NANO LETTERS
Growth and Optical Properties of Strained GaAs−GaxIn1-xP Core−Shell Nanowires
2005 Vol. 5, No. 10 1943-1947
Niklas Sko1 ld,† Lisa S. Karlsson,‡ Magnus W. Larsson,‡ Mats-Erik Pistol,† Werner Seifert,† Johanna Tra1 ga˚rdh,† and Lars Samuelson*,† Solid State Physics/The Nanometer Structure Consortium, Lund UniVersity, Box 118, SE-221 00 Lund, Sweden, and Materials Chemistry/The Nanometer Structure Consortium, Lund UniVersity, Box 124, SE-221 00 Lund, Sweden Received July 8, 2005; Revised Manuscript Received August 19, 2005
ABSTRACT We have synthesized GaAs−GaxIn1-xP (0.34 < x < 0.69) core−shell nanowires by metal−organic vapor phase epitaxy. The nanowire core was grown Au-catalyzed at a low temperature (450 °C) where only little growth takes place on the side facets. The shell was added by growth at a higher temperature (600 °C), where the kinetic hindrance of the side facet growth is overcome. Photoluminescence measurements on individual nanowires at 5 K showed that the emission efficiency increased by 2 to 3 orders of magnitude compared to uncapped samples. Strain effects on the band gap of lattice mismatched core−shell nanowires were studied and confirmed by calculations based on deformation potential theory.
Semiconductor nanowires can serve as waveguides for light as well as for charge carriers. They are therefore suitable building blocks for nanophotonics and nanoelectronics. Lasers,1 photodetectors,2 light-emitting diodes,3 and fieldeffect transistors4 have all been fabricated using individual nanowires. By introduction of heterostructures along the axis of the wire, more complex structures such as single-electron transistors,5 resonant tunneling diodes,6 and quantum dots7 have been realized. Forming a radial shell around the nanowires can significantly improve the performance of such devices. A shell moves surface states away from the core, and if the core and the shell are grown lattice mismatched, pseudomorphic strain induced in the core will offer added flexibility in band structure engineering. A shell can furthermore be used to optimize the optical cavity properties of the nanowire. Since the surface-to-volume ratio of the narrow nanowires is extremely large, surface states will reduce the carrier lifetime and degrade the device performance. Surface passivation is therefore of great importance, especially for GaAs nanowires which oxidize readily. Such passivation can be achieved by forming a shell of a large band gap material around the nanowire so that the surface states are moved away from the charge carriers confined in the core. For an electronics device this will reduce the surface recombination * Corresponding author. E-mail:
[email protected]. † Solid State Physics/The Nanometer Structure Consortium. ‡ Materials Chemistry/The Nanometer Structure Consortium. 10.1021/nl051304s CCC: $30.25 Published on Web 09/13/2005
© 2005 American Chemical Society
current while a photonics device will gain higher emission efficiency. Although the passivation effect is commonly used to motivate the importance of core-shell nanowires,8-10 there has prior to this study not been any comparisons made between the emission efficiency of wires with and without shells. Growing the shell lattice mismatched to the core will induce pseudomorphic strain in the core and thereby offer flexibility in designed band structure engineering. By utilizing the strain, the band gap of the core can be tuned and the valence bands split. Splitting the valence bands and thereby also reducing the effective mass is of importance, e.g., for reducing the lasing threshold current for the nanowires. Strain-induced energy shifts in the core have previously been observed for binary shells,9 but there are no reports on a controlled tuning of the band gap using ternary shells. For photonics applications, a shell can furthermore improve the optical cavity properties of the nanowire.11 A practical lower limit for a nanowire to function as a single mode optical waveguide12 is when (πD/λ)(n12 - n02)1/2 ≈ 1, where D is the nanowire diameter, λ is the wavelength, and n1 and n0 are the refractive indices of the nanowire and the surrounding medium, respectively. For a GaAs nanowire at 300 K (λ ≈ 870 nm, n1 ≈ 3.6) the minimum diameter is 80 nm. When a shell is used, the core diameter can be chosen arbitrarily small while the overall diameter is kept larger to optimize the cavity.
