Physica E 7 (2000) 408412

www.elsevier.nl/locate/physe

Photocurrent spectroscopy of InAs= GaAs self-assembled quantum dots: observation of a permanent dipole moment
P.W. Frya , I.E. Itskevichb; a , D.J. Mowbraya; , M.S. Skolnicka , J.A. Barkerc , E.P. OReillyc , L.R. Wilsona , P.A. Maksymd , M. Hopkinsone , M. Al-Khafajie , J.P.R. Davide , A.G. Cullise , G. Hille , J.C. Clarke
b Institute a Department of Physics, University of She eld, She eld S3 7RH,UK for Solid State Physics, Russian Academy of Sciences, Chernogolovka, Moscow District 142432, Russia c Department of Physics, University of Surrey, Guildford GU2 5XH, UK d Department of Physics, University of Leicester, Leicester LE1 7RH, UK e Department of Electronic and Electrical Engineering, University of She eld, She eld S1 3JD, UK

Abstract The nature of the conned electronic states in InAs=GaAs self-assembled quantum dots is studied using photocurrent spectroscopy measured as a function of applied electric eld. A eld asymmetry of the quantum conned Stark e ect is observed, consistent with the dots possessing a permanent dipole moment. The sign of this dipole indicates that for zero eld the hole wave function lies above that of the electron, in disagreement with the predictions of all recent calculations. Comparison with a theoretical model demonstrates that the experimentally determined alignment of the electron and hole can only be explained if the dots contain a non-zero and non-uniform Ga content. ? 2000 Elsevier Science B.V. All rights reserved. PACS: 73.61.r; 78.66.w; 73.50.Pz; 68.95.+g Keywords: Quantum dots; Optical spectroscopy; Electronic structure

Self-assembled quantum dots (QDs) provide nearly ideal systems for the study of novel physics in 0D structures [1]. Although a number of theoretical studies of the electronic structure of these QDs have been reported [25], uncertainties exist concerning the input parameters, particularly the dot shape, size and com Corresponding author. Tel.: +44-114-222-4561; fax: +44-114-272-8079. E-mail address: d.mowbray@she eld.ac.uk (D.J. Mowbray)

position. Hence, experimental studies are required to provide a test of the underlying assumptions of these theoretical models. In this paper we describe the use of photocurrent spectroscopy to measure absorption-like spectra of InAs QDs and the response of these spectra to large (300 kV=cm) electric elds. An asymmetrical Stark shift of the optical transitions is observed, implying that the dots possess a permanent dipole moment. By comparison with a theoretical model we nd that the sign of this dipole can only be explained if the

1386-9477/00/$ - see front matter ? 2000 Elsevier Science B.V. All rights reserved. PII: S 1 3 8 6 - 9 4 7 7 ( 9 9 ) 0 0 3 5 1 - 3

P.W. Fry et al. / Physica E 7 (2000) 408412

409

Fig. 1. (a) cross-sectional TEM micrograph of an uncapped dot sample, (b) a schematic band diagram of a pin device.

nominal InAs dots have a signicant and non-uniform Ga content. Nominal InAs self-assembled QDs were grown by molecular beam epitaxy on (0 0 1) GaAs sub strates at 500 C. The dots were deposited at 0.01 monolayers per second (ML=s) to give a density 1:5 109 cm2 , base size 18 nm and height 8.5 nm (see Fig. 1a), as determined from transmission electron microscopy. The asymmetric shaped quantum dots, sitting on a 1 ML thick wetting layer, have their apex oriented along the growth direction. Single-dot layers were grown within the intrinsic region of either a pin or nip structure, allowing elds up to 300 kV=cm to be applied either parallel or anti-parallel to the growth direction. Applying a reverse bias to a pin structure (p region at surface) results in an electric eld (F) pointing from substrate to surface (see Fig. 1b). For an nip structure the eld direction is reversed. The total electric eld is given by F = (V + Vbi )=d, where V is the externally applied voltage, Vbi is the built-in voltage ( 1:5 V) and d(=0:3 m) is the intrinsic region width. Monochromated white light ( 3 mW=cm2 , bandwidth 8 meV) from a tungsten

Fig. 2. Photocurrent spectra as a function of applied reverse bias, for temperatures of 5 and 200 K. The inset is a V = 0 V, T = 5 K photocurrent spectrum showing the wetting layer and GaAs absorption.

