Letter pubs.acs.org/NanoLett
Homogeneous Array of Nanowire-Embedded Quantum Light Emitters M. N. Makhonin,*,†,∥ A. P. Foster,†,∥ A. B. Krysa,‡ P. W. Fry,‡ D. G. Davies,† T. Grange,§ T. Walther,‡ M. S. Skolnick,† and L. R. Wilson*,† †
Department of Physics and Astronomy and ‡Department of Electronic and Electrical Engineering, University of Sheffield, Sheffield S3 7RH, United Kingdom § Walter Schottky Institut, Technische Universität München, Am Coulombwall 3, 85748 Garching, Germany S Supporting Information *
ABSTRACT: The potential for scale-up coupled with minimized system size is likely to be a major determining factor in the realization of applicable quantum information systems. Nanofabrication technology utilizing the III−V semiconductor system provides a path to scalable quantum bit (qubit) integration and a materials platform with combined electronic/photonic functionality. Here, we address the key requirement of qubit-site and emission energy control for scale-up by demonstrating uniform arrays of III−V nanowires, where each nanowire contains a single quantum dot. Optical studies of single nanowire quantum dots reveal narrow linewidth exciton and biexciton emission and clear state-filling at higher powers. Individual nanowire quantum dots are shown to emit nonclassically with clear evidence of photon antibunching. A model is developed to explain unexpectedly large excited state separations as revealed by photoluminescence emission spectra. From measurements of more than 40 nanowire quantum dots, we find emission energies with an ensemble broadening of 15 meV. The combination of deterministic site control and the narrow distribution in ensemble emission energy results in a system readily capable of scaling for multiqubit quantum information applications. KEYWORDS: Nanowire, quantum dot, site controlled, InGaAs, homogeneous, single photon
Q
nanowire diameter, growth time, and nanowire density. Several studies of NWQDs grown by a catalyst-assisted method have been reported, showing sharp line emission3,4 and photon antibunching.5 These studies demonstrate the potential of the NWQD approach for providing control over the QD location and dimensions. Site-controlled, catalyst-free NWQD growth has also been achieved, demonstrating narrow linewidths in the InP−InAsP6 and InGaAs−GaAs7 material systems, without study of site-to-site uniformity. A major advantage of catalystfree nanowire growth is the potential for growth of atomically sharp heterostructure interfaces. This is in contrast to catalyzed nanowire growth, where the catalyst reservoir effect can hinder the formation of such sharp interfaces.8,9 A further advantage of
uantum dots (QDs) possess many attractive properties, such as a quantized energy spectrum and ability to electrically control carrier occupation1 and charge state2 that are of great importance in a large number of current areas of semiconductor physics and device development. For many applications, control of both position and emission energy of QDs is desirable. For example, in emerging quantum information applications it is very likely that deterministic site-control of QDs will form a vital requirement for multiquantum bit (qubit) scale-up. The considerable difficulties in controlling QD location and emission properties in the most common self-assembled systems have motivated research into alternative methods of QD formation. One promising system consists of QDs in nanowires (NWQDs) in which fewnanometers thick layers of lower bandgap material are grown within the wider bandgap material of the nanowire. The lateral size, height and density of QDs are then controlled by the © 2013 American Chemical Society
Received: August 17, 2012 Revised: January 6, 2013 Published: February 11, 2013 861
dx.doi.org/10.1021/nl303075q | Nano Lett. 2013, 13, 861−865
Nano Letters
Letter
