OPTICAL STUDIES OF TYPE I AND TYPE II RECOMBINATION IN GaAs-AlAs QUANTUM WELLS

In this paper we report the results of a systematic investigation on the effects of electronic coupling on a series of G&-rmlt' l e quantum well samples in which the thickness of the G& layers was fixed a t s 2 5 L d the AlAs thickness varied between samples from 392 t o 62. Using the techniques of photoluminescence and p h o t o l ~ s c e n c e xcitation spectroscopy we have followed the positions of both the r-related direct gap and the F-X pseudo-indirect gap as a function of A& thickness. A s the A l A s thickness is reduced we observe the system change from a type I1 t o a type I band alignment and measure this crossover a t %l38 of AlAs i n these samples. Intrcduction h G&-Al&ai-,As ml t ip l e quantum well (m) structures it is comnly accepted that the fractional valence band offset l i es i n the region 0.3-0.4 (1). One consequence of this is that when the barrier material i s indirect (X > 0.45), the Al&al-xAs X point lies a t lower energy than the GaAs X point. In this case the AlGaAs acts as a well for X-related states while the GaAs forms a potential well for r-related states. The valence levels are always preferentially localised in the G&. Therefore by choosing appropriate quantum well parameters, GaAs-Alfial-,&i quantum well and superlattice structures may display either a direct r-related f u n h n t a l gap or a r-X pseudo-indirect gap. We shall refer t o these as type I and type I1 systems respectively. Photoluminescence studies have identified type I1 recombination processes in A10.3fiw.6*-A1As heterojunctions (2) and also in GaAs-MO .$W.+ quantum wells under hydrostatic pressure (3). Further evidence for this phenamenm has been provided by studies of narrow G&-& quantum wells (4,5,6,7) . In such structures it has been shown (6,7) that when the GaAs thickness is reduced below %35% the n=1 electron s tate in the GaAs is pushed above the X minimum in the AZAs and the system becomes type 11. Finkman e t a1 (7) and Dawson e t a1 (6) have identified the lowest confined exciton s tate of the type I1 system in the excitation spectrum. Both groups assign the observed features to exciton absorption inwlwhg electrons a t the A1As minimum ie. with momentum parallel to the (001) direction, and holes a t the F-point in the G&. Combining this with the temperature dependence of the type I1 p h o t o l ~ e s c e n c e spectrum mwson et a1 (6) ascribe the emission t o the recombination of localised excitons involving electrons a t the X, minimum, while Finlrman e t a1 (7) assign the main component of the type I1 emission to excitons involving electrons a t the AlAs Xv valleys ie. mnoenta perpendicular to the (001) direction. However, the cornnon conclusion from a l l these reports is that in G&-AlAs WW w i t h GaAs thickness <352 the dominant reconbination process a t low temperatures involves the r-X pseudo-indirect gap. Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:19875112 (25-526 JOURNAL DE PHYSIQUE The structures studied by hwson e t a1 had thin G& layers (<30@ but relatively wide AlAs layers $708 thus ensuring that the electrons in neighbouring GaAs layers were uncoupled. Calculations indicate (8,9) that intrcducing coupling between adjacent G& layers results i n a lowering of the r-point subband minima and hence predict that as the AlAs thickness is reduced the system eventually reverts to a type I band alignment. In this paper we report photoluminescence (PL) and photoluminescence excitation (PLE) spectroscopy measurements on a set of samples w i t h a fixed GaAs thickness b258 in which the AlAs thickness was progressively reduced. In this way we have made a systematic study of the effects of coupling on a type I1 system. Experimental Details The samples reported on in this study were grown by MBE in a Varian Gen I1 system (10). The layers w e r e deposited on (001) orientated semi-insulating G& substrates a t a temperature of 630'~. The growth sequence was as follows, (a) 1.0 um of G& buffer material, (b) 60 periods of G& and AlAs where the G& thickness was fixed and fx AUS Mckness varied in the different samples, and (c) a capping layer of 1000 of G&. Five samples were gram for this work and the thicknesses of the constituent layers were measured by X-ray diffraction techniques (11). The GaAs thickness was determined to be 25*2g for a l l s les while the thicknesses of the AUs layers were measured as 39% 242, 168, 8 . 