• Aucun résultat trouvé

NONCOMMUTATIVE STRUCTURE OF GaAs QUANTUM WELL INTERFACES AND INEQUIVALENT INTERFACE IMPURITY INCORPORATION

N/A
N/A
Protected

Academic year: 2021

Partager "NONCOMMUTATIVE STRUCTURE OF GaAs QUANTUM WELL INTERFACES AND INEQUIVALENT INTERFACE IMPURITY INCORPORATION"

Copied!
5
0
0

Texte intégral

(1)

HAL Id: jpa-00226719

https://hal.archives-ouvertes.fr/jpa-00226719

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.

NONCOMMUTATIVE STRUCTURE OF GaAs QUANTUM WELL INTERFACES AND INEQUIVALENT INTERFACE IMPURITY

INCORPORATION

D. Bimberg, R. Bauer, D. Oertel, D. Mars, J. Miller

To cite this version:

D. Bimberg, R. Bauer, D. Oertel, D. Mars, J. Miller. NONCOMMUTATIVE STRUCTURE OF GaAs QUANTUM WELL INTERFACES AND INEQUIVALENT INTERFACE IMPURITY INCORPO- RATION. Journal de Physique Colloques, 1987, 48 (C5), pp.C5-93-C5-96. �10.1051/jphyscol:1987515�.

�jpa-00226719�

(2)

NONCOMMUTATIVE STRUCTURE OF GaAs QUANTUM WELL INTERFACES ANI) INEQUIVALENT INTERFACE IMPURITY INCORPORATION

D . BIMBERG, R . K . BAUER, D .

OERTEL,

D . MARS* and J . N .

MILLER'

Institut fiir Festkdrperphysik der Technischen Universitdt Berlin, 0-1000 Berlin 12, F.R.G.

' ~ e w l e t t - ~ a c k a r d Laboratories, Palo Alto, C A 94304, U.S.A.

RBsumB

Nous prouvons de manihre directe l a non-commutativitd des interfaces AlGaAsjGaAs e t GaAs/AIGaAs, par une analyse detaillee du p r o f i l des spectres de photoluminescence dans des puits quantiques. Ces puits ont une Bpaisseur L, = 5 nm, i l s sont obtenus par Bpitaxie par jets 1nol6culaires ?I 620°C, avec un taux de croissance de 1 pm/h, avec interruption de l a croissance soit

A

l'une des deux, aux deux ou encore 5 aucune des interfaces. Nous observons une difference dans l a structure cristallographique ainsi que dans I'incorporation des pieges selon l e type dinterface. Ltinterruption de l a croissance sur l'interface AlGaAs diminue l e rendement quantique des puits

.

Durant I'interruption de l a croissance, u n rapide lissage de l a surface ainsi que l a formation d'ilots de croissance de diametre superieur

a

celui de I'exciton bidimensionnel a l i e u pour les surfaces GaAs; au contraire pour les surfaces AlGaAs, aucune formation 8 i l o t s aussi 6tendus n'a pu e t r e observee dans les conditions actuelles de croissance de nos Bchantillons.

Abstract

D i r e c t proof o f the noncommutativity o f GaAsjAlGaAs and AIGaAsjGaAs interfaces o f quantum wells (QW's) o f width L,=5nm is obtained f r o m a careful studv and mathematical analysis o f the structure and s h G e o f photoluminescence (PL) spectra. QW'S were grown by molecular beam epitaxy (MBE) a t 620°C w i t h a r a t e o f l p m j h and the growth was interrupted a t either one, both, o r none o f the interfaces. A large difference i n the microscopic crystallographic and chemical structure o f the t w o types o f interfaces is observed. I n addition we observe inequivalent incorporation o f impurities and traps a t the t w o interfaces. Interrup- t i o n o f growth a t the AlGaAs surface deteriorates the quantum efficiency o f QW's. Analysis o f the P L lineshape gives clear evidence t h a t the AlGaAs growth surfaces are rougher than the GaAs ones. During growth interruption (GRI) a rapid smoothing and the formation o f monolayer growth islands w i t h l a t e r a l sizes larger than the diameter o f the ZD-exciton occurs only f o r GaAs surfaces i n the samples used f o r this study. AlGaAs surfaces also become smoother upon G R I b u t large island formation is not found a t the present set o f experimental and growth parameters.

Crucial properties o f novel two-dimensional devices based on heterostructures are controlled t o a large extent by the structure and chemistry o f their interfaces. F o r example:

-

Availability o f complernentary logic based on 111-V compounds is essential f o r future generations o f computers. Such a family of logic is based on "normalw and "inverted" two- dimensional-electron-gas f i e l d e f f e c t transistors (TEGFET's). Interface roughness scattering was recently discovered by Hong e t al./l/ t o dominate the m o b i l i t y i n inverted InGaAsjAlInAs structures which are amongst the most promising ones f o r such devices.