AsH3 flow rate was kept low at 30 sccm. Radial growth required to produce a passivating GaAsP capping layer was undertaken by increasing the V/III ratio through the introduction of phosphine (PH3) at 300 sccm and increase of AsH3 flow rate to 150 sccm for a 10 s period (for demonstration of capping see Figure S1 in the Supporting Information). A 10 s interrupt before growth of the capping layer allowed the AsH3 flow to be ramped up. Cooling was undertaken in a PH3 overpressure to maintain the nanowire surface by prevention of adatom desorption. Samples without a capping layer but cooled in a PH3 overpressure were also fabricated and also demonstrated high quality InGaAs optical properties, indicating passivation of the surface also occurs in this case. Nanowire growth was characterized using a scanning electron microscope (SEM) with a typical array shown in Figure 1b. The nanowires have a hexagonal morphology with six {110} side facets and a top (111)B facet.12 Transmission electron microscope (TEM) images of individual nanowires reveal a crystal structure typical of catalyst-free GaAs nanowires12,13 (Figure 1c) with a high density of zinc-blende rotational twins observed perpendicular to the growth direction. This is seen as a change in lattice fringe and diffraction contrast upon crossing the twin boundaries, which are separated by only ∼2.7 nm on average. As their contrast will be higher than that of the InGaAs NWQD and their density is so high, they make imaging of the NWQD by electron microscopy challenging. The difficulty in identifying InGaAs quantum well layers formed in GaAs nanowires by TEM has been described elsewhere.14 As previously reported for catalyst-free nanowire growth,15 the nanowire height is inversely related to the nanowire diameter. The nanowires were between 3 and 5 μm in length, for diameters decreasing from 240 to 80 nm. The structure of the GaAs nanowires is shown schematically in Figure 1a, containing a single, thin InGaAs layer which forms the QD. A 45° tilted SEM image of as-grown nanowires (Figure 1b) demonstrates the vertical nature and uniformity of the nanowires. For the growth conditions utilized here, we observe limited radial growth during formation of the nanowire core. Therefore the NWQD radius is nominally that of the initial hole size. Using a linear growth rate approximation, the NWQD height is estimated to be ∼16 nm (∼10 nm) for nanowires 80 nm (240 nm) in diameter. Microphotoluminescence (PL) spectra of the nanowire arrays were measured using a low-temperature piezo system. Diode laser excitation at 650 nm was focused to a 40 NWQDs with a full width half-maximum (fwhm) of ∼15 meV. Nanowire fabrication commenced with the deposition of 20 nm of SiO2 on a GaAs (111)B substrate using plasma enhanced chemical vapor phase deposition. Square arrays of holes, with a constant pitch of 4 μm and diameters of 50−250 nm were then developed in the SiO2 layer using a combination of electron beam lithography (EBL) and reactive ion etching. The patterned substrates were loaded into a horizontal low-pressure (150 Torr) metal−organic vapor phase epitaxy growth reactor and annealed at 780 °C in a hydrogen and arsine (AsH3) environment for 3 min prior to nanowire growth. For nanowire core growth, source materials were trimethyl-gallium (TMGa) and arsine (AsH3). NWQDs were formed through a 2 s addition of trimethyl-indium at 0.012% concentration in an H2 flow of 500 standard cubic centimeters (sccm) halfway through growth of the core. Conditions are calculated to produce a nominal indium incorporation within InGaAs of ∼15%. All growth was undertaken at 750 °C with the bottom and top GaAs barrier layers having equal growth times of 5 min. The flow rate of TMGa corresponds to that used to achieve a growth rate of ∼1.8 Å/s for planar epitaxy on (100) GaAs. The 862
dx.doi.org/10.1021/nl303075q | Nano Lett. 2013, 13, 861−865
Nano Letters
Letter
Figure 2. (a) PL spectra for a single NWQD exhibiting low power exciton (X) line and biexciton (2X) line at higher power (solid lines). Red dashed curve is Gaussian fit to data revealing exciton linewidth of ∼225 μeV. (Inset) Double log plot of PL emission intensity versus laser pump power demonstrates assignment of X (closed circles) and 2X (closed triangles) lines with linear (solid red line) and quadratic (dashed blue line) fitting respectively.
increasing power the biexciton (2X) emission intensity increases, with a measured X-2X splitting of 760 μeV. The assignment of the optical transitions follows from the linear (X) and quadratic (2X) intensity dependence on excitation power (inset of Figure 2). The narrowest linewidth we have observed thus far is 140 μeV, without any optimization of the GaAsP capping layer. At higher excitation powers the PL spectra show clear evidence of QD state filling with recombination observed from up to 6 excited QD states. This data is obtained for a nanowire containing a single InGaAs QD and results in sharp line PL emission from a single location along the nanowire (as demonstrated in Figure 1d, Figure S3 in Supporting Information). When two NWQDs are inserted within the same nanowire, PL from two separate exciton lines is observed (see Figure S4 in Supporting Information). This demonstrates that the formation of NWQDs arises from the incorporation of indium within the nanowire. We can therefore discount the possibility that we are observing wurtzite/zinc blende crystal phase NWQDs as reported by others.17,18 For nanowires of ∼120 nm diameter, a NWQD state separation of the order of 4−8 meV is deduced from PL spectra. The state separation is observed to increase with decreasing nanowire diameter, revealing the role of radial confinement in determining the NWQD optical properties. However, the measured state separation is significantly larger than the 2 meV calculated for 120 nm diameter NWQDs in which the lateral confinement comes solely from the NWQD radial dimensions. It is also observed that the separation energies of successive excited states are approximately constant (Figure 3a), unlike those predicted by an infinite square potential model. These observations can be described by introducing a parabolic potential for electrons and holes across the width of the NWQD. The origin of this potential is thought to be due to a decreasing indium compositional profile from the center to the edge of the nanowire. We model the NWQD as a 2D disk with parabolic variation in bandgap across the QD radius, truncated with infinite potential barriers at the QD radial edge due to the semiconductor−air interface. Figure 3b demonstrates the resulting fit of calculated energy levels with
Figure 3. (a) Power dependent PL spectra for a single NWQD. Vertical scale indicates separation of excited states (Ei) from ground state (E0). (b) Calculated excited state energy levels using a parabolic potential within a 2D disk model of the NWQD. Arrows show fit to experimental data in (a). (c) Detail of disk model of NWQD. A decreasing indium concentration from the center to the edge of the NWQD leads to a parabolic variation in bandgap across the QD. A hard wall boundary (infinite potential) truncates the potential at the edges of the NWQD. (d) Lowest energy electron probability density functions calculated for the potential in (a). Azimuthal (l) and radial (α) state numbers are labeled in the form lα. (e) Measured first excited state to ground state transition energy as a function of nanowire diameter (filled circles). Diameters are determined from SEM images of the individual nanowires. Curves describe theoretical fits to measured data using the potential in (c). Solid red line gives calculation for a flat potential (no indium composition variation within quantum dot). Long-dash black line gives best fit to measured data (conduction band parabolic depth of 29 meV). Pairs of green dot-dash (blue short dash) lines bound 50% (90%) of measured data points about the best fit line.
the power dependence PL data of Figure 3a. The parabolic potential is described in Figure 3c with the associated lowest energy probability density functions shown in Figure 3d. The functions are denoted lα, where l = s,p,d,f... is the azimuthal quantum number and α is the radial quantum number. It is noted that the quantization energy for excitations in the axial direction is calculated to be at significantly higher energy (typically 50 meV) than for the lateral directions, hence axial confinement is not discussed here. Figure 3e compares measured first-excited state to ground state transition energies for over 75 NWQDs with theoretical 863
dx.doi.org/10.1021/nl303075q | Nano Lett. 2013, 13, 861−865
Nano Letters
Letter
