Heteroepitaxial growth of InAs on GaAs(100) mediated by Te at the interface

Heteroepitaxial growth of InAs on GaAs(100) mediated by Te at the interface

Solid State Communications, Vol. 95, No. 12, pp. 873-877, 1995 Copyright 0 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 003...

528KB Sizes 1 Downloads 11 Views

Solid State Communications, Vol. 95, No. 12, pp. 873-877, 1995 Copyright 0 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0038- 1098/95 $9.50+.00 0038-1098(95)00397-5

Pergamon

HETEROEPITAXIAL

GROWTH

OF InAs ON GaAs(100)

MEDIATED

BY Te AT THE INTERFACE

W. N. Rodrigues”, V. H. Etgensb, M. Sauvage-Simkin”, G. Rossid, F. Sirotti, R. Pinchaux” and F. Rochet’

Laboratoire

pour 1’Utilisation du Rayonnement

l?letromagn&ique,

Centre Universitaire Paris-Sud, bgt 209d

91405 Orsay, France.

(Received 3 May 1995 by C.E.T. Gongalves da Silva) The heteroepitaxy of InAs on a Te covered GaAs surface is investigated by photoelectron spectroscopy. Core-level spectra probed with synchrotron radiation show that Te remains at the interface between InAs and GaAs in a concentration much higher than the solubility limit of this element in III-V compounds, suggesting that a new compound is being formed. We propose that this interlayer is responsible for the observed changing in the growth mode of the InAs overlayer from islands to layer by layer. Keywords: A. semiconductors, spectroscopies

A. surfaces and interfaces, B. epitaxy, E. photoelectron

The association of materials with different properties

6.26 x 1014cm-2 ) prior to the InAs growth

is

one of the main objectives in materials science nowadays.

drastically

However,

pseudomorphic

the possible

choices

of materials

leading

to

the

highly perfect epitaxial layers are limited by the strain due

being

to

suppression

lattice

mismatch,

considerations. difference growth

and

by

surface

free

energy

lattice parameter

of islands after the deposition

growth

mode regime

of

this

can change system,

the

of InAs on GaAs(100)

up to 6 monolayers,

and a complete

of the 2D-3D transition was observed.

changes in the growth mode were interpreted

In the case of InAs on GaAs the 7.2 %

in the relative

extended

growth

These

as being due

to a surfactant action of Te on the growth process, based

leads to the

on the intensity of the Te photoemission

of only 1.5 InAs

signal [2]. The Te

a

layer, originally at the GaAs surface, would thus float on

monolayer of a compound AB means a double (001) plane

the InAs growth front, fidfilling the expected behavior of a

of atoms). Recently, it was shown [ 1,2] that the addition of

surfactant,

approximately

Si(OO1) heteroepitaxy

monolayer

at normal

one

growth

(001)

temperature

plane

[l]

(here

of Te on GaAs

(i.e.

as pointed out in the case of As in the Ge on [3].

Our experiments

action of Te in changing the growth

Departamento de Fisica, ICEX , UFMG CP 702 - Belo Horizonte 30161-970, MG Brazil. Laboratoire de MinBralogie-Cristallographie, 4 place Jussieu, F-75252 Paris, Cedex, France. Also at Laboratoire de MinCralogie-Cristallographie, 4 place de Jussieu, F-75252 Paris, Cedex, France. Also at Laboratorium fir Festkhrperphysik, ETH, CH-8093 Ziirich, Switzerland, and Dipartimento di Fisica, UniversitSl di Modena, via Campi 213/A, 4 1100 Modena, Italy. Also at UniversitC Pierre et Marie Curie, 4 place Jussieu, F-75252 Paris, Cedex, France. Laboratoire de Chimie Physique - Universitb Paris VI, 11 rue Pierre et Marie Curie, F-7523 1 Paris, Cedex 05, France.

confirm the

mode of InAs on

GaAs( 100) to a 2-D growth mode. However,

we observe

that about half (001) plane of Te remains at the interface. Such concentration

is many orders of magnitude

higher

than the solubility limit of this element in III-V compounds [4], indicating a probable formation of a buried layer. The fundamental difference between our experiment and that of Grandjean et &. [I] is that the authors of Ref.1 used the (6

x 1)

reconstructed

coverage worked 873

Te:GaAs(OOl) surface [5], with a Te

of 1.3 of a (001) single plane, whereas on the

(*x3)

incommensurate

phase

we

of the

HETEROEPITAXIAL

874

Te:GaAs(OOl) surface [5,6], prepared with a Te coverage

increasing

of 0.9 of a (001) single plane as the precursor state prior to

increase

the InAs deposition.

