AlGaAs and AlGaAs/GaAs/AlGaAs nanowires grown ...

1 downloads 0 Views 1MB Size Report
Nov 8, 2017 - grown by molecular beam epitaxy on silicon ... E-mail: cirlin@beam.ioffe.ru .... mote the growth of all NWs in the array due to the inhibitor effect ...
Journal of Physics D: Applied Physics

Related content

PAPER

AlGaAs and AlGaAs/GaAs/AlGaAs nanowires grown by molecular beam epitaxy on silicon substrates To cite this article: G E Cirlin et al 2017 J. Phys. D: Appl. Phys. 50 484003

- Understanding the vapor–liquid–solid growth and composition of ternary III–V nanowires and nanowire heterostructures V G Dubrovskii - Wurtzite GaAs/AlGaAs core–shell nanowires grown by molecular beam epitaxy H L Zhou, T B Hoang, D L Dheeraj et al. - Nanostructure and strain properties of core-shell GaAs/AlGaAs nanowires Th Kehagias, N Florini, J Kioseoglou et al.

View the article online for updates and enhancements.

This content was downloaded from IP address 18.142.5.48 on 05/12/2017 at 19:31

Journal of Physics D: Applied Physics J. Phys. D: Appl. Phys. 50 (2017) 484003 (6pp)

https://doi.org/10.1088/1361-6463/aa9169

AlGaAs and AlGaAs/GaAs/AlGaAs nanowires grown by molecular beam epitaxy on silicon substrates G E Cirlin1,2,3,4,5, R R Reznik1,2,3,4, I V Shtrom1,2,5, A I Khrebtov1, I P Soshnikov1,2,5, S A Kukushkin6, L Leandro7, T Kasama8 and Nika Akopian7 1

  St. Petersburg Academic University RAS, Khlopina 8/3, 194021 St Petersburg, Russia   Institute for Analytical Instrumentation RAS, Rizhsky 26, 190103, St Petersburg, Russia 3   ITMO University, Kronverkskiy pr. 49, 197101 St Petersburg, Russia 4   Peter the Great St Petersburg Polytechnic University, Polytechnicheskaya 29, 195251, St Petersburg, Russia 5   Ioffe Physical Technical Institute RAS, Politekhnicheskaya 26, 194021, St Petersburg, Russia 6   Institute of Problems of Mechanical Engineering Russian Academy of Science, Bolshoj 61, 199178 St Petersburg, Russia 7   Department of Photonics Engineering, Technical University of Denmark, 2800 Kongens Lyngby, Denmark 8   Center for Electron Nanoscopy, Technical University of Denmark, 2800 Kongens Lyngby, Denmark 2

E-mail: [email protected] Received 18 April 2017, revised 30 August 2017 Accepted for publication 5 October 2017 Published 8 November 2017 Abstract

The data on growth peculiarities and physical properties of GaAs insertions embedded in AlGaAs nanowires grown on different (1 1 1) substrates by Au-assisted molecular beam epitaxy are presented. The influence of nanowires growth conditions on structural and optical properties is studied in detail. It is shown that by varying the growth parameters it is possible to form structures like quantum dots that emit in a wide wavelengths range. These quantum dots show sharp and intense emission lines when an optical signal is collected from a single nanowire. The technology proposed opens new possibilities for integration of direct-band AIIIBV materials on silicon platform. Keywords: nanowires, AlGaAs, quantum dots, molecular beam epitaxy, silicon (Some figures may appear in colour only in the online journal)

1. Introduction

conditions and by supplying different fluxes. Alternatively, AlGaAs heterostructures can be obtained via gold-assisted VLS growth with simultaneous deposition of metal elements. A combination of nanowires (NWs) with quantum dots (QDs) is a promising component for future optoelectronic devices, in particular, single-photon emitters. The most studied epitaxially grown QDs are self-assembled, i.e. grown by island nucleation in the Stranski–Krastanow growth mode. The size, shape and density of self-assembled QDs can be controlled by changing growth parameters such as substrate temperature, growth rate and growth time, but in the end it is a self-organized, strain-induced process, where an independent control

III–V nanowires (NWs) and NW heterostructures are promising building blocks for future close-to-volume-minimizing devices and scalable bottom-up photonic parts, which can be integrated with silicon electronic platform [1, 2]. In particular, AlGaAs core–shell NWs are of the great interest for nano lasers [3, 4], single photon sources [5, 6] and THz emitters [7]. Core–shell GaAs/AlGaAs NW structures have been fabricated either by vapor–liquid–solid (VLS) [3, 8] or selective area epitaxy [9] techniques, where growth of GaAs cores and AlGaAs shells is intentionally separated by changing growth 1361-6463/17/484003+6$33.00

1

© 2017 IOP Publishing Ltd  Printed in the UK

G E Cirlin et al

J. Phys. D: Appl. Phys. 50 (2017) 484003

Figure 1.  SEM images of AlGaAs NWs with different growth time: (a) 5 min, (b) 10 min, (c) 20 min.

of the properties of an array is a challenging task. QDs in nanowires, in contrast, have shown great potential as a highly controllable system. The diameter, height, and density of QDs are defined by the NW diameter, the growth time, and the NW density, respectively, and can be chosen more predictable. Due to a very efficient strain relaxation on the sidewalls, coherent growth can be much easier realized in the NW geometry, where a small footprint is dictated by a metal catalyst particle assisting the NW growth via VLS mechanism. Moreover, it has been recently shown that [1 1 1] grown nanowires, especially heterostructured, are ideal candidates for the generation of entangled photon pairs [10]. In this work we present two key results. First, we show optimal conditions for growing AlGaAs nanowires. Then we demonstrate controlled growth of short GaAs segments inside our nanowires, demonstrating QD-like lines in the PL spectrum taken from a single nanowire.

Ga, As) were opened to promote the growth of the NWs. The reflection high energy electron diffraction (RHEED) technique was used for growth rates calibration (on a separate GaAs(1 0 0) substrate) and to control the evolution of NWs. The nominal Al content x in the solid solution, measured from the ratio of Ga/Al fluxes was varied within the range x  =  0.2– 0.6. The AlGaAs growth rate was maintained constant at 1 monolayer per second. The observation of RHEED patterns dynamics showed, in most cases, pure wurtzite NW crystallographic phase formed already upon a deposition of 50 nm of AlGaAs and remained unchanged during the whole growth run. Another approach was used to form GaAs insertion in AlGaAs NW body. For this particular case, right after the growth of AlGaAs NW part, the Al shutter was controllably closed for several seconds (in a range of 5–15 s, no interruption or substrate temperature change were applied) resulting in the formation of a GaAs nanoisland on the top of a NW, or quantum dot (QD), after embedding inside the AlGaAs NW. To completely embed the QD in the NW the Al shutter was again re-opened for additional 5 min. At the moment, we do not have a representative transmission electron microscopy (TEM) image for the grown structures. This is due to a very small contrast difference between Al and Ga elements during TEM imaging. We, however, have grown a special sample, where an GaAs insertion is located in the top part of a NW, and therefore is only covered with a very thin AlGaAs shell. A TEM image of similar GaAs insertion in AlGaAs NW but grown of GaAs(1 0 0) substrate may be found in e.g. [6]. We, thus, expect to obtain a clear TEM image of GaAs insertion. These TEM measurements are now under investigation. The surface morphology was studied by scanning electron microscopy (SEM). Optical properties of NW arrays were examined using photoluminescence (PL) setup by a continuous-wave neodymium laser (wavelength 532 nm) at an excitation power density of about 10 W cm−2. Hamamatsu R928 photomultiplier was used as a photodetector. PL measurements were taken at a temperature of 10 K with nanowirecontaining samples placed in a closed cycle helium cryostat. In the case of single NW measurements, a specially designed low-temperature µ-PL setup was used (e.g. [12]).

2. Experiments Experimentally, all the samples in the present work were grown by molecular beam epitaxy (MBE). MBE growth is performed using a Riber 21 Compact setup, equipped with an additional vacuum chamber for the deposition of Au (metallization chamber), which allows one to transfer the substrates to the growth chamber with no vacuum brake. Different substrates were used, most of them were Si (1 1 1) wafers and, additionally, several growth runs were performed using hybrid SiC/Si substrates prepared by a special treatment of Si(1 1 1) substrates [11]. At the beginning, we grew a set of samples with just AlGaAs NWs (no GaAs segments) to see how the growth conditions influence on the morphological properties of the NWs. The next set of samples was AlGaAs/GaAs/ AlGaAs hybrid structures where a small, nanometre scale GaAs insertion was embedded inside the AlGaAs NW. Growth of the NWs is performed in two stages. The surface of wafers was treated in a 10:1 aqueous solution of HF, then the samples were loaded to the metallization chamber and heated up to the temperature of 850 °C for 10 min. Next, the sample temperature was decreased down to 550 °C and Au deposition was performed with a total thickness of ~0.1–0.2 nm, keeping an additional 1 min at the same temperature to allow for formation of Au droplets. Then, the samples were cooled down to the room temperature and transferred to the growth chamber. During a second stage, where we grew the NWs, the substrate temperature was increased to the desired temperature (in a range from 510 °C to 570 °C), corresponding shutters (Al,

3.  Results and discussion The first step in growth of high quality AlGaAs/GaAs/ AlGaAs NW based hybrid nanostuctures was the optim­ization of AlGaAs NWs formation procedure by changing the growth conditions. Mostly, we have examined three key technological 2

G E Cirlin et al

J. Phys. D: Appl. Phys. 50 (2017) 484003

Figure 2.  SEM images of AlGaAs NWs with different Al/Ga in gas phase ratio (nominal values): (a) Al0.2Ga0.8As, (b) Al0.3Ga0.7As, (c)

Al0.6Ga0.4As.

parameters: Al content, substrate temperature Ts and NWs growth time. In figures  1(a)–(c) we present SEM images of NWs AlGaAs/Si(1 1 1) arrays grown during different times for a nominal Al content x  =  0.3. It can be seen that in all the cases the NW arrays are standing on the top of a 2D AlGaAs layer. This layer is optically non-active due to the dislocations that are formed as a result of the lattice mismatch between Si and AlGaAs. As follows from the images, 20 min duration of the growth is enough to form nanowires with the length of at least 1 µm, which is the minimal length required for various applications. Our next step was examination of an influence of Al content on morphological properties of AlGaAs/Si(1 1 1) NWs. As follows from the images (figure 2), both the shape and height of the NWs change with increasing molar fraction of Al. The shape changes from a ‘pencil-like’ to a ‘conical’, with the increasing length of the tapered part. The shape becoming almost fully conical at x  =  0.6. The height also consequently decreases from 4 µm (x  =  0.2) to 1.0 µm (x  =  0.6). These changes are common for NWs, which were grown under the regime that was limited by various factors, e.g. growth temper­ature [16]. In our case, this occurs due to the significant difference in the binding energies (and, consequently, in the migration lengths over the surface) of Al and Ga adatoms along the NW sidewalls. The diffusivity of Al on the NW side decreases as Al concentration increases, which is due to a higher density of surface steps where the adatoms incorporated into a tapered shell, and is almost completely suppressed with x  =  0.6. Consequently, the increase of nominal aluminum composition yields more tapered NWs towards higher x. At the same time, the diameter at the very top the NWs remains constant (~20–25 nm) in all cases, which is set by the initial size of Au droplets. A spontaneous AlxGa1−xAs core–shell NWs formation with different compositions x in the cores and shells grown by MBE [13] and in [14] by metalorganic chemical vapor deposition has been recently documented. In particular, for the MBE growth it was found that Al composition in the shell is always higher than that in the core [15]. In table 1 we present the quantative data for the Al content in core and shell estimated from energy dispersive x-ray spectroscopy TEM analysis. To estimate the Al content in the core we measured Al concentration just below of the droplet where the thickness of the shell is negligible whereas the estimation of the Al composition in the shell was performed close to the bottom of the wire where the shell thickness is maximal. The measured values show that Al content in the shell

Table 1.  Al content (in %) in core and shell of AlGaAs NWs

measured with energy dispersive x-ray spectroscopy TEM analysis. Nominal Al/Ga ratio in a gas phase

