1. Sub-bandgap photoluminescence in InGaAs/InAsP heterostructures lattice-mismatched to InP substrates
J. Peter Campbell and Tim Gfroerer Davidson College, Davidson, NC
Mark Wanlass National Renewable Energy Laboratory
Project supported by Research Corporation and the American Chemical Society – Petroleum Research Fund

2. Previous Work
ENERGY
Increasing Mismatch
Previous work provides indirect evidence that increasing lattice mismatch changes the distribution of defect levels.
Fundamental semiconductor theory
Efficiency curves

3. Motivation Can we find direct evidence for these defect levels?
ENERGY
Increasing Mismatch
Phonons (Heat)
Photons (Light)
Can we find direct evidence for these defect levels?
Sub-bandgap PL spectra may reveal the states.
Mechanisms of recombination (Radiative vs. Nonradiative and PL)

4. Fourier Transform IR Spectroscopy
Interference pattern is OPL-dependent, since there is more than one wavelength present. Centerburst is the result of constructive interference of all wavelengths. Sine wave = single wavelength.
Fourier transform separates a waveform into a sum of different frequency sinusoids. Each point of the spectrum is essentially the amplitude of the sinusoid at that frequency.
Vs. scanning monochromator
Advantages: All energy hits sample (increased S:N), Better resolution (proportional to opl)
Fourier
Transform
Interferogram
Spectrum

5. Sample Structure
Bandgap Series: Increased [In] in active layer increases lattice mismatch relative to substrate.
Buffer Series: Varying [As] in InAsP buffer layer optimizes active/buffer layer interface.
Substrate (InP)
InGaAs
DEFECT

6. Bandgap Series Spectra
T = 77K
Samples [In] = .53, .60 contain an identical peak at .35 eV. Phonon-mediated transition?
The other three samples contain no distinguishable peaks.
Sample [In] = .66 may display hypothesized band-edge tail states.
Spectrum of sample [In] = .53 may corroborate previously published energy levels.
PL results suggest that [In] in InGaAs affects distribution of defect states.

7. Temperature Dependence
25 meV is on the order of the characteristic phonon energies for this material. Thus, the peak seems to reflect some sort of phonon-assisted transition. As T increases, more phonons of that energy are available, thus fewer photons are emitted. Resonance.

8. Conclusions and Future Work
Nonradiative transition from mid-gap states appear to be phonon-assisted (phonon energy ~ 30 meV)
Shallow states can result when the buffer/active interface is mismatched.
Transient Capacitance Spectroscopy will be used to further characterize sub-bandgap energy levels. (Summer 2003)

9. Buffer Series Spectra
T = 77K

10. Buffer Series Results
All samples have an unidentified peak near 0.35 eV.
Peak B is strongly [As] dependent.
Other studies show that [As] = 0.52, 0.59 have the highest radiative efficiency.

11. Temperature Dependence
25 meV is on the order of the characteristic phonon energies for this material. Thus, the peak seems to reflect some sort of phonon-assisted transition. As T increases, more phonons of that energy are available, thus fewer photons are emitted. Resonance.

12. Bandgap Series Spectra
Samples [In] = .53, .60 contain an identical peak at .35 eV. Phonon-mediated transition?
The other three samples contain no distinguishable peaks.
Sample [In] = .66 may display hypothesized band-edge tail states.
Spectrum of sample [In] = .53 may corroborate previously published energy levels.
PL results suggest that [In] in InGaAs affects distribution of defect states.