coverage

The growth molecular growth

of InAs

on GaAs

was

beam epitaxy (MBE) at 420°C rate

high-energy

of

0.06 Ml/set,

electron

diffraction

done

by

at a constant

calibrated @HEED)

by reflection specular beam

intensity oscillations during the growth of InJ&4s

Vol. 95, No. 12

GROWTH OF InAs ON GaAs( 100)

layers

on GaAs. Since a III-V crystal in the [ 1001 direction

is

kinetic

InAs

observe

that

agreement

The

the

with

for a

depth at that

of the

Ga3d

is growing

the RHEED

expected

stabilizes

escape

attenuation

overlayer

the

which

higher than the electron energies.

growth

we

of the In4d structures,

indicates

In4d

coverage,

peak

uniformly,

in

observations

during

the

of the InAs. The energy separation

between

the

and Ga3d

coverage,

levels

suggesting

also changes that

the

band

clearly

with

the

off-set

is being

made up of alternate atomic planes of column three and

influenced by a huge doping at the interface introduced by

column

Te, or by a strain effect. An extensive discussion

five elements,

combination each

of two

deposition,

non-intentionally

a monolayer

(Ml) is taken as a

such substrate-like a

0.1

doped

pm GaAs

thick

planes. Prior to buffer

was grown

layer

conditions producing a sharp (2 x 4) reconstructed For the case of Te modified delivered

growth,

surface.

the Te flux was

by a Knudsen cell appended to the main MBE

chamber. Starting with a (2 x4) GaAs surface, we prepared a (*x3) incommensurate phase by exposing the GaAs surface at 540°C to the Te flux, producing an overlayer of (0.9 SO.1) single (001) plane of Te atoms as quantified

by

temperature 420°C

Rutherford

Backscattering

(RBS).

of the sample was then brought

and the InAs growth

performed.

band off-set will be published elsewhere.

of

in standard

The

down to

The RHEED

pattern of the growing InAs displayed 2D features up to

Fig.2 shows the Te3dsiz core-level 700

eV

photons.

Again,

all the

spectra excited by spectra

have

showing

coverage

some

(17 MI), the

diffuse

diffraction

background

for

the

pattern highest

coverage, indicating disorder or a stepped surface. The photoemission

measurements

levels.

The integral

intensity

of the Te3ds,z

core-level normalized to the signal at zero InAs coverage is shown in Fig.3 as a function of the overlayer

in a

coverage.

Also included are the data for the Ga3d core-level and the data

measured

et

by Massies

d.

[2]

for

InAs

on

Te:GaAs(lOO). From the Ga3d intensity decay we confirm the layer by layer growth mode of the InAs film, by fitting a simple exponential

decay to the data points. The best fit

The Te3d intensity stabilization

h, = 8 A.

decreases

initially tending

to a

at higher coverages.

Such a feature

could

indicate island formation or a Te segregation

were performed

been

normalized by the counts at the high kinetic energy side of the core

gave a escape depth for the Ga3d photoelectrons the highest

on the

growth front. The former interpretation

towards the

can be discarded,

UHV chamber equipped with a RIBER MAC II electron energy analyzer, and connected equipped

to the SU7 undulator line

with a 10 meters TGM at SuperACOiLURE.

I

I

InAs on Te:GaAs( 100) I x 3

The transfer of the samples between the growth chamber and the photoemission

I

I

I

XPS

17OeV

system was done in a portable UHV

vessel equipped with an ion pump and a transfer rod. The time needed to commute

was typically 45 minutes at a

pressure always smaller than 10dp mbar. In Fig.1 we show the Ga3d and In4d core levels, excited

by

170 eV photons.

The

spectra

have

been

normalized by the count rate at the highest kinetic energy. The strong contrast between

the intensities

In4d levels arises from cross-section photon energy, near the Cooper-minimum

of Ga3d and

differences

at that

of the In4d level

[7]. Besides the strong attenuation of the Ga3d peak with

i

1

144

146

I

14x

I

I

150

Kinetic Energy (eV) Fig. 1. Spectra for the Ga3d and In4d core-levels with 170 eV photons for increasing InAs coverage.

taken

Vol. 95, No. 12

~TEROEF~~

875

GROWTH OF InAs ON GaAs(100)

r XI’S

IIL

I10

114

I.!”

