Letter pubs.acs.org/NanoLett
Direct Imaging of Single Au Atoms Within GaAs Nanowires Maya Bar-Sadan,†,‡,* Juri Barthel,† Hadas Shtrikman,§ and Lothar Houben† †
Ernst Ruska-Centre for Microscopy and Spectroscopy with Electrons, Forschungszentrum Jülich GmbH, 52425 Jülich, Germany Chemistry Department, Ben Gurion University of the Negev, Be’er Sheba, Israel § Braun Center for Submicrometer Research, Weizmann Institute of Science, Rehovot 76100, Israel ‡
S Supporting Information *
ABSTRACT: Incorporation of catalyst atoms during the growth process of semiconductor nanowires reduces the electron mean free path and degrades their electronic properties. Aberration-corrected scanning transmission electron microscopy (STEM) is now capable of directly imaging single Au atoms within the dense matrix of a GaAs crystal, by slightly tilting the GaAs lattice planes with respect to the incident electron beam. Au doping values in the order of 1017−18 cm3 were measured, making ballistic transport through the nanowires practically inaccessible. KEYWORDS: Single atom detection, scanning transmission electron microscopy, dopants, nanowires, transport phenomena
S
observed that dopant incorporation can trigger a change in the overall structure of a nanowire.13 For these reasons, imaging of buried single dopant atoms is a long-sought goal. The location of dopant atoms as well as unintentional impurities is extremely important for the understanding of their interaction with the host lattice.14,15 Atom probe tomography was applied lately to detect minute quantities of 100 ppm of Au dopant atoms in InAs,8 but in other cases, the gold concentration could not be determined by this technique.9 Another technique which has been used to probe single dopant atoms within a matrix is high-angle annular dark-field (HAADF) scanning transmission electron microscopy (STEM).10,16,17 For atomic resolution STEM, the electron beam is focused to a fine probe with diameters of less than 1 Å and scanned over the sample in steps of a few picometers. Inside the sample the electron probe is scattered to high angles mainly by the electrostatic potentials of the atomic nuclei. For each scan point, the high-angle scattering signal is integrated by an annular detector below the object and registered as the intensity of one image pixel. In general the HAADF signal intensity is enhanced with increasing atomic number producing chemically sensitive “Zcontrast”. In a simplified but frequently applied approach,18 the intensity of the HAADF-STEM signal is estimated to be proportional to Z1.6−1.8. The difference between the atomic numbers Zhost of the host lattice and Zdopant of the foreign atom enables the identification of dopant atoms within the matrix. Compared to previous works on systems with much larger
emiconductor nanowires, produced by the vapor−liquid− solid (VLS) method, are considered for applications with sensors, field effect transistors, and energy harvesting devices.1−6 Nanowires of III−V semiconductors are produced in specific compositions which enable the tuning of the nanowire’s bandgap. In addition, these nanomaterials provide high electron mobility, hence making them exceptionally interesting for photovoltaic devices as well as for other nanoelectronic and nanophotonic devices. The nanowires realize one-dimensional quantum structures where an electrical current can flow along one direction, the growth direction, extending over a few micrometers. At the same time the electronic states are confined in the orthogonal dimensions within tens of nanometers. However, many fundamental growth-related aspects of nanowires are still not well understood.7 The control over unintentional incorporation of catalyst atoms during the growth is an important issue when producing high-quality wires for transport experiments. In particular, incorporation of atoms from the Au catalyst droplet is known to be a major problem in Si nanowires growth and is a source for debates concerning InAs and GaAs as well.8−10 In particular, it is an obstacle for transport experiments since it makes the ballistic transport practically inaccessible.11 This problem is tackled by the development of self-catalyzed synthetic modes, which do not as yet provide a full alternative to the gold assisted growth of III−V nanowires.12 Incorporation of single dopant atoms is sometimes beneficial for the use in devices and applications. Dopants may act as donors or acceptors, introducing shallow states just below the conduction edge or just above the valence edge, respectively. However, surface and ridge structural morphologies could affect dopant incorporation during growth. Moreover, it was also © 2012 American Chemical Society
