Silicon band gap modeling becomes more demanding as temperature and charge-carrier density rise. These variables are linked: a narrowing band gap can increase the population of electrons and holes, while high carrier concentrations can further affect the gap. For engineers working with semiconductor thermal processes, laser-material interactions, or high-temperature models, capturing that feedback matters. A model based only on classical Maxwell-Boltzmann statistics may not adequately describe silicon once its electron gas begins to approach degeneracy.
Why the silicon band gap matters in thermal models
The band gap is the energy separation between the valence band and the conduction band. Electrons that gain sufficient energy can move into the conduction band, leaving holes in the valence band. Both populations contribute to electrical behavior, so the band gap helps determine the number of available charge carriers.
In silicon, increasing temperature narrows the band gap and raises the carrier concentration. At sufficiently high concentrations, the electron and hole populations no longer behave as a nondegenerate gas. The Fermi level—the energy that describes the distribution of occupied states—then becomes important to calculating carrier density. Because the gap, Fermi level, and carrier concentration influence one another, they must be treated as coupled quantities in the model.
This is relevant to high-temperature analysis because experimental measurements of silicon’s band gap cover a more limited range than may be needed for studying conditions near melting. The modeling work considered intrinsic silicon in thermodynamic equilibrium, with equal electron and hole concentrations and electroneutrality maintained.
Why Fermi-Dirac statistics are needed
The Fermi-Dirac distribution describes the probability that an energy state is occupied by an electron. At low carrier concentrations, it approaches the Maxwell-Boltzmann approximation, which is simpler to use. At higher concentrations, however, that classical approximation becomes less suitable because it does not fully account for the quantum-statistical behavior of a degenerate carrier gas.
To calculate electron and hole concentrations across both nondegenerate and degenerate conditions, the model uses Fermi-Dirac integrals. These integrals connect the carrier population to the reduced Fermi level and the density of available states in each band. Since the relevant half-order integral cannot generally be evaluated in a simple closed form, the authors use an analytic approximation that supports calculations across a range of degeneracy.
The model also applies the electroneutrality condition to locate the Fermi level. In practice, this requires a numerical solution: an inner iteration determines the Fermi level, while an outer iteration updates the carrier concentration and band gap until the coupled values converge. This approach allows the calculation to transition smoothly between classical and quantum-statistical regimes.
Representing temperature- and carrier-dependent narrowing
The temperature component of the model follows a Varshni-type relationship. This captures the established temperature dependence of the silicon band gap, including the influence of lattice expansion and electron-lattice interactions. A further term represents narrowing associated with collective carrier interactions at high concentrations. In the formulation, that contribution depends on the cube root of carrier density and includes a fitting parameter.
That fitting parameter is consequential: the study evaluates two values, including one selected so that the modeled band gap reaches zero at silicon’s equilibrium melting temperature. This is a modeling assumption, not a general experimental measurement of the gap at melting. The results should therefore be read as predictions under the stated model and parameter choices, rather than as a single definitive band-gap curve.
What the calculations indicate near melting
At lower temperatures, up to approximately the silicon Debye temperature, the calculated gap curves using Fermi-Dirac and Maxwell-Boltzmann treatments are reported to agree with one another and with the cited experimental data. At higher temperatures, the predictions separate as carrier concentration and degeneracy become more significant.
The reported calculations show weak degeneracy emerging around 1,000 K, with differences between the quantum-statistical and classical carrier-density estimates. In the Fermi-Dirac results, strong degeneracy begins around 1,600 K for holes and at a higher temperature for electrons. At the equilibrium melting temperature of 1,687 K, the case using the larger fitting parameter predicts a zero band gap and a carrier concentration of about 1.7 × 1020 cm-3. With a smaller fitting parameter, the modeled gap reaches zero at a higher temperature, around 2,000 K.
By comparison, the Maxwell-Boltzmann calculation keeps the gap positive at the melting point in the reported results. This difference illustrates why the statistical treatment matters when modeling high-temperature silicon: classical and quantum approaches can produce substantially different estimates of both the gap and carrier density once degeneracy develops.
Practical implications for engineering analysis
For engineers developing thermal or process models involving silicon, the central lesson is to check whether the assumed carrier statistics remain appropriate across the full operating range. A model that is adequate at moderate temperatures may not remain reliable as carrier density increases and the Fermi level approaches a band edge.
- Track coupled variables. Temperature, band gap, Fermi level, and carrier concentration should be updated consistently rather than treated as independent inputs.
- Use quantum statistics where needed. Fermi-Dirac integrals provide a way to account for weak and strong degeneracy, while the Maxwell-Boltzmann approximation is more appropriate in the nondegenerate limit.
- Document assumptions. High-temperature predictions depend on the selected narrowing model and its fitted parameter, particularly near melting and beyond.
The study also discusses implications for nonequilibrium laser heating, where carrier populations may rise before the lattice reaches its melting temperature. Those conditions differ from the equilibrium calculations described here, so they require an appropriate nonequilibrium treatment rather than direct use of equilibrium predictions.
Conclusion
Silicon band gap modeling at elevated temperatures requires attention to the interaction between carrier density, band-gap narrowing, and Fermi-level position. Fermi-Dirac statistics offer a framework for representing the transition from nondegenerate to degenerate carrier behavior, while numerical iteration resolves the coupled quantities. For industrial and semiconductor process analysis, the practical priority is to match the statistical model to the conditions being simulated and to make parameter assumptions explicit—especially when extrapolating toward melting or other high-temperature regimes.