Opt Quant Electron DOI 10.

1007/s11082-011-9460-0

Analysis of wetting layer effect on electronic structures of truncated-pyramid quantum dots

Qiuji Zhao Ting Mei Daohua Zhang Oka Kurniawan

Received: 27 September 2010 / Accepted: 16 March 2011 Springer Science+Business Media, LLC. 2011

Abstract We carry out a theoretical analysis of wetting layer effect on band-edge proles and electronic structures of InAs/GaAs truncated-pyramid quantum dots, including the strain effect. A combination of an analytical strain model and an eight-band Fourier transform-based k p method is adopted in the calculation. Strain modied band-edge proles indicates that wetting layer widens the potential well inside the dot region. Wetting layer changes groundstate energy significantly whereas modies probability density function only a little. The main acting region of wetting layer is just underneath the base of the dot. Wetting layer redistributes probability density functions of the lowest electron state and probability density functions of highest hole state differently because of the different action of quantum connement on electrons and holes. Keywords Electronic structures Fourier-transform based k p method InAs/GaAs quantum dots Wetting layer effect

1 Introduction Self-assembled quantum dots in the application of optoelectronics are usually fabricated by the Stranski-Krastanov (SK) epitaxial growth model (Bimberg et al. 1999). Taking the InAs/GaAs structure shown in Fig. 1 as an example, about a 1-2ML InAs wetting layer (WL)

Q. Zhao D. Zhang School of Electrical and Electronic Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore T. Mei (B ) Institute of Optoelectronic Material and Technology, South China Normal University, 51063 Guangzhou, Peoples Republic of China e-mail: ting.mei@ieee.org O. Kurniawan Institute of High Performance Computing, A*STAR, Singapore 117528, Singapore

123

Q. Zhao et al. Fig. 1 Geometric model of a truncated-pyramid InAs/GaAs quantum-dot structure. The bottom dimension, the top dimension, and the height of dot are 12 nm, 12 f nm, and 6(1 f ) nm, respectively. The thickness of WL is 0.5 nm. f is the truncation factor

z[001]

GaAs y[010] InAs x[100]

is grown on a substrate, followed by a spontaneous coherent InAs island formation process. Finally, a deposition of an additional GaAs material, which is usually the same material as the substrate, is adopted to cover the quantum-dot island. Deposition of the thin WL leads to mist strain at the interface of WL and substrate, and the associated strain energy increases rapidly as the WL thickens. As a critical thickness of WL is created, formation of the dot island can occur to relieve the strain energy (Matthews 1975). Therefore, there must be inuence of WL on strain distribution and electronic structures of quantum dots. However, WL is usually omitted or seldom discussed throughly by considering the very thin thickness of the WL in the past works (Pryor 1998; Andreev et al. 1999; Stier et al. 1999; Schlliwa et al. 2007). Therefore, it is necessary to examine the effect of WL completely, especially for those structures with high aspect ratios. WL effect on electronic structures of a very shallow InAs/GaAs lens-shape quantum dot has been investigated using the atomic strain model and the empirical tight-binding model that are much more complicated than the continuum strain model and the k p model (Lee et al. 2004). Consequently, we adopt an eight-band Fourier transform-based k p method (FTM) (Mei 2007; Zhao and Mei 2011) together with an analytical continuum strain model (Andreev et al. 1999) to study effect of WL on electronic structures of a series of truncated-pyramid InAs/GaAs quantum dots in this article. We rst investigate effect of WL on ground-state energies of quantum dots with different heights, and then we select half-truncated structures to nd the detail information of WL effect.

2 Computation method 2.1 Strain and band-edge proles For the strain calculation, we utilize the analytical method developed with Fourier-transform technique (Andreev et al. 1999). To see the detail position of conduction band (CB), HH band, LH band, and SO band after strain modication, we investigate the band-edge proles with following expressions (Chuang 1995), E c = E g + ac eh E H H = av eh beb E L H = av eh + beb E S O = av eh , (1)

where ac , av , b, and are the CB deformation potential, hydrostatic valence band deformation potential, shear deformation potential and spin-orbit splitting energy, respectively. eh

123

Analysis of wetting layer effect on electronic structures of truncated-pyramid quantum dots

and eb are the hydrostatic strain and biaxial strain expressed as, eh = ex x + e yy + ezz eb = ezz ex x + e yy /2. 2.2 Electronic structures Electronic structures of quantum dots are usually obtained by solving equation including envelop function F(r) and position-dependent Hamiltonian matrix H(r;k) with differential operators, H r; k x , k y , k z F(r ) = EF(r) , (3) (2)

where E is the energy. The envelop function F(r) of a QD superlattice can be expanded in plane waves along x, y, and z, F(r) = 1 eik x x+ik y y+ikz z d x d y dz cn x n y n z ei (n x x x+n y y y+n z z z ) ,
nx ny nz 2 d , a