Al content in core

Error

Al content in shell

Error

Al0.3Ga0.7As Al0.4Ga0.6As Al0.5Ga0.5As Al0.6Ga0.4As

16 18 28 38

3 2 3 2

24 38 48 55

2 4 3 5

is close to the nominal Al/Ga fluxes ratio (determined from the calibration on a separate GaAs(1 0 0) substrate), while the Al content in the core is systematically smaller (typically twice), than the nominal one. This observation is attributed to the lower aluminum surface diffusivity (aluminum collection length 250 nm against 780 nm for gallium at the substrate temperature 510 °C for x  =  0.2, 8 nm and 160 nm, when the nominal aluminum content is raised to 0.6) [16]. Also, aluminum leaves the droplet at least 100 times faster than gallium [16, 17]. These two findings lead to a self-organized formation of a core/shell structure consisting of a cylinder-like core with the diameter equal to the droplet size and a conical-like surrounding shell with the tapering angle dependent on the Al content x. This rather big difference between the bonding rates of aluminum and gallium with arsenic (at least 2 orders of magnitude) [17] should be very advantageous for obtaining sharp NW heterointerfaces by switching the group III fluxes. Figures 3(a)–(c) show SEM images of AlGaAs/Si(1 1 1) NWs arrays grown at different substrate temperatures (510– 570 °C) for a nominal Al content x  =  0.4. It is seen that an increase in the growth temperature leads to a decrease in the total length of NWs with corresponding increase of the 2D layer formed on the top of a Si(1 1 1) substrate. Additionally, a rather big difference of the surface density of NWs (figures 1 and 3) is observed. Theoretically, the NWs surface density is equal to the number of Au seeds on the substrate. Since we have used the same parameters to deposit gold on all substrates (both, Si and SiC/Si), the number of the nuclei has to be the same. But, since different growth conditions for the NWs growth were applied, the resulted NWs density may vary. For example, if the total thickness is not enough to promote the growth of all NWs in the array due to the inhibitor effect, the total density at the starting stage may be lower in comparison to the final one. The same argument is also valid for the elevated temperatures, where the competition between the growth of a 2D layer and NWs is dictated by the difference in the chemical potentials in liquid and layer phases. If 3

G E Cirlin et al

J. Phys. D: Appl. Phys. 50 (2017) 484003

Figure 3.  SEM images of AlGaAs NWs grown at different temperatures: (a) 510 °C, (b) 550 °C, (c) 570 °C.

Figure 4.  Low temperature (10 K) PL spectra from AlGaAs/GaAs NWs arrays with 12 s GaAs insertions, grown at different substrate temperatures.

the chemical potential in a liquid phase is higher than that for a solid layer, no NW growth occurs. PL intensity form these arrays (not shown here) also decreases significantly with the increase in Ts. To prove the tendency, AlGaAs/GaAs/AlGaAs based heterostuctures with GaAs insertions grown for 12 s were additionally grown under the same growth conditions (Al content x  =  0.4, Ts  =  (510–550) °C), and the corresponding PL spectra are presented in figure 4. Typically, we used a defocused laser excitation to get a signal from a big number of NWs, more than 1000, except the case when we investigate single NWs, then the laser was specially focused down to 1 µm2. As we described before, AlGaAs ‘parasitic’ layer is defected due to the lattice mismatch difference between the layer and the substrate. This layer is optically non-active and we do not detect any noticeable signal from it. It was checked for separate samples with just AlGaAs layer grown on Si(1 1 1) substrate to make sure that our PL signal originates from NWs only. All the spectra consist of two parts—a higher energy part, corresponding to the emission from AlGaAs NWs and a lower energy part, which is due to the luminescence from GaAs insertions (QDs). As substrate temperature increases the total PL intensity decreases for both lower and higher energy bands. In addition, the PL band which is associated with QDs is shifted towards lower energies (~30 meV shift) at Ts  =  550 °C and completely disappeares at Ts  =  570 °C (not shown). An interesting peculiarity in optical properties of AlGaAs NWs arrays is observed when Si(1 1 1) substrate was replaced with hybrid SiC/Si(1 1 1) one. The growth temperatures for both samples were set to 510 °C. In figure 5 we compare low

