HAL Id: jpa-00226737
https://hal.archives-ouvertes.fr/jpa-00226737
Submitted on 1 Jan 1987
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
GROWTH, CHARACTERIZATION AND OPTICAL STUDIES OF InxGa1-xAs / InyAl1-yAs
STRAINED-LAYER SUPERLATTICES ON InP
J. Gerard, J. Marzin, J. Primot
To cite this version:
J. Gerard, J. Marzin, J. Primot. GROWTH, CHARACTERIZATION AND OPTICAL STUDIES
OF InxGa1-xAs / InyAl1-yAs STRAINED-LAYER SUPERLATTICES ON InP. Journal de Physique
Colloques, 1987, 48 (C5), pp.C5-169-C5-173. �10.1051/jphyscol:1987533�. �jpa-00226737�
GROWTH, CHARACTERIZATION AND OPTICAL STUDIES OF In,Ga,-,As / In,Al,-,As STRAINED-LAYER SUPERLATTICES ON InP
J.M. GERARD, J.Y. MARZIN and J. PRIMOT
Centre National d l E t u d e s des T6l~communications, Laboratoire d e Bagneux, 196, Avenue Henri Ravera, F-92220 Bagneux, France
Des superr6seaux contraints Inx Gal-xAs/InyAll-yAs, rhlisbs par 6pitaxie par jets moleculaires, ont fait l'objet d'une etude structurale (double diffraction X ) et optique (photoluminescence,transmission). Les resultats obtenus pennettent d'analyser sbparhnt les effets lies aux contraintes et au confinement.
InxGal-xAs/InyAl,_yAs strained layer superlattices have been grown by molecular beam epitaxy. Structural (X-ray double diffraction) and optical (photoluminescence, transmission) studies have been performed. These data allow us to analyse separetely strain and confinement related effects in this syst.em.
111-V compounds strained-layer superlattices (SLS) have been extensively studied in the last years. In these structures, the smaller band-gap material can either be under in-plane biaxial tension (as for example in AlyInl-YAs/GaAs [I]
system) or compression (as for GalIn,__As/ GaAs [2] superlattices). Generally, for a given substrate and system, only of the two strained configurations can be obtained
.
This restriction does not hold for In Ga As / In yAll-yAs superlattices grown on InP substrate, and the versatility of these structures make them very attractive for the study of the strain induced effects on the superlattice band structure. We report in this paper a structural characterization and optical study of this newly investigated system, for a wide range of sublayer thicknesses and compositions.
We focused our work on InxGa~-xAs/InyALl-yAs SLS (layer thicknesses Lx, Ly )
in self mechanical equilibrium when grown lattice matched to (001) InP substrate.
This property, which minimizes the density of strain-related defects within the structure, imposes the following relation between x,y,L-,La,:
* 7
(1): as=(axx Lx+ayx Ly)/(Lx+ Ly)
where ax,a ,as are the unstrained alloys and substrate lattice parameters. Such SLS are generally perfectly described in the elastic limit, if L1, - L_ are sma.ller than , the predicted sublayer "critical" thicknesses [3].
Growth.
The samples were grown by Molecular Beam Epitaxy in a Riber 2300 system. We first deposited a .5 pm (1nGaAl)As quaternary alloy buffer layer on the (001) InP substrate, before growing an N (6<%10) periods multiquantum well from InxGal-xAs and InyAll-yAs alternate layers
.
The substrate temperature was kept constant during the growth of the whole structure, in the range 520'C-540'C. It allows simultaneously to avoid indium desorption and optimize the cptical quality of the SLS. The &,flux was regulated to ensure As- stabilisation of the surface.
We -use two indium effusion cells (fluxes Fixand F ) to grow the different
i Y
sublayers [4]. Denoting F and Fa the I11 species fluxes delivered by tthc gallium and
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1987533
C5-170 JOURNAL DE PHYSIQUE
aluminium cells, the compositions x and y are respectively determined by F /Fix and Fa/FiY ratios. A supplementary, non restrictive, proportional relationships between designed layer thicknesses and growth rates allows us, actor-ding to (1) and Vegard's law, to grow a buffer layer lattice matched to InP, by just mixing these four fluxes. This two indium cells method appears therefore as a very flexible process to design and grow InxGe~~xAs/InyAl,~yAs SLS.
X-ray investir@tion.