Received: January 25, 2012 Revised: April 12, 2012 Published: April 12, 2012 2352
dx.doi.org/10.1021/nl300314k | Nano Lett. 2012, 12, 2352−2356
Nano Letters
Letter
Zdopant/Zhost ratios,10,17,19,20 e.g., ZAu/ZSi = 5.6, the detection of dopant atoms in the present case where ZAu/ZGa = 2.5 can be considered as a very challenging task. A Z ratio of 2.5 is mentioned to be insufficient for the imaging of single dopants within a matrix in zone-axis configurations,16 where the crystalline lattice is oriented with its atom columns parallel to the optic axis. However, estimating dopant detectability from the pure Zdopant/Zhost ratio neglects dynamical diffraction. The aim to resolve and possibly identify dopant atoms in the Au-GaAs system by HAADF-STEM imaging requires deep understanding of the interaction between the sample and the electron wave, as already pointed out by Voyles et al. in their work on Sb-doped Si.16 The usually applied zone-axis configuration, where HAADF-STEM is able to resolve individual atomic columns, is drawn schematically in Figure 1a. In this situation multiple elastic scattering events repeatedly
In this configuration, the excess signal of Au atoms is enhanced over the background signal of adjacent Ga atoms since the high angle scattering from the column strongly decreases.21 GaAs nanowires were grown by the procedure described in ref 22 using molecular beam epitaxy (MBE) and assisted by a Au droplet acting as a catalyst. The wires were sonicated in ethanol and spread onto a carbon holey grid. The sample was heated slightly to 70 °C in vacuum for 2 h to remove residual hydrocarbons and reduce carbon deposited contamination under the electron beam inside the TEM. The samples were examined in a FEI probe-corrected Titan 80−300 TEM equipped with a double-hexapole corrector at an acceleration voltage of 300 kV. The aberrations up to third order were corrected to small values, eliminating resolution limiting aberration effects within the illumination aperture of 25 mrad radius. The probe current was approximately 60 pA. HAADF images were recorded with an inner collection angle of 80 mrad and a sampling of 0.02 nm per pixel. Our HAADF image simulations are based on an atomic structure model of hexagonal GaAs, where a single Au atom was placed on a substitutional Ga site in the GaAs matrix. All images were simulated for a 300 kV aberration-corrected electron probe with a semiconvergence angle of 25 mrad using the multislice algorithm for elastic scattering.23 Small object tilts were considered by applying a tilted propagator function between the slices of atomic layers. The calculations include thermal diffuse scattering by averaging over frozen lattice states.24 The HAADF signal was calculated for a collection angle of 80−150 mrad, and the resulting images were convoluted with a Gaussian point spread to account for partial spatial coherence due to the finite geometrical width of the electron source. We show the effect of such a small tilt on the detection sensitivity of a dopant atom by a simulation study of the elastic interaction between the electron probe and the atomic lattice using multislice calculations.25 The details of the applied simulation parameters and approximations are discussed in the Supporting Information. The calculations indeed show that the channeling of the electron wave along atomic columns is drastically reduced for sample orientations off a high-symmetry zone axis. Consequently, the matrix signal is reduced, and the detectability of single Au atoms inside the GaAs matrix is improved. Figure 1b,c,e displays the distribution of the detected intensity in the crystal and highlights the positions of the atomic nuclei where the signal is generated. When the electron probe is positioned on a pure Ga column in zone-axis orientation as in Figure 1b, the high-angle scattering power decays along the column. Additionally, the distribution of scattering power along the column oscillates with a clear minimum at approximately 4 nm depth (marked by the white arrow). This observation deviates drastically from the uniform distribution of the scattering power, which is assumed in the simplified Z-contrast interpretation. The simulation shows that most of the scattered electrons initiate from interactions within the upper third of the structure (approximately 3 nm). When a single Ga atom is replaced by a Au atom, the cascade of the high-angle scattering events, as shown in Figure 1c, significantly changes compared to a pure Ga column. The Au atom apparently produces a very strong signal, but still a large fraction of the beam continues to channel along the column, leading to additional contributions to the detected signal.
