Because it can be grown lattice matched to InP, the ternary GaAsSb alloy system is of potential interest for optical device applications working in the 1.3– 1.7μm range. Of course, because of the high difference in refractive index which separates GaSb from AlSb, the most targeted issue is distributed Bragg reflectors (DBR) which will be made of different admixtures of GaSb and AlSb related compounds. Along this line, good quality DBR have been already demonstrated using MBE [1]. The active layers were made of GaAsSb and AlAsSb with lattice-matching compositions. The resulting difference in refractive index was about 0.5 and only 8 pairs of undoped material were necessary to get a peak reflectivity of 80% at 1.5 μm. Because of the large miscibility gap which limits the compositional range [2], the heteroepitaxial growth of GaAsSb on InP has not been very much investigated. Most of the previous works concern MBE [1, 3–5] and only a few reports deal with MOCVD [6, 7]. In the present paper, we show that good quality GaAs0.5Sb0.5 can be grown on InP using MOCVD. However, there is obviously a very narrow range of growth conditions which allows deposition of uniform layers. Indeed, in many cases, we find a periodic modulation of the final composition. This results in strained superlattices (S-SLs) structures. To estimate the corresponding compositional changes, they have been investigated in great detail. We have used a self-consistent approach to determine most structural parameters from the optical properties and, then, checked versus the X-ray diffraction pattern. All the GaAsSb layers investigated in this work were deposited on 〈100〉 InP wafers. We used a prototype horizontal MOCVD system, equiped with a vent-run manifold. The carrier flow was 14 slm of palladium-purified hydrogen and the reactant sources were TMGa, TMSb, TBAs and 5%-PH3. This latter source was only used to stabilize the substrate surface when heating the susceptor. Indeed, using the results of S-correlated theory [8, 9], we find that the optimized interface layer (IL) should have the average composition In0.68Ga0.32As0.59P0.41. This defines a lattice-matched IL which fits both the InP substrate and the topmost GaAs0.5Sb0.5 overlayer. It ensures also a continuity of the elastic-density and structural properties through the two-heterointerface system which cannot be fulfilled by a single abrupt heterointerface like GaAs0.5Sb0.5/InP or any As-rich IL like InAsyP1−y . Investigating various growth conditions in the range 560◦C≤ Tg≤ 620◦C; 0.9≤V/III ratio≤ 2 and 0.5 10−4≤6PIII≤ 3 10−4 atmosphere, we have optimized the results in terms of X-ray measurements. Our best data have been found using a total gas pressure of 700 torr, a growth temperature of 600 ◦C and, finally, a V/III ratio of 1.3. The typical growth rate was about 2 μm/hr. In Fig. 1, we show a typical X-ray diffraction spectrum collected on a thick GaAsSb layer deposited on top of InP. This shows very good lattice-matching, with both layer and substrate peaks superimposed to give a final (total) FWHM equal to 140 arcseconds. Compared with the one (30 arcseconds) which was collected on a piece of uncovered substrate, this is about 4 times larger. Such a ratio is typical of very good quality material [8]. At 2 K, the corresponding PL spectra display single line features. This is shown in Fig. 2. The peak position is 723 meV and the FWHM is equal to 14 meV. Both values compare well with results reported for MBE material [4] (8 meV at 2K). In some cases, we have found from PL an increased intensity, with respect to the standard bulk material. This, systematically, correlates with a weak evidence of superlattice structures in the DDX spectra (see Fig. 3a) which suggests that a superperiod has been formed. Checking for the origin of this S-SLs, we have two different possibilities. First, a natural demixing of the metastable alloy on both sides of the miscibility gap [2]. Second, a regular effect of ordering-disordering modulation [6, 10]. To find, quantitatively, the corresponding compositional changes we have investigated in detail the structural and optical properties of these spontaneously grown GaAsy1Sb1−y1/GaAsy2Sb1−y2 SLs. The problem is very similar to the one encountered in the case of Inx1Ga1−x1As/Inx2Ga1−x2As SLs [11]. In order to evaluate the average magnitude of the wells and barriers thickness and composition, one has to play with 4 structural parameters. None of them is known from the
Search from 250 M+ research papersSearch from 250 M+ research papers
Nov 11, 2020
S S Mohanty + 3
Open Access