Conventional X-ray double diffraction was performed, using a Cu Ka source and InP as a first crystal. As clearly shown by Quillec et al. [5], this powerful enalysis method is particularly adapted to study SLS. Fig.1 shows a typical profile obtained on a 10 period SLS; the diffraction intensity was mesured in the vicinity of the 004 reflection of the substrate, which accounts for the main peak, at the center of the profile. The two sets of satellites corresponding to the SLS diffraction are well separated, on both sides of this central peak. lhis is clearly FIG.1. Typical X-ray double
diffraction profile obtained on a ten periods I-/ InAlAs SLS (sample 11.41, near InP (004)
$
reflection. The indicated "zero
:
order" diffraction position of the E SLS intense does peak: not -it correspond is therefore to an
$ 6
deduced from the location of the
observed satellites.
A[A
,'layer buffer gunderstood, since the lattice parameters of each sublayer and of the substrate significantly differ in the growth direction z .
We used fits of these profiles to check the design parameters of our samples, following quite closely the method described in ref 5. The corrected caracteristics of the samples involved in this study are listed in table 1.
For strains smaller than ?I%, the experimental satellites widths are quite close to the theoretical prediction, thus demonstrating the high structural quality of the grown SLS. When a sublayer experiences tensile strains of about I%, a clear broadening of the diffraction satellites in its related set was evidenced. For larger strains (sample 0), a total desappearence of the SLS typical profile occurs and a single peak is then related to the layer. This last feature could be correlated with a slow but drastic deterioration of the HHEED pattern during the growth of the SLS. The bad cristallinity of the grown alloys under high tensile stress can certainly explain some of these features. We further used X-ray double diffraction to obtain reflection topography of these samples, selecting p r t of the diffraction profile of the layer. No network of misfit dislocations could be detected, as was expected from the evaluated critical thicknesses. High quality growth is therefore limited in this system by the structural quality of the strained alloys rather than dislocation nucleation, for strains larger than 1%.
Optical investigation.
Fig 2 shows the 8K transmission spectra obtained on these SLS. Clear excitonic structures, persistent at roao temperature, show up against the steplike background caracteristic of two dimensional systems, thus proving a high optical quality. The observed photoluminescence (PL) is typically 10-15 meV broad, and related to the first levels seen in absorption. Smll fluctuations of the thickness or composition of the well, and recombination processes related to residual impurities could account for the 5 to 20 meV Stokes shift between PL and the excitonic absorption edge.
The energies of the allowed optical transitions were estimated in an envelope function approximation [ 6 ] , based on the relevant zone center wave functions 13/2,T3/2>, ,(3/2,T1/2>. and )1/2,71/2>
,
of both (direct gap) strained hosts. They are labeled and indicated by arrows in Fig 2. All calculations are conducted in the same frame as in ref.2. We use the same value (75%) for the offsetquite independent of the choice of the precise value of the band discontinuities.The good agreement between experimental and theoretically calculated transition energies Table I : SLS corrected indium compositions x
,
y and sublayer thicknesses Lx,
L .. are deduced from a fit of the X-ray double diffraction profiles, for samples 1 to 7. nb designs the number of periods, and P the SLS period,
directly deduced from the profile.FIG.B. Transmission spectra obtained at 8 K on the studied SLS (see table I). The arrows indicate the calculated transitions associated with heavy (---+) and light
( - +) holes; the exciton binding energy is not included in these calculations. The
related PL syctrum is given for samples 4 and 5, and its maximum marked by vertical bars for the other samples. 8--+ shows the buffer layer absorption edge.
.- C
'5 t
C
e
-
L , , , , ,
.7 .8 .9 1. 1.1
E(eV)
.8 .9 1 1.1 1.2 13
E(eV)
allows us to study more precisely the optical properties of these SLS.
Heavy holes (HH) and light holes (LH) related bandgaps are plotted in fig.3 for a bulk Inx Gal-. As or InyAll-yAs layer strained on InP. The built in biaxial strain is equivalent to the sum of an hydrostatic pressure and an uniaxial stress along the z direction, whose effects are respectively: i) a shift of the zone center bandgap, ii)a lifting of the zone center valence band ( a ) degeneracy and coupling between LH and split-off bands. Owing to this M3 splitting, heavy and light holes are confined within the SLS by different potentials. Numerous band extrema configurations can be surmized if the band offset parameter is unknown. Due to the magnitude of the LH and HH related transitions (fig.2), the electron heavy hole as well as electron light hole systems are of type I. The two band edge line up possibilities corresponding to such confinement of electrons and holes in the small band gap material are displayed fig.4 a) and b).
