & Kleiman, R. N. Decoupling recombination mechanisms and trap state localization in direct bandgap semiconductors using photoluminescence decay. Journal of Applied Physics 122, 095705 (2017). URL https://doi.org/10.1063/1.5001128.
  42. [42]Timoshenko, J. & Roldan Cuenya, B. In Situ / Operando Electrocatalyst Characterization by X-ray Absorption Spectroscopy. Chemical Reviews 121, 882–961 (2021). URL https://pubs.acs.org/doi/10.1021/acs.chemrev.0c00396.DOI
  43. [43]Rehr, J. J. & Albers, R. C. Theoretical approaches to x-ray absorption fine structure. Reviews of Modern Physics 72, 621–654 (2000). URL https://link.aps.org/doi/ [44] Woicik, J. C. et al. Lattice vibrations and energy landscape of the isoelectronic semiconductor series CuBr, ZnSe, GaAs, and Ge: The special case of CuBr and its d -level chemistry. Physical Review B 108, 195202 (2023). URL https://link.aps.org/doi/10.1103/PhysRevB.108.195202.
  44. [45]Jin, X., Chen, S., Lemkan, C. & Li, T. Role of local atomic short-range order distribution in alloys: Why it matters in Si-Ge-Sn alloys. Physical Review Materials 7, L111601 (2023). URL https://link.aps.org/doi/10.1103/PhysRevMaterials.7.L111601.DOI
  45. [46]Jin, X., Chen, S. & Li, T. Coexistence of two types of short-range order in Si–Ge–Sn medium-entropy alloys. Communications Materials 3, 1–9 (2022). URL https://www.nature.com/articles/s43246-022-00289-5.DOI
  46. [47]Monkhorst, H. J. & Pack, J. D. Special points for Brillouin-zone integrations. Physical Review B 13, 5188 (1976).DOI
  47. [48]Tran, F. & Blaha, P. Accurate Band Gaps of Semiconductors and Insulators with a Semilocal Exchange-Correlation Potential. Phys. Rev. Lett. 102, 226401 (2009). URL https://link.aps.org/doi/10.1103/PhysRevLett.102.226401.DOI
  48. [49]Kresse, G. & Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Physical Review B 54, 11169–11186 (1996). URL https://link.aps.org/doi/10.1103/PhysRevB.54.11169.DOI
  49. [50]Eckhardt, C., Hummer, K. & Kresse, G. Indirect-to-direct gap transition in strained and unstrained snₓge₁−ₓ alloys. Phys. Rev. B 89, 165201 (2014). URL https://link.aps.org/doi/10.1103/PhysRevB.89.165201.
  50. [51]Polak, M. P., Scharoch, P. & Kudrawiec, R. The electronic band structure of Ge₁−ₓSnₓ in the full composition range: Indirect, direct, and inverted gaps regimes, band offsets, and the Burstein–Moss effect. Journal of Physics D: Applied Physics 50, 195103 (2017). URL https://dx.doi.org/10.1088/1361-6463/aa67bf.
  51. [52]Popescu, V. & Zunger, A. Effective Band Structure of Random Alloys. Physical Review Letters 104, 236403 (2010). URL https://link.aps.org/doi/10.1103/PhysRevLett.104.236403.
  52. [53]Rubel, O., Bokhanich, A., Ahmed, S. J. & Assmann, E. Unfolding the band structure of disordered solids: From bound states to high-mobility Kane fermions. Physical Review B 90, 115202 (2014). URL https://link.aps.org/doi/10.1103/PhysRevB.90.115202.DOI
  53. [54]Meng, A. C. et al. Core-Shell Germanium/Germanium–Tin Nanowires Exhibiting Room-Temperature Direct- and Indirect-Gap Photoluminescence. Nano Letters 16, 7521–7529 (2016). URL https://doi.org/10.1021/acs.nanolett.6b03316.
  54. [55]Meng, A. C. et al. Growth mode control for direct-gap core/shell ge/GeSn nanowire light emission. Materials Today 40, 101–113 (2020). URL https://www.sciencedirect.com/science/article/pii/S136970212030208X.DOI
  55. [56]Assali, S. et al. Kinetic Control of Morphology and Composition in ge/GeSn Core/Shell Nanowires. ACS Nano 14, 2445–2455 (2020). URL https://doi.org/10.1021/acsnano.9b09929.
  56. [57]Gencarelli, F. et al. Crystalline Properties and Strain Relaxation Mechanism of CVD Grown GeSn. ECS Journal of Solid State Science and Technology 2, P134 (2013). URL https://iopscience.iop.org/article/10.1149/2.011304jss/meta.
  57. [58]Meng, A. C. et al. Bending and precipitate formation mechanisms in epitaxial Ge-core/GeSn-shell nanowires. Nanoscale 13, 17547–17555 (2021). URL https://pubs.rsc.org/en/content/articlelanding/2021/nr/d1nr04220c.DOI
  58. [59]Ravel, B. & Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: Data analysis for X-ray absorption spectroscopy using IFEFFIT. Journal of Synchrotron Radiation 12, 537–541 (2005). URL //scripts.iucr.org/cgi-bin/paper?ph5155.pubmed.ncbi.nlm.nih.gov/15968136
  59. [60]Shirley, E. L. & Woicik, J. C. Revisiting the K-edge X-ray absorption fine structure of Si, Ge–Si alloys, and the isoelectronic series: CuBr, ZnSe, GaAs, and Ge. Physical Chemistry Chemical Physics 24, 20742–20759 (2022). URL https://pubs.rsc.org/en/content/articlelanding/2022/cp/d2cp00912a.DOI
  60. [61]Schubert, E. F., Göbel, E. O., Horikoshi, Y., Ploog, K. & Queisser, H. J. Alloy broadening in photoluminescence spectra of AlₓGa₁−ₓAs. Physical Review B 30, 813–820 (1984). URL https://link.aps.org/doi/10.1103/PhysRevB.30.813.DOI
  61. [62]Ouadjaout, D. & Marfaing, Y. Localized excitons in II-VI semiconductor alloys: Density-of-states model and photoluminescence line-shape analysis. Physical Review B 41, 12096–12105 (1990). URL https://link.aps.org/doi/10.1103/PhysRevB.41.12096.

Paper details

vlnc:W_takczsts
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents