NANO LETTERS
Semiconductor Nanocrystal Quantum Dots on Single Crystal Semiconductor Substrates: High Resolution Transmission Electron Microscopy
2005 Vol. 5, No. 5 969-973
Atul Konkar,* Siyuan Lu, and Anupam Madhukar Nanostructure Materials and DeVices Laboratory, Departments of Materials Science and Physics, UniVersity of Southern California, Los Angeles, California 90089-0241
Steven M. Hughes and A. Paul Alivisatos Department of Chemistry, UniVersity of California, Berkeley, California 94720 Received February 9, 2005; Revised Manuscript Received March 19, 2005
ABSTRACT We report on high-resolution transmission electron microscope structural studies of InAs colloidal semiconductor nanocrystal quantum dots (NCQDs) on ultrathin GaAs (001) semiconductor single-crystal substrates. We employ a benign method for preparing electron transparent specimens that is suitable for the study of such fragile samples. The image contrast comprises contributions from electron scattering from both the NCs and the GaAs substrate. Long-term electron exposure studies reveal different damage mechanisms operative in the nanocrystals and the substrate.
Three-dimensionally confined nanoscale semiconductor volumes, called quantum dots, exhibit unique properties that have been exploited in the fabrication of novel semiconductor devices.1 The two dominant types of semiconductor quantum dots synthesized via purely growth control are (a) heteroepitaxically grown self-assembled island quantum dots (SAQDs)1-3 and (b) colloidal nanocrystal quantum dots (NCQDs) in solution.4 In the former approach, the SAQDs are a part of the substrate crystal. The vast infrastructure and techniques used in wafer processing can be directly applied to SAQD containing substrates, as can also concepts of band-gap and stress engineering.1 Optical or charge carrier based communication with SAQDs is straightforward.1 However, as the SAQD formation is driven by latticemismatch induced strain energy,2 the ability to tailor their size and spatial distribution is limited. Some success in manipulating SAQD size and shape has been achieved by controlling growth5,6 and the nature of the substrate, e.g., via patterning,7 lattice mismatch,8 substrate crystallographic orientation.9 The NCQDs, by contrast, are synthesized typically in organic solvent using finely tuned precursor reaction kinetics, so-called arrested precipitation. In principle, the NCQD size can thus be varied freely to tune their electronic energy levels over a large range. When placing the NCQDs in a solid environment, their spatial arrangement * Corresponding author. E-mail:
[email protected]. 10.1021/nl0502625 CCC: $30.25 Published on Web 04/12/2005
© 2005 American Chemical Society
can be independently controlled even though no epitaxial relationship to the solid environment (i.e., the solid matrix) is automatically realized. Research efforts on a variety of approaches are underway.10 The advantages of each of these types of quantum dots would be significantly enhanced in terms of their applications potential if the NCQDs could be integrated into single-crystal semiconductor substrates and/or matrices.11 Such integration poses challenges relating to the fundamental conceptual understanding of the traditional notions of epitaxy as applied/ extended to curved surfaces/interfaces, as well as relating to the implementation strategies, given the inherently different synthesis environments.11 Here we report on the first HRTEM study of the structural nature of a semiconductor nanocrystal adsorbed onto a crystalline semiconductor substrate. We note that a lowered thermodynamic stability of nanocrystals due to their high surface curvature associated energy has been known for some time as the “Kelvin effect”.12 Enhanced surface energy also gives rise to higher reactivity. Recent studies of attachment between nanocrystals in their solution environment show that, when oriented appropriately, a low-temperature pathway for covalent bonding results due, as expected, to the maximizing of at least the first nearest neighbor interactions at the bonding interface.13 The individual nanocrystal properties are intimately linked to their structure, including size, shape, bulk (interior), and surface bonding. Their large surface-to-volume ratio
makes them sensitive to environmental conditions. Indeed recent experiments have reported changes in the structure of NCQDs resulting from changes in solvent14 and in packing density.15 In this paper we report on high-resolution transmission electron microscope (HRTEM) based structural studies of the behavior of NCQDs when placed on single-crystal semiconductor substrates. This constitutes a first step toward the eventual goal of integration. Previous structural studies of NCQDs on solid substrates have predominantly used X-ray diffraction or TEM of NCQDs dispersed on amorphous C film. Some TEM studies of NCQDs deposited on polycrystalline metallic materials are extant.16 In this work we have chosen to examine the structural behavior of InAs NCQDs on single-crystal GaAs(001) substrate as a canonical NCQD/ crystalline substrate system. This material combination