GaAs nanoneedles grown on sapphire
APPLIED PHYSICS LETTERS 98, 123101 共2011兲
GaAs nanoneedles grown on sapphire Linus C. Chuang, Michael Moewe, Kar Wei Ng, Thai-Truong D. Tran, Shanna Crankshaw, Roger Chen, Wai Son Ko, and Connie Chang-Hasnaina兲 Department of Electrical Engineering and Computer Sciences, and Applied Science and Technology, University of California at Berkeley, Berkeley, California 94720, USA
共Received 15 November 2010; accepted 9 January 2011; published online 21 March 2011兲 Heterogeneous integration of dissimilar single crystals is of intense research interests. Lattice mismatch has been the most challenging bottleneck which limits the growth of sufficient active volume for functional devices. Here, we report self-assembled, catalyst-free, single crystalline GaAs nanoneedles grown on sapphire substrates with 46% lattice mismatch. The GaAs nanoneedles have a 2–3 nm tip, single wurtzite phase, excellent optical quality, and dimensions scalable with growth time. The needles have the same sharp, hexagonal pyramid shape from ⬃100 nm 共1.5 min growth兲 to ⬃9 m length 共3 h growth兲. © 2011 American Institute of Physics. 关doi:10.1063/1.3567492兴 Heterogeneous integration of dissimilar single crystals has been of interest because it enables monolithic, largescale integration of functional devices that cannot be achieved with one material system alone. Mismatch in lattice constant and thermal expansion coefficient presents a major constraint that limits the growth of thick, high-quality films on foreign substrate,1 thereby preventing them from attaining the active volumes needed for functional devices. For a given lattice mismatch, there exists a critical thickness.2 When growth exceeds the critical thickness, misfit dislocations form to reduce the strain energy imposed by the mismatch. This drastically degrades the optical and electrical properties of the materials,1 rendering them not suitable for device applications. Recently, researchers have increased the critical thickness by laterally confining the deposition areas in the form of, for example, nanowires with small footprints.3,4 In such case, the nanowire diameter becomes the critical dimension, which decreases with lattice mismatch. For an 11.6% mismatch, the critical diameter is only 26 nm,4 making it challenging for device engineering. Here, we report high-quality, single-phase GaAs structures directly grown on a c-plane sapphire substrate with an extremely large 46% lattice mismatch. The growth assumes a tapered nanoneedle structure with dimensions that scale linearly with growth time over two orders of magnitude without apparent critical dimensions. The GaAs nanoneedle maintains a 2–3 nm tip throughout its growth and is formed spontaneously from the very beginning without catalysts. Its excellent optical quality with bright room-temperature 共RT兲 luminescence and narrow Raman linewidths attest to its coherent crystalline phase. The wurtzite nanoneedles are hexagonal pyramids with 7°–11° sidewall-to-sidewall angles that taper to an ultrasharp tip. The combination of an exceedingly sharp tip and wide, strong base makes them excellent candidates for applications use in field emitter, microscopic, and nonlinear optical signal generation.5 Figure 1共a兲 shows the side view of GaAs nanoneedles on sapphire, on which most GaAs nanoneedles are aligned vertically to the 关0001兴 substrate orientation. Slanted nanon¯ 02典 directions. eedles are identified along one of the three 具11 The nanoneedles are grown by low temperature 共400 ° C兲 a兲
Electronic mail: [email protected].
0003-6951/2011/98共12兲/123101/3/$30.00
metal-organic chemical vapor deposition without substrate surface treatment. The substrate goes through a 3 min pregrowth annealing at 600 ° C with tertiarybutylarsine 共TBA兲, followed by growth at 400 ° C. The sources are triethylgallium and TBA and the mole fractions are set as 1.12⫻ 10−5 and 5.42⫻ 10−4, respectively, in a 12 l/min hydrogen carrier gas flow. Sharp GaAs nanoneedles with 9° tip and smooth sidewall facets are observed for 82 min growth time 关Fig. 1共b兲兴. A hexagonal cross section is seen with the top-down scanning electron microscope 共SEM兲 image in Fig. 1共c兲. The ¯ 00兴 direction is rotated by 30° with respect to nanoneedle 关11 ¯ the 关1100兴 direction of the sapphire substrate 关Fig. 1共c兲兴, determined by transmission electron microscopy 共TEM兲. This 30° rotation is similar to the report of thin-film GaN on sapphire due to the alignment of Ga with Al.6 The lattice mismatch is, hence, calculated to be 46% with compressive
FIG. 1. 共Color online兲 共a兲 Side-view SEM image of as-grown GaAs nanoneedles. 共b兲 30°-tilt SEM image of a nanoneedle with 82 min growth time. ¯ 00其 and 共0001兲 terraces. 共c兲 Top-down view of The sidewalls are made of 兵11 a nanoneedle showing a hexagonal cross section. The in-plane orientation of the nanoneedle shows a 30° rotation with respect to the sapphire substrate. 共d兲 30°-tilt SEM images of nanoneedles with different growth times. The image of the nanoneedle with 1.5 min growth time is enlarged by 20⫻. 共e兲 Schematic showing the growth of single-crystalline nanoneedle body and the polycrystalline connecting layer. Inset is a cross-section SEM image for 60 min growth. A thin, 200 nm connecting layer is seen.
