---
title: Short-Period ScxAl1-xN/GaN Superlattices
url: https://www.emergentmind.com/topics/short-period-scxal1-xn-gan-superlattices
type: topic
---

# Short-Period ScxAl1-xN/GaN Superlattices

Searching arXiv for the cited paper and closely related ScAlN/GaN superlattice work to ground the article.
{"query":"arXiv:2507.08951 OR \"Structural optimization of lattice-matched Sc0.14Al0.86N/GaN superlattices for photonic applications\"","max_results":5,"sort_by":"submittedDate","sort_order":"descending"}
{"query":"ScAlN GaN superlattice photonic applications ScxAl1-xN/GaN arXiv","max_results":10,"sort_by":"relevance","sort_order":"descending"}
Short-period Sc$_x$Al$_{1-x}$N/GaN superlattices are III-nitride heterostructures in which alternating Sc$_x$Al$_{1-x}$N and GaN layers are repeated on the nanometer scale to form a periodic wide-bandgap stack. In the reported molecular-beam epitaxy study, the system is examined as a lattice-matched platform for photonic applications, with particular emphasis on the condition $x=0.14$ in Sc$_x$Al$_{1-x}$N, where reciprocal-space mapping confirms in-plane matching to GaN and coherent growth through total thicknesses up to approximately $600\,\mathrm{nm}$ [2507.08951]. The same study identifies growth temperature, well thickness, and interface intermixing as the main structural variables governing coherence, defect formation, and the effective band-structure profile along the growth direction.

## 1. Superlattice architecture and design space

The reported superlattices consist of $N=50$ repeats of a Sc$_x$Al$_{1-x}$N barrier layer with thickness $t_B=6\,\mathrm{nm}$ and a GaN well layer with thickness $t_W$. Two well-thickness regimes are distinguished: “thick”-well superlattices with $t_W=6\,\mathrm{nm}$ and “ultra-thin”-well superlattices with $t_W=2\,\mathrm{nm}$ [2507.08951]. The corresponding nominal periods are

$$
d_{SL}=t_B+t_W=
\begin{cases}
12\,\mathrm{nm} & \text{(thick-well)}\\
8\,\mathrm{nm} & \text{(ultra-thin well)}
\end{cases}
$$

and the overall film thickness is

$$
H=N\cdot d_{SL}\simeq 600\,\mathrm{nm} \text{ (12 nm period)} \quad \text{or} \quad 400\,\mathrm{nm} \text{ (8 nm period)}.
$$

| Superlattice type | Nominal period $d_{SL}$ | Overall thickness $H$ |
|---|---:|---:|
| Thick-well ($t_W=6\,\mathrm{nm}$) | $12\,\mathrm{nm}$ | $\simeq 600\,\mathrm{nm}$ |
| Ultra-thin-well ($t_W=2\,\mathrm{nm}$) | $8\,\mathrm{nm}$ | $\simeq 400\,\mathrm{nm}$ |

Within this geometry, the stated objective is not merely periodic layering, but the realization of heterostructures that are simultaneously lattice-matched and capable of strong piezoelectric or intersubband optical functionality. The barrier/well terminology is structural as well as functional: Sc$_x$Al$_{1-x}$N is treated as the barrier constituent and GaN as the well constituent, with the resulting periodic composition profile intended for photonic and piezoelectric device concepts.

## 2. Lattice matching, Vegard interpolation, and strain sign

The Sc composition was varied from $x=0.12$ to $x=0.18$ to identify the lattice-matching point with GaN. Within a linear-interpolation, or Vegard’s, approximation, the wurtzite in-plane lattice constant of Sc$_x$Al$_{1-x}$N is written as [2507.08951]

$$
a(x)=x\cdot a_{\mathrm{ScN}}+(1-x)\cdot a_{\mathrm{AlN}},
$$

with $a_{\mathrm{AlN}}\approx 3.112\,\text{\AA}$ and $a_{\mathrm{ScN}}$ taken from prior ScAlN calibration, described as approximately $4.5\,\text{\AA}$ in the hypothetical rock-salt phase extrapolation but adjusted by experiment for the wurtzite-derived lattice. From high-resolution X-ray diffraction and Rutherford backscattering in prior studies, the condition $x=0.14\pm0.01$ gives an in-plane lattice constant coincident with the GaN template lattice constant $a_{\mathrm{GaN}}\approx 3.189\,\text{\AA}$.

