RAMAN SCATTERING IN MBE GaInAs-InP QUANTUM WELLS

HAL Id: jpa-00226684

https://hal.archives-ouvertes.fr/jpa-00226684

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.

RAMAN SCATTERING IN MBE GaInAs-InP QUANTUM WELLS

M. Watt, C. Sotomayor Torres, P. Hatton, H. Vass, P. Claxton, J. Roberts

To cite this version:

M. Watt, C. Sotomayor Torres, P. Hatton, H. Vass, P. Claxton, et al.. RAMAN SCATTERING IN MBE GaInAs-InP QUANTUM WELLS. Journal de Physique Colloques, 1987, 48 (C5), pp.C5-483- C5-486. �10.1051/jphyscol:19875102�. �jpa-00226684�

JOURNAL DE PHYSIQUE

Colloque C5, supplbment au n0ll, Tome 48, novembre 1987

R A M A N SCATTERING IN MBE GaInAs-InP QUANTUM WELLS

M. WATT, C.M. SOTOMAYOR TORRES, P.D. HATTON*, H. VASS*, P.A. C L A X T O N * ~ and J.S. R O B E R T S * ~

Physics Dept., University of St Andrews, GB-St' Andrews, KY16 9SS, Fife, Great-Britain

'physics Dept., University of Edinburgh, GB-Edinburgh. EH9 3 3 2 , Great-Britain

* * Electronics and Electrical Engineering Dept., University of Sheffield, GB-Sheffield, S1 3 J D , Great-Britain

Abstract

The phonon modes in GaInAs-InP heterostructures were studied by means of Rarnan scatterin

.

The mode frequencies were found to occur at the bulk values and to be well-defined even for layers of 251.

The phonon intensity ratios were found to depend very strongly on the incident laser power and we attribute this to screening of the modes by photoexcited carriers.

Introduction

Rarnan scattering studies have been made of two MBE-grown G Q . ~ ~ I ~ ~ , ~ ~ A S - I ~ P quantum well structures: a single and a multiple quantum well. These structures are relevant for optoelectronic device applications and are therefore of current interest. Tri le crystal x-ray work on the sin le quantum well is reported elsewhere [l] and yielded a value of 230bPfor the well thickness and 6 8 0 i for the cap layer thickness (the nominal figures given by the growth time were 160A and 500A respectively). The multiple quantum well consisted of 50 periods of 25A well and l00A barrier with a l000A cap layer (nominal values). We studied the sample in standard backscattering geometry using the 4880A (2.54eV) line of an Argon ion laser and the 48251 (2.57eV) line of a Krypton ion laser. These lines lie close to the El (2.62eV) resonance of the GalnAs and thus allow an enhanced signal from the buried layer. They also provide a reasonable penetration of the sample estimated at about 700A in InP and about 200A in GaInAs [2]. The scattered light was analysed by a l m double spectrometer and detected by a cooled photomultiplier tube operating in photon counting mode. The etch used to remove the InP was 3: 1 H3PO4 : HC1 and that for the GaInAs was 80% 3:l H2S04 :

ffiO

There is still uncertainty about the nature of the phonon modes in GaInAs. Kakimoto and Katoda [3] showed clear evidence for two-mode behaviour across the composition range of the alloy and this is endorsed by Landa et a1 [4]. Pearsall et al[5] demonstrated a "mixed-mode" behaviour, however, with one clear GaAs-like L 0 mode, one clear InAs-like TO mode and two indeterminate modes. In our bulk GaInAs sample we observe two strong modes (see Fig. la) and we identify the dominant mode as the GaAs-like L 0 phonon at 265cm-1 and the second mode as the InAs-like L 0 phonon at 232cm-l, after Kakimoto and Katoda [3]. Both peaks obey the L 0 selection rule for backscattering: i.e. allowed in the z(x,y)z configuration and forbidden in the z(x,x)z configuration. Although in both configurations scattering from TO phonons is forbidden, it is usually observed in the Raman spectra of most 111-V alloys.