98, 123101-1
© 2011 American Institute of Physics
123101-2
Chuang et al.
FIG. 2. 共Color online兲 共a兲 TEM image showing a nanoneedle tip with less than 3 nm in diameter. Inset is the FFT image showing the wurtzite crystal¯ 00兴 zone axis. The nanoneedle orientation during the line structure with 关11 inspection is schematically shown. 共b兲 Selective-area diffraction pattern for ¯ 10兴 zone axis. 共c兲 An HRTEM image, with 关11 ¯ 00兴 a nanoneedle with 关12 zone axis, showing that the nanoneedle sidewall ridge consists of atomic steps which result in the taper of a nanoneedle. 共d兲 HRTEM of a GaAs nanoneedle grown on GaAs with a sharp tip of only three lattice spacings.
strain. This fixed in-plane orientation relationship suggests the nanoneedle growth being epitaxial. The needles are sharp from the beginning and remain the same geometry for long growth durations. Figure 1共d兲 shows nanoneedles with growth time ranging from 1.5 to 180 min. The distinctive nanoneedle shape is already observed after 1.5 min of growth. The base diameter and height increase linearly with growth time and maintain a constant taper angle. We estimate the growth rates to be 0.3 m / h and ¯ 00典 and 关0001兴 crystal orientations, 3.6 m / h for the 具11 respectively. The nanoneedle growth starts with seed nucleation, followed by two-dimensional thin-film deposition on the preferred six sidewall facets in a core-shell mode. This growth mode is the same as our GaAs nanoneedles on roughened Si substrates.7 Nearly the same growth is observed with a substrate having ten times larger mismatch. In addition to the prominent nanoneedle growth, a thin connecting layer between individual nanoneedles is observed 关Figs. 1共a兲–1共d兲兴. The cross-section SEM image of a 60 min GaAs nanoneedle growth run 关Fig. 1共e兲, inset兴 shows a 200 nm thick connecting layer covering the entire sapphire substrate surface except where nanoneedles grow. The same connecting layer is seen for GaAs nanoneedles grown on Si7 and identified as polycrystalline GaAs by TEM, similar to low-temp growth of GaAs on sapphire.8 The singlecrystalline GaAs nanoneedle and the polycrystalline connecting layer are thought to coalesce in the nanoneedle root region as schematically shown in Fig. 1共e兲. At the beginning of the growth, a nanoneedle seed forms a small singlecrystalline connection to the sapphire substrate with elsewhere covered by a thin polycrystalline film 共both shown in red兲. Each incremental growth 共shown by different colors兲 continues to deposit an additional single-crystalline film 共at ⬃0.3 m / h兲 onto the nanoneedle sidewall and a polycrystalline film 共at ⬃0.2 m / h兲 onto the connecting layer,
Appl. Phys. Lett. 98, 123101 共2011兲
FIG. 3. 共Color online兲 共a兲 4 K -PL spectra for a nanoneedle with various excitation powers. Two peaks were seen. 共b兲 Multi-Lorentzian fit for the 4 K spectrum with 100 W excitation power with peaks at 1.505 and 1.519 eV having respective linewidths of 20 and 18 meV. 共c兲 Peak intensity vs excitation power for the two peaks. The ⬃1.519 eV peak 共free exciton兲 intensity increases with the excitation power while the ⬃1.505 eV peak 共impurity related兲 starts to saturate at 500 W. 共d兲 300 K -PL spectra for a nanoneedle with various excitations.