provides compatibility in terms of bonding (both materials are in the III-V class of compound semiconductors) and simultaneously allows exploration of effects of the 7% lattice mismatch strain between the respective bulk lattices. It also is compatible with the InAs/GaAs SAQDs that have been shown to offer quantum dot devices such as lasers and detectors.1,2 Electron transparent GaAs (001) membrane substrates were fabricated via a three step procedure. (1) MBE growth of a lattice-matched heterostructure on a 500 µm GaAs (001) substrate consisting of 0.5µm GaAs buffer layer, 50 nm of AlGaAs (Al mole fraction 0.6), and a 15 nm or 20 nm GaAs layer was accomplished. (2) The samples were then cleaved into ∼ 2 mm square pieces and bonded face down to standard Cu grids using a resin. The pieces were mechanically polished from the back side down to ∼100 µm and subsequently dimpled to a thickness of ∼40 µm. (3) The dimpled specimens were serially etched from the backside in two selective etchants, the first to selectively remove GaAs and the second to selectively remove the AlGaAs layer. Citric acid (CA) solution was prepared by dissolving anhydrous citric acid in DI water in a 1:1 proportion by weight. The resulting CA solution was mixed with hydrogen peroxide in 5:1 proportion by volume to yield the first etchant. For selective AlGaAs etching, 30% HF in water was used. The area of the resulting GaAs membrane can be controlled by adjusting the etching time in the first solution. Typically, circular regions with diameter on the order of ∼1 mm were prepared. Such large regions are especially important in the study of nanostructures since they allow statistically significant sampling given the extreme sensitivity of NCQDs to electron beam damage, a point elaborated further in the following. With this method of preparation we can control the local and global thickness of the membrane down to 1 ML precision. And the exclusive use of chemical etching to prepare the final membrane avoids artifacts that might be caused by mechanical thinning or ion-beam based methods. This issue is especially critical in the study of such fragile samples. InAs colloidal NCQDs with ∼8 nm diameter were prepared in toluene solution via a new approach that we call punctuated growth17 which allows greater control on size 970
Figure 1. (a) Plan-view HRTEM image of GaAs (001) membrane. The lattice fringes correspond to the orthogonal {220} planes. (b) In-situ AFM image of a typical MBE grown sample surface. The averaged height profile across a monolayer step from the boxed region is shown below the image.
distribution compared to the established procedure for InAs NCQD synthesis.18 The as-synthesized NCQD surface is covered with tri-octyl phosphine (TOP) molecules that act as surfactant and prevent coagulation. They also passivate the surface, as manifest in their luminescence spectra in solution. The NCQDs were deposited on the GaAs substrate by dip coating prior to step (2) noted above. During the TEM specimen preparation, the resin curing temperature was kept low (∼150 °C), much below the NCQD synthesis temperature (∼270 °C) to minimize any thermally driven changes. The TEM studies were done on an Akashi EM-002B microscope operating at 200 kV with high-resolution objective pole piece. The point-to-point resolution of this objective pole piece is 0.184 nm. All imaging reported here was done using an objective aperture that only allows spatial frequencies up to 5.7 nm-1. The images were captured on film and subsequently digitized. Only the brightness and contrast has been optimized for these images. No other processing, Fourier or other filtering, has been done. As a reference, Figure 1, panel (a) shows a plan view HRTEM image of the GaAs membrane without the nanocrystal quantum dots, taken along the [001] zone axis. The lattice fringes in the image correspond to the orthogonal 〈220〉 GaAs planes. The spacing between the planes is 0.2 nm. Insitu atomic force microscope (AFM) imaging (see Figure 1b) of as grown samples show that the typical surface consists of ∼200 nm wide (001) terraces separated by monolayer high steps. Figure 2a shows a low magnification image of the InAs NCQDs on an ultrathin GaAs membrane. Due to the small dispersion in their sizes the NCs are seen to form large patches of close packed monolayer structures. In addition, small patches composed of few to tens of NCQDs and isolated NCQDs are also observed. The NCQDs have a nearly spherical shape and an average diameter of ∼8 nm. A selected area diffraction pattern is shown in Figure 2b. The bright spots originate from electron diffraction from the GaAs substrate and the superimposed rings from the InAs NCQDs. The intensity of the rings is uniform along their circumference indicating that the NCQDs do not have a preferred orientation on such atomically flat substrates. We speculate that this lack of preferred orientation might be due to the TOP molecules covering the NC surface. Another Nano Lett., Vol. 5, No. 5, 2005
Figure 2. (a) Low magnification image of InAs NCs on GaAs (001); (b) corresponding selected area electron diffraction pattern.