In this paper we report on growth of GaAs nanowires with a GaxIn1-xP shell using low-pressure metal-organic vapor phase epitaxy (MOVPE). We investigate the optical properties of these nanowires with a focus on the passivation and strain effects of the shell. The common method for producing nanowires is to catalyze growth using a metallic seed particle, usually Au. Growth should ideally only take place underneath the Au and not on the substrate and side facets of the nanowire. Therefore, the nanowires are grown at a low temperature where growth on the uncatalyzed surfaces is kinetically hindered.13 The unidirectional growth is usually explained by the vapor-liquid-solid mechanism where a liquid Au droplet works as a preferential sink for the growth elements, which precipitate at the liquid-solid interface as the droplet supersaturates. Recently, investigations have indicated that, under some circumstances, the mechanism may instead be a vapor-solid-solid mechanism.14 In our approach Au aerosols, size selected to control the core diameter of the wire, were randomly dispersed on (111)B GaAs substrates. The samples were then transferred to the MOVPE reactor cell and annealed at 580 °C for 10 min prior to growth in group V ambient (AsH3 at a molar fraction of 5 × 10-4 in 6 L/min H2 at 100 mbar pressure) to desorb the surface oxide. After annealing, the temperature was ramped down to 450 °C for the nanowire growth which started when trimethylgallium (TMG) was supplied to the reactor cell. At this low temperature very little growth takes place at the uncatalyzed surfaces. The nanowires grow in the 〈111〉B direction with the Au particle on top, but only little growth takes place on the {110} side facets of the nanowire.15 After 9 min the growth was terminated by switching off the TMG flow, and the temperature was subsequently increased to 600 °C, in ambient AsH3/H2, where the kinetic hindrance of the side facet growth is overcome and the GaxIn1-xP shell could be deposited. AsH3 was switched off and PH3 (at a molar fraction of 1.5 × 10-2) switched on 8 s before the TMG and the trimethylindium (TMI) to prevent any Ga/In exchange taking place in the GaAs core. The TMG molar fraction was kept constant at 2 × 10-5 while the TMI molar fraction was varied between 8 × 10-6 and 3 × 10-5 for the different samples. Samples with four different shell compositions were produced. Figure 1a shows a scanning electron micrograph (SEM) of the nanowires tilted 45° from the e-beam. A top view of a single nanowire is seen in Figure 1b. The shell growth was clearly epitaxial forming well-defined {110} side facets. Figure 1c shows a high-angle annular dark field scanning transmission electron microscope (STEM) image of the top of a wire, the arrows are guides for the eye to identify the core. The image reveals that the GaxIn1-xP shell grows both on the {110} side facets and underneath the Au in the 〈111〉B direction. The growth rate of the shell is approximately the same on the side facets as underneath the Au. This indicates that the growth rate at 600 °C, the temperature at which the shell was grown, is limited by mass transport rather than by surface reactions. X-ray energy dispersive spectrometer (XEDS) line scans across the 1944
Figure 1. (a, b) SEM images of core-shell nanowires. (a) Wires seen at a 45° angle, scale bar 1 µm. (b) Top view of a single wire. The round Au particle is seen in the center and the inner hexagon is the {110} side facets while the outer hexagonal shape is the base of the wire. Scale bar 100 nm. (c) High-angle annular dark field STEM image of the top of a wire. The growth rate of the shell is approximately the same on the side facets as underneath the Au. Scale bar 50 nm. (d) XEDS line scan across a nanowire, the lower trace represents the As signal from the core while the upper trace represents the P signal from the shell.
nanowires (Figure 1d) confirmed the core-shell structure. The lower trace in Figure 1d represents the As signal from the core while the upper trace represents the P signal from the shell. According to XEDS point measurements the Ga content in the shell was 0.34, 0.48, 0.58, and 0.69 for the four samples, respectively. Thus the lattice mismatch of these core-shell nanowires ranged from -1.3% to 1.3%. Photoluminescence (PL) measurements were performed in order to study the effect of the shell on the electronic structure as well as on the luminescence efficiency. The wires were transferred to a patterned SiO2 or Au surface where single wires could be localized and studied. Measurements were carried out at 5 K. For excitation below the band gap of the shell, a tunable Ti:sapphire laser emitting at 700 nm was used while the 458 nm line from an Ar+ laser was used for excitation above the band gap of the shell. The emission Nano Lett., Vol. 5, No. 10, 2005
Figure 2. (a) PL image of a core-shell nanowire, the wire functions as a waveguide and emission is only observed from the ends. (b) SEM image of the same wire vertically shifted in relation to part a. Scale bar for (a) and (b) is 1 µm. (c) PL from 60 nm thick wires with and without the shell. The shell improved the emission efficiency 2 to 3 orders of magnitude. Measurements were performed at 5 K.
