- Email: [email protected]
Resonant photocurrent-spectroscopy of individual CdSe quantum dots
Physica E 42 (2010) 2521–2523
Contents lists available at ScienceDirect
Physica E journal homepage: www.elsevier.com/locate/physe
Resonant photocurrent-spectroscopy of individual CdSe quantum dots M. Panfilova n, S. Michaelis de Vasconcellos, A. Pawlis, K. Lischka, A. Zrenner University of Paderborn, Department of Physics and Center of Optoelectronics and Photonics Paderborn (CeOPP), Warburger Straße 100, 33098 Paderborn, Germany
a r t i c l e in fo
abstract
Article history: Received 14 September 2009 Received in revised form 29 November 2009 Accepted 9 January 2010 Available online 15 January 2010
Here we report on investigations on CdSe quantum dots incorporated in ZnSe based Schottky photodiodes with near-field shadow masks. Photoluminescence and photocurrent of individual quantum dots were studied as a function of the applied bias voltage. The exciton energy of the quantum dot ground state transition was shifted to the excitation energy by using the Stark effect tuning via an external bias voltage. Under the condition of resonance with the laser excitation energy we observed a resonant photocurrent signal due to the tunnelling of carriers out of the quantum dots at electric fields above 500 kV/cm. & 2010 Elsevier B.V. All rights reserved.
Keywords: CdSe/ZnSe quantum dots Photodiode Quantum confined Stark Effect Photocurrent II–VI Semiconductors
1. Introduction The concept of the optical and electrical manipulation of quantum-states on the nano-scale and their applications in quantum information systems has recently become a challenging topic in physics. One approach to investigate semiconductor qubits makes use of a two-level system, which is formed by the exciton ground state in a single quantum dot (QD). With individual quantum dots enclosed in a Schottky photodiode (PD) with near-field shadow masks on a semi-transparent contact, electrical and optical access is provided to probe and to modify the quantum state of QD-excitons. With well-defined resonant laser pulse (i.e., a p-pulse) acting on the ground state of a QD, the exciton-qubit can be controlled. After the qubit preparation, readout of the quantum state is possible by measuring the resulting photocurrent (PC) of the PD. Recently, clear evidence for the successful manipulation and readout of the quantum state of an exciton in a single InGaAs/ GaAs QD PD was demonstrated [1–3]. The relevant time scale for the coherent manipulations is basically determined by the coherence time T2, the radiative recombination time Trec and the time constant of the tunnelling escape Tesc. For efficient electric readout, the tunnelling process must be significantly faster than the radiative recombination. To perform a maximum number of coherent manipulations within the dephasing time, the duration of the excitation pulse has to be as short as possible. Furthermore, simultaneous resonant pumping of the exciton and two-photon
n
Corresponding author. Tel.: + 49 5251 6058 33; fax: + 49 5251 6058 43. E-mail address: [email protected] (M. Panfilova).
1386-9477/$ - see front matter & 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.physe.2010.01.013
biexciton transitions has to be avoided to preserve the integrity of the single exciton-qubit, which is considered here. Therefore the spectral line width of the excitation pulse has to be significantly smaller than the exciton/biexciton energy splitting. Thus, the excitation laser pulse width is limited to the ps-range in InGaAs/ GaAs QD PDs (corresponding to a typical biexciton binding energy of about 3 meV). In CdSe QDs the biexciton binding energy is known to be considerably higher, about 25 meV. This allows a significant decrease of the excitation pulse duration by nearly a factor of ten. Due to the higher confinement energies in CdSe/ZnSe QDs [4], the operation of such a device may also be possible at considerably higher temperatures [5] as compared to typical GaAs-based III–V quantum systems. In this paper we present a detailed study of the general parameters of CdSe/ZnSe QD PDs. In particular we determine the critical electric field for the transition from photoluminescence (PL) to PC in CdSe/ZnSe QD PDs.