Results and Discussion

The Raman spectrum of the single quantum well was markedly different from that of the bulk; both L 0 phonon modes had been quenched to the extent that they had become indistinguishable from the forbidden TO modes (see Fig. lb). This quench was a consistant feature of the spectra from the single quantum well and persisted when we varied the frequency of the exciting light. Initially, the quench was thought to be due to a mechanism similar to that found in the magnetophonon experiments of Portal et a1 [6], where the authors found that as the electron confinement increased, so the electrons penetrated further into the barrier layers and interacted wlth the InP L 0 phonons rather than the GaAs-like L 0 phonons. This

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:19875102

C5-484 JOURNAL DE _PHYSIQUE

mechanism does not explain the total quench of the L 0 modes, nor does it explain why this should occur for such a large well width. In order to investigate the influence of the InP barriers, we selectively etched off the InP cap layer. The spectrum of the etched sample is shown in Fig. lc. After removal of the cap, the phonon modes of the alloy layer resembled-those of the bulk material with both the L 0 modes occurring at the bulk frequencies. The complete removal of the quench suggested that the barriers did not play the major role in causing the quench and hence an explanation other than penetration of the barriers must be sought.

The effect of the incident laser power on the intensity of the GaAs-like L 0 phonon mode is shown in Fig. 2a. This suggested that the laser powers used in the experiment might be sufficiently high so as to create a photoexcited electron hole plasma (Em) within the well which would screen the L 0 phonons. This was c o n f i e d by monitoring both the half width at half maximum and the integrated intensity of the photoluminescence (PL) emission from the well as a function of incident laser power (see Figs. 2b and 2c).

For the Raman intensity and the PL half width, the onset of a saturation behaviour occurred with a power of between 50 and 80mW incident on the sample. The Raman spectra from the single quantum well were recorded within this saturation regime, thus strongly suggesting a screening effect of the L 0 mode. This saturation is not evident in the plot of the PL integrated intensity .as a function of incident power and we conclude that the carrier density within the excited volume increases to a saturation value after which the plasma expands so as to keep a constant density.

In order to make a simple estimate of the saturation density of the photoexcited EHP, we used the assumptions that each absorbed photon creates one electron-hole pair, that these decay to the bottom of the conduction band within 10 picoseconds and then remain there for 10 nanoseconds before recombining. This yields a value of 2.4~1010 electron-hole pairs1Watt. From the absorption lengths in GaInAs and InP [2] we were able to determine where the carriers would be created. For the 57mWatts of our experiment, this yields a density of 1.9~1017 cm-3 in the InP cap layer and a 2D density of 1.3~1012 cm-2 in the well. This density is certainly of the correct order of magnitude to populate the well and screen the phonons [8], and about ten times that determined by Schubnikov-de Haas measurements.

Fig. l

Raman scattering spectra of: (a) bulk Gao,471no.53As. The peaks are identified in the text. (b) 230A G Q . ~ ~ I ~ ~ . ~ ~ A ~ single quantum well with 680A InP cap layer. -(c) as (b) but with the cap layer removed. (d) as (c) but with the well layer also removed.

The dominant peaks in (a) are quenched in (b);

the strong mode at 350cm-1 is the InP L 0 phonon from the cap layer. With the InP cap layer removed by etching (c) the modes are no longer quenched thus suggesting a mechanism in which the barriers do not play a dominant role. Finally, in spectrum (d) it can be seen that once the alloy layer is removed, there is no visible structure in the 200-300cm-1 region of the spectrum.

l I

200 250 300 350

Frequency (cm-' )

''InP '~)/I(ClaAs-l~ke LO) HWHM (meV) 15.

10

30 20

10 (b)

>

0 50 100 1% 200 Power Into sample (mW)

Power Into sample (rnW

Integrated Intensity

T

4

Fig. 2 3-

(a) Intensity of the GaAs-like L 0 mode normalised to the InP L 0 mode of the cap _2.

layer. (b) PL half width at half maximum. (c) PL integrated intensity. Each is plotted as a function of incident laser power. 1.

I -O

.

100 200 300 400 500 600 >

Power Into samp(e (mW)

The effect on the surface depletion width of such an electron concentration was estimated using the expressions given by Schubert et a1 [9]. Making an order of magnitude assumption about the surface potential (1 eV), the depletion width was calculated to be about 500A in the InP cap layer. This puts the quantum well beyond the surface depletion width thus suggesting that it could be sufficiently populated to screen the L 0 phonon scattering.

