Home
Search
Collections
Journals
About
Contact us
My IOPscience
Fabrication and optical characterization of large scale membrane containing InP/AlGaInP quantum dots
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2015 Nanotechnology 26 235201 (http://iopscience.iop.org/0957-4484/26/23/235201) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 132.203.227.62 This content was downloaded on 14/06/2017 at 00:08 Please note that terms and conditions apply.
You may also be interested in: Single-photon emission from electrically driven InP quantum dots epitaxially grown on CMOS-compatible Si(001) M Wiesner, W-M Schulz, C Kessler et al. Visible single-photon generation from semiconductor quantum dots Thomas Aichele, Valéry Zwiller and Oliver Benson In-assisted deoxidation of GaAs substrates for the growth of single InAs/GaAs quantum dot emitters Tian Xia (), YongJin Cho, Michele Cotrufo et al. Engineered quantum dot single-photon sources Sonia Buckley, Kelley Rivoire and Jelena Vukovi Time resolved emission at 1.3 mu m of a single InAs quantum dot by using a tunable fibre Bragg grating G Muñoz-Matutano, D Rivas, A L Ricchiuti et al. Quantum optics with single quantum dot devices Valéry Zwiller, Thomas Aichele and Oliver Benson Template method for nano-order positioning and dense packing of quantum dots for optoelectronic device application Kohki Mukai, Akinobu Hirota, Yuta Shimizu et al. Epitaxially Grown Indium Phosphide Quantum Dots on a Virtual Ge Substrate Realized on Si(001) Michael Wiesner, Moritz Bommer, Wolfgang-Michael Schulz et al.
Nanotechnology Nanotechnology 26 (2015) 235201 (5pp)
doi:10.1088/0957-4484/26/23/235201
Fabrication and optical characterization of large scale membrane containing InP/ AlGaInP quantum dots H Niederbracht, F Hargart, M Schwartz, E Koroknay, C A Kessler, M Jetter and P Michler Institut fuer Halbleiteroptik und Funktionelle Grenzflaechen, University of Stuttgart, Allmandring 3, 70569 Stuttgart, Germany E-mail: [email protected] Received 12 January 2015, revised 16 March 2015 Accepted for publication 30 March 2015 Published 21 May 2015 Abstract
Single-photon sources with a high extraction efficiency are a prerequisite for applications in quantum communication and quantum computation schemes. One promising approach is the fabrication of a quantum dot containing membrane structure in combination with a solid immersion lens and a metal mirror. We have fabricated an 80 nm thin semiconductor membrane with incorporated InP quantum dots in an AlGaInP double hetero barrier via complete substrate removal. In addition, a gold layer was deposited on one side of the membrane acting as a mirror. The optical characterization shows in detail that the unique properties of the quantum dots are preserved in the membrane structure. Keywords: semiconductor membrane, InP quantum dot, III/V semiconductor (Some figures may appear in colour only in the online journal) 1. Introduction
in the InAs/GaAs [4] material system. On the other hand, the complete substrate can be removed as shown for InAs quantum dots in an InP membrane [5, 6]. They are used, e.g., in photonic crystal structures to enhance the confinement in the z-direction [7] and to create high Q-cavities [8] and waveguides [9]. Therefore, these kinds of structures are suitable to enhance the unique properties of semiconductor QDs [10]. In this paper, we report on a process to fabricate largesized (cm) AlGaInP membranes containing InP QDs via substrate removal. These QDs have already proven their ability to emit non-classical light [11], have confirmed singlephoton emission up to elevated temperature of 100 K [12] and showed variable emission energies from red to green [13]. The additional benefit of the QDs used in this paper is the red spectral emission range where the detection efficiency for Sibased avalanche photo diodes (Si-APDs) is the highest to date [14]. To facilitate efficient collection from single nanostructures, InP QDs with reduced spatial density were embedded into the membrane. Comparison of micro-photoluminescence (μ-PL) measurements of bare and processed
The need for higher collection efficiency of single-photon components has led to a variety of different device approaches [1]. One approach is to fabricate a thin semiconductor membrane containing quantum dots (QDs) in combination with a solid immersion lens (SIL) and a metal mirror as suggested by Chen et al [2]. Recently the same group reported about reaching 99% collection efficiency for giant CdSe/CdS QDs with a dielectric antenna sample in combination with a thin membrane [3]. Nowadays, several techniques to fabricate a semiconductor membrane have been established. The fabrication of such membrane structures typically includes electron beam lithography combined with dry- and wet-chemical etching steps, which limit the processed size to several millimeters. The main issue concerning the process is the concerted removal of the layers below the active membrane region. Basically, two approaches are possible. The first one is realized by partly removing a sacrificial layer, e.g., AlGaAs, by wet chemical etching to create a freestanding membrane part. This was demonstrated, for example, 0957-4484/15/235201+05$33.00
1
© 2015 IOP Publishing Ltd Printed in the UK
H Niederbracht et al
Nanotechnology 26 (2015) 235201
samples revealed the integrity of the QDs in the membrane structure. Furthermore, time-resolved PL measurements demonstrate nearly the same carrier dynamics in both samples, and the second-order autocorrelation proves the emission from a single QD. The design of our structure is inspired by the scheme of Chen et al [2]. They forecast the realization of 99% collection efficiency for a suitably processed membrane antenna sample. For a semiconductor-based QD, a membrane with a metal mirror has to be fabricated with an adapted refractive index. As our structure fulfills these requirements, we anticipate, in combination with a SIL with a suitable refractive index, a higher collection efficiency for further measurements.