Kinetic Energy

700 cv

(eV)

Fig.2. Spectra for the Te3dsn core-level eV photons for increasing coverage.

taken with 700

based on the evidences of a layer by layer growth found by RHEED,

Ga3d

mi~os&opy

spectra

images

intensity decreases

and

high-resolution

of our samples

electron

[8]. Since the Te

However, the RBS measurements

indicate that the amount of Te originally deposited GaAs surface prior to the InAs overgrowth

on the

remains in the

sample. Diffusion into the GaAs bulk is also not expected due to the low temperatures arguments

interface.

In

incorporated

involved in the InAs epitaxy.

support

involve partial segregation. that are not at

I - exp (--Z(iv+ 1)dA) 1 - exp f-2d*/h,)

it could be an indication of segregation

with loss by desorption.

These

Fig.3. Core-level intensity as a function of the InAs coverage, referred to the spectra at zero coverage. Also shown are the data from Massies ef al. [2]. The lines are the best fits considering the models discussed in the text. Oreis the absolute Te coverage of the GaAs surface prior to the InAs deposition.

two

alternative

models

that

In the first model, the Te atoms

the surface remain at the InAsKiaAs the

second

model,

these

atoms

are

in the growing film.

For the first model (partial segregation

plus Te fixed at

the interface), the intensity of a core-level referenced to the signal at zero overlayer coverage,

i,,

can be described by

here

Zb = 1 -Z,

incorporated

is

the

contribution

from

I the

(2) Te

in the film.

Both fimctions have been fitted to the Te data points in Fig.3 using a non-linear least square fitting procedure.

The

results of the fittings are shown in Table I. Both models gave good results as indicated

by the small x2 values.

However the physical meaning of the values found for the different cont~butio~

in each model must be discussed.

For the second model, 74 %

of the Te atoms in the

sample should be incorporated

in the 17 monolayers thick

InAs layer. This concentration

surpasses by a factor of 10

the limit of solubility of this eiement conditions

in InAs [4]_ The

met in the MBE growth process do not favor

such excess doping. This could only happen if the growth

the function: I,.(N) = Z, +Ziexp(-2dsNlL)

(1)

front advanced faster than the allowed by the diffusion of the present chemical species. Therefore,

we can rule out

where Z, is the contribution from the segregated fraction of Te, and Z, = 1 -I, is the ~n~bution from the Te atoms remaining at the interface, do is the distance between

this model.

planes in the [ 1001 crystalline

63 % of the Te atoms, i.e. 0.57 kO.07 monolayer of Te,

direction

and is equal to

The results obtained

for the first model indicate that

1.5146 A for InAs, and N is the number of deposited

remained at the interface. The data measured by Massies

monolayers of 1nA.s.

et al. [2] could also be fitted by the same function

In

the

incorporation is:

second,

model

(partial

segregation

in the overlayer) the corresponding

plus

function

as

above, the best fitting ~dicating Zi = 0.21. This means that around 0.3 Ml remained at the interface,

considering

the

coverage values indicated by those authors. The observed

876

HETEROEPITAXIAI.

Vol. 95, No. 12

GROWTH OF InAs ON GaAs( 100)

Table I. Rest&s c&fitting the intensity decay of the Te3d data with the two proposed models. ii is the c~~~b~tioo from the Te fraction at the interface, fb is the Te fraction inco~rated in the volume of the growing film, I,is the fraction segregated to the surface, 5 is the electron mean free path .

r,

4 Model I

0.63 f 0.02

Model II

discrepancy between

0.74 f: 0.04

our Ii values and that of

et al.could arise from differences absolute

coverages.

measurements

to

concentrations

solubility

qu~ti~

case

the

we

used

coverages.

RBS

The

the interface

three orders of magnitude

limit

semiconductors

segregated,

our

of

Te

found at the interface in both experiments

are approximately the

In

Massies

in the determination

[4]. forms

of

this

element

in

0.26 + 0.04

7&l

0.00058

case of the Te modified growth characteristics demonstrated

are not met, Massies and Grandjean

have

clearly the reduction of the surface mobility

at the GaAs(IO0) suggest

such standard surfactant

surface

the broadening

by the Te atoms of the

surfactant

[Il].

We

concept

by

phenomena at the interface rather than at the growth front.