was collected by an optical microscope, dispersed through a spectrometer and detected by a liquid N2 cooled charge coupled device (CCD) camera. For time-resolved measurements a mode-locked Ti:sapphire laser was used as excitation source. The laser light was frequency doubled to provide pulses shorter than 250 fs at 440 nm. A streak camera was used for detection. The time resolution of the system was about 10 ps. To investigate the effect of the shell on the emission efficiency, nanowires with and without shell were excited at 700 nm. The wavelength was chosen so that excitation only occurred in the core. Wires with a core diameter of 20, 40, and 60 nm were studied; for the core-shell wires the shell thickness was 70-100 nm. All wires with shell were luminescing intensely regardless of the core diameter, while for the wires without shell only the 60 nm wires had a luminescence detectable with our setup. A typical luminescence image of a core-shell nanowire is seen in Figure 2a. Emission is only observed from the ends of the nanowire, illustrating its waveguiding properties. A scanning electron microscopy (SEM) image of the wire is seen in Figure 2b. Figure 2c shows a comparison between two 60 nm wires, one without a shell and one with an almost lattice matched Ga0.48In0.52P shell. The emission efficiency was improved 2 to 3 orders of magnitude for the 60 nm wires when a passivating shell was used. For thinner wires the effect of the shell is expected to increase as the surface-to-volume ratio of the core increases. Strain effects on the band gap of the core were measured for the four different shell compositions using photolumiNano Lett., Vol. 5, No. 10, 2005
Figure 3. (a) PL spectra from four wires with different shell composition, x ) 0.34, 0.48, 0.58, and 0.69, from bottom to top. The core diameter was 40 nm for all the wires while the shell thickness varied between 70 and 100 nm. By straining the core the band gap was tuned from 1.37 to 1.61 eV. The multiple peaks in the shell spectra indicate alloy ordering. Measurements were performed at 5 K. (b) The band gap of the core and shell respectively as a function of x, the Ga content in the shell. Experimental values are plotted as filled and empty dots while solid and dashed lines represent calculated values for the core and the shell, respectively. (c) Schematic illustration of the strain induced in the core by the lattice mismatched shell. Figure only indicates radial strain effects.
nescence. We also computed the shift of the band gap due to strain both for the core and for the shell. The calculations were based on linear deformation potential theory where mixing of eight bands was included. The strain was computed using continuum elasticity theory using a grid of 120 × 120 × 120 elements and solved by finite difference methods. More information regarding the calculations can be found in ref 16. Figure 3a shows the spectra of four representative wires with different composition of the GaxIn1-xP shell (x ) 0.34, 0.48, 0.58, and 0.69, from bottom to top); the wires were excited at 458 nm. The core diameter was 40 nm while the shell thickness varied between 70 and 100 nm. We have 1945
Figure 4. Time-resolved PL from a 40 nm wire with Ga0.48In0.52P shell. The intensity is normalized and the spectra from the core and the shell are translated for clarity. The luminescence decay of the core was on the order of 100 ps while the shell generally showed a slower decay. Measurements were performed at 5 K.
found that the strain situation hardly changes for a shell/ core thickness ratio higher than 2, meaning that for these samples the core is fully strained. In Figure 3a, the spectra from the shell show multiple peaks which indicate alloy ordering; hence composition and band gap vary from region to region in the shell. When the wires were excited in the shell, using photons with energy exceeding the band gap of the shell, the carrier transport from the shell to the core was poor. Due to the ordering, localized potential minima in the band structure of the shell are expected to collect the charge carriers and the peak originating from the core was therefore not so pronounced. For each sample in Figure 3a the core spectrum has been enhanced 5-10 times in comparison to the spectrum from the shell. By straining the core the band gap was tuned from 1.37 to 1.61 eV. For the sample with the largest compressive strain, the PL peak from the core broadened substantially which we attribute to the inhomogeneous strain from the alloy ordered shell. Regions in the shell with different composition will exert different amounts of strain on the nanowire core. In Figure 3b experimental values of the band gap of the core and the shell respectively are plotted as a function of x, the Ga content in the shell, and compared to calculations. For