(AFM) image (figure 1) of an uncapped sample with similar growth conditions. The lateral dimension of the large QDs is in the order of ≈50 nm, whereas it is ≈30 nm for the small QDs. After completion of the ripening, the first upper barrier was grown at the same growth temperature as the QDs to prevent the material intermixing between the different layers, while the lower and the second upper barrier were deposited at 710 °C. The barriers consist of (AlxGa1 −x ) 0.51In 0.49P with an aluminum content of x = 0.5 directly surrounding the nanostructure with a thickness of 10 nm and a 30 nm thick outer barrier with x = 0.7 aluminum content. A schematic of the structure can be seen in figure 2. In the next process step 200 nm of SiO2 were sputtered on top of the semiconductor. After this 5 nm of Cr and 200 nm of Au are evaporated. The Cr serves as adhesion promoter for the Au, which acts as a mirror and is prerequisite for the high efficiency source described before. This mirror provides a measured reflectivity between 93.1 and 95.6% in the range of 630 to 670 nm. Then the sample is mechanically thinned to approximately 100 μm and glued with nail polish to an Si carrier wafer. The nail polish is sufficiently resistant to the following wet chemical etching steps and also does not negatively affect them. Afterwards, the sample is etched with an etchant consisting of H2SO4:H2O:H2O2 at 70 °C to a total thickness of 10 μm. This etch is non-selective; therefore, some sidewall protection of the sample through the polish is favorable. For the last 10 μm we change to a NH4OH:H2O2 mixture to selectively etch the GaAs down to the (Al 0.7Ga 0.3) 0.51In 0.49P barrier, which acts as an etch-stopping layer. After this step, we end with a 80 nm thick semiconductor membrane combined with a gold mirror at the bottom. A scanning electron microscopy (SEM) picture of the fully processed sample is shown in figure 3. The nicely flat and clean etched surface can be seen on top of the layer package. As this process only uses mechanical thinning and wet etching, it is cost efficient, easy and has no limitations in sample size.
2. Epitaxy and process
3. Optical characterization
The samples were deposited by metal-organic vapor-phase epitaxy (MOVPE) on a standard GaAs substrate with 6 ◦ miscut towards the [111]A direction, which is to prevent ordering in the AlGaInP layers [15]. The InP QDs were grown self-assembled in the Stranski–Krastanow growth mode [16] at a temperature of 610 °C. To reduce the spatial density of the nanostructures, the phosphine (PH3) stabilization during the Ostwald ripening phase was reduced directly after the InP material deposition. This step was introduced to enhance the redistribution of the In atoms due to large surface migration [17]. Thus, the spatial density of the targeted smaller QDs is reduced (≈ 2 × 10 9 cm−2 ) at the expense of larger sized QD formation. The bimodal nature [18] is therefore shifted to larger structures, which has to be taken into account for the following growth steps in order to guarantee an epitaxial flat surface of the complete structure. The distribution of large and small QDs originating from the bimodal growth can be seen in an atomic force microscope
After the process, we studied and compared the optical characteristics of our QDs inside a processed membrane sample, later referred to as ‘thinned,’ and in an unprocessed sample. This also gives us feedback of the process-induced influences on the properties of the QDs. Further characterization of these kinds of QDs without the membrane structure can be found in reference [19]. Both samples are from the same epitaxial run and directly comparable. The samples are placed in a He flow cryostat to keep the samples at nominal 4 K. We have used a Fianium white light laser for pulsed excitation of the samples. With an acousto-optic tunable filter the desired wavelength can be selected. For cw excitation we used a RLTMSL laser emitting at 532 nm. The excitation power is adjusted using a variable attenuator. The laser light is focused down onto the sample using a 50 × microscope objective with a numerical aperture (NA) of 0.45 to an approximately 1.5 μm spot diameter. The carriers were generated in the barrier and subsequently captured by the QDs.
Figure 1. AFM image (4 μm2) of a similar InP quantum dot sample without capping. The bimodal distribution of large and small quantum dots is observable [19].
2
H Niederbracht et al
Nanotechnology 26 (2015) 235201
Figure 2. Sketch of the sample structure with the double heterobarriers, the different aluminum content and with the corresponding deposition
temperatures.