-

The spontaneous emission p r o f i l e and the gain p r o f i l e o f a quantum w e l l (QW) laser o f QW w i d t h L, = 6-8 n m are dominated by interface roughness induced broadening o f the QW energy levels. F o r a given f i x e d i n t e r f a c e roughness of T 1 monolayer the s t a t i s t i c a l contribution t o the luminescence h a l f w i d t h o f the (e,hh)-recombination increases f r o m

= 10

meV for L, = 6.5 n m t o

> l 0 0 meV f o r LZ

-

0.0 n m i n AlGaAsjGaAs QW's w i t h an A I content o f 41.2 % 121. The quantum efficiency o f spontaneous recombination on the other hand does not seem t o be

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1987515

(3)

JOURNAL DE PHYSIQUE

1

LW' GRDYlH INIERRPlIm4 R I R FREE SR=m (X

GO fls

BOTH7 G.? RIPR

":

strongly influenced by "intrinsic" interface roughness but by trap incorporation a t the inter- faces 131.

Growth kinetics and surface structure o f a semiconductor A are i n general n o t expected t o be identical t o that o f a chemically d i f f e r e n t semiconductor B. Thus A/B (A up, B down) and B / A interfaces are also not expected t o be identical. Evidence f o r such inequivalence was recently reported. The reports are based on evaluations o f RHEED /4,5/ and luminescence experiments 121. Zhang e t al. reemphasized recently that RHEED is purely qualitative and very d i f f i c u l t t o interpret 161. On the other hand we have demonstrated /7/ t h a t interface roughness distribution functions can be determined w i t h sub-A precision from an analysis o f QW luminescence lineshapes. O f course, a detailed understanding o f the character o f the luminescence (singlet, doublet) and the physics governing luminescence lineshapes is a prerequisite f o r such determina- tions /8/.

The purpose of the present paper is twofold. F i r s t we would l i k e t o demonstrate clearly the enormous s t r u c t u r a l difference between interfaces o f type A/B and B/A. We w i l l show how the contributions o f both interfaces t o the broadening o f optical spectra o f QW's can be separated by a clever combination o f growth and luminescence experiments. Secondly, detailed results f o r the roughness o f and i m p u r i t y incorporation a t AlGaAs and GaAs growth surfaces grown by MBE as a function o f the interruption t i m e o f the growth are reported.

ioo The combination GaAs/AlGaAs is

presently the most important one f o r

rl semiconductor heterostructure devices

D 75 and can be regarded as a model f o r

0

U

-

50 other heterostructures. M u l t i p l e QW's o f AlGaAs/GaAs/-

U

.-

AlGaAs were grown by MBE w i t h 0.28

01 nm/s growth r a t e a t 620°C on undoped

C

25 GaAs substrates (100 orientation). The

C A1 content i n the barrier is 40 %. The

H thickness o f the QW's is nominally 5

D nm, and they are separated b y 18 n m

1 60 i 62 i 62 i 63 AlGaAs layers. Growth was interrup- E n e r g y C e V l ted f o r 100 S a t none, one o r b o t h o f the interfaces. D e t a i l s o f the growth Fig. 1: Comparison o f P L spectra o f AIGaAs/- are published elsewhere /3/.The GaAs QW's w i t h 100s growth interruption at the photoluminescence was excited w i t h AIGaAs, a t the GaAs, a t both and a t neither o f the 647 n m line o f a Kr-laser w i t h a the growth surfaces. The FWHM are 4.3meV, typical excitation density o f 0.9 10.9meV, 3.7meV and 6.3meV, respectively. . . W/cm2. The s ~ e c t r a l resolution o f the

0.5 m mono=hromator was 0.06 nm.

o

I

p--.l-

v

The P L s ~ e c t r a o f four such otherwise

~m OROWTH wmnumw identical samples are shown i n Fig. 1.

AT A F= SURFACE OF The spectrum o f t h e sample without

arowth interruption (GRI) is broad, w i t h a F W H M equal t o 7 meV. G R I on the AlGaAs surface changes the spec- t r u m only quantitative1y:it shifts t o lower energies and i t s FWHM i s redu- ced t o 4 meV. I f g r o w t h is interrupted a t the GaAs surface we observe a qualitative change o f the spectrum:

t w o ( s t i l l broad) maxima appear. These t w o maxima sharpen considerably i f

3.60 (.M the sample is grown w i t h G R I on both

Magi IeVl interfaces. Growth interruption also causes enhanced i m p u r i t y incorpora- Fig. 2: Comparison o f P L spectra o f 5nm t i o n a t the interfaces /3/. These impu- AIGaAs/GaAs QWS w i t h 100s growth interruption rities are responsible for a number of at the AlGaAs and the GaAs surfaces on a extrinsic emission peaks shown in semilogarithmic scale. Fig.2 i n a semilogarithmic p l o t f o r a

(4)

pronounced i m p u r i t y incorporation, whereas during G R I a t the GaAs surface the i m p u r i t y incorporation i s much weaker. Some extrinsic luminescence can be identified i n Fig. 1 already on a linear scale. While f o r the samples without GRI these features are hardly observable they become clearly visible f o r samples w i t h G R I a t the AlGaAs and at both surfaces.

