Modeling neutral defects in III-V ternary alloys with a special quasirandom structure: Analysis of As- and III-site point defects in InGaAs

While first-principles density functional theory modeling has become a vital tool to investigate defect properties in semiconductors, the lack of crystalline periodicity in pseudobinary random composition alloys, such as <a:math xmlns:a="http://www.w3.org/1998/Math/MathML"><a:mrow><a:msub><a:mi>In</a:mi><a:mrow><a:mn>1</a:mn><a:mo>−</a:mo><a:mi>x</a:mi></a:mrow></a:msub><a:msub><a:mi>Ga</a:mi><a:mi>x</a:mi></a:msub><a:mi>As</a:mi></a:mrow></a:math>, complicates such analyses. We present a simulation strategy to systematically take into account the variability in the local defect environment in order to predict statistical properties of neutral intrinsic defects in <b:math xmlns:b="http://www.w3.org/1998/Math/MathML"><b:mrow><b:msub><b:mi>In</b:mi><b:mrow><b:mn>1</b:mn><b:mo>−</b:mo><b:mi>x</b:mi></b:mrow></b:msub><b:msub><b:mi>Ga</b:mi><b:mi>x</b:mi></b:msub><b:mi>As</b:mi></b:mrow></b:math>. We use a comprehensive sampling from a modest-sized 64-atom special quasirandom structure (SQS) to define a statistically representative set of defects, and use a 512-atom hypercell, a <c:math xmlns:c="http://www.w3.org/1998/Math/MathML"><c:mrow><c:mn>2</c:mn><c:mo>×</c:mo><c:mn>2</c:mn><c:mo>×</c:mo><c:mn>2</c:mn></c:mrow></c:math> supercell of SQS supercells, to achieve cell-size convergence. We articulate an equivalent site principle and describe how it constrains atomic chemical reference energies in computation of defect formation energies in pseudobinary alloys. A simple protocol for estimating reference energies for the Ga and In atoms sharing the III site succeeds in obtaining the equivalence of defects at Ga-sites and In sites in the SQS supercell, (&lt;30 meV differences in average formation energies). For III-site defects, such as the As antisite <d:math xmlns:d="http://www.w3.org/1998/Math/MathML"><d:msub><d:mi>As</d:mi><d:mtext>III</d:mtext></d:msub></d:math>, the statistical variability in formation energies is modest, <e:math xmlns:e="http://www.w3.org/1998/Math/MathML"><e:mrow><e:mo>≈</e:mo><e:mn>0.1</e:mn><e:mo>–</e:mo><e:mn>0.2</e:mn><e:mspace width="0.16em"/><e:mi>eV</e:mi></e:mrow></e:math>. The variability in formation energy at As-site defects, such as the As vacancy <g:math xmlns:g="http://www.w3.org/1998/Math/MathML"><g:msub><g:mi>v</g:mi><g:mtext>As</g:mtext></g:msub></g:math>, can be much larger, &gt;1 eV. The As antisite is shown to be a low-energy defect and the most likely to be present in as-grown materials, just as in GaAs. All other defects are higher-energy defects unlikely to be important in native material, but potentially important in radiation-damaged material. With a strong variability in defect energies, especially on the As-site, explicit consideration of statistical variability due to compositional randomness will be imperative for meaningful and quantitative comparisons to experiment.