3 a n d 5.78. The photoluminescence and excitation spectra were recorded with the samples mxlnted on the cold finger of a variable temperature (4-300K) continuous flow cryostat. The samples were excited by either an Ar ion laser or a lamp and uxmochromator. The luminescence was collected and analysed by a double-grating Spex 1404 spectrometer and detected w i t h a cooled G& photormltiplier (C31034) and associated photoncounting electronics. The PLE experiments were performed by exciting the sample with radiation from either a lamp and scanning spectrometer or a tunable dye (IXX) laser, and monitoring the emission intensity as a function of excitation energy. rimental Results and Discussion Z t l y , consider the quantum well structure which has AUS layers b398. ~k low temperature PL spectrum from this sample is s h in figure 1. The nature of the emission has already been discussed (6) and we identify no-phonon and phonon-assisted exciton transitions h a l v i n g a hole a t the r-point in the G& and an electron a t the X-point in the adjacent AUS layer. In the same sample we can observe the energy transitions associated with the type I subbands by m i t o r i n g the intensity of one component of the type I1 emission and recording the PLE spectrum. As shuwn in figure 2 we observe strong type I exciton peaks involving n=l electrons confined a t the direct subband mininann (Elr) of the G& and either n=l heavy holes (HHl) or n=1 light holes (W). Conbining these experimental techniques enables us to follow the positions of both the r-related direct gap and the r-X indirect gap as we reduce the AlAs thickness. Consider now samples in which the AlAs thickness is progressively reduced to 242 and 168. The law temperature PL spectra of these samples are illustrated in figure 1 and the corresponding PLE spectra are shown in figure 2. In each case the ldnescence can again be ascribed to type 11 recombination. Notice however that the emission shif ts t o higher energy with decreasing A U s thickness. This reflects the increase in the electron confinement at the X-min imm in the AlAs. In the PLE spectra (figure 2) we can identify the type I exciton transitions (Elr-HH1) and (El r-LH1). A s the MAs thickness is reduced we observe both a shif t of the type I transitions t o lower energy and also a decrease in the splitt ing of heavy and light-hole peaks. We believe this arises as a result of coupling adjacent GaAs layers and because the effects are most dramatic on the particles of smaller mass ie. the electrons and light-holes, the light-hole subband mves to lower energy more rapidly than the heavy-hole subband. When the AlAs thickness is reduced still further to$8.48 W observe a dramatic shif t of the luminescence peak to lower energy (figure 1). In this sample we assign the emission as due to type I recombination of electrons and holes both confined in the G&. lhis assignment has been confirmed by recording the PLE spectrum. The (Elr-HH1) exciton peak in the PLE spectrum and the emission peak are coincident within lmeV. A further decrease in the AlAs thickness to % 5.7% results in a shif t of the type I emission spectrum to still lower energy; figure 1. This reflects a continued lowering of the r-point subband mknimm with increased coupling. 1.70 1.75 1.80 1.85 1.90 Emission energy (eV) I I I J 1.80 1.90 2.00 2 .l0 Excitation energy (eV) Fiwe 1 Low temperature (7K) PL -e 2 Low temperature (7K) spectra. The emission from samples w i t h PLE=tra showing the type I AlAs thicknesses of 39x, 242 and 16% is transitions for the samples associated w i t h a type 11 process. When described in figure l. 'lhe y axis the AlAs thickness is reduced t o 8.4g and has been displaced for clarity. 5.7% the emission becomes type I. The y axis has been displaced for clarity. 1.945 of the (ElP-HHl) exciton peak as measured by PLE spectroscopy. . Ihe broken line shows the K --variation of the type I1 emission a--* l ine w i t h AlAs thickness. The type 11-type I crossover occurs a t %138. 1.70 I 0 10 20 30 40 50 Thickness of ALAS (a)