The integrated PL intensities o f the samples w i t h G R I a t the GaAs surface and without G R I are nearly but n o t completely equal, whereas G R I a t the AlGaAs surface reduces these intensities t o 45 % o f the i n i t i a l value. This strong drop o f the PL emission can be explained only b y rapid formation o f nonradiative centers a t the AlGaAs growth surface which reduces the quantum efficiency

11

and decay time. Our earlier observation /3/ o f a strong reduction o f 11 upon GRI at both interfaces is thus shown t o be predominantly related t o the AlGaAs growth surface.

Apparently t h e strong chemical r e a c t i v i t y o f A I gives rise t o a much stronger incorporation o f impurities and traps a t t h e AlGaAs surface than a t the GaAs surface.

The width o f the recombination and absorption bands o f QW1s are strongly affected b y the chemical and crystallographic roughness o f t h e i r interfaces /3,9,10/. We have demonstrated that luminescence lineshapes o f QW's can be described perfectly by a product o f the 2 0 density o f states step function convoluted b y a Gaussian describing the L, (i.e. energy) distribution caused by interface roughness w i t h a carrier distribution function. A f i t o f theoretical t o experimental iineshapes yields therefore; a/ the variance o f the energy distribution function U (E), b/ the mean transition energy E(L,) and c/ the carrier temperature.

The a (E) values thus obtained can be used t o calculate t h e variance o t t h e quantum w e l l width distribution function G Q (LZ). (IQ is equal t o 0.079 n m f o r the QW1s without GRI. With G R I a t the GaAs surface i t is reduced t o 0.066 nm. G Q becomes s t i l l smaller upon GRI a t the AlGaAs surface (0.056 nm) and a t both surfaces (0.036 nm). These results shall be f i r s t used t o test numerically our model and the consistency o f our method o f evaluation.

Ib\ GROWTH INTERRU-"^" lllTERFP

I

.

VALUES

Xf..RR>n 1

2 2 DISCRETE

p LUMINESCENCE

i o r . c. VEAKS

Fig. 3: Model f o r the interface disorder o f an AIGaAslGaAs QW and the energy states o f the X(e,hh) exciton i n such a w e l l grown (a) with- out and (b) w i t h interruption o f the growth.

The statistical distribution of QW-width results f r o m s t a t i s t i c a l distribution o f the t w o interface positions (see Fig. 3).

where G G and GA is the variance of the GaAs and the AlGaAs surface distribution functions, respectively. Now

G&OS,OS)- G :(100s,0s)= U;(OS)- G &(100s)=0.19h2

and G; (OS, OS)

-

G; (OS, 100s) = G: (OS)

- G L

(100s) = 0.51

W2.

The sum o f these

(5)

JOURNAL DE PHYSIQUE

t w o values [ G & (0 S)

-

ob, (100 S)]

+ [

G ~ ( O S)- ~ i ( 1 0 0 S)] = 0.50 must be equal

t o G; (0 S, 0 S)

-

G; (100 S, 100 S). Indeed we get from the results given above also 0.50

i2.

We w i l l use the GQ-values now t o obtain separate estimates of the roughness o f the uninterrupted GaAs and AlGaAs growth surfaces. F o r convenience we assume t h a t a f t e r 100 S

GR1 a t only one growth surface the contribution f r o m the corresponding i n t e r f a c e t o the width distribution function o f the QW given by Equ. 1 can be neglected. This assumption is justified by our observation /7/ t h a t Q almost shows no variation any more a t GRI times larger than 30 S.

We obtain G G (0 S) = 0.043nm and GA (0 S) = 0.056nm. A s t i l l more refined lineshape theory which includes Lorentzian broadening e f f e c t s yields somewhat smaller absolute values and a larger difference between GaAs and AlGaAs growth surface roughness.. These results w i l l be given elsewhere

1111.

The mean transition energy E(L,) o f the recombination o f n = 1 heavy-hole excitons depends f o r a given A I mole f r a c t i o n only on the QW-width L, i n the theoretical framework o f isolated undoped QW's between undoped barriers. The knowledge o f E (L,) allows us t o calculate the actual L,-value. F o r the noninterrupted QW's this value does not correspond t o an integral number o f GaAs monolayers b u t t o the mean value o f the L, distribution function.