halogen lamp was focussed onto 400 m diameter, annular contact mesas and the resultant photocurrent was detected using lock-in techniques (fmod 200 Hz), allowing the detection of very low ( 1 pA) currents. The low incident power results in extremely low dot carrier occupancies ( 1) and hence many carrier e ects can be neglected [6]. Fig. 2 shows photocurrent spectra, recorded as a function of applied reverse bias from 0 to 8 V (total electric eld F from 50 to 320 kV=cm) for a pin sample at temperatures of 5 and 200 K. At low temperatures and elds, carriers are unable to escape from the dots and a photocurrent signal is only observed from the InAs wetting layer (1.41 eV) and the GaAs barriers (1.5 eV) (see inset). However, as the eld is increased carriers are able to escape from the dots by tunnelling and four well-dened features in the energy range 1.11.3 eV (at 5 K) are observed. These features arise from inter-band QD transitions between conned electron and hole states. Similar behaviour is observed at 200 K, although the QD transitions are observed at increasingly lower elds as the temperature

410

P.W. Fry et al. / Physica E 7 (2000) 408412

Fig. 3. Transition energies as a function of electric eld for nip (negative eld) and pin (positive eld) samples at a temperature of 200 K. The data for the nip sample has been rigidly shifted by 7 meV:1 The solid line is a theoretical t to the experimental data. The dashed line is a quadratic t to the data.

is raised, re ecting a transition in the dominant carrier escape mechanism from tunnelling at low temperatures to thermal excitation at high temperatures. With increasing eld, and at all temperatures, all the QD transitions shift strongly to lower energy (by 30 meV at 8 V ( 300 kV=cm)), the signature of the quantum conned Stark e ect [7], but without any qualitative change in the overall form of the spectra. The 200 K ground state transition energies for a pin and an nip 1 sample are plotted as a function of eld, calculated using Vbi = 1:5 V, in Fig. 3. The combination of these two samples allows both negative and positive elds to be applied to the QDs, with positive eld corresponding to reverse bias of the pin structure. The transition energy exhibits a marked asymmetry about F = 0, with the maximum value obtained for a non-zero eld of 90 kV=cm. This asymmetry implies that the self-assembled quantum dots have a permanent dipole moment (p). A non-zero Stark shift at F = 0 of the ground state tran1 A small shift of 7 meV to higher energy has been applied to the results of sample B, in order to obtain a continuous variation of the peak positions between positive and negative electric elds. Samples A and B were grown consecutively in order to obtain the minimum possible run-to-run variation in dot parameters between samples. The observed energy di erence of 7 meV corresponds to only a 2:5% variation in dot base size.

sition has also been observed in the PL of individual InAlAsAlGaAs QDs [8] which could only be studied over the more limited eld range of 60 kV=cm due to quenching of the PL at high elds. In contrast, the present photocurrent measurements allow both ground and excited state transitions to be studied over a much larger eld range. The eld dependence of the transition energies (E) in Fig. 3 can be described to a very reasonable approximation (dashed line in Fig. 3) by the equation E = E0 + pF + F 2 where E0 is the transition energy at zero eld, the second term arises from the non-zero dipole moment (p), and the third term () arises from polarisation of the dots in the applied eld (the quantum conned Stark e ect). By tting the above equation to the experimental data in Fig. 3 a value of p = (7 2) 1029 Cm is determined, corresponding to an electronhole separation of r = 4:0 1 A obtained from p = er. A permanent dipole moment is predicted from theoretical modelling of self-assembled QDs, due to their non-uniform shape along the eld direction [2]. However, the sign of the dipole moment deduced from the present measurements is opposite to that predicted by previous theoretical studies of pure InAs dots. The maximum transition energy in Fig. 3 occurs for a negative eld, corresponding to a eld direction from apex to base. For this eld direction the electron is attracted to the apex (hole attracted to the base) of the dots. This result implies that the electron charge density distribution lies closer to the base than that of the hole at F = 0, with the resultant dipole pointing from base to apex. This nding, to the best of our knowledge, is in contrast to that predicted by all accurate theoretical modelling [25] of InAs QDs which predict hole localisation towards the base of the dots, below the electrons. This alignment, which occurs for pure InAs dots and for any shape for which the lateral size decreases from base to apex (e.g. the pyramidal shape used in the models of Refs. [25]) results from the strain-induced form of the valence band edge prole [3] and the ratio of the e ective masses along the growth direction (m m ). e hh To determine the conditions necessary to reverse the relative alignment of the electron and hole, the QD shape, size and compositional dependence of the permanent dipole moment (p) and the quadratic eld coe cient were calculated using the envelope function