Figure 4. Overview μ-PL for a thinned and an unprocessed sample Figure 3. SEM picture of the sample after substrate removal. A clean
and flat surface was achieved.
from the same epitaxy. The GaAs peak (λ ≈ 820 nm) is clearly missing for the thinned sample.
For detection we use a spectrometer with 0.5 m dispersion length and a 1800 l/mm grating. The spectrometer is equipped with a Peltier-cooled charge-coupled device (CCD) chip for detection. To prove the non-classical nature of the light emitted by a single QD, the dispersed light is sent to a Hanbury–Brown and Twiss (HBT) setup [20] in order to perform autocorrelation measurements. The HBT basically consists of a 50/50 beam splitter, two detectors and electronics. In principle, the time delay between two detection events on the different detectors (start and stop) is measured. The SiAPDs have a time resolution of 500 ps. For time-correlated single-photon counting (TCSPC) we used as a start signal an APD with a time resolution of 40 ps for the QD signal and as a stop signal the laser trigger as we run the setup in the reversed mode. In the following the samples were excited at 530 nm with an excitation power density of 2.35 W cm−2. In figure 4, an overview micro-photoluminescence spectrum of the thinned and unprocessed sample can be seen. The emission from the GaAs at λ≈ 820 nm is missing in the case of the thinned sample, which indicates that the whole substrate was successfully removed during the process. Also the PL from
large and small QDs is clearly observable. The PL emission of the large QDs is less pronounced for the unprocessed sample, as the charge carriers generated in the barrier might be lost to the substrate. This loss mechanism is not available for the processed sample. For further studies, we concentrate on the small dots, which emit in the targeted red spectral range around 650 nm. We carried out TCSPC measurements to compare the charge carrier dynamics for the QDs of the thinned and unprocessed sample. The measured data can be described by a biexponential fit function. The first decay time displays the carrier lifetime within the QDs. The second decay time includes carrier recapture processes of the QD which are well known for these structures [18]. Material disorder inside the quaternary barrier layers may lead to lower energetic traps, which can store and release charge carriers. An explanation of these traps can be given by looking in more in detail into the growth process. During the fabrication of the lowdensity sample, the phosphine flow is reduced to enhance the In-surface migration. This leads on the one side to a lower spatial density of small sized QDs but also forms some large sized QDs. These large QDs support a large strain field that 3
H Niederbracht et al
Nanotechnology 26 (2015) 235201
Figure 5. Time-resolved measurements for the thinned and the
unprocessed sample. Only a slight change in the decay times is observable. We used excitation energies varying from 11.76 to 23.53 W cm−2.
influences the growth of the quaternary AlGaInP barrier. Therefore, either structural defects or material ordering can lead to low-energetic traps for the charge carriers. Especially during a pulsed excitation these traps capture charge carriers, which will be released a certain time after the excitation pulse and therefore reexcite the QD. The first decay time of these small InP QDs is smaller than 1 ns [21], and only slight changes are observable within the samples as shown in figure 5. We get values for the first decay time for the thinned and unprocessed sample between τ1,thinned = 536 and 1007 ps and τ1,unprocessed = 569 and 940 ps, respectively. This slight change might come from surface traps, which have their origin in the thinning process, leading to a band bending due to electric fields. Another possibility could be strain in our thinned sample because of the different thermal expansion coefficient of SiO2 and the semiconductor, which lead to piezoelectric fields [22, 23] and thus reduces the overlap of the wave functions and therefore reduces the recombination rate. We carried out HBT measurements for the thinned and unprocessed sample under cw excitation. In figure 6(a) and (b) a μ-PL measurement of an exemplary single QD of the thinned and unprocessed sample is plotted. The quantum dots are emitting at 635.01 nm with a full width at half maximum (FWHM) of 285 μeV (thinned) and 638.01 nm with a FWHM of 90 μeV (unprocessed), respectively. In general we get for the unprocessed sample linewidths between 90 and 436 μeV and for the thinned sample between 120 and 474 μeV. If we compare these linewidths, almost no change in the FWHM is observable. We want to mention that the count rates of the unprocessed sample are rather poor, as no mirror enhances the emission. The corresponding autocorrelation measurements are displayed in figure 6(c) and (d). The peak on the right side of the zero at around 10 ns originates from unwanted crosstalk between the two APDs during the measurement. For the following analysis this area was not taken into account to not overestimate the Poisson level. Without any correction we get and g(2) . a g(2) thinned (0) = 0.30 ± 0.05 unprocessed (0) = 0.46 ± 0.11 If we apply standard corrections for dark counts and background [24], we end with a g(2) and thinned,corr (0) = 0.27
Figure 6. (a) and (b) are μ-PL of the thinned and unprocessed
sample. (c) and (d) are the corresponding autocorrelation measurement of a single quantum dot. All measurements were done with the same excitation power. For the μ-PL we used a 1800 grating, the QDs are emitting at 635.01 nm and 638.01 nm, respectively.
respectively. Possible additional g(2) unprocessed,corr (0) = 0.33 defects might occur at the top of the processed sample, which can lead to a reduced density of active QDs. This leads to a reduced background in the emission spectrum and results in a slightly better g(2) (0). In addition we estimated the collection efficiency into the first lens of the processed sample to be 0.9 ± 0.1%.
