Effect of Concentration Fluctuations on Material Properties of Disordered Alloys

This paper demonstrates that standard Special Quasi-random Structure (SQS) calculations can significantly underestimate the bandgaps of disordered alloys due to wavefunction localization in rare minority configurations, and proposes a density-of-states fitting (DOSF) method to extract bandgaps from majority configurations to resolve the long-standing discrepancy between theoretical predictions and experimental results.

The Gist

The Big Picture: Mixing the Perfect Smoothie

Imagine you are trying to make the perfect fruit smoothie by mixing two types of fruit: Apples (A) and Bananas (B). You want a smoothie that is exactly 50% apple and 50% banana.

In the world of materials science, scientists do this with atoms to create alloys (like mixing Zinc and Tin in a crystal). They want to predict how this "smoothie" will behave—specifically, how much energy it takes to make it conduct electricity (this is called the bandgap).

For a long time, scientists had a tool called SQS (Special Quasi-Random Structure). Think of SQS as a blender that creates a small cup of smoothie that looks perfectly mixed on average. If you take a spoonful, it tastes like 50/50. This tool worked great for predicting things like the weight or volume of the smoothie.

The Problem: The "Bad Apple" in the Cup

However, when scientists tried to predict the electronic properties (the bandgap) using this tool, they hit a wall.

Here is the issue: Even if your smoothie is 50/50 on average, randomness means that sometimes, by pure chance, you might get a spoonful that is 100% apple or 100% banana.

• In a tiny cup (a small computer model), you rarely get these extreme spoonfuls.
• But as you make the cup bigger (to get a more accurate model), you start to find these rare, extreme clusters.

The Analogy:
Imagine you are looking for the "deepest point" in a swimming pool to measure its depth.

• The Standard Method: You drop a probe and measure the distance to the very bottom of the deepest hole.
• The Reality: In a disordered alloy, the "bottom" isn't a smooth floor. Because of random mixing, there are tiny, deep, isolated holes (caused by rare clusters of atoms) that look like deep pits.
• The Mistake: The computer sees these tiny, deep pits and says, "Aha! The pool is incredibly deep!" But in reality, those pits are just rare, isolated defects. The actual water level (the property that matters for the material) is much higher.

As the computer model got bigger, it found more of these "deep pits" (rare atomic clusters). This caused the calculated "depth" (bandgap) to shrink and eventually disappear, even though real-world experiments showed the material still had a healthy, usable bandgap.

The Solution: Ignoring the "Outliers"

The authors of this paper realized that the standard way of calculating the bandgap was flawed because it was too sensitive to these rare, weird clusters.

They proposed a new method called DOSF (Density-of-States Fitting).

The Analogy:
Instead of looking for the single deepest point in the pool, imagine you are trying to determine the "average depth" of the pool floor.

1. You ignore the tiny, isolated holes (the rare defects).
2. You look at the shape of the floor where the majority of the water is.
3. You draw a smooth line through the main floor to see where it naturally starts and ends.

By "fitting" the data to the shape of the majority, they filtered out the noise caused by the rare, defect-like clusters.

The Results

When they applied this new "smoothing" method to their computer models:

1. Stability: The calculated bandgap stopped shrinking as the model got bigger. It settled at a stable number (about 1.0 eV).
2. Agreement: This new number matched real-world experiments much better than the old method.
3. Versatility: They also showed that this method works for materials that are partially mixed (not 100% random), which is how most real materials are made.

Why This Matters

Before this paper, there was a huge disagreement between what computers predicted and what scientists measured in the lab. It was like the computer saying, "This material is a perfect insulator," while the lab said, "No, it conducts electricity just fine."

This paper fixes that disagreement. It teaches us that when studying messy, disordered materials, we shouldn't get distracted by the rare, weird outliers. Instead, we should focus on the "average" behavior of the majority. This opens the door to designing better solar cells, LEDs, and computer chips using disordered alloys.

Summary in One Sentence

The paper fixes a long-standing error in computer simulations of mixed materials by teaching the computer to ignore rare, weird atomic clusters and focus on the "average" behavior, finally making the math match the real-world experiments.

Technical Summary

1. Problem Statement

Disordered semiconductor alloys (e.g., $A_{1-x}B_xX$) are crucial for tuning material properties, but their theoretical modeling is hindered by the lack of periodicity. The Special Quasi-Random Structure (SQS) method is the standard approach, using finite supercells to mimic the averaged atomic correlation functions of an ideal random alloy.