P.W. Fry et al. / Physica E 7 (2000) 408412

411

method, treating the electrons and holes with separate one-band Hamiltonians. The strain distribution for a given dot shape was calculated using a Greens function technique which provides an analytical expression in the form of a Fourier series for the strain tensor [9]. The band gaps and o sets were calculated using model solid theory [10], including hydrostatic strain e ects; the heavy-hole Hamiltonian included the spatial variation of the biaxial strain deformation potential and the directional dependence of the heavy-hole mass. Carrier e ective masses, determined using 3-band k p theory, and band o sets were assumed to vary linearly with composition. For pure InAs pyramidal dots our calculations give good agreement with previous theories [25] with the hole wave function always below that of the electron. Reversal of this alignment is found to be possible only for dots having a graded Inx Ga1x As composition, with x increasing from base to apex (the holes tend to be localised in the region with the largest In composition). In addition, it is also necessary to severely truncate the pyramid to achieve a dipole of the correct sign (strain e ects localise the hole strongly below the electron until the truncation factor is greater than 0:6). The continuous line in Fig. 3 shows the best t to the experimental data. This was calculated using a pyramid of base length 15.5 nm, height 22 nm, of which the top 75% is truncated to give a height of 5.5 nm, and an In mole fraction which varies linearly from 50% at the base to 100% at the (truncated) top surface. has a value 2:5 107 eV kV2 cm2 . These parameters give a good t to the experimental data. Although other combinations of size, shape and composition may give a similar quality of t, the present shape represents a good approximation to that obtained from structural measurements (see Fig. 1). However, the main conclusion, namely that a non-zero and non-uniform Ga composition and a truncated shape are required to give the correct vertical alignment of the electron and hole wave functions, is independent of the precise parameters used for the t. A permanent dipole moment, having a sign identical to that determined in the present work, has been determined from PL measurements of individual InAlAsAlGaAs self-assembled QDs [8]. Although agreement was obtained between the experimentally determined sign of the dipole and the predictions of a theoretical model, this agreement is entirely

fortuitous, arising from the neglect of strain e ects in the model. However, the experimental results of Ref. [8] support the generality of the present ndings. Evidence for the presence of Ga in nominal InAs dots has recently been reported by Joyce et al. [11] from STM studies. For the growth temperature used for the present QDs (500 C), the total volume of the dots is found to be greater than that of the deposited InAs, consistent with an indi usion of material from the surrounding GaAs. The grading required to explain the sign of the experimentally observed dipole moment (a greater Ga concentration at the dot base) has been observed in STM studies of In0:3 Ga0:7 As QDs [12] and attributed to In segregation e ects during growth. In conclusion, a photocurrent spectroscopic study of self-assembled InAs=GaAs QDs, as a function of applied electric eld, has demonstrated that the QDs possess a permanent dipole moment. Although such a dipole moment is expected due to the asymmetrical shape of the dots, the sign of the dipole is opposite to that predicted by all recent calculations. From a theoretical study of the dependence of the dot properties on shape, size, and composition, we demonstrate that the correct sign of the dipole moment can only be obtained if it is assumed that the dots contain a non-zero and non-uniform Ga composition. Acknowledgements We wish to thank Dr. A.D. Andreev for useful discussions and acknowledge the nancial support of the Engineering and Physical Sciences Research Council (EPSRC) UK (Grant GR=L28821) and the Russian Foundation for Fundamental Research (Grant 97-02-17802) and GNTP Nanostructures (Russia, grant 97-1068). References
[1] D. Bimberg, M. Grundmann, N.N. Ledentsov, Quantum Dot Heterostructures, Wiley, New York, 1998. [2] M. Grundmann, O. Steir, D. Bimberg, Phys. Rev. B 52 (1995) 11 969. [3] M. Cusack, P.R. Briddon, M. Jaros, Phys. Rev. B 54 (1996) R2300. [4] C. Pryor, Phys. Rev. B 57 (1998) 7190. [5] J. Kim, L.-W. Wang, A. Zunger, Phys. Rev. B 57 (1998) R9408.

412

P.W. Fry et al. / Physica E 7 (2000) 408412 [9] A.D. Andreev, J.R. Downes, D.A. Faux, E.P. OReilly, J. Appl. Phys., unpublished. [10] M.P.M.C. Krijn, Semicond. Sci. Technol. 6 (1991) 27. [11] P.B. Joyce, T.J. Krzyzewski, G.R. Bell, B.A. Joyce, T.S. Jones, Phys. Rev. B 58 (1998) R15 981. [12] N. Grandjean, J. Massies, O. Tottereau, Phys. Rev. B 55 (1997) R10 189.

[6] R. Warburton, C.S. Durr, K. Karrai, J.P. Kotthaus, G. Mideiros-Ribeiro, P.M. Petro , Phys. Rev. Lett. 79 (1997) 5282. [7] D.A.B. Miller, D.S Chemla, T.C. Damen, A.C. Gossard, W. Wiegmann, T.H. Wood, C.A. Burrus, Phys. Rev. B 32 (1985) 1043. [8] S. Raymond, J.P. Reynolds, J.L. Merz, S. Fafard, Y. Feng, S. Charbonneau, Phys. Rev. B 58 (1998) R13 415.