4. Conclusion In conclusion we have shown a process without size limitation to fabricate a thin membrane containing III/V semiconductor QDs emitting in the red spectral range with reduced density. SEM as well as μ-PL measurements confirm the complete substrate removal. We observed only a slight change in the decay time and no degradation of single-photon emission, which is confirmed with a g(2) -value of 0.30 ± thin (0) 0.05. Also almost no change in the linewidth is observed for the thinned sample compared to the unprocessed sample. Our approach is not limited to a special sample size. This allows a large-scale application transferring the whole membrane. For example, the membrane could be wafer bonded to a wafer with Si3N4 waveguide structures and then several QDs could be used as emitters and coupled to the waveguides to accomplish quantum operations on a chip like a controlledNOT gate or even more complex schemes. Also, a combination with a an SIL should be possible to increase the collection efficiency in future. 4
H Niederbracht et al
Nanotechnology 26 (2015) 235201
Acknowledgments
[10] Michler P, Kiraz A, Becher C, Schoenfeld W V, Petroff P M, Zhang L, Hu E and Imamoglu A 2000 Science 290 2282–5 [11] Roßbach R, Reischle M, Beirne G J, Jetter M and Michler P 2008 Appl. Phys. Lett. 92 071105 [12] Bommer M, Schulz W M, Roßbach R, Jetter M, Michler P, Thomay T, Leitenstorfer A and Bratschitsch R 2011 J. Appl. Phys. 110 063108 [13] Roßbach R, Schulz W, Reischle M, Beirne G, Jetter M and Michler P 2007 J. Cryst. Growth 298 595–98 [14] Eisaman M D, Fan J, Migdall A and Polyakov S V 2011 Rev. Sci. Instrum. 82 071101 [15] Lin J F, Jou M J, Chen C Y and Lee B J 1992 J. Cryst. Growth 124 415–19 [16] Stranski I N and Krastanow L 1938 Sitz. Ber. Akad. Wiss., Math.-naturwiss. Kl. Abt. IIb 146 797 [17] Murata H, Ho I H, Su L C, Hosokawa Y and Stringfellow G B 1996 J. Appl. Phys. 79 6895–9 [18] Schulz W M, Roßbach R, Reischle M, Beirne G J, Bommer M, Jetter M and Michler P 2009 Phys. Rev. B 79 035329 [19] Koroknay E 2013 Epitaxial processes for low-density quantum dots in III-V semiconductors PhD Thesis University of Stuttgart [20] Brown R H and Twiss R Q 1956 Nature 177 27–29 [21] Reischle M, Beirne G J, Roßbach R, Jetter M and Michler P 2008 Phys. Rev. Lett. 101 146402 [22] Pryor C, Pistol M E and Samuelson L 1997 Phys. Rev. B 56 10404–11 [23] Tanaka T, Takano K, Tsuchiya T and Sakaguchi H 2000 J. Cryst. Growth 221 515–9 [24] Brouri R, Beveratos A, Poizat J P and Grangier P 2000 Opt. Lett. 25 1294–6
The authors would like to thank M Ubl for help with the process, E. Kohler for support with the MOVPE and they gratefully acknowledge partial funding by SFB TRR 21.
References [1] Gregersen N, Kaer P and Mork J 2013 IEEE J. Selected Topics Quantum Electron. 19 1–16 [2] Chen X W, Götzinger S and Sandoghdar V 2011 Opt. Lett. 36 3545–7 [3] Chu X L, Brenner T J K, Chen X W, Ghosh Y, Hollingsworth J A, Sandoghdar V and Götzinger S 2014 Optica 1 203–8 [4] Reese C, Becher C, Imamoglu A, Hu E, Gerardot B D and Petroff P M 2001 Appl. Phys. Lett. 78 2279–81 [5] Frederick S, Dalacu D, Aers G, Poole P, Lapointe J and Williams R 2006 Physica E 32 504–7 [6] Dalacu D, Frederick S, Bogdanov A, Poole P J, Aers G C, Williams R L, McCutcheon M W and Young J F 2005 J. Appl. Phys. 98 023101 [7] Scherer A, Painter O, DUrso B, Lee R and Yariv A 1998 J. Vac. Sci. Technol. B 16 3906–10 [8] Hennessy K, Badolato A, Winger M, Gerace D, Atature M, Gulde S, Falt S, Hu E L and Imamoglu A 2007 Nature 445 896–9 [9] Lodahl P, Mahmoodian S and Stobbe S 2013 http://arxiv.org/ abs/1312.1079
5
Search
Collections
Journals
About
Contact us
My IOPscience
Fabrication and optical characterization of large scale membrane containing InP/AlGaInP quantum dots
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2015 Nanotechnology 26 235201 (http://iopscience.iop.org/0957-4484/26/23/235201) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 132.203.227.62 This content was downloaded on 14/06/2017 at 00:08 Please note that terms and conditions apply.