where as the ratio of the two growth rates 共1.5⫻兲 determines the single-crystalline and polycrystalline boundary. As a result, the single-crystalline part of an as-grown nanoneedle should have a reversely tapered root. This is verified for GaAs nanoneedles grown on Si. For GaAs nanoneedles on sapphire, the interface is still under investigation. Figure 2共a兲 shows a high-resolution TEM 共HRTEM兲 image of a nanoneedle tip 2–3 nm in diameter. The zone axis is ¯ 00兴, with a nanoneedle sidewall facet approximately per关11 pendicular to it, due to the small but finite nanoneedle taper angle. The fast Fourier transform 共FFT兲 of the TEM image is shown as the inset. Since this FFT pattern can be with either ¯ 00兴 or zinc-blende 关21 ¯¯1兴 zone axis, the same wurtzite 关11 nanoneedle was rotated by 30° about its growth axis for a second diffraction pattern. In this second case, the resulting selective-area diffraction pattern 关Fig. 2共b兲兴 unambiguously identifies the wurtzite structure, which is coherent throughout the nanoneedle. The nanoneedle axes shown in Figs. 2共a兲 and 2共b兲, are along the 关0001兴 direction and hence it suggests the nanoneedle growth on c-sapphire being epitaxial. Close inspection of the tapered nanoneedle sidewall ridge reveals single-atomic steps 关Fig. 2共c兲兴, which result in the taper. The ¯ 00其 and tapered sidewall facets therefore consist of 兵11 共0001兲 terraces. It is interesting to note that the sharp nanoneedle tip, the geometry and crystalline phase are identical for needles grown on the 46% mismatched system and those on GaAs,7 for which we show the sharpest single crystal ever reported with only three lattice spacings at the tip 关Fig. 2共d兲兴. Single-needle microphotoluminescence 共-PL兲 is investigated using a 514 nm argon laser focused to ⬃1.5 m in diameter. Figure 3共a兲 shows typical 4 K -PL spectra for a range of excitation powers. A fit of the PL spectrum at 100 W excitation power by multiple Lorentzian line shapes 关Fig. 3共b兲兴 shows two peaks at 1.519 and 1.505 eV. The intensities of the two peaks as a function of the excitation power are plotted in Fig. 3共c兲. The peak intensity at ⬃1.519 eV has a superlinear dependence on excitation,
123101-3
Appl. Phys. Lett. 98, 123101 共2011兲
Chuang et al.
FIG. 4. 共Color online兲 共a兲 4 K backscattering -Raman spectrum of a GaAs nanoneedle. An LO line at 293.5 cm−1 with a narrow 2.6 cm−1 linewidth as well as a 2LO line are seen. The PL background from this resonant Raman scattering is subtracted out for clarity. The inset shows the as-measured spectrum for comparison. The pump wavelength is 794.94 nm. 共b兲 LO phonon intensity as a function of incident photon energy. The LO intensity peaks when the incident photon energy coincides with the free-exciton energy.
whereas the ⬃1.505 eV peak begins to saturate at 500 W. The former is attributed to free exciton emission and the latter to impurity-related emission. The linewidth at 1.519 eV is very narrow at 18 meV 关Fig. 3共b兲兴 comparing to that of wurtzite GaAs nanowires grown on a GaAs substrate.9 The 1.519 eV free-exciton energy for wurtzite GaAs nanoneedles is 4 meV larger than that of zinc-blende GaAs. This is close to the 7 meV reported by Martelli et al.10 from a wurtzite GaAs nanowire measurement. RT -PL was also measured 关Fig. 3共d兲兴 and only one single peak could be observed. This is expected since the impurity level mentioned above is too shallow to be seen with RT thermal energy kT ⬃ 25 meV. This single peak does not saturate as the excitation power increases and thus is the bandedge emission. The RT peak energy at 500 W is 1.432 eV, which is 8 meV larger than the bulk 300 K zinc-blende GaAs bandgap of 1.424 eV. The experimental difference between wurtzite and zinc-blende bandgaps is small comparing to a theoretical value of 33 meV.11 Overall, free-exciton recombination dominates 4 K PL with a narrow linewidth. This testifies to the excellent nanoneedle crystal quality. The coherent crystalline structure is further corroborated by 4 K backscattering micro-Raman 共-Raman兲 characterization of a single nanoneedle using a tunable single-mode Ti:sapphire laser.12 The GaAs nanoneedles with 3 h growth time 共⬃9 m long兲 are mechanically transferred to another substrate such that they lie flat to facilitate the measurements, where the polarizations of the pump laser and backscattered light are aligned to the orthogonal and parallel directions to the needle growth direction, respectively. Figure 4共a兲 shows a -Raman spectrum with 794.94 nm pump. A sharp longitudinal optical 共LO兲 phonon line at 293.5 cm−1 and a 2LO line at 587 cm−1 are seen.13 The LO phonon linewidth is only 2.6 cm−1 and is comparable with that of thin film epitaxial GaAs.14 The PL background from this resonant -Raman scattering configuration is already fitted and then subtracted from the figure for clarity. The inset shows the as-measured spectrum for comparison. The previously identified free-exciton and impurity-related PL peaks can also be seen. The intensity of the LO phonon line is measured as a function of the incident photon energy 共pump energy兲 and is shown in Fig. 4共b兲. The LO intensity peaks when the incident photon energy coincides with the freeexciton energy but not the impurity-related photon energy. This further confirms our previous assignments for the PL peaks since the total number of free-exciton states should be more than that of the impurity states hence the Raman scattering resonates more strongly with the free-exciton states.