Figure 5.  PL spectra measured at 10 K from AlGaAs NWs arrays grown on Si(1 1 1) and SiC/Si(1 1 1) substrates.

temper­ature PL spectra taken from Al0.3Ga0.7As NW arrays (Al/Ga nominal ratio) grown on these substrates under identical growth conditions, equal to those we used for Si(1 1 1) so far. It is seen that low energy peaks (related to NWs) are similar in position (1.61 eV and 1.59 eV, for Si and SiC/Si respectively). At the same time, for SiC/Si based sample a new pronounced band appears at higher energies, which is currently under our more extended investigation. A possible reason for that might be a formation of space inhomogeneity in the solid solutions during the growth of the core/shell ternary NWs. This phenom­ enon has been found during the growth, e.g. GaAs/AlGaAs core/shell NWs when AlGaAs shell was deposited on GaAs core [18–20]. In these works a formation of (i) AlGaAs of a higher Al concentration than Al content in the shell on of the facets borders and (ii) AlGaAs of a lower than the shell Al content nanoscale insertions (or QDs) having typical size 4–5 nm at the facet corners. Due to the quantization effect, PL emission from these QDs was shifted towards 1.8–1.95 eV, which is similar to our observation in AlGaAs NWs grown on SiC/ Si(1 1 1) substrate. Relatively broad spectral widths of the PL bands (~120 meV) are also typical for QDs ensembles having rather large island sizes and/or composition distributions. As it was shown before, if the Al shutter is closed for an appropriate time (typically, 5–20 s) during the growth of AlGaAs NWs and then re-opened again a new PL band appeared at lower energy (in comparison to the PL band originated from AlGaAs NWs), which is associated with the formation of GaAs nanoscale insertion inside AlGaAs NWs. It is found that, by varying growth conditions, a position of both AlGaAs-related PL signal and GaAs QD related band can be changed. Typical PL spectra taken from AlGaAs/GaAs/ 4

G E Cirlin et al

J. Phys. D: Appl. Phys. 50 (2017) 484003

Figure 6.  PL spectra measured at T  =  10 K for hybrid AlGaAs/GaAs/AlGaAs NWs with (a) different Al contents in a solid solution and different growth times of GaAs QDs: x  =  0.4, 12 s (1); x  =  0.5, 15 s (2) and (b) the same Al contents x  =  0.6 in a solid solution, but different growth times of GaAs QDs: 15 s (1) and 11 s (2).

like quantum dots that emit in a wide wavelength range, and to engineer band profiles for both, quantum dots (GaAs insertions) and barriers (AlGaAs nanowires). Additionally, we performed optical studies of single nanowire quantum dots grown in the range of 5–20 s for the Al content x  =  0.3 in AlGaAs NWs. In figure 7 we show a photoluminescence of a single hybrid nanowire—quantum dot with GaAs insertion grown for 15 s. We observe sharp and intense lines around 1.58 eV, corresponding to various excitonic complexes in the quantum dot. We clearly see (not show here, will be presented in a separate paper [21]) that by changing the quantum dot growth time from 5 to 20 s we control the emission wavelength from 750 to 800 nm, correspondingly, supporting the PL data for NW arrays presented here. In conclusion, we demonstrate growth of AlGaAs nanowires on silicon substrates, and show dependence of their structural and optical properties on the key growth parameters. We also demonstrate a controlled growth of short GaAs segments inside AlGaAs nanowires. Photoluminescence of such segments shows sharp and intense QD-like emission lines. Our work, therefore, opens new prospects for integration of direct bandgap semiconductors and single-photon sources on silicon platform for various applications in the fields of silicon photonics and quantum information technology.

Figure 7.  Low temperature (1.5 K) PL of a single GaAs quantum dot in AlGaAs nanowire under off-resonant (532 nm) cw excitation. Sharp lines correspond to various excitonic complexes in the quantum dot. The quantum dot was grown for 15 s.