filter was used to select the exciton line, with some contribution from the biexciton also measured. Figure 4b shows a representative NWQD g (2) (t) measurement result. An exponential fit to this data reveals an exciton lifetime of ∼13.1 ns, which is in good agreement with the lifetime obtained in Figure 4a. The contribution of background signal to the measurement was accounted for by accumulating PL signal after a slight rotation of the bandpass filter to adjust the transmission wavelengths of the filter and thus remove the exciton line. A further background correction was applied in the manner of ref 23. This revealed a g(2)(t = 0) value of ∼50%. This is not unexpected as the contribution to the signal from biexciton recombination leads to an increased detection probability of multiphoton emission. This result therefore provides clear evidence of photon antibunching in the emission from these NWQDs. Figure 5a summarizes PL measurements from a representative sample of 20 NWQDs, demonstrating the uniformity of
values obtained from the parabolic potential model. The failure of a uniform indium concentration to describe the observed PL state separations is highlighted clearly (see line showing calculated fit for flat potential model). It is found that 50% of the NWQDs show state separations described by a parabolic conduction band potential depth of between 19 and 38 meV. These values correspond to a variation in the indium fraction from 15% at the center to 13 and 10.5% at the radial edge of the NWQD respectively. The model calculations of ground to excited state energy separations described here do not account for strain within the InGaAs NWQD, as described elsewhere for nanowire heterostructures.19 Our approach is justified as internal strain affects the ground state PL energy but has a much weaker influence on ground to excited state transition energies. Strain effects may also arise due to the presence of a GaAsP radial capping layer.11,20 The influence of strain on the GaAs PL emission energy due to the GaAsP capping layer is not observed. This suggests that capping-induced strain has little effect on PL emission from the NWQD. Time-resolved PL measurements were performed to study carrier recombination dynamics of single NWQDs. Excitation was achieved using a 405 nm pulsed laser operating at 20 MHz with PL emission collected using a fiber-coupled avalanche photo diode (APD). We typically observe long PL lifetimes of ∼10 ns for NWQDs and lifetimes of ∼6.5 ns associated with carrier recombination in the GaAs region of the nanowire (Figure 4a). PL lifetimes in this range are consistent with
Figure 5. (a) Representative PL spectra for 20 nanowires. Spectra shifted in y for clarity. (b) Ground-state energy statistical distribution for 42 NWQDs with Gaussian fit to complete distribution (blue line) revealing an ensemble broadening with fwhm of ∼15 meV.
emission wavelength in the array. These spectra were obtained at a constant power, explaining the presence of biexciton and charged state features in some of the NWQD spectra. Sharp features present within the GaAs emission at ∼1.5 eV support the conclusion that long GaAs PL lifetimes are associated with type-II recombination. Analysis of 42 NWQDs with diameter of 120−140 nm leads to a PL ensemble broadening of 15 meV (Figure 5b). At present, this is slightly larger than achieved by the best Stranski-Krastanow QD growth24 but has a significant advantage in that the emission is position controlled. Our measured ensemble broadening value compares favorably with a recent report of 14.4 meV ensemble broadening for sitecontrolled InGaAs quantum dot growth25 but is not yet optimized in the manner of the inverted pyramid approach to deterministic quantum dot growth.26 One contribution to the observed broadening originates with the variation of the initial hole opening size in the SiO2 mask which influences the NWQD height through its effect on growth rate. Further optimization of the EBL fabrication will enable more uniform nanowire growth rates resulting in a decrease in ensemble PL broadening. The approach reported here has significant scope for further improvement and exploitation. Control of emission wavelength is possible through tailoring of the source fluxes during growth. Reduced exciton linewidths of NWQDs can be expected
Figure 4. (a) PL lifetime data with monoexponential fitting (solid lines) for NWQD and GaAs in the same NW revealing lifetimes of 10.2 and 6.5 ns, respectively. (b) Second order autocorrelation measurement of g(2)(t) for typical NWQD exciton with accompanying exponential fit revealing a lifetime of ∼13.1 ns.