The nanoneedle nucleation is believed to start via direct vapor-solid cluster formation since no liquid-phase catalyst is seen on the tip at any time during the growth 关Fig. 1共d兲兴. The clusters are thought to be nucleated with a preferred aspect ratio which accounts for the nanoneedle shape due to a large strain from the lattice mismatch.15 By defining the nanoneedle aspect ratio as length over base diameter, the experimental aspect ratio of ⬃5.2– 8.2 agrees the theoretical number of 7.8 for three-dimensional stress-driven GaAs islands nucleated on a sapphire substrate.15 Exact nature of how the GaAs nanoneedle nucleates and maintain the aspect ratio with growth time are still under investigation. In conclusion, we report the catalyst-free, selfassembled, single-crystalline growth of epitaxial wurtzite GaAs nanoneedles on a 46% lattice-mismatched sapphire substrate. Sharp needle shapes form at the very beginning and their dimensions scale with growth time. No critical dimension is observed for this nanoneedle growth mode. Bright PL was seen despite the large lattice mismatch. A sharp 4 K LO phonon line of 2.6 cm−1 linewidth was also observed, testifying excellent crystal quality. Our results illustrate a growth mechanism for heterogeneous integration of highly mismatched materials. The authors acknowledge discussions with V. G. Dubrovskii and assistance by C. Kisielowski; funding by DoD NSSEFF, UCB EECS Fellowship, and NSF-IGERT 共DGE0333455兲. 1
R. M. Lum, J. K. Klingert, R. B. Bylsma, A. M. Glass, A. T. Macrander, T. D. Harris, and M. G. Lamont, J. Appl. Phys. 64, 6727 共1988兲. 2 J. Zou, D. J. H. Cockayne, and B. F. Usher, J. Appl. Phys. 73, 619 共1993兲. 3 F. Glas, Phys. Rev. B 74, 121302 共2006兲. 4 L. C. Chuang, M. Moewe, C. Chase, N. P. Kobayashi, and C. ChangHasnain, Appl. Phys. Lett. 90, 043115 共2007兲. 5 R. Chen, S. Crankshaw, T. Tran, L. C. Chuang, M. Moewe, and C. ChangHasnain, Appl. Phys. Lett. 96, 051110 共2010兲. 6 T. Lei, K. F. Ludwig, and T. D. Moustakas, J. Appl. Phys. 74, 4430 共1993兲. 7 M. Moewe, L. C. Chuang, S. Crankshaw, C. Chase, and C. ChangHasnain, Appl. Phys. Lett. 93, 023116 共2008兲. 8 J. L. Kenty, J. Electron. Mater. 2, 239 共1973兲. 9 T. B. Hoang, A. F. Moses, H. L. Zhou, D. L. Dheeraj, B. O. Fimland, and H. Weman, Appl. Phys. Lett. 94, 133105 共2009兲. 10 F. Martelli, M. Piccin, G. Bais, F. Jabeen, S. Ambrosini, S. Rubini, and A. Franciosi, Nanotechnology 18, 125603 共2007兲. 11 M. Murayama and T. Nakayama, Phys. Rev. B 49, 4710 共1994兲. 12 S. Crankshaw, L. C. Chuang, M. Moewe, and C. Chang-Hasnain, Phys. Rev. B 81, 233303 共2010兲. 13 A. Mooradian and G. B. Wright, Solid State Commun. 4, 431 共1966兲. 14 G. Abstreiter, E. Bauser, A. Fischer, and K. Ploog, Appl. Phys. A 16, 345 共1978兲. 15 V. G. Dubrovskii, N. V. Sibirev, X. Zhang, and R. A. Suris, Cryst. Growth Des. 10, 3949 共2010兲.
