Temperature Dependent Optical Band Gap Measurements of III-V films by Low Temperature Photoluminescence Spectroscopy Linda M. Casson, Francis Ndi and Eric Teboul HORIBA Scientific, 3880 Park Avenue, Edison, NJ, 08820, USA ABSTRACT Photoluminescence (PL) spectroscopy is a powerful technique for probing the structures of many types of III-V semiconductor materials. When a semiconductor material is excited at a particular wavelength, electron-hole pairs are generated. The most intense radiative transition is between the conduction band and valence band, and this measurement is used to determine the material band gap. Radiative and non-radiative transitions in semiconductors also involve localized defect levels. The photoluminescence energy associated with these levels can be used to identify specific defects, and the amount of photoluminescence can be used to determine their concentration, and thus predict device quality. At ambient temperatures, the PL signal is typically broad, as much as 100 nm in width. When cooled, structural details may be resolved, and a small spectral shift between 2 samples may represent a change in a structural parameter. Thus a system with high spectral resolution is required. In this paper, a modular Low Temperature Photoluminescence system (LTPL) for measuring optical band gap as a function of temperature is described. Results show that the optical band gap shifts towards higher energy as the sample temperature decreases. INTRODUCTION Temperature dependent photoluminescence spectroscopy (PL) is a powerful optical method used for characterizing materials. It can be used to identify defects and impurities in Silicon and group III-V element semiconductors, and determine semiconductor band gaps (Eg). When light with sufficient energy hits a semiconductor, the material absorbs light, creating an electron-hole pair. An electron from the valence band jumps to the conduction band leaving a hole in the valence band. The electron relaxes back down to the lowest energy level in the Conduction Band. The electron drops back across the bandgap recombines with the hole. A photon is emitted during this recombination process; the emission of light is called photoluminescence. The photon emitted upon recombination corresponds to the energy difference between the valence and conduction bands or band gap (Eg), and is hence lower in energy than the excitation photon, so that the luminescence is red-shifted with respect to the excitation light. This is also called a Stokes shift. In addition to the main emission band, especially at low temperatures side bands may be observed, which may correspond to emission from defect states, indicating defects or impurities in the material.
EXPERIMENTS Two separate samples were characterized on the photoluminescence system described below. Sample 1 was made of a silicon doped (1 x 10 18 cm -3 ) 400 micron thick InP substrate, a 1 micron thick InP (undoped) buffer layer, a 3 micron thick absorbing layer of lattice-matched In 0.53 Ga 0.47 As deposited by organometallic vapor phase epitaxy, and a top InP cap layer ~0.5 micron thick. Sample 2 was made of a silicon doped (1.5 x 10 19 cm -3 ) 600 micron thick InP substrate and a 0.375 micron thick absorbing layer of lattice-matched In 0.53 Ga 0.47 As deposited by organometallic vapor phase epitaxy. Samples were mounted in a closed-cycle cryostat cooled to 6 K and excited with a 647nm diode laser. The PL emission was collected with a HORIBA Scientific Low-Temperature Cryostat Interface with reflective optics, and coupled into HORIBA Scientific ihr550 spectrometer for analysis. PbS and InGaAs photodetectors were used to measure the PL signal, with chopper and lock-in amplifier for synchronous detection. A schematic of the experimental setup is shown in Figure 1. ACH-C Figure 1. Schematic of experimental setup for temperature dependent PL measurements RESULTS AND DISCUSSION As the sample temperature increases, the PL spectrum broadens and shifts to lower energy. The spectral broadening may be quantified by the full-width at half-maximum (FWHM) of the PL signal, which is proportional to kt at temperatures above 40 K. As the sample temperature increases, the intensity of spectrum also decreases, due to energy lost to nonradiative processes and interactions with the surroundings. Equation 1 describes the theoretical relationship between intensity, PL signal width and temperature. Figure 1 shows PL spectra of Sample 1, measured at various temperatures between 6 K and room temperature. Figure 2 shows PL signal FWHM plotted as a function of temperature and above 40 K shows a linear relationship [1]. As described by Pecharapa et al., PL spectral broadening is due to two
components: a temperature-independent inhomogeneous broadening due to interface roughness and fluctuations in alloy binding energies (observed at temperatures < 40 K) and a temperatureindependent homogeneous broadening due to thermal motion, collisions and phonon interactions (observed at temperatures > 40 K) [2]. I(hυ) α (hυ-eg) 1/2 exp{-(hυ Eg)/kT} (1) 350000 300000 250000 InGaAs/InP Sample 1 6K Intensity, arb. 200000 150000 100000 296K 10x 200K 5x 100K 2x 50000 0 0.74 0.76 0.78 0.80 0.82 0.84 Energy (ev) Figure 2. PL spectra of InGaAs/InP Sample 1 measured between 6K and 296K. 40 PL spectrum FWHM (mev) 35 30 25 20 15 10 FWHM α kt 5 0 0 50 100 150 200 250 300 Temperature (K) Figure 3. PL signal FWHM plotted as a function of temperature.
The material band gap can be measured by determining the peak intensity of the PL signal, and decreases as the sample temperature increases. The experimental data were fitted to the Varshni equation (equation 2) [3, 4], and coefficients α and β were determined using a nonlinear regression. The fitted values of α and β were α = 4.82 x 10-4 ev/k and β = 430.05 K. Eg(T) = E 0 αt 2 /(T + β) (2) 0.80 Eg vs. Temperature 0.81 Experimental Varshni Model 0.79 Eg (ev) 0.78 0.77 0.76 0.75 0 50 100 150 200 250 300 Temperature (K) Figure 4. Plot of Eg as a function of temperature. Figure 5. PL spectrum of Sample 1 at 6 K. Side-bands indicate defects or impurities.
Figure 6. PL spectrum of Sample 2 at 6 K. Side-bands indicate defects or impurities. CONCLUSIONS The optical band gap of InGaAs used for channel materials in high electron mobility transistor is a critical parameter in the proper functioning of lighting devices. The results presented in this paper, carried out with high resolution and temperature dependent photoluminescence spectroscopy, demonstrate excellent agreement between the theory given by the Varshni equation and the experimental results on the inverse relationship between band gap and temperature. Additionally, the use of a high resolution spectrometer of 0.5 meter or longer focal length was necessary to give evidence for the potential presence of material defects that could impact the specifications of the final device. ACKNOWLEDGEMENTS The authors would like to thank Dr. M. Samant, J. Brockman and N. Aetukuri at IBM Almaden Research Center for assistance in PL measurements; W. Pinkston of ElectroOptical Systems, Inc. and Dr. Tatsuo Ishikawa of Horiba Scientific for providing samples; Dr. Li Yan of Horiba Scientific and Prof. V.G. Stoleru of University of Delaware for valuable discussions. REFERENCES: 1. M. Fox, Optical Properties of Solids, Oxford University Press, 2001. 2. W. Pecharapa, W. Techitheera, P. Thanomgam and J. Hukeaw, Temperature-dependent photoluminescence investigation of narrow well-width InGaAs/InP single quantum well, Proc. SPIE Vol. 6793, 67930C (2008). 3. Y. P. Varshni, Physica 43, 149 (1967). 4. K. P. O Donnell and X. Chen, Temperature dependence of semiconductor band gaps, Appl. Phys. Lett. 58, (25), 2924-2926 (2001).