The in-plane lattice mismatch is defined as

$$
\delta=\frac{a_{SL}-a_{\mathrm{GaN}}}{a_{\mathrm{GaN}}},
$$

where $a_{SL}$ is the average in-plane lattice constant of the superlattice and, for pseudomorphic growth, equals $a(x)$. Accordingly,

$$
\delta(x)=\frac{a(x)-a_{\mathrm{GaN}}}{a_{\mathrm{GaN}}},
$$

so that $\delta=0$ when $x\approx 0.14$. For $x<0.14$, $\delta<0$, leading to tensile strain in the ScAlN layers; for $x>0.14$, $\delta>0$, producing compressive strain. This composition-dependent sign change organizes the subsequent defect phenomenology: under-tuned Sc content produces tensile-driven cracking, whereas over-tuned Sc content promotes compressive relaxation mechanisms.

## 3. Molecular-beam epitaxy and temperature optimization

All samples were grown by plasma-assisted molecular-beam epitaxy on Fe-doped semi-insulating c-plane GaN/sapphire templates. The process sequence included a $150\,\mathrm{nm}$ GaN buffer grown at $740\,^\circ\mathrm{C}$, followed by superlattice growth using N-flux from a $250\,\mathrm{W}$ RF source at $0.25\,\mathrm{sccm}$ N$_2$, Ga-rich GaN growth at approximately $3.6\,\mathrm{nm/min}$, and metal-poor ScAlN growth at approximately $2.6\,\mathrm{nm/min}$ [2507.08951].

The reported optimum substrate temperature depends on the GaN well thickness. For superlattices with $6\,\mathrm{nm}$ GaN wells, the optimal $T_{\mathrm{sub}}$ is approximately $600\,^\circ\mathrm{C}$. For superlattices with ultra-thin wells of $\leq 2\,\mathrm{nm}$, $T_{\mathrm{sub}}$ must be lowered to approximately $550\,^\circ\mathrm{C}$. The temperature window is constrained on both sides. At $T_{\mathrm{sub}}\lesssim 550\,^\circ\mathrm{C}$, HAADF-STEM image analysis shows interface widths of approximately $1.5\,\mathrm{nm}$, corresponding to about $5$ monolayers, whereas at $600\,^\circ\mathrm{C}$ the same interfaces broaden to approximately $1.7$–$2.0\,\mathrm{nm}$, or $6$–$7$ monolayers, due to enhanced intermixing. Conversely, $T_{\mathrm{sub}}$ below approximately $500\,^\circ\mathrm{C}$ degrades overall crystallinity, as indicated by broad XRD peaks and absence of atomic steps in AFM, while $T_{\mathrm{sub}}$ above approximately $625\,^\circ\mathrm{C}$ induces ScAlN surface roughening.

The optimization problem is therefore explicitly multivariate. High temperature improves some aspects of epitaxy but broadens interfaces; low temperature sharpens interfaces but eventually compromises crystallinity; and the acceptable compromise shifts with well thickness. For ultra-thin wells, where a few monolayers materially alter confinement and composition gradients, the lower optimum temperature is structurally significant.

## 4. Reciprocal-space mapping and coherent growth

High-resolution X-ray diffraction reciprocal-space maps of the asymmetric $(10\ 1\ 5)$ reflection provide the principal evidence for lattice matching and coherence. For $x=0.14$, the superlattice satellite streaks are strictly aligned along $Q_x$ with the GaN template peak, confirming $\delta\approx 0$ and full in-plane coherence through $H\approx 600\,\mathrm{nm}$ [2507.08951]. The abstract further states that this lattice matching is observed regardless of the thickness of the GaN interlayers, as evidenced by symmetric superlattice satellites aligned in-plane with the underlying substrate peak.

The non-matched compositions show distinct reciprocal-space signatures. At $x=0.12$, the satellites shift to larger $Q_x$, corresponding to the tensile-strained regime, and crack formation is observed by optical microscopy. At $x=0.18$, the $6\,\mathrm{nm}$-well superlattice remains coherent, with strain partitioning by thick GaN mitigating relaxation, but superlattices with $2\,\mathrm{nm}$ wells show two satellite sets, one coherent and one relaxed. The period $d_{SL}$ can be extracted from the angular spacing $\Delta\theta$ between symmetric $(0002)$ fringes through Bragg’s law,

$$
n\lambda = 2d_{SL}\sin\theta_n,
$$

where $n$ is the fringe order and $\lambda$ the Cu K$\alpha$ wavelength.