GaAs nanoneedles grown on sapphire Linus C. Chuang, Michael Moewe, Kar Wei Ng, Thai-Truong D. Tran, Shanna Crankshaw, Roger Chen, Wai Son Ko, and Connie Chang-Hasnaina兲 Department of Electrical Engineering and Computer Sciences, and Applied Science and Technology, University of California at Berkeley, Berkeley, California 94720, USA
共Received 15 November 2010; accepted 9 January 2011; published online 21 March 2011兲 Heterogeneous integration of dissimilar single crystals is of intense research interests. Lattice mismatch has been the most challenging bottleneck which limits the growth of sufficient active volume for functional devices. Here, we report self-assembled, catalyst-free, single crystalline GaAs nanoneedles grown on sapphire substrates with 46% lattice mismatch. The GaAs nanoneedles have a 2–3 nm tip, single wurtzite phase, excellent optical quality, and dimensions scalable with growth time. The needles have the same sharp, hexagonal pyramid shape from ⬃100 nm 共1.5 min growth兲 to ⬃9 m length 共3 h growth兲. © 2011 American Institute of Physics. 关doi:10.1063/1.3567492兴 Heterogeneous integration of dissimilar single crystals has been of interest because it enables monolithic, largescale integration of functional devices that cannot be achieved with one material system alone. Mismatch in lattice constant and thermal expansion coefficient presents a major constraint that limits the growth of thick, high-quality films on foreign substrate,1 thereby preventing them from attaining the active volumes needed for functional devices. For a given lattice mismatch, there exists a critical thickness.2 When growth exceeds the critical thickness, misfit dislocations form to reduce the strain energy imposed by the mismatch. This drastically degrades the optical and electrical properties of the materials,1 rendering them not suitable for device applications. Recently, researchers have increased the critical thickness by laterally confining the deposition areas in the form of, for example, nanowires with small footprints.3,4 In such case, the nanowire diameter becomes the critical dimension, which decreases with lattice mismatch. For an 11.6% mismatch, the critical diameter is only 26 nm,4 making it challenging for device engineering. Here, we report high-quality, single-phase GaAs structures directly grown on a c-plane sapphire substrate with an extremely large 46% lattice mismatch. The growth assumes a tapered nanoneedle structure with dimensions that scale linearly with growth time over two orders of magnitude without apparent critical dimensions. The GaAs nanoneedle maintains a 2–3 nm tip throughout its growth and is formed spontaneously from the very beginning without catalysts. Its excellent optical quality with bright room-temperature 共RT兲 luminescence and narrow Raman linewidths attest to its coherent crystalline phase. The wurtzite nanoneedles are hexagonal pyramids with 7°–11° sidewall-to-sidewall angles that taper to an ultrasharp tip. The combination of an exceedingly sharp tip and wide, strong base makes them excellent candidates for applications use in field emitter, microscopic, and nonlinear optical signal generation.5 Figure 1共a兲 shows the side view of GaAs nanoneedles on sapphire, on which most GaAs nanoneedles are aligned vertically to the 关0001兴 substrate orientation. Slanted nanon¯ 02典 directions. eedles are identified along one of the three 具11 The nanoneedles are grown by low temperature 共400 ° C兲 a兲
Electronic mail: [email protected].
0003-6951/2011/98共12兲/123101/3/$30.00
metal-organic chemical vapor deposition without substrate surface treatment. The substrate goes through a 3 min pregrowth annealing at 600 ° C with tertiarybutylarsine 共TBA兲, followed by growth at 400 ° C. The sources are triethylgallium and TBA and the mole fractions are set as 1.12⫻ 10−5 and 5.42⫻ 10−4, respectively, in a 12 l/min hydrogen carrier gas flow. Sharp GaAs nanoneedles with 9° tip and smooth sidewall facets are observed for 82 min growth time 关Fig. 1共b兲兴. A hexagonal cross section is seen with the top-down scanning electron microscope 共SEM兲 image in Fig. 1共c兲. The ¯ 00兴 direction is rotated by 30° with respect to nanoneedle 关11 ¯ the 关1100兴 direction of the sapphire substrate 关Fig. 1共c兲兴, determined by transmission electron microscopy 共TEM兲. This 30° rotation is similar to the report of thin-film GaN on sapphire due to the alignment of Ga with Al.6 The lattice mismatch is, hence, calculated to be 46% with compressive
FIG. 1. 共Color online兲 共a兲 Side-view SEM image of as-grown GaAs nanoneedles. 共b兲 30°-tilt SEM image of a nanoneedle with 82 min growth time. ¯ 00其 and 共0001兲 terraces. 共c兲 Top-down view of The sidewalls are made of 兵11 a nanoneedle showing a hexagonal cross section. The in-plane orientation of the nanoneedle shows a 30° rotation with respect to the sapphire substrate. 共d兲 30°-tilt SEM images of nanoneedles with different growth times. The image of the nanoneedle with 1.5 min growth time is enlarged by 20⫻. 共e兲 Schematic showing the growth of single-crystalline nanoneedle body and the polycrystalline connecting layer. Inset is a cross-section SEM image for 60 min growth. A thin, 200 nm connecting layer is seen.