(4)
a

where dx , d y , and dz are periods along x, y, and z directions, =

T nx n y nz

< k <

for = x, y, z, cn x n y n z = c1 c2 . . . c8 , and a is the lattice constant. Envelop function F(r) can be approximated using Fourier series and complex exponential components, and the Fourier transform of the spatially-varying Hamiltonian H(r;k) can be obtained by entering analytical Fourier transform of quantum dot shape functions into the Hamiltonian matrix H(r;k) of k p theory. Therefore, the matrix of Fourier domain Hamiltonian M for the eigen-problem can be formulated by a bra-ket operation to (3) utilizing the orthogonality of complex exponential function in F(r), and the new eigen-problem can be written as (Mei 2007; Zhao and Mei 2011), [Mst ][ct ] = E[ct ], where M is constructed with Fourier series of H(r;k) as follows, Mst = 1 2
x,y,z ,

(5)

1 () Huv, qi = m i n i K K + 2

x,y,z

() Huv, qi = m i n i K (6)

(0) + Huv, qi = m i n i , and, K K = 2k k + k (m + n ) + k (m + n ) + (m n + m n ) K = 2k + (m + n ) .

(7)

The dimension of M is only determined by the truncated order of Fourier frequencies of H(r;k), i.e. V (2N x + 1)(2N y + 1)(2Nz + 1) V (2N x + 1)(2N y + 1)(2Nz + 1), V is the number of total bands involved in the calculation, and N ( = x, y, z) is the truncated Fourier order. Compared the calculation using the tight-binding model, calculation with FTM is much easier and more convenient. The drawback of spurious solutions in k p theory can also be avoided by truncating the higher order of Fourier series (Zhao and Mei 2011).

123

Q. Zhao et al.

3 Results and discussion In this section, we rst discuss the inuence of WL on ground-state energies of the structure shown in Fig. 1 with different truncation factors ( f = 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9). Provided the same material, critical thickness of WL is not affected by the height of the geometry. Therefore, the WL with a thickness of 0.5 nm was chosen for all the structures. To make a clear illustration, we also calculate ground-state energies of the dot (structure without WL) with same truncation factors. Figure 2 shows ground-state energies of electron (E1) and heavy-hole (HH1) of dots with (solid lines) and without (dashed lines) WL under different truncation factors. It can be seen that the difference between the case with WL and the case without WL is obvious only at higher truncation ( f 0.5), especially for HH1due to the heavier effective mass of heavy-hole. That is to say, significant WL effect only appears when the thickness of WL is comparable with the height of the dot. To further investigate the detail information about effect of WL, we calculate ground-state energy of the half-truncated dot ( f = 0.5) with a cube just underneath its bottom (denoted as QDC). The thickness of cube is also 0.5 nm. Ground-state energy E1 and HH1 of QDC are tabulated in Table 1 with those of half-truncated structure shown in Fig. 1 (denoted as QDW) and half-truncated dot (denoted as QD). Figure 3 plots the strain components ex x and ezz of QD (dashed lines) and QDW (solid lines) along the [001] direction. Corresponding band-edge proles of QD (dashed lines) and QDW (solid lines) along the [001] direction after strain modication are also shown in Fig. 4. Since WL just causes the change of the initial

Ground-state energy E1 and HH1(eV)

0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

E1
Solid lines: with WL Dashed lines: without WL

HH1

Truncation factor f
Fig. 2 Ground-state energies E1 and HH1 of a series of truncated-pyramid InAs/GaAs dots with (solid lines) and without (dashed lines) inclusion of WL Table 1 Ground-state energy E1 and HH1 of QD, QDC, and QDW

Ground-state energy

Structures in calculation QD QDC 0.5099 0.6308 QDW 0.4972 0.6223

E1 (eV) HH1 (eV)

0.5472 0.6437

123

Analysis of wetting layer effect on electronic structures of truncated-pyramid quantum dots

0.04 0.02 0.00

exx

Strain along [001] direction

-0.02 -0.04 -0.06 0.04 0.02 0.00 -0.02 -0.04 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6

ezz

Distance(nm)
Fig. 3 Strain components ex x and ezz of QD (dashed lines) and QDW (solid lines) along the [001] direction
1

Band-edge profiles along [001](eV)