You may also be interested in: Single-photon emission from electrically driven InP quantum dots epitaxially grown on CMOS-compatible Si(001) M Wiesner, W-M Schulz, C Kessler et al. Visible single-photon generation from semiconductor quantum dots Thomas Aichele, Valéry Zwiller and Oliver Benson In-assisted deoxidation of GaAs substrates for the growth of single InAs/GaAs quantum dot emitters Tian Xia (), YongJin Cho, Michele Cotrufo et al. Engineered quantum dot single-photon sources Sonia Buckley, Kelley Rivoire and Jelena Vukovi Time resolved emission at 1.3 mu m of a single InAs quantum dot by using a tunable fibre Bragg grating G Muñoz-Matutano, D Rivas, A L Ricchiuti et al. Quantum optics with single quantum dot devices Valéry Zwiller, Thomas Aichele and Oliver Benson Template method for nano-order positioning and dense packing of quantum dots for optoelectronic device application Kohki Mukai, Akinobu Hirota, Yuta Shimizu et al. Epitaxially Grown Indium Phosphide Quantum Dots on a Virtual Ge Substrate Realized on Si(001) Michael Wiesner, Moritz Bommer, Wolfgang-Michael Schulz et al.
Nanotechnology Nanotechnology 26 (2015) 235201 (5pp)
doi:10.1088/0957-4484/26/23/235201
Fabrication and optical characterization of large scale membrane containing InP/ AlGaInP quantum dots H Niederbracht, F Hargart, M Schwartz, E Koroknay, C A Kessler, M Jetter and P Michler Institut fuer Halbleiteroptik und Funktionelle Grenzflaechen, University of Stuttgart, Allmandring 3, 70569 Stuttgart, Germany E-mail: [email protected] Received 12 January 2015, revised 16 March 2015 Accepted for publication 30 March 2015 Published 21 May 2015 Abstract
Single-photon sources with a high extraction efficiency are a prerequisite for applications in quantum communication and quantum computation schemes. One promising approach is the fabrication of a quantum dot containing membrane structure in combination with a solid immersion lens and a metal mirror. We have fabricated an 80 nm thin semiconductor membrane with incorporated InP quantum dots in an AlGaInP double hetero barrier via complete substrate removal. In addition, a gold layer was deposited on one side of the membrane acting as a mirror. The optical characterization shows in detail that the unique properties of the quantum dots are preserved in the membrane structure. Keywords: semiconductor membrane, InP quantum dot, III/V semiconductor (Some figures may appear in colour only in the online journal) 1. Introduction
in the InAs/GaAs [4] material system. On the other hand, the complete substrate can be removed as shown for InAs quantum dots in an InP membrane [5, 6]. They are used, e.g., in photonic crystal structures to enhance the confinement in the z-direction [7] and to create high Q-cavities [8] and waveguides [9]. Therefore, these kinds of structures are suitable to enhance the unique properties of semiconductor QDs [10]. In this paper, we report on a process to fabricate largesized (cm) AlGaInP membranes containing InP QDs via substrate removal. These QDs have already proven their ability to emit non-classical light [11], have confirmed singlephoton emission up to elevated temperature of 100 K [12] and showed variable emission energies from red to green [13]. The additional benefit of the QDs used in this paper is the red spectral emission range where the detection efficiency for Sibased avalanche photo diodes (Si-APDs) is the highest to date [14]. To facilitate efficient collection from single nanostructures, InP QDs with reduced spatial density were embedded into the membrane. Comparison of micro-photoluminescence (μ-PL) measurements of bare and processed
The need for higher collection efficiency of single-photon components has led to a variety of different device approaches [1]. One approach is to fabricate a thin semiconductor membrane containing quantum dots (QDs) in combination with a solid immersion lens (SIL) and a metal mirror as suggested by Chen et al [2]. Recently the same group reported about reaching 99% collection efficiency for giant CdSe/CdS QDs with a dielectric antenna sample in combination with a thin membrane [3]. Nowadays, several techniques to fabricate a semiconductor membrane have been established. The fabrication of such membrane structures typically includes electron beam lithography combined with dry- and wet-chemical etching steps, which limit the processed size to several millimeters. The main issue concerning the process is the concerted removal of the layers below the active membrane region. Basically, two approaches are possible. The first one is realized by partly removing a sacrificial layer, e.g., AlGaAs, by wet chemical etching to create a freestanding membrane part. This was demonstrated, for example, 0957-4484/15/235201+05$33.00
1
© 2015 IOP Publishing Ltd Printed in the UK
H Niederbracht et al
Nanotechnology 26 (2015) 235201
samples revealed the integrity of the QDs in the membrane structure. Furthermore, time-resolved PL measurements demonstrate nearly the same carrier dynamics in both samples, and the second-order autocorrelation proves the emission from a single QD. The design of our structure is inspired by the scheme of Chen et al [2]. They forecast the realization of 99% collection efficiency for a suitably processed membrane antenna sample. For a semiconductor-based QD, a membrane with a metal mirror has to be fabricated with an adapted refractive index. As our structure fulfills these requirements, we anticipate, in combination with a SIL with a suitable refractive index, a higher collection efficiency for further measurements.