Figure 3. HRTEM image showing lattice fringes in the NCQD and the GaAs corresponding to the {111} InAs and {220} GaAs planes, respectively.
possibility is that the NCQDs do not have sizable welldefined facets. Figure 3 shows HRTEM image of a single isolated InAs NCQD. Well defined lattice fringes from both the NCQD (spacing 0.35 nm) and GaAs substrate (spacing 0.2 nm) are seen in the image. The contrast in the image can be qualitatively understood as resulting from two independent interference effects: the forward scattered beam and (a) the {111} InAs diffracted beam from the NCQD or (b) the {220} GaAs diffracted beam from the GaAs substrate. A dark halo surrounding the NCQD is also observed. The width of the halo is ∼1 to 1.5 nm. The origin of this contrast is not clear at this time. However, as the length of TOP from the P atom to the 3 terminal C atoms is ∼1.3 nm, we speculate that the layer of TOP molecules surrounding the NCQD is responsible. If true, this will be consistent with the observed lack of preferred NCQD orientation, as noted above. The TOP layer may be able to effectively screen the otherwise expected covalent interaction between the NCQD atoms and the GaAs substrate. We are currently studying approaches to remove the surfactant layer and examine the resulting NCQD/ substrate structures in this material system. One may qualitatively understand the contrast seen in Figure 3 based on the interference effects in InAs NCQD and GaAs substrate occurring independently. However, in the general the image contrast is due to interference between Nano Lett., Vol. 5, No. 5, 2005
Figure 4. HRTEM image of NCQDs showing multiple beam interference effects.
multiple beams accepted by the objective aperture. An example of such multiple beam interference is shown in Figure 4. The image is from a region containing a bunch of NCQDs. Faint, closely spaced, fringes from {220} GaAs planes are visible in the background of the entire image. Fringes are visible within a few NCQDs as well. Here we focus on understanding the fringe contrast within the NC marked by an arrow. Two sets of fringes are visible. The fringes corresponding to the smaller spacing are from the 971
Figure 5. HRTEM image of In nanocrystals that result from extended electron beam exposure. Inset shows higher magnification image of one such nanocrystal.
{111} InAs planes. The other set of fringes, running almost orthogonal to the {111} fringes, have a spacing of 1.4 nm. This set of fringes is an example of interference between beams diffracted by InAs NCQD and GaAs substrate, resulting in the so-called Moire´ fringes. Given the relative orientation of the NCQD with respect to the substrate, the observed spacing and orientation of the Moire´ fringes is consistent with that calculated due to interference between the {111} diffracted beam from InAs NCQD and {200} diffracted beam from the GaAs membrane. We note that a simple kinematical calculation would predict Moire´ fringes with a spacing of 0.9 nm within the NCQD shown in Figure 3. However, no such fringes are apparent. A full multibeam image simulation that accounts for the diffraction condition of both the NCQD and the substrate, substrate thickness, and the three-dimensional NCQD shape is thus needed to understand the detailed image contrast in such structures. A major concern in electron beam imaging of nanostructures is the damage resulting from the imaging process. New damage mechanisms can become operative in such structures due to bottlenecks in the channels for energy dissipation and also because of the large surface-to-volume ratio. In structures under investigation here, some electron beam induced electronic excitation can be trapped within the NCQDs since the TOP layer offers a heterojunction barrier to the transfer of carriers to the GaAs substrate. These long-lived excitations can effectively reduce activation barriers to a variety of chemical processes, thus resulting in chemical, and attendant structural, changes in the NCQDs. A striking example of such a phenomenon is manifest in Figure 5, which shows a HRTEM image from a region after prolonged exposure to the electron beam. The corresponding selected area diffraction pattern is shown in Figure 6. The electron dose is more than an order of magnitude higher compared to that used for obtaining the images shown in figures 2, 3, and 4. Overall the diffraction pattern shows the bright spots from the GaAs (001) substrate and a superimposed ring pattern. However, the ring pattern spacing has changed compared to that seen 972
Figure 6. Electron diffraction pattern from region after prolonged electron beam exposure.