III-V

Te is

in the growing InAs film

[9] indicate that 0.5 monolayer

Te at the interface reduces the GaAs/InAs interface strain

We understand accommodating

we consider the

the role of Te at the interface

the lattice

mismatch

between

as

the two

materials. In this process the Te bonds at the interface are stretched

and

pseudomorphic energetic

energy. Based on the above cited arguments

x2

bulk

limited by its solid solubility (i.e.2 x 10rpcmS )[4]. Recent first principles c~culations

0.00027

including also the systems where the growth is modified by

and the excess

with an incorporation

UQ 14+1

higher than

These results suggest that the Te at a compound

1, 0.37 + 0.02

bent

when

growth

the

thickness

is attained. One approach

related to stretching

solids is presented

critical

and bending

by Harrison

for

to the

of bonds in

[IL?] by introduc~g

the

force constants Co and Cr , describing the bond stretching first model as the most adequate. indicate that a non-negligible

Therefore,

concentration

our results

of Te is present

at the interface between the InAs overlayer and the GaAs substrate. Massies et al.[1,2] have considered at the surface

as responsible

only the Te

for the modiied

growth,

ascribing to it a su~actant role as found for As in the GelSi case

[3].

Their

observation

inte~retation

was

supported

by the

of around one monolayer of Te floating on the

InAs growth front. By preparing the InAs film on a GaAs surface

with

a Te coverage

slightly

lower

than

one

monolayer, we found a smaller amount of Te at the surface than at the interface. We propose and not the Te at the surface,

that this Te interlayer, plays the major role in

pointed

out

that

a necessary

[ 121, indicating a softer Te bonding. This chemical trend is systems

InAsiGaAs

has

and

la~ice-mismatch around

lattice-mismatch

InAs works

efficient

surface

segregation,

energies of both the substrate

reducing

requirement the sufiace

and overlayer

is an free

[lo]. In the

former

a

of 7.2 % and a ZD-3D transition for only [I]. The second has a

of 7.9 % and a ZD-3D transition

that the Te interlayer

embedded

for

supports the idea

between the GaAs and

as an elastic buffer to reduce

the energy

increase associated with the strain by allowing the bending of the bonds to the InAs layer.

In summary, we confirmed

evidenced

Another

The

about 6 Ml of ZnTe 1131. This iinther

incorporation,

or overlayer.

ZnTeiGaAs.

1.5 MI of InAs thickness

the epitaxial

either substrate

are

also exemplified by comparing the heteroepit~al

requeriment for a su~a~tant is a sufficient mobility to avoid so that it does not preferentiaIIy adsorb to

These force constants

related to the elastic constants ciiand ~1s. The force constants Cl for the Te-based II-VI compounds are about 40 % lower as compared to As-based III-V compounds

and stretching

changing the growth mode of InAs on GaAs. Cope1 et nE. [lo]

and bond bending respectively.

growth

the action of Te allowing

of InAs on GaAs,

the localization

and we have

of the Te at the interface,

excluding a standard surfactant action of that element as leading to the observed InAs on GaAs.

changes of the growth mode of

Vol. 95, No. 12 Acknowledgement

HETEROEPITAXIAL - We wish to thank

GROWTH

J. Massies, Y.

Petroff, H. Chacham, A. C. Ferraz and the permanent staff of LURE

for helpful

discussions

and strong

OF InAs ON GaAs( 100)

W.N.R. and V.H.E. were supported by CEE/CNPq-RHAE grants.

support. REFERENCES

1. N. Grandjean, J. Massies, and V.H. Etgens, Phys. Rev. Lett. 69, 799 (1992). 2. J. Massies, N. Grandjean, and V. H. Etgens, Appl. Phys. Lett. 61, 99 (1992). 3. M. Cope1 etnl., Phys. Rev. B 42, 11682 (1990). 4. see Landolt-Bortstein, JH 22-b: Impurities and Defects in Group IV and III-V Compounds; ed. M. Schulz; Springer-Verlag, Berlin, 1989. 5. Y. Gobil, J. Cibert, K. Saminadayar, and S. Tatarenko, Surface Sci. 211/212, 969 (1989). 6. J. Cibert, K. Saminadayar, S. Tatarenko, and Y. Gobil, Phys. Rev. B 39, 12047 (1989). 7. J.J. Yeh and I. Lindau, Atomic Data and Nuclear Data Tables 32, 1 (1985).

8. S. Tatarenko, P. H. Jouneau, and V. H. Etgens; to be published. 9. R.H.Miwa, A.C.Ferraz, W.N.Rodrigues and H.Chachan; to be published. 10.M. Copel, M. C. Reuter, Efthimios Kaxiras, and R. M. Tromp, Phys. Rev. Lett. 63, 632 (1989). 11 .J. Massies and N. Grandjean, Phys. Rev. B 48, 8502 (1993). 12.W. A. Harrison, “Electronic Structure and the Properties of Solids”, W. H. Freeman and Company, San Francisco, 1980; cap.8, pp 194-196. 13.V. H. Etgens, Dr. Thesis, Universite Pierre et Marie Curie - Paris VI, Paris, France, 1991.

877