previous studies of nanowires exhibiting polytypism between wurtzite and zinc-blende GaAs.21 We attribute this to the high density of rotational twins observed within our nanowires, with each twin boundary forming a wurtzite layer.22 The resulting staggered band-offset results in spatially indirect recombination and PL lifetimes of ∼10 ns. This explanation holds for the InGaAs NWQD as well as for the bulk GaAs nanowire, as the nominal height of the NWQD is greater than the mean separation of rotational twins. To demonstrate the potential of this system for quantum information applications, measurements of the second order correlation function g(2)(t) were undertaken for a single NWQD utilizing a Hanbury-Brown Twiss setup. NWQD photoluminescence under continuous wave (CW) laser excitation at 650 nm was detected by two free space APDs and counted by a time-correlated single photon counting module with a time resolution of ∼40 ps. A 10 nm bandpass 864
dx.doi.org/10.1021/nl303075q | Nano Lett. 2013, 13, 861−865
Nano Letters
Letter
(7) Tatebayashi, J.; Ota, Y.; Ishida, S.; Nishioka, M.; Iwamoto, S.; Arakawa, Y. Appl. Phys. Lett. 2012, 100, 263101. (8) Dick, K. A.; Bolinsson, J.; Mattias Borg, B.; Johansson, J. Nano Lett. 2012, 12, 3200−3206. (9) Paladugu, M.; Zou, J.; Guo, Y.-N.; Zhang, X.; Kim, Y.; Joyce, H. J.; Gao, Q.; Hoe Tan, H.; Jagadish, C. Appl. Phys. Lett. 2008, 93, 101911. (10) Loss, D.; DiVincenzo, D. P. Phys. Rev. A 1998, 57 (1), 120−126. (11) Hua, B.; Motohisa, J.; Kobayashi, Y.; Hara, S.; Fukui, T. Nano Lett. 2009, 9, 112−116. (12) Motohisa, J.; Noborisaka, J.; Takeda, J.; Inari, M.; Fukui, T. J. Cryst. Growth 2004, 272, 180−185. (13) Hiruma, K.; Yazawa, M.; Katsuyama, T.; Ogawa, K.; Haraguchi, K.; Koguchi, M.; Kakibayashi, H. J. Appl. Phys. 1995, 77 (2), 447−462. (14) Yang, L.; Motohisa, J.; Takeda, J.; Tomioka, K.; Fukui, T. Nanotechnology 2007, 18, 105302. (15) Noborisaka, J.; Motohisa, J.; Fukui, T. Appl. Phys. Lett. 2005, 86, 213102. (16) Dal Don, B.; Zhao, H.; Moehl, S.; Ziegler, C.; Kalt, H. Phys. Status Solidi C 2003, 0, 1237−1241. (17) Akopian, N.; Patriarche, G.; Liu, L.; Harmand, J.-C.; Zwiller, V. Nano Lett. 2010, 10, 1198−1201. (18) Dick, K. A.; Thelander, C.; Samuleson, L.; Caroff, P. Nano Lett. 2010, 10, 3494−3499. (19) Pistol, M.-E.; Pryor, C. Phys. Rev. B 2009, 80, 035316. (20) Montazeri, M.; Fickenscher, M.; Smith, L. M.; Jackson, H. E.; Yarrison-Rice, J.; Kang, J. H.; Gao, Q.; Hoe Tan, H.; Jagadish, C.; Guo, Y.; Zou, J.; Pistol, M.-E.; Pryor, C. E. Nano Lett. 2010, 10, 880−886. (21) Spirkoska, D.; Arbiol, J.; Gustafsson, A.; Conesa-Boj, S.; Glas, F.; Zardo, I.; Heigoldt, M.; Gass, M. H.; Bleloch, A. L.; Estrade, S.; Kaniber, M.; Rossler, J.; Peiro, F.; Morante, J. R.; Abstreiter, G.; Samuelson, L.; Fontcuberta i Morral, A. Phys. Rev. B 2009, 80, 245325. (22) Dick, K. A.; Caroff, P.; Bolinsson, J.; Messing, M. E.; Johansson, J.; Deppert, K.; Wallenberg, L. R.; Samuelson, L. Semicond. Sci. Technol. 2010, 25, 024009. (23) Brouri, R.; Beveratos, A.; Poizat, J.-P.; Grangier, P. Opt. Lett. 2000, 25 (17), 1294−1296. (24) Mi, Z.; Bhattacharya, P. J. Appl. Phys. 2005, 98, 023510. (25) Huggenberger, A.; Heckelmann, S.; Schneider, C.; Höfling, S.; Reitzenstein, S.; Worschech, L.; Kamp, M.; Forchel, A. Appl. Phys. Lett. 2011, 98, 131104. (26) Mohan, A.; Gallo, P.; Felici, M.; Dwir, B.; Rudra, A.; Faist, J.; Kapon, E. Small 2010, 6 (12), 1268−1272. (27) Friedler, I.; Sauvan, C.; Hugonin, J. P.; Lalanne, P.; Claudon, J.; Gerard, J. M. Opt. Express 2009, 17 (4), 2095−2110. (28) Tomioka, K.; Motohisa, J.; Hara, S.; Hiruma, K.; Fukui, T. Nano Lett. 2010, 10 (5), 1639−1644. (29) Scofield, A. C.; Shapiro, J. N.; Lin, A.; Williams, A. D.; Wong, P.S.; Liang, B. L.; Huffaker, D. L. Nano Lett. 2011, 11, 2242−2246.