98, 123101-1
© 2011 American Institute of Physics
123101-2
Chuang et al.
FIG. 2. 共Color online兲 共a兲 TEM image showing a nanoneedle tip with less than 3 nm in diameter. Inset is the FFT image showing the wurtzite crystal¯ 00兴 zone axis. The nanoneedle orientation during the line structure with 关11 inspection is schematically shown. 共b兲 Selective-area diffraction pattern for ¯ 10兴 zone axis. 共c兲 An HRTEM image, with 关11 ¯ 00兴 a nanoneedle with 关12 zone axis, showing that the nanoneedle sidewall ridge consists of atomic steps which result in the taper of a nanoneedle. 共d兲 HRTEM of a GaAs nanoneedle grown on GaAs with a sharp tip of only three lattice spacings.
strain. This fixed in-plane orientation relationship suggests the nanoneedle growth being epitaxial. The needles are sharp from the beginning and remain the same geometry for long growth durations. Figure 1共d兲 shows nanoneedles with growth time ranging from 1.5 to 180 min. The distinctive nanoneedle shape is already observed after 1.5 min of growth. The base diameter and height increase linearly with growth time and maintain a constant taper angle. We estimate the growth rates to be 0.3 m / h and ¯ 00典 and 关0001兴 crystal orientations, 3.6 m / h for the 具11 respectively. The nanoneedle growth starts with seed nucleation, followed by two-dimensional thin-film deposition on the preferred six sidewall facets in a core-shell mode. This growth mode is the same as our GaAs nanoneedles on roughened Si substrates.7 Nearly the same growth is observed with a substrate having ten times larger mismatch. In addition to the prominent nanoneedle growth, a thin connecting layer between individual nanoneedles is observed 关Figs. 1共a兲–1共d兲兴. The cross-section SEM image of a 60 min GaAs nanoneedle growth run 关Fig. 1共e兲, inset兴 shows a 200 nm thick connecting layer covering the entire sapphire substrate surface except where nanoneedles grow. The same connecting layer is seen for GaAs nanoneedles grown on Si7 and identified as polycrystalline GaAs by TEM, similar to low-temp growth of GaAs on sapphire.8 The singlecrystalline GaAs nanoneedle and the polycrystalline connecting layer are thought to coalesce in the nanoneedle root region as schematically shown in Fig. 1共e兲. At the beginning of the growth, a nanoneedle seed forms a small singlecrystalline connection to the sapphire substrate with elsewhere covered by a thin polycrystalline film 共both shown in red兲. Each incremental growth 共shown by different colors兲 continues to deposit an additional single-crystalline film 共at ⬃0.3 m / h兲 onto the nanoneedle sidewall and a polycrystalline film 共at ⬃0.2 m / h兲 onto the connecting layer,
Appl. Phys. Lett. 98, 123101 共2011兲
FIG. 3. 共Color online兲 共a兲 4 K -PL spectra for a nanoneedle with various excitation powers. Two peaks were seen. 共b兲 Multi-Lorentzian fit for the 4 K spectrum with 100 W excitation power with peaks at 1.505 and 1.519 eV having respective linewidths of 20 and 18 meV. 共c兲 Peak intensity vs excitation power for the two peaks. The ⬃1.519 eV peak 共free exciton兲 intensity increases with the excitation power while the ⬃1.505 eV peak 共impurity related兲 starts to saturate at 500 W. 共d兲 300 K -PL spectra for a nanoneedle with various excitations.
