Photoluminescence and double-crystal x-ray study of InGaAs/InP: Effect of mismatch strain on band gap

Abstract

Epitaxial layers of InGaAs on InP are the building blocks in optoelectronic device fabrication, where the dependence of the band gap on composition is utilized in device design. The band gap can be determined from the photoluminescence peak energy and composition from lattice size. This work reports a detailed correlation between both the room-temperature (300 K), and low-temperature (7 K) photoluminescence peak energy of epitaxial InGaAs, and the lattice mismatch relative to InP as measured by x-ray double-crystal diffraction. Nominally undoped 1- and 2-μm-thick layers of high quality InGaAs were grown on InP (001) by metalorganic chemical vapor deposition. The relaxed mismatch for these coherent layers was between −0.18% and 0.12%. The observed dependence of the 7-K photoluminescence energy on lattice mismatch confirms the theory of People [Appl. Phys. Lett. 50, 1604 (1987); Phys. Rev. B 32, 1405 (1985)] and Kuo et al. [J. Appl. Phys. 57, 5428 (1985)] which includes the effect of strain on the J= (3)/(2) valence band. The 7 K photoluminescence energy of zero mismatch InGaAs grown on semi-insulating InP substrates ([Fe]=1016 cm−3) was 0.804±0.002 eV and of zero mismatch InGaAs grown on n-type ([S]=2×1019 cm−3) substrates was 0.801±0.002 eV. This difference is attributed to the difference in absolute lattice constant for the two types of substrates. The correlation was extended to room-temperature photo- luminescence where the peak recombination energy depends on the excitation conditions. Simple spectral line-shape analysis showed that the λ 1/2 max (taken from the low-energy side of the peak) was a reliable figure of merit and could be used to estimate the degree of lattice mismatch independent of excitation conditions. This algorithm is applied to the nondestructive mapping of whole wafers.

A convenient band-gap interpolation technique and an improved band line-up model for InGaAlAs on InP

“…The dotted curve is near the median of the data at 300 K, whereas the dash-dot curve shows a better match to the data near 0 K. Therefore, in this work, we had fitted the results in such a way that it passes near the median of the data at both near 0 and 300 K, and the result is shown as the solid curve. The solid curve is within the experimental error of the PL measurement on zero mismatch InGaAs grown on semi-insulating InP substrates at 7 K [25], and the 300 K band-gap energy was fitted to the widely accepted value of 0.75 eV [26,27]. The values of all fitting parameters used are listed in Table 2 for the ease of reference.…”

Fotoluminescência do ingaas/inp crescido pela técnica de epitaxia por feixe molecular

“…Sistemas InGaAs/InP são particularmente atrativos, devido a sua aplicação em optoeletrônica e em dispositivos eletrônicos de alta velocidade (PEARSALL, 1993), sendo materiais básicos em vários dispositivos tais como: fotodetetores (SENGUPTA et al, 1997;WEISS et al, 2000), guia de ondas (WEISS et al, 2000;LEWÉN et al, 2002;DORREN et al, 2000), moduladores (WEISS et al, 2000;DORREN et al, 2000;GEDDO;BELLANI;GUIZZETTI, 1994), conversores termofotovoltáicos (TPV) (GINIGE et al, 2004), "lasers" de grande comprimento de onda (WEISS et al, 2000;BASSIGNANA;MINER;PUETZ, 1989;LOUATI et al, 1987) e "lasers" de cascata quântica (QCL) (TROCCOLI et al, 2003). Em particular, a energia de "bandgap" em torno de 0,75 eV, à temperatura ambiente, torna o InGaAs/InP estratégico em sistemas de comunicação operando entre 1,1 e 1,7 µm (CHEN; KIM, 1981), região que abrange a janela de baixa dispersão (1,3 µm) e a de mínima atenuação (1, 5 µm) das fibras ópticas (MORRAL et al, 2003).…”

Fourier transform-luminescence spectroscopy of semiconductor thin films and devices

“…3. The observation that excitation density at constant excitation power varies as a function of material composition in Ga x In 1-x As also serves to emphasize the importance of utilizing a reproducible method, as in [24], for calculating band energies from room temperature PL spectra.…”

“…The PL band energies are greater by about 20 -25 meV in the absorber layer than the predicted values. Lattice strain in the biaxially compressed LMM layers [18] would be expected to increase the PL band energy [24], requiring modifications to the optical bowing parameters in Eq. (1) [19].…”

“…7. We used the method of Bassignana, et al [24] to estimate the PL band energies as the energy at half of the band maximum intensity (measured on the low-energy side of the bands), rather than the energies of the PL band maxima. This approach minimizes the effect of variations in excitation density on the room-temperature PL measurements.…”