through optimization of the GaAsP capping layer to reduce the effect of spectral diffusion of photogenerated charges. Optimizing radial nanowire dimensions may allow for an enhancement of the axial single photon emission rate in the manner of ref 27. From a device perspective, we anticipate potential for all-electrical control of individual NWQDs by adapting established electrical contacting schemes for nanowire arrays.28 Arrays of heterostructured nanowires have also been demonstrated to show promise as the building blocks of dielectric-rods in air photonic crystals.29 In conclusion, we have demonstrated arrays of single InGaAs quantum dots in GaAs nanowires with resulting high quality optical properties. An ensemble broadening of 15 meV compares favorably with other deterministic quantum dot growth schemes. Exciton linewidths down to 140 μeV have been observed, along with distinct photon antibunching behavior. Larger than expected energy splitting between ground and higher order energy states has been explained using a variable compositional profile for the indium within each NWQD. The deterministic site control provided by this NWQD growth approach, allied with narrow ensemble distribution of PL emission energy, results in a system readily capable of scaling for multiqubit quantum information applications.
■
ASSOCIATED CONTENT
■
AUTHOR INFORMATION
S Supporting Information *
Additional information and figures. This material is available free of charge via the Internet at http://pubs.acs.org. Corresponding Author
*E-mail: (M.N.M.) m.makhonin@sheffield.ac.uk; (L.R.W.) luke.wilson@sheffield.ac.uk. Author Contributions ∥
M.N.M. and A.P.F. contributed equally to this work.
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by EPSRC Grants EP/G001642 and EP/J007544. A.P.F. acknowledges funding through the Research Councils U.K. Energy Programme, via the E-Futures DTC at the University of Sheffield, U.K. T.G. acknowledges support from the Austrian Science Fund FWF (SFB IR-ON).
■
REFERENCES
(1) Drexler, H.; Leonard, D.; Hansen, W.; Kotthaus, J. P.; Petroff, P. M. Phys. Rev. Lett. 1994, 73 (16), 2252−2255. (2) Ediger, M.; Dalgarno, P. A.; Smith, J. M.; Gerardot, B. D.; Warburton, R. J.; Karrai, K.; Petroff, P. M. Appl. Phys. Lett. 2005, 86, 211909. (3) van Weert, M. H. M.; Akopian, N.; Perinetti, U.; van Kouwen, M. P.; Algra, R. E.; Verheijen, M. A.; Bakkers, E. P. A. M.; Kouwenhoven, L. P.; Zwiller, V. Nano Lett. 2009, 9 (5), 1989−1993. (4) Dalcau, D.; Mnaymneh, K.; Wu, X.; Lapointe, J.; Aers, G. C.; Poole, P. J.; Williams, R. L. Appl. Phys. Lett. 2011, 98, 251101. (5) Borgstrom, M. T.; Zwiller, V.; Müller, E.; Imamoglu, A. Nano Lett. 2005, 5 (7), 1439−1443. (6) Dorenbos, S. N.; Sasakura, H.; van Kouwen, M. P.; Akopian, N.; Adachi, S.; Namekata, N.; Jo, M.; Motohisa, J.; Kobayashi, Y.; Tomioka, K.; Fukui, T.; Inoue, S.; Kumano, H.; Natarajan, C. M.; Hadfield, R. H.; Zijlstra, T.; Klapwijk, T. M.; Zwiller, V.; Suemune, I. Appl. Phys. Lett. 2010, 97, 171106. 865
dx.doi.org/10.1021/nl303075q | Nano Lett. 2013, 13, 861−865