where as the ratio of the two growth rates 共1.5⫻兲 determines the single-crystalline and polycrystalline boundary. As a result, the single-crystalline part of an as-grown nanoneedle should have a reversely tapered root. This is verified for GaAs nanoneedles grown on Si. For GaAs nanoneedles on sapphire, the interface is still under investigation. Figure 2共a兲 shows a high-resolution TEM 共HRTEM兲 image of a nanoneedle tip 2–3 nm in diameter. The zone axis is ¯ 00兴, with a nanoneedle sidewall facet approximately per关11 pendicular to it, due to the small but finite nanoneedle taper angle. The fast Fourier transform 共FFT兲 of the TEM image is shown as the inset. Since this FFT pattern can be with either ¯ 00兴 or zinc-blende 关21 ¯¯1兴 zone axis, the same wurtzite 关11 nanoneedle was rotated by 30° about its growth axis for a second diffraction pattern. In this second case, the resulting selective-area diffraction pattern 关Fig. 2共b兲兴 unambiguously identifies the wurtzite structure, which is coherent throughout the nanoneedle. The nanoneedle axes shown in Figs. 2共a兲 and 2共b兲, are along the 关0001兴 direction and hence it suggests the nanoneedle growth on c-sapphire being epitaxial. Close inspection of the tapered nanoneedle sidewall ridge reveals single-atomic steps 关Fig. 2共c兲兴, which result in the taper. The ¯ 00其 and tapered sidewall facets therefore consist of 兵11 共0001兲 terraces. It is interesting to note that the sharp nanoneedle tip, the geometry and crystalline phase are identical for needles grown on the 46% mismatched system and those on GaAs,7 for which we show the sharpest single crystal ever reported with only three lattice spacings at the tip 关Fig. 2共d兲兴. Single-needle microphotoluminescence 共-PL兲 is investigated using a 514 nm argon laser focused to ⬃1.5 m in diameter. Figure 3共a兲 shows typical 4 K -PL spectra for a range of excitation powers. A fit of the PL spectrum at 100 W excitation power by multiple Lorentzian line shapes 关Fig. 3共b兲兴 shows two peaks at 1.519 and 1.505 eV. The intensities of the two peaks as a function of the excitation power are plotted in Fig. 3共c兲. The peak intensity at ⬃1.519 eV has a superlinear dependence on excitation,
123101-3
Appl. Phys. Lett. 98, 123101 共2011兲
Chuang et al.
FIG. 4. 共Color online兲 共a兲 4 K backscattering -Raman spectrum of a GaAs nanoneedle. An LO line at 293.5 cm−1 with a narrow 2.6 cm−1 linewidth as well as a 2LO line are seen. The PL background from this resonant Raman scattering is subtracted out for clarity. The inset shows the as-measured spectrum for comparison. The pump wavelength is 794.94 nm. 共b兲 LO phonon intensity as a function of incident photon energy. The LO intensity peaks when the incident photon energy coincides with the free-exciton energy.
whereas the ⬃1.505 eV peak begins to saturate at 500 W. The former is attributed to free exciton emission and the latter to impurity-related emission. The linewidth at 1.519 eV is very narrow at 18 meV 关Fig. 3共b兲兴 comparing to that of wurtzite GaAs nanowires grown on a GaAs substrate.9 The 1.519 eV free-exciton energy for wurtzite GaAs nanoneedles is 4 meV larger than that of zinc-blende GaAs. This is close to the 7 meV reported by Martelli et al.10 from a wurtzite GaAs nanowire measurement. RT -PL was also measured 关Fig. 3共d兲兴 and only one single peak could be observed. This is expected since the impurity level mentioned above is too shallow to be seen with RT thermal energy kT ⬃ 25 meV. This single peak does not saturate as the excitation power increases and thus is the bandedge emission. The RT peak energy at 500 W is 1.432 eV, which is 8 meV larger than the bulk 300 K zinc-blende GaAs bandgap of 1.424 eV. The experimental difference between wurtzite and zinc-blende bandgaps is small comparing to a theoretical value of 33 meV.11 Overall, free-exciton recombination dominates 4 K PL with a narrow linewidth. This testifies to the excellent nanoneedle crystal quality. The coherent crystalline structure is further corroborated by 4 K backscattering micro-Raman 共-Raman兲 characterization of a single nanoneedle using a tunable single-mode Ti:sapphire laser.12 The GaAs nanoneedles with 3 h growth time 共⬃9 m long兲 are mechanically transferred to another substrate such that they lie flat to facilitate the measurements, where the polarizations of the pump laser and backscattered light are aligned to the orthogonal and parallel directions to the needle growth direction, respectively. Figure 4共a兲 shows a -Raman spectrum with 794.94 nm pump. A sharp longitudinal optical 共LO兲 phonon line at 293.5 cm−1 and a 2LO line at 587 cm−1 are seen.13 The LO phonon linewidth is only 2.6 cm−1 and is comparable with that of thin film epitaxial GaAs.14 The PL background from this resonant -Raman scattering configuration is already fitted and then subtracted from the figure for clarity. The inset shows the as-measured spectrum for comparison. The previously identified free-exciton and impurity-related PL peaks can also be seen. The intensity of the LO phonon line is measured as a function of the incident photon energy 共pump energy兲 and is shown in Fig. 4共b兲. The LO intensity peaks when the incident photon energy coincides with the freeexciton energy but not the impurity-related photon energy. This further confirms our previous assignments for the PL peaks since the total number of free-exciton states should be more than that of the impurity states hence the Raman scattering resonates more strongly with the free-exciton states.