2. Experimental The CdSe QD samples were grown on n-doped (2 1018 cm 3) (0 0 1) GaAs substrate using standard MBE. Fig. 1 shows a schematic view of the multilayer structure. At first, a 20 nm thick fluorine doped n-ZnSe buffer layer was deposited to decrease the band bending between ZnSe and GaAs. Next, 40 nm intrinsic ZnSe was deposited followed by the Stranski–Krastanov (SK) dot formation [6] from 2.1 monolayers of CdSe. Finally, the QDs were capped by 30 nm undoped ZnSe. To decrease the Schottky diode area, silicon dioxide was deposited by plasma enhanced chemical vapour deposition and holes with about 50 mm diameter etched
2522
M. Panfilova et al. / Physica E 42 (2010) 2521–2523
Fig. 1. Schematic cross-section drawing of a single QD PD on the basis of a n-i-Schottky diode. Optical access to single QDs is provided by a shadow mask.
into the oxide. The Schottky barrier was formed by a 6 nm thick semi-transparent gold layer on top of the structures. For the spatial isolation of individual QDs an aluminium shadow mask was fabricated using e-beam lithography. The Ohmic back contacts were processed by depositing indium on the GaAs substrate. Using this diode design we have obtained leakage currents of less than 10 pA at room temperature and less than 0.5 pA at 4 K up to a reverse bias voltage of 5 V, corresponding to an internal electric field of approximately 700 kV/cm. We applied a negative bias voltage between the Schottky contact and the substrate to investigate the states of individual QDs and to perform PC experiments. In particular at low bias voltages the single particle electron levels of the QDs are shifted already above the Fermi level of the n-doped back contact (the charging regime is not considered further here). We performed m-PL spectroscopy for the optical characterization of the QDs. Therefore, the sample was placed in a lowtemperature microscope at T=4.2 K. The laser excitation was focused onto the shadow mask by a NA= 0.75 microscope objective, which also collected the PL in a confocal geometry. A diode laser with an emission wavelength of 404 nm was used for the above bandgap cw excitation of the sample. Furthermore a laser pumped optical parametric oscillator (OPO) was used for pulsed and tunable excitation between 515 and 560 nm. The laser pulses had a repetition frequency of 160 MHz, a temporal width of about 3 ps, and a spectral width of about 3 meV. The PL was dispersed by a 0.5 m spectrograph and detected by a liquidnitrogen-cooled charge coupled device (CCD). The spectral resolution of the setup was about 250 meV.
3. Results and discussion The PL spectrum observed in the unstructured region of the sample is inhomogeneously broadened to 70 meV as shown in Fig. 2. Separated lines appear at the low-energy tail of the PL spectrum. By using the shadow masks, we restricted the active region of the sample to apertures of about 500 nm diameter and observed clearly separated lines. For the given QD density of about 109 cm 2 the use of shadow mask is essential to select individual QDs for a detailed PC spectroscopy. PL measurements with different applied bias voltages were performed to investigate the Stark shift of QD-excitons. We found that the tuning behaviour of the PL signal depends on the excitation energy. Under the condition of cw excitation at 404 nm (above the bandgap of ZnSe) no Stark shift of the exciton lines was observed for reverse bias voltages between 0 and 4 V. This effect is obviously caused by screening of the internal electric field. The exact mechanism for this screening (internal charge accumulation, lateral potential drop, etc.) remains unclear until now and is, as a matter of fact, not in the focus of our investigation
Fig. 2. PL spectrum of an ensemble of CdSe QDs.
Fig. 3. Voltage-dependent PL mapping of individual QDs. The QCSE is observed for each of the lines within the spectral region with increasing reverse bias voltage.
presented here. For the interesting regimes of nearly resonant excitation (typically 200 meV above the ground state) or resonant excitation of the CdSe QDs, the above-described screening effects are almost absent. Fig. 3 shows a two-dimensional map of the PL energy of the QD-excitons for nearly resonant excitation as a function of the bias voltage. We observe several transition lines with different strengths, most likely caused by contributions from different dots. In the following we discuss the strongest exciton line labelled (X), which we assign to the ground state of an individual QD. As shown in Fig. 3, the quantum confined Stark effect (QCSE) of the exciton transition (X) causes an almost linear redshift of the QD ground state with increasing reverse bias voltage. Between 0 and 3 V bias, we observe a Stark shift of about 4.4 meV. Since this Stark shift is almost linear, we can further conclude that the QD exciton has an almost constant induced dipole moment p =er, where e is the elementary charge and r is the separation between the centers-of-gravity of the electron and hole probability distributions. Based on the model by Skolnick et al. [7] we calculated a separation r = 0.13 nm from the observed Stark shift. This value is considerably smaller than the previously observed value in typical InGaAs/GaAs devices. The intensity of the neutral exciton (X) first increases up to a reverse bias voltage of about 1 V and further decreases for still higher reverse bias voltage. The initial increase of intensity could be caused by charging effects (competing emission from charged
M. Panfilova et al. / Physica E 42 (2010) 2521–2523
2523
depletion layer of our device. The field-induced ionization of QD-excitons occurs in our experiment at an internal electric field of about 600 kV/cm, making the subsequent avalanche multiplication of the PC very likely. In comparison to the wellknown InGaAs/GaAs [8] structures, CdSe/ZnSe PDs hence require internal electric fields for PC extraction, which are almost 20 times larger. Both the low leakage current at room temperature and the enhanced PC output could make widegap CdSe/ZnSe PDs very attractive for future coherent applications.