Depending on the strain configuration within the SLS, the strain and confinement induced splittings may add or partly compensate each other. If the well is under compressive biaxial stress, the SLS bandgap corresponds to a transition involving the first electron and heavy hole subbands, as in InxGa,-xA~ /W system 123. For a well under tension this bandgap may either correspond to a HH or LH
C5-172 JOURNAL DE PHYSIQUE
related transition, depending on the relative strength of strain and confinement effects, as demonstrated in Alxlnl-xAs/GaAs system [I].
Two series of samples were grown in order to analyse separately strain and confinement related effects in InxGal-xAs/InyAll_yAs system.
.- Samples 1-2-3 were designed with same Lx, Ly, close to 100 A, and strain states varying for the well from -1% (compression) to 0.7%. The band gap shift (see fig.2) is mostly due to the variation of the well indium content, but strain related effects are clearly evidenced. The first (HH related) transition is indeed well separated from upperlying transitions in sample 1, whereas HH and LH first subbands lie at much closer energies for a well under tensile stress (sample 3).
Gn the other hand the comparison between sample 5,6 and 7xdisplays the effects
--
of increasing confinement for a given strain state (0.7% tension for InGaAs). The well thickness was reduced from 1208, for sample 5 to 6M (sample 6) and 308. (sample 7), the barrier thickness being proportionally decreased. The confinement energies raise induces a general upwards shift on the transmission spectra. This effect is more prononced for the LH related transitions, light holes having a much smaller mass than heavy holes in the z direction. For sample 7, only one absorption edge is evidenced ,corresponding to both single LH and HH related transitions. Our theoretical approach predicted a HH to conduction bandgap, as the first HH subband FIG.3 8K light hole (---) and heavy hole (-) related bandgaps are calculated for a bulk InGaAs or lnAlAs alloy strained to InP, and plotted as a function of x or y.
The unstrained bandgap (---) is also shown for comparison. The scale of experienced biaxial strains, comnon to both alloys, is indicated at the center of the figure.
In composition y
.51 73
.
.63 .53 .43 .33 IIn composition x
Cfl
InGaAs a) InGaAs b)
----.I ,.--..a
Hti LH
InAl As
"Y.n
InGaAs ,)FIG.4 Three possible band extrema --
configurations in an InGaAs/InAlAs SLS. CB refers to the conduction band, a d VB to the valence band, split by the strain into heavy (HH) and light (LH) hole bands
.
should lie a few meV higher than the LH one. The increase in energy of the first LH related transiticn is limited indeed in sample 7 by the coupling between the wells due to the reduction of the barriers thicknesses. Such reversal in the bandgap nature would therefore be more clearly seen for a series of samples designed with thicker barriers.
Another band extrema configuration, described in fig.4 c), is likely tooccur in this system, when the well experiences large tensile strains. Such situation would lead to the confinement of heavy holes in the large bandgap InyAl!-yAs layers and to the disappearance of the HH related transitions in the absorptlon spectra.
'he calculated critical strain, beyond which this configuration occurs, strongly depends on the choosen band discontinuities, and can therefore be used to estimate the offset parameter in this system. ( This parameter defines the part of the bandgaps difference A g that is accomodated by the conduction band discontinuity; as in ref.2, Og includes the bandgap modifications induced by the hydrostatic component
large tensile strains regime.
To summarize, we have investigated InxGa,-xAs/InyAll-yAs superlattices,grown by MBE on InP substrate. In a wide range of alloy compositions and sublayer thicknesses, providing the lattice matching to InP, they exhibit high quality structural and optical properties
.
The first valence subband can display a light or heavy hole character, depending on the design parameters. This system enlarges therefore the choice of materials that can be used on InP substrate.Acknowledgments: The authors want to thank M. Buillec for advices on the MBE growth, and fruitful connnents and discussions
.
I M. Sauvage et al.
,
Surf. Science 174,
573 (1986) J.Y. Marzin et al.,
Phys. Rev. B 31,
8298 (1985)J.W. Matthews and A.E. Blakeslee, J.Cryst.Growth 32,216 (1976) L. Goldstein et al.
,
J. Cryst. Growth 81,
396 (1987)M. Quillec et al. J. Appl. Phys. 55 ,2904 (1984) G. Bastard, Phys. Rev. B 24, 5693 (1981)
L. Goldstein et al., Int. Phys. Conf. Ser. n'74, 133 (1984)