The nanoneedle nucleation is believed to start via direct vapor-solid cluster formation since no liquid-phase catalyst is seen on the tip at any time during the growth 关Fig. 1共d兲兴. The clusters are thought to be nucleated with a preferred aspect ratio which accounts for the nanoneedle shape due to a large strain from the lattice mismatch.15 By defining the nanoneedle aspect ratio as length over base diameter, the experimental aspect ratio of ⬃5.2– 8.2 agrees the theoretical number of 7.8 for three-dimensional stress-driven GaAs islands nucleated on a sapphire substrate.15 Exact nature of how the GaAs nanoneedle nucleates and maintain the aspect ratio with growth time are still under investigation. In conclusion, we report the catalyst-free, selfassembled, single-crystalline growth of epitaxial wurtzite GaAs nanoneedles on a 46% lattice-mismatched sapphire substrate. Sharp needle shapes form at the very beginning and their dimensions scale with growth time. No critical dimension is observed for this nanoneedle growth mode. Bright PL was seen despite the large lattice mismatch. A sharp 4 K LO phonon line of 2.6 cm−1 linewidth was also observed, testifying excellent crystal quality. Our results illustrate a growth mechanism for heterogeneous integration of highly mismatched materials. The authors acknowledge discussions with V. G. Dubrovskii and assistance by C. Kisielowski; funding by DoD NSSEFF, UCB EECS Fellowship, and NSF-IGERT 共DGE0333455兲. 1
R. M. Lum, J. K. Klingert, R. B. Bylsma, A. M. Glass, A. T. Macrander, T. D. Harris, and M. G. Lamont, J. Appl. Phys. 64, 6727 共1988兲. 2 J. Zou, D. J. H. Cockayne, and B. F. Usher, J. Appl. Phys. 73, 619 共1993兲. 3 F. Glas, Phys. Rev. B 74, 121302 共2006兲. 4 L. C. Chuang, M. Moewe, C. Chase, N. P. Kobayashi, and C. ChangHasnain, Appl. Phys. Lett. 90, 043115 共2007兲. 5 R. Chen, S. Crankshaw, T. Tran, L. C. Chuang, M. Moewe, and C. ChangHasnain, Appl. Phys. Lett. 96, 051110 共2010兲. 6 T. Lei, K. F. Ludwig, and T. D. Moustakas, J. Appl. Phys. 74, 4430 共1993兲. 7 M. Moewe, L. C. Chuang, S. Crankshaw, C. Chase, and C. ChangHasnain, Appl. Phys. Lett. 93, 023116 共2008兲. 8 J. L. Kenty, J. Electron. Mater. 2, 239 共1973兲. 9 T. B. Hoang, A. F. Moses, H. L. Zhou, D. L. Dheeraj, B. O. Fimland, and H. Weman, Appl. Phys. Lett. 94, 133105 共2009兲. 10 F. Martelli, M. Piccin, G. Bais, F. Jabeen, S. Ambrosini, S. Rubini, and A. Franciosi, Nanotechnology 18, 125603 共2007兲. 11 M. Murayama and T. Nakayama, Phys. Rev. B 49, 4710 共1994兲. 12 S. Crankshaw, L. C. Chuang, M. Moewe, and C. Chang-Hasnain, Phys. Rev. B 81, 233303 共2010兲. 13 A. Mooradian and G. B. Wright, Solid State Commun. 4, 431 共1966兲. 14 G. Abstreiter, E. Bauser, A. Fischer, and K. Ploog, Appl. Phys. A 16, 345 共1978兲. 15 V. G. Dubrovskii, N. V. Sibirev, X. Zhang, and R. A. Suris, Cryst. Growth Des. 10, 3949 共2010兲.