4. Conclusions
Fig. 4. Photocurrent spectra of an individual QD when excited resonantly to the ground state energy at 4 V (red curve). The blue curve corresponds to the photocurrent of non-resonant excitation of the QD and the grey curve is the resulting dark current without excitation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
states). For even larger reverse bias voltages, the intensity of the neutral exciton is reduced, since tunnelling of carriers out of the QD is now competing with the radiative decay of the excitons. For reverse bias conditions exceeding 3 V corresponding to the internal electric fields in the order of 500 kV/cm, the PL is effectively quenched due to tunnelling of carriers out of the QD (transition to the PC regime). Fig. 4 shows the results of PC measurements taken at the same aperture of the shadow mask as the PL data (see Fig. 3). The grey curve shows a very low leakage current without excitation. For a non-resonant excitation condition (blue curve), we observe a PC of less than 10 pA, even at the reverse bias voltage up to 5 V. This background PC is most likely due to the contribution of QDs, which are more remote from the aperture. The red curve shows the PC for resonant excitation of the QD. Using the QCSE, the QD exciton transition energy was tuned into resonance with the laser energy at a bias voltage of about 4 V. We observe a strong PC signal at about 4.1 V applied bias voltage. This signal has, as a matter of fact, a clear resonant signature as expected for a QD exciton. The peak amplitude of this PC signal (about 100 pA) is considerably higher as expected even for p-pulse excitation (about 25 pA for the given laser pulse repetition frequency of 160 MHz) [1]. We interpret the strongly enhanced peak amplitude of the PC signal as a strong evidence for the appearance of avalanche multiplication of photoelectrons and holes in the
The optoelectronic properties of ZnSe Schottky diodes containing SK self assembled CdSe QDs were investigated. At quasiresonant excitation of individual QDs, we observed a pronounced redshift of the PL emission with increasing negative bias voltage due to the QCSE. At a bias voltage of 3 V (corresponding to an applied electric field of about 500 kV/cm) the PL is quenched due to depletion of the exciton ground state by tunnelling. For resonant excitation of the QDs by 3 ps laser pulses, we observe a pronounced photocurrent signal at a bias voltage of about 4.1 V, which we consider as a first demonstration of electric readout of widegap CdSe QDs incorporated in a ZnSe based Schottky diode.
Acknowledgements The authors like to acknowledge financial support by the BMBF via 01BM917 and 01BM466 and German Research Foundation (DFG), Project GRK 1464. References [1] A. Zrenner, E. Beham, S. Stufler, F. Findeis, M. Bichler, G. Abstreiter, Nature 418 (2002) 612. [2] S.J. Boyle, A.J. Ramsay, A.M. Fox, M.S. Skolnick, A.P. Heberle, M. Hopkinson, Phys. Rev. Lett 102 (2009) 207401. ¨ [3] A. Zrenner, P. Ester, S. Michaelis de Vasconcellos, M.C. Hubner, L. Lackmann, S. Stufler, M. Bichler, J. Phys.: Condens. Matter 20 (2008) 454210. [4] X. Zhong, R. Xie, Y. Zhang, T. Basche, W. Knoll, Chem. Mater. 17 (2005) 4038. [5] K. Sebald, P. Michler, T. Passow, D. Hommel, G. Bacher, A. Forchel, Appl. Phys. Lett. 81 (2002) 2920. [6] T. Makino, R. Andre´, J.-M. Ge´rard, R. Romestain, Le Si Dang, M. Bartels, K. Lischka, D. Schikora, Appl. Phys. Lett 82 (2003) 2227. [7] R. Oulton, J.J. Finley, A.D. Ashmore, I.S. Gregory, D.J. Mowbray, M.S. Skolnik, M.J. Steer, S.-L. Liew, M.A. Migliorato, A.J. Cullis, Phys. Rev. B 66 (2002) 045313. [8] A. Zrenner, S. Stufler, P. Ester, S. Michaelis de Vasconcellos, M.C. Huebner, M. Bichler, Phys. Status Solidi (B) 243 (2006) 3696.