the shell the centroid of the multiple peaks was taken as the average band gap. Experimental values are plotted as filled and empty dots while solid and dashed lines represent calculated values for the core and the shell, respectively. Due to the alloy ordering we tend to underestimate the band gap of the shell. The small band gap regions contribute more to the spectra than the ones with large band gap. Figure 3c is a schematic illustration of the strain induced in the core by the lattice mismatched shell. The left figure illustrates tensile strain corresponding to the lower spectrum in (a), the middle figure illustrates lattice matching nearly corresponding to the second spectrum from the bottom, and the right figure illustrates compressive strain corresponding to the two top spectra. Time-resolved PL measurements were performed on the wires with a 40 nm core and Ga0.48In0.52P shell, Figure 4. The decay of the luminescence intensity from the core was 1946
exponential with a time constant on the order of 100 ps although some of the investigated wires displayed an additional faster decay component. The luminescence decay of the shell was nonexponential, and for most of the investigated wires the decay was slower than the decay for the core. The short lifetime indicates that although the GaAs surface states were removed by the shell, there were still a large number of nonradiative recombination centers left. Previous studies17 on pure GaAs nanowires have shown radiative lifetimes on the order of nanoseconds and a decrease in the surface recombination rate after surface passivation treatment. Due to the low emission efficiency of the uncapped wires, we have not been able to do any timeresolved measurements on these wires. In summary, we have synthesized GaAs-GaxIn1-xP (0.34 < x < 0.69) core-shell nanowires. SEM and STEM studies showed that the shell grows epitaxially. PL studies showed that the emission efficiency of a GaAs nanowire can be improved by at least 2 to 3 orders of magnitude when a passivating shell is used. By straining the GaAs nanowires using lattice mismatched shells, the band gap of the core was tuned over a range of 240 meV. The measured straininduced shifts of the band gap were in good agreement with calculated values. Alloy ordering of the shell reduced the homogeneity of the strain and the carrier transport from the shell to the core. Time-resolved measurements showed that the luminescence intensity decayed with a time constant on the order of 100 ps for the core, indicating more nonradiative recombination centers in addition to the surface states which had been passivated by the shell formation. Acknowledgment. This work was carried out within the Nanometer Structure Consortium in Lund and was supported by the Swedish Foundation for Strategic Research (SSF), the Swedish Research Council (VR), the NoE SANDiE (EU Grant No. NMP4-CT-2004 500101), Office of Naval Research (ONR), and the Knut and Alice Wallenberg Foundation. References (1) Huang, M. H.; Mao, S.; Feick, H.; Yan, H.; Wu, Y.; Kind, H.; Weber, E.; Russo, R. Yang, P. Science 2001, 292, 1897. (2) Wang, J.; Gudiksen, M. S.; Duan, X.; Cui, Y.; Lieber, C. M. Science 2001, 293, 1455. (3) Duan, X.; Huang, Y.; Cui, Y.; Wang, J.; Lieber, C. M. Nature 2001, 409, 66. (4) Cui, Y.; Lieber, C. M. Science 2001, 291, 851. (5) Thelander, C.; Mårtensson, T.; Bjo¨rk, M. T.; Ohlsson, B. J.; Larsson, M. W.; Wallenberg, L. R.; Samuelson, L. Appl. Phys. Lett. 2003, 83, 2052. (6) Bjo¨rk, M. T.; Ohlsson, B. J.; Thelander, C.; Persson, A. I.; Deppert, K.; Wallenberg, L. R.; Samuelson, L. Appl. Phys. Lett. 2002, 81, 4458. (7) Panev, N.; Persson, A. I.; Sko¨ld, N.; Samuelson, L. Appl. Phys. Lett. 2003, 83, 2238. (8) Lauhon, L. J.; Gudiksen, M. S.; Wang, D.; Lieber, C. M. Nature 2002, 420, 57. (9) Lin, H.-M.; Chen, Y.-L.; Yang, J.; Liu, Y.-C.; Yin, K.-M.; Kai, J.J.; Chen, F.-R.; Chen, L.-C.; Chen, Y.-F.; Chen, C.-C. Nano Lett. 2003, 3, 537. (10) Tateno, K.; Gotoh, H.; Watanabe, Y. Appl. Phys. Lett. 2004, 85, 1808. (11) Choi, H.-J.; Johnson, J. C.; He, R.; Lee, S.-K.; Kim, F.; Pauzauskie, P.; Goldberger, J.; Saykally, R. J.; Yang, P. J. Phys. Chem. B 2003, 107, 8721.
Nano Lett., Vol. 5, No. 10, 2005
(12) Duan, X.; Huang, Y.; Agarwal, R.; Lieber, C. M. Nature 2003, 421, 241. (13) Stringfellow, G. B. Organometallic Vapor-Phase Epitaxy; Academic Press: New York, 1999; pp 324-347. (14) Persson, A. I.; Larsson, M. W.; Stenstro¨m, S.; Ohlsson, B. J.; Samuelson, L.; Wallenberg, L. R. Nat. Mater. 2004, 3, 677.
Nano Lett., Vol. 5, No. 10, 2005
(15) Borgstro¨m, M.; Deppert, K.; Samuelson, L.; Seifert, W. J. Cryst. Growth 2004, 260, 18. (16) Pryor, C. E.; Pistol, M.-E. xxx.lanl.gov/cond-mat/0501090. (17) Ogawa, K.; Haraguchi, K.-I.; Hiruma, K.; Fujisaki, Y.; Katsuyama, T.; Fasol, G. J. Lumin. 1992, 53, 387.
NL051304S
1947