Screening of the L 0 phonon by free carriers in bulk semiconductors is a well-understood phenomenon [see, eg., 101; with increased carrier density, the plasma frequency increases, first approaching and then becoming greater than the phonon frequency. This results in an anticrossing behaviour and yields the standard W+ and W- modes. In two dimensions (2D), this effect is different. The reduced dimensionality yields slab-like plasma oscillations of the 2D electron gas (2DEG) in the plane of the layer. In order to couple to these, one must have a component of scattering wavevector, q, parallel to the 2D layer [l l]. This is not the case for the backscattering geometry of our experiment. Since the electron energy levels perpendicular to the plane of the layer are quantised, we would expect to couple to intersubband excitations which have the parallel component of q approximately equal to zero. Previous experimental studies [12,13], have investigated this effect in GaAs-AlGaAs heterostructures, where it was demonstrated that the collective intersubband excitations coupled strongly to the L 0 phonon modes. We suggest that it is this coupling that causes the quench of the L 0 modes in our Rarnan spectra.

Our experimental parameters are inappropriate to observe this electronic scattering since our experiments were all done at ambient temperature. Liquid helium temperatures are needed to ensure that the quantised well energy levels are free from thermal broadening which tends to smear out any structure. This constraint would also apply to the 2D plasma oscillations [l l]. Low temperature work is in progress.

In the multiple quantum well sample, the thick cap layer of InP (1000A) was removed together with the first well leaving l00A of InP on the sample surface. The spectrum observed is shown in fig 3a.

The InP barrier was subsequently etched off and the spectrum observed is shown in fig 3b. It is clear that these spectra do not significantly differ in either mode frequency or intensity ratio. Moreover, the phonon modes of the alloy layers have the same frequencies as those in the bulk, there is no shift of the modes towards the spectrum of the quaternary alloy, GaInAsP, thus suggesting that the layers are well-defined and the interfaces are sharp. There is also no evidence of a GaAs-like L 0 mode quench which is further evidence that the InP barrier layers do not play a significant role in this process.

It is worth noting that a 2cm-1 shift to lower frequencies of the InP L 0 phonon mode is observed

JOURNAL DE PHYSIQUE

Fig. 3

Raman scattering spectra of a

, Gao.471no.53As-InP multiple quantum well sample with 50 periods of 25A well and l00A barrier. (a) l00A InP barrier on surface. (b) 25A GaInAs well on surface. These spectra do not significantly differ in either mode frequency or intensity ratio.

I l

250 300 350

Frequency (cm-' )

in scattering from a cap or barrier layer with respect to a buffer layer or bulk sample. We have looked at various possible explanations for this shift including strain effects induced by variations in alloy composition, As contamination of cap and barrier Iayers and the formation of GaInAsP interface layers.

Each of these possible mechanisms would, however, also give rise to other, much stronger, effects than the shift of the InP L 0 mode and we do not observe any of these other effects. The origin of the shift of the InP L 0 mode is not yet clear and is under further investigation.

Conclusions

We have observed well-defined phonon frequencies from a GaInAs single and multiple quantum well and these occur at the bulk GaInAs values. There is no evidence of graded interface regions even in 25A layers. The phonon intensity ratios strongly depend on the free carrier concentration with high laser powers generating a sufficiently dense EHP to screen the dominant L 0 phonon mode in the well.

Acknowledgements

We acknowledge financial support from the SERC for this work. One of us (M Watt) is supported by a SERC-IT studentship.

References

1. T W Ryan, P D Hatton, S Bates, M Wan, C M Sotomayor Torres, P A Claxton, J S Roberts: Semicond Sci & Techno 2 241:1987

2. D E Aspnes, A A Studna: Phys Rev B-= 985: 1983 3. K Kakimoto, T Katoda: Appl Phys Lett

a

4. G Landa, R Carles, J B Renucci: Proc 18th ICPS, Stockholm: 1986 5. T P ~earkall, R ~ G l e s , J C Portal: Appl Phys L& Q 436: 1983 6. 3 C Portal, G Gregoris, M A Brummell, R J Nicholas, M Razeghi,

M A di Forte-Poisson, K Y Cheng, A Y Cho: Proc EP8,Tokyo:1985 7. A J Taylor, J M Wiesenfeld: Phys Rev B 3 2321: 1987

8. M S Skolnick. K J Nash. P R Ta~ster. L ,D J Mowbrav, S J Bass. A D Pitt:

P ~ ~ S R ~ V B

&

9. E F Schubert, K Ploog, H Dambkes, K Heime: Appl Phys 63: 1984 10. A Mooradian, A L McWhorter: Phys Rev Lett

B

11. A Pinczuk, M G Lamont, A C Gossard: Phys Rev Lett 2092: 1986 12. A Pinczuk, J Shah, A C Gossard, W Wiegmann: Phys Rev Lett 46 1341:1981 13. Ch Zeller, G Abstreier, K Ploog: Surf Sci 113 85: 1982