(AFM) image (figure 1) of an uncapped sample with similar growth conditions. The lateral dimension of the large QDs is in the order of ≈50 nm, whereas it is ≈30 nm for the small QDs. After completion of the ripening, the first upper barrier was grown at the same growth temperature as the QDs to prevent the material intermixing between the different layers, while the lower and the second upper barrier were deposited at 710 °C. The barriers consist of (AlxGa1 −x ) 0.51In 0.49P with an aluminum content of x = 0.5 directly surrounding the nanostructure with a thickness of 10 nm and a 30 nm thick outer barrier with x = 0.7 aluminum content. A schematic of the structure can be seen in figure 2. In the next process step 200 nm of SiO2 were sputtered on top of the semiconductor. After this 5 nm of Cr and 200 nm of Au are evaporated. The Cr serves as adhesion promoter for the Au, which acts as a mirror and is prerequisite for the high efficiency source described before. This mirror provides a measured reflectivity between 93.1 and 95.6% in the range of 630 to 670 nm. Then the sample is mechanically thinned to approximately 100 μm and glued with nail polish to an Si carrier wafer. The nail polish is sufficiently resistant to the following wet chemical etching steps and also does not negatively affect them. Afterwards, the sample is etched with an etchant consisting of H2SO4:H2O:H2O2 at 70 °C to a total thickness of 10 μm. This etch is non-selective; therefore, some sidewall protection of the sample through the polish is favorable. For the last 10 μm we change to a NH4OH:H2O2 mixture to selectively etch the GaAs down to the (Al 0.7Ga 0.3) 0.51In 0.49P barrier, which acts as an etch-stopping layer. After this step, we end with a 80 nm thick semiconductor membrane combined with a gold mirror at the bottom. A scanning electron microscopy (SEM) picture of the fully processed sample is shown in figure 3. The nicely flat and clean etched surface can be seen on top of the layer package. As this process only uses mechanical thinning and wet etching, it is cost efficient, easy and has no limitations in sample size.
2. Epitaxy and process
3. Optical characterization
The samples were deposited by metal-organic vapor-phase epitaxy (MOVPE) on a standard GaAs substrate with 6 ◦ miscut towards the [111]A direction, which is to prevent ordering in the AlGaInP layers [15]. The InP QDs were grown self-assembled in the Stranski–Krastanow growth mode [16] at a temperature of 610 °C. To reduce the spatial density of the nanostructures, the phosphine (PH3) stabilization during the Ostwald ripening phase was reduced directly after the InP material deposition. This step was introduced to enhance the redistribution of the In atoms due to large surface migration [17]. Thus, the spatial density of the targeted smaller QDs is reduced (≈ 2 × 10 9 cm−2 ) at the expense of larger sized QD formation. The bimodal nature [18] is therefore shifted to larger structures, which has to be taken into account for the following growth steps in order to guarantee an epitaxial flat surface of the complete structure. The distribution of large and small QDs originating from the bimodal growth can be seen in an atomic force microscope
After the process, we studied and compared the optical characteristics of our QDs inside a processed membrane sample, later referred to as ‘thinned,’ and in an unprocessed sample. This also gives us feedback of the process-induced influences on the properties of the QDs. Further characterization of these kinds of QDs without the membrane structure can be found in reference [19]. Both samples are from the same epitaxial run and directly comparable. The samples are placed in a He flow cryostat to keep the samples at nominal 4 K. We have used a Fianium white light laser for pulsed excitation of the samples. With an acousto-optic tunable filter the desired wavelength can be selected. For cw excitation we used a RLTMSL laser emitting at 532 nm. The excitation power is adjusted using a variable attenuator. The laser light is focused down onto the sample using a 50 × microscope objective with a numerical aperture (NA) of 0.45 to an approximately 1.5 μm spot diameter. The carriers were generated in the barrier and subsequently captured by the QDs.
Figure 1. AFM image (4 μm2) of a similar InP quantum dot sample without capping. The bimodal distribution of large and small quantum dots is observable [19].
2
H Niederbracht et al
Nanotechnology 26 (2015) 235201
Figure 2. Sketch of the sample structure with the double heterobarriers, the different aluminum content and with the corresponding deposition
temperatures.