Contents lists available at ScienceDirect
Physica E journal homepage: www.elsevier.com/locate/physe
Resonant photocurrent-spectroscopy of individual CdSe quantum dots M. Panfilova n, S. Michaelis de Vasconcellos, A. Pawlis, K. Lischka, A. Zrenner University of Paderborn, Department of Physics and Center of Optoelectronics and Photonics Paderborn (CeOPP), Warburger Straße 100, 33098 Paderborn, Germany
a r t i c l e in fo
abstract
Article history: Received 14 September 2009 Received in revised form 29 November 2009 Accepted 9 January 2010 Available online 15 January 2010
Here we report on investigations on CdSe quantum dots incorporated in ZnSe based Schottky photodiodes with near-field shadow masks. Photoluminescence and photocurrent of individual quantum dots were studied as a function of the applied bias voltage. The exciton energy of the quantum dot ground state transition was shifted to the excitation energy by using the Stark effect tuning via an external bias voltage. Under the condition of resonance with the laser excitation energy we observed a resonant photocurrent signal due to the tunnelling of carriers out of the quantum dots at electric fields above 500 kV/cm. & 2010 Elsevier B.V. All rights reserved.
Keywords: CdSe/ZnSe quantum dots Photodiode Quantum confined Stark Effect Photocurrent II–VI Semiconductors
1. Introduction The concept of the optical and electrical manipulation of quantum-states on the nano-scale and their applications in quantum information systems has recently become a challenging topic in physics. One approach to investigate semiconductor qubits makes use of a two-level system, which is formed by the exciton ground state in a single quantum dot (QD). With individual quantum dots enclosed in a Schottky photodiode (PD) with near-field shadow masks on a semi-transparent contact, electrical and optical access is provided to probe and to modify the quantum state of QD-excitons. With well-defined resonant laser pulse (i.e., a p-pulse) acting on the ground state of a QD, the exciton-qubit can be controlled. After the qubit preparation, readout of the quantum state is possible by measuring the resulting photocurrent (PC) of the PD. Recently, clear evidence for the successful manipulation and readout of the quantum state of an exciton in a single InGaAs/ GaAs QD PD was demonstrated [1–3]. The relevant time scale for the coherent manipulations is basically determined by the coherence time T2, the radiative recombination time Trec and the time constant of the tunnelling escape Tesc. For efficient electric readout, the tunnelling process must be significantly faster than the radiative recombination. To perform a maximum number of coherent manipulations within the dephasing time, the duration of the excitation pulse has to be as short as possible. Furthermore, simultaneous resonant pumping of the exciton and two-photon
n
Corresponding author. Tel.: + 49 5251 6058 33; fax: + 49 5251 6058 43. E-mail address: [email protected] (M. Panfilova).
1386-9477/$ - see front matter & 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.physe.2010.01.013
biexciton transitions has to be avoided to preserve the integrity of the single exciton-qubit, which is considered here. Therefore the spectral line width of the excitation pulse has to be significantly smaller than the exciton/biexciton energy splitting. Thus, the excitation laser pulse width is limited to the ps-range in InGaAs/ GaAs QD PDs (corresponding to a typical biexciton binding energy of about 3 meV). In CdSe QDs the biexciton binding energy is known to be considerably higher, about 25 meV. This allows a significant decrease of the excitation pulse duration by nearly a factor of ten. Due to the higher confinement energies in CdSe/ZnSe QDs [4], the operation of such a device may also be possible at considerably higher temperatures [5] as compared to typical GaAs-based III–V quantum systems. In this paper we present a detailed study of the general parameters of CdSe/ZnSe QD PDs. In particular we determine the critical electric field for the transition from photoluminescence (PL) to PC in CdSe/ZnSe QD PDs.
