Pii: S0026-2692(98)00137-2

The exciton dynamics in In0.15Ga0.85As/GaAs quantum wells grown on (111)B and (100) GaAs substrates are studied by the time-resolved photoluminescence (PL). We have found that the piezoelectric fields in (111)B samples affect the transient behavior of PL spectra. Compared with the reference (100) samples, we have confirmed that the piezoelectric effect induces slower exciton relaxation in (111)B strained quantum wells. q 1999 Elsevier Science Ltd. All rights reserved. In recent years, studies of epitaxial layers on (111)B GaAs substrates increased rapidly. One of the most interesting characteristics in such layers is the piezoelectric effect caused by the strain layers on GaAs substrates of this direction [1]. A lot of works have been devoted to the growth of the strain quantum wells and the study of piezoelectric fields [2–5]. However, there are still very few studies on the carrier-dynamic of the strain quantum wells under strong piezoelectric fields. The carrier dynamics of quantum structures is important in many applications, especially in optoelectronic devices. In this work, we found some unique dynamic behaviors in (111)B InGaAs/GaAs quantum wells which might be related to the piezoelectric effect. We have measured time-resolved photoluminescence (PL) of samples with two InGaAs/GaAs quantum wells on both (111)B and (100) GaAs substrates, grown side by side. The structures we used are two In0.15Ga0.85As/GaAs quantum wells, separated by a 150 nm GaAs space layer and capped by a 50 nm GaAs layer. The well widths are 2 nm and 3 nm. There are two kinds of arrangement: structure A is with the 3 nm quantum well above the 2 nm quantum well, and structure B is with the 2 nm quantum well above the 3 nm quantum well. We use QWA and QWB to represent these two kind of structures. Both structures were grown by molecular beam epitaxy on (111)B and (100) GaAs substrates placed side by side in the chamber. We labeled these four samples as QWA0, QWA1, QWB0 and QWB1, with index ‘‘0’’ representing (100) samples, and ‘‘1’’ for (111)B samples. The GaAs layers were grown at 5908C and the InGaAs layers were grown at 5258C. The samples were excited by an amplified mode locked Titanium Sapphire laser from Spectra Physics with the laser energy at 3.1 eV. The time-resolved PL spectra were measured by a Hamamatsu streak camera C4334 with a time resolution of 5 ps. The measurements were performed at 4.2 K. The PL intensity versus wavelength and time both were obtained at the same time. One example of the measured results is shown in Fig. 1. We can integrate the data over the exciton lifetime to get an ordinary PL spectra, or integrate over a short range of wavelength to get the transient behavior in that wavelength range. Figs. 2 and 3 show the integrated results of the structure B samples. Fig. 2 is the time-integrated PL spectrum, while Fig. 3 is the transient behavior which is the wavelength-integrated result over the range of each peak shown in the spectra of Fig. 2. In order to distinguish the eight quantum wells of four samples more clearly, the quantum wells were labeled QWA0t, QWA0b, QWA1t, QWA1b, QWB0t, QWB0b, QWB1t and QWB1b, where the second index, ‘‘t’’ represents the top quantum wells, and ‘‘b’’ is for the bottom quantum wells. In those PL spectra, the peaks at the shorter wavelength side are from the 2 nm quantum wells, while those at longer wavelengths are from the 3 nm quantum wells. All the peaks show strong intensity with linewidth no larger than 5 meV, indicating Microelectronics Journal 30 (1999) 367–371 Microelectronics Journal 0026-2692/99/$ see front matter q 1999 Elsevier Science Ltd. All rights reserved. PII: S0026-2692(98)00137-2 q This work was supported by the National Science Council under contract NSC87-2215-E009-010. * Corresponding author. E-mail address: [email protected] (C.P. Lee) good sample quality. However, in Fig. 3, the dynamic behaviors of the two quantum wells show significant differences in (111)B samples. Obviously, the rise and decay time depends on the position of the quantum well. The fitted decay time constants of peaks from all quantum wells are listed in Table 1. For the quantum wells near surface, the decay time constants are around 500 ps, while the time constants of the quantum well beneath are well above 1 ns. For the (100) samples of the same structures, which were grown at the same time, the decay time differences are much smaller. There are several possible mechanisms could be involved in the dynamic behaviors observed. The first mechanism is carrier diffusion. The incident laser energy we used for the PL measurement is 3.1 eV. The absorption coefficient of GaAs at this energy is around 10 cm. Therefore, the number of carriers excited in GaAs decreases a lot at the location more than 100 nm away from surface. Many carriers diffused toward the bottom quantum wells and were captured. This mechanism is called ‘‘vertical ambipolar transport’’ [6] and exists in both (100) and (111)B samples. We considered it as the basic mechanism that induced a longer lifetime of the peak from the quantum well located farther from surface. The second mechanism is the wavefunction separation [7] of electrons and holes inside the quantum wells of (111)B samples, because of the high piezoelectric fields. It could be easily found in Table 1 that for both (100) and (111)B samples of the same structure, the time constant of the (111)B quantum well is larger than that of the (100) quantum well located at the same position in the sample. For the quantum wells in the same (111)B samples, the piezoelectric fields inside the quantum wells near surface (QWA1t and QWB1t) were compensated by the surface F.Y. Tsai et al. / Microelectronics Journal 30 (1999) 367–371 368 Fig. 1. One example of the measured results on the (111)B sample of structure B. The PL intensity versus wavelength and time both were shown at the same