Figure 4. Overview μ-PL for a thinned and an unprocessed sample Figure 3. SEM picture of the sample after substrate removal. A clean
and flat surface was achieved.
from the same epitaxy. The GaAs peak (λ ≈ 820 nm) is clearly missing for the thinned sample.
For detection we use a spectrometer with 0.5 m dispersion length and a 1800 l/mm grating. The spectrometer is equipped with a Peltier-cooled charge-coupled device (CCD) chip for detection. To prove the non-classical nature of the light emitted by a single QD, the dispersed light is sent to a Hanbury–Brown and Twiss (HBT) setup [20] in order to perform autocorrelation measurements. The HBT basically consists of a 50/50 beam splitter, two detectors and electronics. In principle, the time delay between two detection events on the different detectors (start and stop) is measured. The SiAPDs have a time resolution of 500 ps. For time-correlated single-photon counting (TCSPC) we used as a start signal an APD with a time resolution of 40 ps for the QD signal and as a stop signal the laser trigger as we run the setup in the reversed mode. In the following the samples were excited at 530 nm with an excitation power density of 2.35 W cm−2. In figure 4, an overview micro-photoluminescence spectrum of the thinned and unprocessed sample can be seen. The emission from the GaAs at λ≈ 820 nm is missing in the case of the thinned sample, which indicates that the whole substrate was successfully removed during the process. Also the PL from
large and small QDs is clearly observable. The PL emission of the large QDs is less pronounced for the unprocessed sample, as the charge carriers generated in the barrier might be lost to the substrate. This loss mechanism is not available for the processed sample. For further studies, we concentrate on the small dots, which emit in the targeted red spectral range around 650 nm. We carried out TCSPC measurements to compare the charge carrier dynamics for the QDs of the thinned and unprocessed sample. The measured data can be described by a biexponential fit function. The first decay time displays the carrier lifetime within the QDs. The second decay time includes carrier recapture processes of the QD which are well known for these structures [18]. Material disorder inside the quaternary barrier layers may lead to lower energetic traps, which can store and release charge carriers. An explanation of these traps can be given by looking in more in detail into the growth process. During the fabrication of the lowdensity sample, the phosphine flow is reduced to enhance the In-surface migration. This leads on the one side to a lower spatial density of small sized QDs but also forms some large sized QDs. These large QDs support a large strain field that 3
H Niederbracht et al
Nanotechnology 26 (2015) 235201
Figure 5. Time-resolved measurements for the thinned and the
unprocessed sample. Only a slight change in the decay times is observable. We used excitation energies varying from 11.76 to 23.53 W cm−2.
influences the growth of the quaternary AlGaInP barrier. Therefore, either structural defects or material ordering can lead to low-energetic traps for the charge carriers. Especially during a pulsed excitation these traps capture charge carriers, which will be released a certain time after the excitation pulse and therefore reexcite the QD. The first decay time of these small InP QDs is smaller than 1 ns [21], and only slight changes are observable within the samples as shown in figure 5. We get values for the first decay time for the thinned and unprocessed sample between τ1,thinned = 536 and 1007 ps and τ1,unprocessed = 569 and 940 ps, respectively. This slight change might come from surface traps, which have their origin in the thinning process, leading to a band bending due to electric fields. Another possibility could be strain in our thinned sample because of the different thermal expansion coefficient of SiO2 and the semiconductor, which lead to piezoelectric fields [22, 23] and thus reduces the overlap of the wave functions and therefore reduces the recombination rate. We carried out HBT measurements for the thinned and unprocessed sample under cw excitation. In figure 6(a) and (b) a μ-PL measurement of an exemplary single QD of the thinned and unprocessed sample is plotted. The quantum dots are emitting at 635.01 nm with a full width at half maximum (FWHM) of 285 μeV (thinned) and 638.01 nm with a FWHM of 90 μeV (unprocessed), respectively. In general we get for the unprocessed sample linewidths between 90 and 436 μeV and for the thinned sample between 120 and 474 μeV. If we compare these linewidths, almost no change in the FWHM is observable. We want to mention that the count rates of the unprocessed sample are rather poor, as no mirror enhances the emission. The corresponding autocorrelation measurements are displayed in figure 6(c) and (d). The peak on the right side of the zero at around 10 ns originates from unwanted crosstalk between the two APDs during the measurement. For the following analysis this area was not taken into account to not overestimate the Poisson level. Without any correction we get and g(2) . a g(2) thinned (0) = 0.30 ± 0.05 unprocessed (0) = 0.46 ± 0.11 If we apply standard corrections for dark counts and background [24], we end with a g(2) and thinned,corr (0) = 0.27
Figure 6. (a) and (b) are μ-PL of the thinned and unprocessed
sample. (c) and (d) are the corresponding autocorrelation measurement of a single quantum dot. All measurements were done with the same excitation power. For the μ-PL we used a 1800 grating, the QDs are emitting at 635.01 nm and 638.01 nm, respectively.
respectively. Possible additional g(2) unprocessed,corr (0) = 0.33 defects might occur at the top of the processed sample, which can lead to a reduced density of active QDs. This leads to a reduced background in the emission spectrum and results in a slightly better g(2) (0). In addition we estimated the collection efficiency into the first lens of the processed sample to be 0.9 ± 0.1%.