A qualitatively d i f f e r e n t situation prevails i f the G R I takes place a t the GaAs surface. The origin o f the

two

exciton peaks which now appear, can be explained, i f one assumes a larger surface mobility o f Ga atoms as compared t o A I atoms. This m o b i l i t y leads t o a more rapid formation of "large" growth islands d i f f e r i n g i n height by roughly one monolayer (ML) on GaAs surfaces. The size o f these islands must be larger than the 2D-exciton diameter ( = l 7 nm) otherwise they escape observation. The difference o f the mean transition energies o f the t w o exciton bands is therefore AE (LZ) = E (L,)

-

E (L,

+

1 ML). I n the present case these t w o energies are 1.619 eV and 1.613 eV f o r samples w i t h G R I a t the GaAs surfaces and 1.614 eV and 1.6092 eV f o r the samples w i t h G R I a t both surfaces. The calculated energies f o r LZ1s equal t o 18 and 19 ML's (5.088 n m and 5.3706 nm) are 1.615 eV and 1.608 eV. Their difference is 7 meV, which compares favourably w i t h the t w o experimental differences o f 6.0 meV and 5.7 meV.

Since QW's w i t h G R I a t the AlGaAs surface have emission spectra w i t h only one maximum, we conclude t h a t growth islands larger than 17 n m do not develop a t this surface during the interruption. The surface diffusion velocity o f the A I and Ga atoms a t the present growth conditions is nevertheless high enough t o e f f i c i e n t l y reduce the i n i t i a l roughness o f the AlGaAs surface.

We would l i k e t o acknowledge helpful discussions w i t h J.Christen. The work a t B e r l i n was founded by D F G i n the framework o f SFB6.

REFERENCES

Ill W.P. Hona. J. Sinah. and P.K. Bhattacharva: <, <,

.

, ADD^. Phvs. Lett. i n print

D. Bimberg, D. Mars, J.N. Miller, R. Bauer,

6:

~ e r i e l , and J. christen; Superl. Microstr.

3, 79 (1987)

D. Bimberg, D. Mars, J.N. Miller, R. Bauer, and D. Oertel; J. Vac. Technol.

84,

1014 (1986) F. Voillot, A. Madhukar, J.Y. Kim, P. Chen, N.M. Cho, W.C. Tang, and P.G. Newman;

Appl. Phys. Lett.

48,

1009 (1986)

T. Fukunaga, K.L. l. Kobayashi, and H. Nakashima; Jpn. J. Appl. Phys. 24, L 510 (1985) J. Zhang, J.H. Neave, P.J. Dobson, and B.A. Joyce; Appl. Phys.

S,

31x1987)

D. Bimberg, J. Christen, T. Fukunaga, H. Nakashima, D.E. Mars, and J.N. Miller; Proc.

PCSI 14, Salt Lake C i t y 1987, t o be published i n J. Vac. Sci. Techn.,

85.

(1987)

R.K. Bauer, D. Bimberg, J. Christen, D. Oertel, D. Mars, J.N. Miller, T. Fukunaga, and H.

Nakashima; Proc. 18th Intern. Conf. Semic., Stockholm (ed.: 0. Engstrom), World Scientific

-

Singapore,

1,

525 (1986)

C. Weisbuch, R. Dingle, A.C. Gossard, and W. Wiegmann; Solid State Comm., 2 , 7 0 9 (1981) A. Juhl, D. Oerte1,'C. Maczey, D. Bimberg, K. Carey, R. Hull, and G.A. Reid; Superl.

Micrastr. Vo1.3,No.3,p.205(1987)

.

J. Christen and D. Bimberg, t o be published.

Références

Documents relatifs

We may note from the figures that for a given concentration x the values of the exciton binding energies have the same qualitative behavior as those previously obtained for

Abstract.- The temperature behavior of the photoluminescence from GaAs-GaAIAs multiple quantum well structures indicates that the maintaining of the excitonic properties of

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

On the other hand, the experimental determination of S(0) from the structure factor data yields a value of 0.03 at the lowest after a supplementary rough isotropic correction for

Fig.. modes of the SSQW expected on the basis of presently known behavior of the GaAs and A10,32Ga0.68As optical phonon branches. A detailed discussion of the

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

We have calculated the binding energies of electrons, holes and excitons on a quantum well interface defect, modelled by a cylindrical protusion.. The effect of

&amp;tween SD and UD sample is observed. The lifetime shortening arises frcm the enhancement of the radiative recanbinaticn rate. This means that 2-dimensional electron