2. Experimental The CdSe QD samples were grown on n-doped (2 1018 cm 3) (0 0 1) GaAs substrate using standard MBE. Fig. 1 shows a schematic view of the multilayer structure. At first, a 20 nm thick fluorine doped n-ZnSe buffer layer was deposited to decrease the band bending between ZnSe and GaAs. Next, 40 nm intrinsic ZnSe was deposited followed by the Stranski–Krastanov (SK) dot formation [6] from 2.1 monolayers of CdSe. Finally, the QDs were capped by 30 nm undoped ZnSe. To decrease the Schottky diode area, silicon dioxide was deposited by plasma enhanced chemical vapour deposition and holes with about 50 mm diameter etched
2522
M. Panfilova et al. / Physica E 42 (2010) 2521–2523
Fig. 1. Schematic cross-section drawing of a single QD PD on the basis of a n-i-Schottky diode. Optical access to single QDs is provided by a shadow mask.
into the oxide. The Schottky barrier was formed by a 6 nm thick semi-transparent gold layer on top of the structures. For the spatial isolation of individual QDs an aluminium shadow mask was fabricated using e-beam lithography. The Ohmic back contacts were processed by depositing indium on the GaAs substrate. Using this diode design we have obtained leakage currents of less than 10 pA at room temperature and less than 0.5 pA at 4 K up to a reverse bias voltage of 5 V, corresponding to an internal electric field of approximately 700 kV/cm. We applied a negative bias voltage between the Schottky contact and the substrate to investigate the states of individual QDs and to perform PC experiments. In particular at low bias voltages the single particle electron levels of the QDs are shifted already above the Fermi level of the n-doped back contact (the charging regime is not considered further here). We performed m-PL spectroscopy for the optical characterization of the QDs. Therefore, the sample was placed in a lowtemperature microscope at T=4.2 K. The laser excitation was focused onto the shadow mask by a NA= 0.75 microscope objective, which also collected the PL in a confocal geometry. A diode laser with an emission wavelength of 404 nm was used for the above bandgap cw excitation of the sample. Furthermore a laser pumped optical parametric oscillator (OPO) was used for pulsed and tunable excitation between 515 and 560 nm. The laser pulses had a repetition frequency of 160 MHz, a temporal width of about 3 ps, and a spectral width of about 3 meV. The PL was dispersed by a 0.5 m spectrograph and detected by a liquidnitrogen-cooled charge coupled device (CCD). The spectral resolution of the setup was about 250 meV.
3. Results and discussion The PL spectrum observed in the unstructured region of the sample is inhomogeneously broadened to 70 meV as shown in Fig. 2. Separated lines appear at the low-energy tail of the PL spectrum. By using the shadow masks, we restricted the active region of the sample to apertures of about 500 nm diameter and observed clearly separated lines. For the given QD density of about 109 cm 2 the use of shadow mask is essential to select individual QDs for a detailed PC spectroscopy. PL measurements with different applied bias voltages were performed to investigate the Stark shift of QD-excitons. We found that the tuning behaviour of the PL signal depends on the excitation energy. Under the condition of cw excitation at 404 nm (above the bandgap of ZnSe) no Stark shift of the exciton lines was observed for reverse bias voltages between 0 and 4 V. This effect is obviously caused by screening of the internal electric field. The exact mechanism for this screening (internal charge accumulation, lateral potential drop, etc.) remains unclear until now and is, as a matter of fact, not in the focus of our investigation
Fig. 2. PL spectrum of an ensemble of CdSe QDs.
Fig. 3. Voltage-dependent PL mapping of individual QDs. The QCSE is observed for each of the lines within the spectral region with increasing reverse bias voltage.
presented here. For the interesting regimes of nearly resonant excitation (typically 200 meV above the ground state) or resonant excitation of the CdSe QDs, the above-described screening effects are almost absent. Fig. 3 shows a two-dimensional map of the PL energy of the QD-excitons for nearly resonant excitation as a function of the bias voltage. We observe several transition lines with different strengths, most likely caused by contributions from different dots. In the following we discuss the strongest exciton line labelled (X), which we assign to the ground state of an individual QD. As shown in Fig. 3, the quantum confined Stark effect (QCSE) of the exciton transition (X) causes an almost linear redshift of the QD ground state with increasing reverse bias voltage. Between 0 and 3 V bias, we observe a Stark shift of about 4.4 meV. Since this Stark shift is almost linear, we can further conclude that the QD exciton has an almost constant induced dipole moment p =er, where e is the elementary charge and r is the separation between the centers-of-gravity of the electron and hole probability distributions. Based on the model by Skolnick et al. [7] we calculated a separation r = 0.13 nm from the observed Stark shift. This value is considerably smaller than the previously observed value in typical InGaAs/GaAs devices. The intensity of the neutral exciton (X) first increases up to a reverse bias voltage of about 1 V and further decreases for still higher reverse bias voltage. The initial increase of intensity could be caused by charging effects (competing emission from charged
M. Panfilova et al. / Physica E 42 (2010) 2521–2523
2523
depletion layer of our device. The field-induced ionization of QD-excitons occurs in our experiment at an internal electric field of about 600 kV/cm, making the subsequent avalanche multiplication of the PC very likely. In comparison to the wellknown InGaAs/GaAs [8] structures, CdSe/ZnSe PDs hence require internal electric fields for PC extraction, which are almost 20 times larger. Both the low leakage current at room temperature and the enhanced PC output could make widegap CdSe/ZnSe PDs very attractive for future coherent applications.