0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -1.2 -6 -4 -2 0

CB

HH LH SO
2 4 6

Distance (nm)
Fig. 4 Band-edge proles of QD (dashed lines) and QDW (solid lines) along [001] direction

lattice mismatch at the interface of dot and WL, WL just redistributes the strain prole in the WL region other than changes strain magnitude inside the dot area so much. Consequently, WL widens potential wells but keeps positions of band edges unchanged almost inside the dot region. As a result, ground-state energy of QDW in Table 1 varies a lot from that of QD, especially for ground-state energy of electron owing to the smaller electron mass. Additionally, it can be seen that discrepancy of ground-state energy between QD and QDW is much more than that between QDW and QDC. This indicates that the main inuence of WL on ground-state energy comes from the area just beneath the bottom of the dot. Figure 5a describes probability density functions (PDFs) corresponding to E1 of QD and QDW. After the hybrid 2D/0D WL connement (Cornet 2005), the center of the PDF of QDW shifts

123

Q. Zhao et al.

Fig. 5 PDFs corresponding to ground-state energy E1 of QD and QDW (Fig. 5a) and PDFs corresponding to ground-state energy HH1 of QD and QDW. The black lines make up the shape section of QD and QDW

towards the WL region and a very small portion of PDF in the barrier region moves into the region just beneath the bottom of the dot but not the whole WL region. Consequently, effect of WL to electronic structures mainly comes from the region just beneath the bottom of the dot. PDFs corresponding to HH1 of QD and QDW are shown in Fig. 5b. Similar as that in the Fig. 5a, the center of the PDF of QDW also moves towards the WL region. The most significant difference in the PDFs of E1 and HH1 is that that heavy-hole are conned more obvious than electron, this can be inferred from the smaller PDF volume and bigger PDF value on the colorbar. As a result, very small portion of PDF of HH1 lies in the barrier region and hence the PDF portion that locates in the WL is mainly from the dot region.

4 Conclusions In summary, we have adopted a continuum strain model and eight-band FTM to investigate the effect of WL on electronic structures of truncated-pyramid InAs/GaAs quantum dots. Compared with the tight-bind method, FTM with a simple and neat Hamiltonian matrix simplies the computation process, and reduces the computation time as well. FTM also can avoid spurious solutions in the k p theory handily. WL effect becomes obvious only as the thickness of WL is comparable with the height of the dot. Connement of WL changes the ground-state electron energy up to 0.05 eV. Due to the different quantum connement acting on electron and hole, WL drives a small portion of PDF of E1from barrier region into WL but drives a small portion of PDF of HH1 from the dot region in the WL. WL also widens the potential well and thus reduces the energy band-gap,

123

Analysis of wetting layer effect on electronic structures of truncated-pyramid quantum dots

the main inuence of WL on electronic structures is just from the region underneath the bottom of the truncated-pyramid dot. We therefore can conclude that our work has supported the previous study done by Lee et al. very well and also has supplemented the their study from a very different point of view.
Acknowledgments This work is supported by the Singapore National Research Foundation under CRP Award No. NRF-G-CRP 2007-01. Q. J. Zhao would like to thank the institute above for nancial support.

References
Andreev, A.D. et al.: Strain distributions in quantum dots of arbitrary shape. J. Appl. Phys. 86, 297305 (1999) Bimberg, D. et al.: Quantum Dots Heterostuctures. Wiley, Chichester (1999) Chuang, S.L.: Physics of Optoelectronic Device. Wiley, New York (1995) Cornet, C. et al.: Approach to wetting-layer-assisted lateral coupling of InAs/InP quantum dots. Phys. Rev. B 72, 035342035346 (2005) Lee, S. et al.: Effect of wetting layer on the strain and electronic structure of InAs self-assembled quantum dots. Phys. Rev. B 70, 125307125313 (2004) Matthews, J.W.: Epitaxial Growth. Academic Press, New York (1975) Mei, T.: Fourier transform-based kp method of semiconductor supperlattice electronic structure. J. Appl. Phys. 102, 053708053712 (2007) Pryor, C.: Eight-band calculations of strained InAs/GaAs quantum dots compared with one-, four-, and six-band approximations. Phys. Rev. B 57, 71907195 (1998) Schlliwa, A. et al.: Impact of size, shape, and composition on piezoelectric effect and electroni properties of In(Ga)As/GaAs quantum dots. Phys. Rev. B 76, 205324205340 (2007) Stier, O. et al.: Electronic and optical properties of strained quantum dots modeled by 8-band kp theory. Phys. Rev. B 59, 56885701 (1999) Zhao, Q.J., Mei, T., J_ID: J. Appl. Phys. (in press) (2011). doi:10.1063/1.3549686

123