A plausible implication is that composition alone does not determine whether a Sc$_x$Al$_{1-x}$N/GaN superlattice remains coherent. The reported contrast between $x=0.18$ structures with $6\,\mathrm{nm}$ and $2\,\mathrm{nm}$ wells suggests that relative layer thicknesses participate directly in strain partitioning and relaxation behavior.

## 5. Interface chemistry, intermixing, and delayed Sc incorporation

Cross-sectional HAADF-STEM combined with STEM-EDX mapping of Ga, Al, and Sc reveals significant atomic-scale intermixing at each interface, and the study identifies this temperature-dependent intermixing as a major factor in setting the nitride composition variation and implicitly the band-structure profile along the growth direction [2507.08951]. At $600\,^\circ\mathrm{C}$, the bottom GaN/ScAlN interface width is approximately $1.7\,\mathrm{nm}$, whereas the top interface is approximately $2.0\,\mathrm{nm}$. At lower temperature, specifically around $550\,^\circ\mathrm{C}$, the Sc-incorporation delay is reduced and the interfaces are chemically sharper.

The EDX line profiles show that the Al/Ga half-intensity points coincide, thereby defining the “chemical” interface, but the Sc onset and termination are offset: the delay is approximately $4$ monolayers at the top of ScAlN and approximately $3$ monolayers at the bottom. Lowering $T_{\mathrm{sub}}$ to $550\,^\circ\mathrm{C}$ reduces this Sc-incorporation delay to less than $1$ monolayer, yielding sharper three-metal profiles. The study identifies this sharpening as critical for ultra-thin quantum-well band-edge definition.

These measurements qualify an assumption often made in idealized superlattice design. The reported data suggest that nominal barrier and well thicknesses do not by themselves define the realized composition profile; the actual profile depends strongly on interface broadening and on the fact that Sc incorporation exhibits delays relative to Al at both onset and termination. In short-period and especially ultra-thin structures, this distinction is not cosmetic, because the chemical interface and the Sc profile are not identical.

## 6. Defect formation, critical thickness, and functional consequences

When $x$ deviates from $0.14$, the in-plane mismatch accumulates and eventually exceeds the critical thickness for coherent growth. For $x<0.14$, tensile strain is relieved by crack formation, observed optically for $x=0.12$, consistent with classic fracture-based relaxation. For $x>0.14$, compressive strain tends to relax via misfit dislocations once the Matthews–Blakeslee critical thickness is exceeded, expressed as [2507.08951]

$$
h_c \propto |f|^{-1}\ln(h_c/b),
$$

where $f=\delta$ is the misfit and $b$ the Burgers vector. In thin-well superlattices, the lower $h_c$ leads to partial plastic relaxation, evident in broadened and split superlattice satellites in reciprocal-space maps as well as alternating strain fields mapped by geometric phase analysis in STEM.

Although the focus of the study is structural, the reported photonic measurements directly connect structural optimization to device-relevant response. Infrared intersubband absorption measurements show that the lattice-matched $x=0.14$ superlattice exhibits four-times larger total absorption and a sharper resonance at $E_{\max}\approx 645.6\,\mathrm{meV}$ with $\mathrm{FWHM}\approx 62\,\mathrm{meV}$, compared with the partially relaxed $x=0.18$ superlattice, which shows $E_{\max}\approx 673.0\,\mathrm{meV}$ and $\mathrm{FWHM}\approx 78\,\mathrm{meV}$ [2507.08951]. The same structural quality is presented as a basis for exploiting the high piezoelectric coefficient $e_{33}$ of Sc$_x$Al$_{1-x}$N, described in continuum theory by

$$
e_{33}(x)\simeq (1-x)e_{33}^{\mathrm{AlN}} + x e_{33}^{\mathrm{ScN}},
$$

and stated to exceed that of pure AlN by over $200\%$.

Taken together, the reported results define the governing variables of the system as the precise choice of $x$ to achieve $\delta=0$ at $x\approx 0.14$, the relative layer thicknesses of $6\,\mathrm{nm}$ barriers with either $6\,\mathrm{nm}$ or $2\,\mathrm{nm}$ wells, and the substrate temperature of approximately $600\,^\circ\mathrm{C}$ for thick-well superlattices or approximately $550\,^\circ\mathrm{C}$ for ultra-thin wells. This suggests that short-period Sc$_x$Al$_{1-x}$N/GaN superlattices are best understood not as a single lattice-matched composition problem, but as a coupled problem of mismatch, thermal intermixing, and thickness-dependent relaxation.

Source: https://www.emergentmind.com/topics/short-period-scxal1-xn-gan-superlattices