in Figure 2b and the intensity along the periphery is not uniform. The observed symmetry and spacing is consistent with that of face centered cubic (fcc) indium. Previous studies have shown that transformation from the face centered tetragonal (fct) structure of bulk indium to a fcc structure occurs in spherical indium nanocrystals below a diameter ∼5 nm.19 The continuous set of Moire´ fringes observed in the Figure 5 are thus from individual In nanocrystals. Some of these In nanocrystals are seen to have sizes up to 15 nm, potentially indicating coalescence and growth of In originating from different InAs NCQDs. The modulation of intensity around the ring periphery is a result of averaging over smaller number of In nanocrystals given their larger size and perhaps some contribution of local environment dependent kinetics of growth and coalescence leading to preferred orientation. Nano Lett., Vol. 5, No. 5, 2005
These observations clearly evidence preferential loss of arsenic from the InAs NCQDs. The observation of crystalline In nanoparticles indicates that the local temperature has not exceeded 156 °C, the In bulk melting temperature. At these lower temperatures the InAs nanocrystals are expected to be quite stable. Thus the arsenic loss is not predominantly caused by electron beam induced local heating. Also, no such arsenic loss is apparent in the GaAs substrate which has a bond enthalpy similar to that of InAs. We suggest that the electron beam induced excitation trapped within the NCQDs substantially lowers the energy barrier to arsenic desorption, the electron stimulation facilitating the desorption process at lower temperatures. A similar proposal has been suggested in the case of electron beam induced changes in GaSb nanocrystals.20 In summary, we have demonstrated lattice imaging of colloidal semiconductor NCQDs on single-crystal semiconducting substrates. A rich variety in the image contrast, caused by multiple electron beam interference effects in these structures, is observed. The surfactant molecules covering the NCQD surface perhaps play a central role in screening interactions between the NCQD and the substrate. The novel TEM specimen preparation approach employed here will allow further nanoscale studies of such advanced structures while minimizing artifacts inherent in other conventional methods. Acknowledgment. We thank Prof. U. Banin (The Hebrew University of Jerusalem, Israel) for helpful discussions regarding InAs NCQD synthesis and Ms. Yi Zhang for the AFM studies. This work was supported by the DoD, under the DURINT (Defense University Research Initiative on Nanotechnology) program, through DARPA/AFOSR Grant No. F49620-01-1-0474.
Nano Lett., Vol. 5, No. 5, 2005
References (1) See, e.g., Nano-Optoelectronics, Grundman, M., Ed.; SpringerVerlag: Berlin 2002. (2) Guha, S.; Madhukar, A.; Rajkumar, K. C. Appl. Phys. Lett. 1990, 57, 2110. Leonard, D.; Krishnamurthy, M.; Reaves, C. M.; Denbaars, S. P.; Petroff, P. M. Appl. Phys. Lett. 1993, 63, 3203. Moison, J. M.; Houzay, F.; Barthe, F.; Leprince, L.; Andre´, E.; Vatel, O. Appl. Phys. Lett. 1994, 64, 196. (3) Xie, Q.; Madhukar, A.; Chen, P.; Kobayashi, N. P. Phys. ReV. Lett. 1995, 75, 2541. (4) Alivisatos, A. P. Science 1996, 271, 933. Coe, S.; Woo, W-K.; Bawendi, M.; Bulovi, V. Nature 2002, 420, 800. (5) Mukhametzhanov, I.; Heitz, R.; Zheng, J.; Chen, P.; Madhukar, A. Appl. Phys. Lett. 1998, 73, 1841. (6) Mukhametzhanov, I.; Wei, Z.; Heitz, R.; Madhukar, A. Appl. Phys. Lett. 1999, 75, 85. (7) Konkar, A.; Madhukar, A.; Chen, P. Appl. Phys. Lett. 1998, 72, 220. (8) Yeoh, T. S.; Swint, R. B.; Elarde, V. C.; Coleman, J. J. Appl. Phys. Lett. 2004, 84, 3031. (9) Gong, Q.; No¨tzel, R.; Hamhuis, G. J.; Eijkemans, T. J.; Wolter, J. H. Appl. Phys. Lett. 2002, 81, 3254. (10) Rogach, A. L.; Talapin, D. V.; Shevchenko, E. V.; Kornowski, A.; Haase, M.; Weller, H. AdV. Funct. Mater. 2002, 12, 653. (11) Madhukar, A.; Lu, S.; Konkar, A.; Zhang, Y.; Ho, M.; Hughes, S. M.; Alivisatos, A. P. Nano Lett. 2005, 5, 479. (12) Thompson, W. (Kelvin) Philos. Mag. 1871, 42, 449. (13) Huang, F.; Zhang, H.; Banfield, J. F. J. Phys. Chem. B 2003, 107, 10470. (14) Zhang, H.; Gilbert, B.; Huang, F.; Banfield, J. F. Nature 2003, 424, 1025. (15) Huang, F.; Gilbert, B.; Zhang, H.; Banfield, J. F. Phys. ReV. Lett. 2004, 92, 155501. (16) Colvin, V. L.; Goldstein, A. N.; Alivisatos, A. P. J. Am. Chem. Soc. 1992, 114, 5221. (17) Lu, S.; Madhukar, A.; Hughes, S. M.; Alivisatos, A. P., to be published. (18) Guzelian, A. A.; Banin, U.; Kadavanich, A. V.; Peng, X.; Alivisatos, A. P. Appl. Phys. Lett. 1996, 69, 1432. (19) Yokozeki, A.; Stein, G. D. J. Appl. Phys. 1978, 49, 2224. (20) Yasuda, H.; Mori, H.; Lee, J. G. Phys. ReV. Lett. 2004, 92, 135501.
NL0502625
973