4. Conclusion In conclusion we have shown a process without size limitation to fabricate a thin membrane containing III/V semiconductor QDs emitting in the red spectral range with reduced density. SEM as well as μ-PL measurements confirm the complete substrate removal. We observed only a slight change in the decay time and no degradation of single-photon emission, which is confirmed with a g(2) -value of 0.30 ± thin (0) 0.05. Also almost no change in the linewidth is observed for the thinned sample compared to the unprocessed sample. Our approach is not limited to a special sample size. This allows a large-scale application transferring the whole membrane. For example, the membrane could be wafer bonded to a wafer with Si3N4 waveguide structures and then several QDs could be used as emitters and coupled to the waveguides to accomplish quantum operations on a chip like a controlledNOT gate or even more complex schemes. Also, a combination with a an SIL should be possible to increase the collection efficiency in future. 4
H Niederbracht et al
Nanotechnology 26 (2015) 235201
Acknowledgments
[10] Michler P, Kiraz A, Becher C, Schoenfeld W V, Petroff P M, Zhang L, Hu E and Imamoglu A 2000 Science 290 2282–5 [11] Roßbach R, Reischle M, Beirne G J, Jetter M and Michler P 2008 Appl. Phys. Lett. 92 071105 [12] Bommer M, Schulz W M, Roßbach R, Jetter M, Michler P, Thomay T, Leitenstorfer A and Bratschitsch R 2011 J. Appl. Phys. 110 063108 [13] Roßbach R, Schulz W, Reischle M, Beirne G, Jetter M and Michler P 2007 J. Cryst. Growth 298 595–98 [14] Eisaman M D, Fan J, Migdall A and Polyakov S V 2011 Rev. Sci. Instrum. 82 071101 [15] Lin J F, Jou M J, Chen C Y and Lee B J 1992 J. Cryst. Growth 124 415–19 [16] Stranski I N and Krastanow L 1938 Sitz. Ber. Akad. Wiss., Math.-naturwiss. Kl. Abt. IIb 146 797 [17] Murata H, Ho I H, Su L C, Hosokawa Y and Stringfellow G B 1996 J. Appl. Phys. 79 6895–9 [18] Schulz W M, Roßbach R, Reischle M, Beirne G J, Bommer M, Jetter M and Michler P 2009 Phys. Rev. B 79 035329 [19] Koroknay E 2013 Epitaxial processes for low-density quantum dots in III-V semiconductors PhD Thesis University of Stuttgart [20] Brown R H and Twiss R Q 1956 Nature 177 27–29 [21] Reischle M, Beirne G J, Roßbach R, Jetter M and Michler P 2008 Phys. Rev. Lett. 101 146402 [22] Pryor C, Pistol M E and Samuelson L 1997 Phys. Rev. B 56 10404–11 [23] Tanaka T, Takano K, Tsuchiya T and Sakaguchi H 2000 J. Cryst. Growth 221 515–9 [24] Brouri R, Beveratos A, Poizat J P and Grangier P 2000 Opt. Lett. 25 1294–6
The authors would like to thank M Ubl for help with the process, E. Kohler for support with the MOVPE and they gratefully acknowledge partial funding by SFB TRR 21.
References [1] Gregersen N, Kaer P and Mork J 2013 IEEE J. Selected Topics Quantum Electron. 19 1–16 [2] Chen X W, Götzinger S and Sandoghdar V 2011 Opt. Lett. 36 3545–7 [3] Chu X L, Brenner T J K, Chen X W, Ghosh Y, Hollingsworth J A, Sandoghdar V and Götzinger S 2014 Optica 1 203–8 [4] Reese C, Becher C, Imamoglu A, Hu E, Gerardot B D and Petroff P M 2001 Appl. Phys. Lett. 78 2279–81 [5] Frederick S, Dalacu D, Aers G, Poole P, Lapointe J and Williams R 2006 Physica E 32 504–7 [6] Dalacu D, Frederick S, Bogdanov A, Poole P J, Aers G C, Williams R L, McCutcheon M W and Young J F 2005 J. Appl. Phys. 98 023101 [7] Scherer A, Painter O, DUrso B, Lee R and Yariv A 1998 J. Vac. Sci. Technol. B 16 3906–10 [8] Hennessy K, Badolato A, Winger M, Gerace D, Atature M, Gulde S, Falt S, Hu E L and Imamoglu A 2007 Nature 445 896–9 [9] Lodahl P, Mahmoodian S and Stobbe S 2013 http://arxiv.org/ abs/1312.1079
5