However, a significant discrepancy exists between theoretical calculations and experimental measurements regarding the bandgap of these alloys:

• The Issue: As SQS supercell sizes increase to better model randomness, the calculated bandgap (defined conventionally as the energy difference between the Lowest Unoccupied State, $LUS$, and Highest Occupied State, $HOS$) often decreases drastically, sometimes vanishing entirely.
• The Cause: In larger supercells, statistically rare local concentration fluctuations occur (e.g., regions that are nearly pure $A$ or pure $B$). These "minority configurations" act as defects, introducing localized states near the band edges. These states cause wavefunction localization, artificially lowering the $LUS-HOS$ gap.
• The Conflict: Experimental measurements reflect the electronic states of the majority atomic environments, not the rare defect-like tails. Consequently, standard SQS calculations fail to reproduce experimental bandgaps, leading to a long-standing controversy in computational materials science.

2. Methodology

The authors propose a new computational framework to resolve this discrepancy, moving away from the conventional $E_{LUS-HOS}$ definition toward a Density-of-States Fitting (DOSF) approach.

• System Studied: The paper uses the random alloy $Zn_{0.5}Sn_{0.5}P$ as a case study, comparing 64-atom and 512-atom SQS supercells.
• Computational Tools:
  • Structural relaxation and electronic structure calculations were performed using Density Functional Theory (DFT) via the VASP code.
  • Functionals used: SCAN for geometry and HSE06 for accurate electronic structure.
  • Models included standard SQS (random) and Special Quasi-Disorder Structures (SQDS) to model partially disordered alloys with varying long-range order parameters ($\eta$).
• The DOSF Approach:
  • Instead of taking the raw energy difference between $HOS$ and $LUS$, the authors fit the calculated Density of States (DOS) near the band edges to a theoretical function that accounts for non-parabolicity.
  • The dispersion relation is modeled as $\frac{\hbar^2 k^2}{2m^*_d} = \gamma(E) = E + c_1E^2 + c_2E^3 + \dots$.
  • The DOS is derived as $N(E) \propto \gamma^{1/2}(d\gamma/dE)$.
  • By fitting the DOS ($N^2(E)$) near the Fermi level and extrapolating to where $N^2(E) = 0$, the true bandgap ($E_{g,DOS}$) is determined. This method effectively filters out the localized tail states caused by rare minority motifs.

3. Key Contributions

1. Identification of the Root Cause: The paper demonstrates that the poor convergence of the bandgap in large SQS cells is not a failure of the SQS method itself, but a consequence of the conventional definition of the bandgap ($E_{LUS-HOS}$) being dominated by rare, defect-like minority configurations.
2. Proposal of DOSF Method: The authors introduce the DOSF method as a robust way to define the bandgap of disordered alloys. This aligns the theoretical definition with experimental reality by focusing on the majority configurations and filtering out artificial band tails.
3. Resolution of the Convergence Paradox: The study shows that while $E_{LUS-HOS}$ decreases with supercell size (due to increased probability of rare motifs), the $E_{g,DOS}$ converges rapidly to a stable value, independent of supercell size.
4. Extension to Partially Disordered Alloys: The method is successfully applied to alloys with varying degrees of long-range order ($\eta$), showing that the DOSF bandgap follows a predictable $\eta^2$ trend between ordered and random limits.

4. Key Results

• Supercell Size Dependence:
  • In a 64-atom supercell, the conventional gap ($E_{LUS-HOS}$) is 0.63 eV.
  • In a 512-atom supercell, the conventional gap drops to 0.10 eV due to the appearance of extreme local motifs (e.g., $Zn_{10}Sn_2$ or $Zn_1Sn_{11}$).
  • However, using the DOSF method, the bandgap converges to ~1.0 eV for both cell sizes, showing size independence.
• Comparison with Experiment:
  • Experimental bandgaps for disordered $Zn_{0.5}Sn_{0.5}P$ are reported between 1.2 eV and 1.5 eV.
  • Previous theoretical calculations (using $E_{LUS-HOS}$) yielded values between 0 and 0.7 eV, creating a massive discrepancy.
  • The new DOSF result (~1.0 eV) is significantly closer to experimental values and resolves the underestimation issue.
• Partially Disordered Alloys:
  • For alloys with a long-range order parameter $\eta = 0.5$, the calculated DOSF bandgap is 1.22 eV, which agrees well with the experimental value of 1.39 eV.
  • The results confirm that the DOSF method correctly captures the transition from ordered to disordered states, whereas the $E_{LUS-HOS}$ method fails due to the dominance of localized states.

5. Significance

This work addresses a fundamental challenge in the computational modeling of disordered materials. By distinguishing between statistical quantities (like total energy and volume, which are unaffected by rare motifs) and non-statistical quantities (like bandgaps, which are sensitive to local wavefunction localization), the authors provide a corrected framework for electronic structure calculations.

• Theoretical Impact: It resolves the long-standing disagreement between theory and experiment for semiconductor alloy bandgaps.
• Practical Application: The DOSF method allows researchers to use larger, more accurate supercells without fear of artificial bandgap closure.
• Broader Scope: The authors suggest this approach is applicable to any material property dependent on localized wavefunctions, offering a pathway for the accurate design of advanced disordered semiconductors for technological applications.