AlGaAs hybrid NW arrays are presented in figures 6(a) and (b). The QDs wavelength emission is always shifted relative to the bulk GaAs to higher energies, which indicates that a quantum structure is formed inside the wire. It also becomes possible to obtain emission from GaAs QDs at the same wavelength in the case of different AlGaAs shell compositions (figure 6(a)). For this particular case it is necessary just to increase the GaAs insertion growth time (15 s and 12 s of GaAs growth for the nominal Al compositions x of 0.5 and 0.4, respectively), because an increase in AlGaAs composition results in a decrease in the deposition rate of GaAs. Another important circumstance is that the spectral position of a QD related peak can be controllably changed by varying the size of GaAs insertions, while keeping the same Al content in a NW. We do it by simply changing the growth time of GaAs insertion. In figure  6(b) we present PL spectra for AlGaAs NWs having nominal Al composition x  =  0.6 with GaAs growth time of 11 s and 15 s. It is seen that high energy peaks (related to NWs) are very close each other (~1.82 eV), but the QD related peaks have significantly different spectral positions (1.62 eV and 1.71 eV for 15 s and 11 s of GaAs growth time, respectively). To sum up, it is shown that by varying the growth parameters it is possible to form structures

Acknowledgments We are grateful for the support of the Ministry of Education and Science of Russian Federation (state task, project No. 16.2483.2017/4.6). The nanowire samples were grown under the support of Russian Science Foundation (Project No 14-1200393). The work was also supported by Villum Fonden (Project no. VKR023444). References [1] Hyun J K, Zhang S and Lauhon L J 2013 Ann. Rev. Mat. Res. 43 451 [2] Ng K W, Ko W S, Tran T-T D, Chen R, Nazarenko M V, Lu F, Dubrovskii V G, Kamp M, Forchel A and ChangHasnain C J 2013 ACS Nano 7 100 5

G E Cirlin et al

J. Phys. D: Appl. Phys. 50 (2017) 484003

[14] Chen C, Shehata S, Fradin C, LaPierre R, Couteau C and Weihs G 2007 Nano Lett. 7 2584 [15] Lim S K, Tambe M J, Brewster M M and Gradečak S 2008 Nano Lett. 8 1386 [16] Dubrovskii V G, Shtrom I V, Reznik R R, Samsonenko Yu B, Khrebtov A I, Soshnikov I P, Rouvimov S, Akopian N, Kasama T and Cirlin G E 2016 Cryst. Growth Des. 16 7251 [17] Priante G, Glas F, Patriarche G, Pantzas K, Oehler F and Harmand J C 2016 Nano Lett. 16 1917 [18] Heiss M et al 2013 Nat. Mater. 5 439 [19] Rudolph D et al 2013 Nano Lett. 13 1522 [20] Jeon N, Loitsch B, Morkoetter S, Abstreiter G, Finley J, Krenner H J, Koblmueller G and Lauhon L J 2015 ACS Nano 8 8335 [21] Leandro L, Pedersen C P, Reznik R, Jöns K D, Samsonenko Y, Khrebtov A, Kasama T, Zwiller V, Cirlin G and Akopian N Nanowire quantum dots tuned to atomic resonances (In preparation)

[3] Mayer B et al 2013 Nat. Commun. 4 2931 [4] Boland J L et al 2015 Nano Lett. 15 1336 [5] Heinrich J, Huggenberger A, Heindel T, Reitzenstein S, Höfling S, Worschech L and Forchel A 2010 Appl. Phys. Lett. 96 211117 [6] Kats V N et al 2012 Semicond. Sci. Technol. 27 015009 [7] Trukhin V N, Bouravleuv A D, Mustafin I A, Kakko J P, Huhtio T, Cirlin G E and Lipsanen H 2015 Appl. Phys. Lett. 106 252104 [8] Kang J-H et al 2011 Cryst. Growth Des. 11 3109 [9] Tomioka K, Kobayashi Y, Motohisa J, Hara S and Fukui T 2009 Nanotechnology 20 145302 [10] Singh R and Bester G 2009 Phys. Rev. Lett. 103 063601 [11] Kukushkin S A and Osipov A V 2014 J. Phys. D: Appl. Phys. 47 313001 [12] Akopian N, Patriarche G, Liu L, Harmand J-C and Zwiller V 2010 Nano Lett. 10 1198 [13] Dubrovskii V G, Sibirev N V, Cirlin G E, Tchernycheva M, Harmand J C and Ustinov V M 2008 Phys. Rev. E 77 031606

6