4. Conclusions
Fig. 4. Photocurrent spectra of an individual QD when excited resonantly to the ground state energy at 4 V (red curve). The blue curve corresponds to the photocurrent of non-resonant excitation of the QD and the grey curve is the resulting dark current without excitation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
states). For even larger reverse bias voltages, the intensity of the neutral exciton is reduced, since tunnelling of carriers out of the QD is now competing with the radiative decay of the excitons. For reverse bias conditions exceeding 3 V corresponding to the internal electric fields in the order of 500 kV/cm, the PL is effectively quenched due to tunnelling of carriers out of the QD (transition to the PC regime). Fig. 4 shows the results of PC measurements taken at the same aperture of the shadow mask as the PL data (see Fig. 3). The grey curve shows a very low leakage current without excitation. For a non-resonant excitation condition (blue curve), we observe a PC of less than 10 pA, even at the reverse bias voltage up to 5 V. This background PC is most likely due to the contribution of QDs, which are more remote from the aperture. The red curve shows the PC for resonant excitation of the QD. Using the QCSE, the QD exciton transition energy was tuned into resonance with the laser energy at a bias voltage of about 4 V. We observe a strong PC signal at about 4.1 V applied bias voltage. This signal has, as a matter of fact, a clear resonant signature as expected for a QD exciton. The peak amplitude of this PC signal (about 100 pA) is considerably higher as expected even for p-pulse excitation (about 25 pA for the given laser pulse repetition frequency of 160 MHz) [1]. We interpret the strongly enhanced peak amplitude of the PC signal as a strong evidence for the appearance of avalanche multiplication of photoelectrons and holes in the
The optoelectronic properties of ZnSe Schottky diodes containing SK self assembled CdSe QDs were investigated. At quasiresonant excitation of individual QDs, we observed a pronounced redshift of the PL emission with increasing negative bias voltage due to the QCSE. At a bias voltage of 3 V (corresponding to an applied electric field of about 500 kV/cm) the PL is quenched due to depletion of the exciton ground state by tunnelling. For resonant excitation of the QDs by 3 ps laser pulses, we observe a pronounced photocurrent signal at a bias voltage of about 4.1 V, which we consider as a first demonstration of electric readout of widegap CdSe QDs incorporated in a ZnSe based Schottky diode.
Acknowledgements The authors like to acknowledge financial support by the BMBF via 01BM917 and 01BM466 and German Research Foundation (DFG), Project GRK 1464. References [1] A. Zrenner, E. Beham, S. Stufler, F. Findeis, M. Bichler, G. Abstreiter, Nature 418 (2002) 612. [2] S.J. Boyle, A.J. Ramsay, A.M. Fox, M.S. Skolnick, A.P. Heberle, M. Hopkinson, Phys. Rev. Lett 102 (2009) 207401. ¨ [3] A. Zrenner, P. Ester, S. Michaelis de Vasconcellos, M.C. Hubner, L. Lackmann, S. Stufler, M. Bichler, J. Phys.: Condens. Matter 20 (2008) 454210. [4] X. Zhong, R. Xie, Y. Zhang, T. Basche, W. Knoll, Chem. Mater. 17 (2005) 4038. [5] K. Sebald, P. Michler, T. Passow, D. Hommel, G. Bacher, A. Forchel, Appl. Phys. Lett. 81 (2002) 2920. [6] T. Makino, R. Andre´, J.-M. Ge´rard, R. Romestain, Le Si Dang, M. Bartels, K. Lischka, D. Schikora, Appl. Phys. Lett 82 (2003) 2227. [7] R. Oulton, J.J. Finley, A.D. Ashmore, I.S. Gregory, D.J. Mowbray, M.S. Skolnik, M.J. Steer, S.-L. Liew, M.A. Migliorato, A.J. Cullis, Phys. Rev. B 66 (2002) 045313. [8] A. Zrenner, S. Stufler, P. Ester, S. Michaelis de Vasconcellos, M.C. Huebner, M. Bichler, Phys. Status Solidi (B) 243 (2006) 3696.