Figure 1. Schematic setup for the simulation of HAADF-STEM images of a 18 nm thick GaAs nanowire with 96 atomic layers of Ga in (a) zone-axis configuration and (d) with a small object tilt. The primary electron beam direction is marked by the green arrow, while the black arrow denotes the orientation of GaAs lattice planes. The images (b,c,e) display x−z volume projections of calculated high-angle electron wave components, which are collected by the HAADF detector. Spots with a high intensity indicate probable centers of highangle scattering events along the atom columns in the GaAs nanowire. The physical z/x aspect ratio of the projections is increased by a factor of 18 for a better visibility. The dotted hexagon represents the orientation of the wire. The calculation result for a pure Ga column aligned parallel to the beam direction is presented in (b). The result for a similar column with one Au substitution at the 9th atomic layer is shown in (c). The same column composition was used for (e), where we additionally applied an object tilt of 3° and artificially enhanced the image contrast to visualize the dissipation of the beam onto adjacent atom columns.
focus the electron wave to the column. This so-called channeling effect enhances the wave amplitude at certain sites along the column resulting in a stronger high-angle scattering from the atoms in the sample column.21 However, the depths where the beam is refocused depends, e.g., on the chemical composition and on the distance between the atoms along the column. Therefore, the sensitivity for the detection of single dopant atoms in the zone-axis orientation is highly nonlinear and depends on numerous material and instrument parameters. Our approach is to apply a small tilt to the lattice planes relative to the incident beam direction, as shown in Figure 1d. 2353
dx.doi.org/10.1021/nl300314k | Nano Lett. 2012, 12, 2352−2356
Nano Letters
Letter
For a small object tilt as in Figure 1e, the interaction of the probe wave with the Au atom remains strong, but the channeling is greatly reduced, consequently the Au signal is enhanced over the signal from the Ga atoms. We calculate the HAADF signal enhancement (SE) due to a gold atom as the ratio between the image intensity obtained from a column with a gold atom relative to the image intensity from adjacent pure Ga columns, see Figure 2a,b. The SE in zone-axis orientation is
Figure 3. The SE due to a single Au atom relative to a pure column with 24 Ga atoms in a 9 nm thick nanowire with a 3° object tilt (b). The dashed line denotes the uncertainty level of the calculated SE. The SE is calculated with four different focus values of the electron probe, as marked in (a). For a focus of Z = 0 nm, the probe is focused at the top of the nanowire, and for Z = 9 nm, the probe is focused at the bottom surface. The foci Z = 2 and 4.5 nm correspond to focus planes inside the object.
the dopant is situated far away from the focus of the beam, the calculated SE is smaller than the calculation’s uncertainty, which is marked by the dashed line. Two important observations from the simulations are: (i) With a tilted sample, a very strong SE is obtained when the Au atom is close to the focus plane of the electron probe; and (ii) the detection of Au atoms with a sub-angstrom probe stems from a fraction of the depth of the sample when using single focus images. The effect of the focus spread (caused by fluctuations of the electron wavelength and of the electron− optical lens currents) is a slight broadening of the depth sensitivity curves in Figure 3, but the results show only minor changes to the above-mentioned calculated SE. More details about the influence of microscope parameters on the detectability of single Au atoms are presented in the Supporting Information. The experimental images in Figure 4 were taken with a focal plane at the top surface, since the simulation showed maximum SE for this configuration when a Au atom is located close to the surface and down to a depth of a few nanometers. Au atoms further inside the nanowire can also be detected with a probe focused further downward, but the expected SE is slightly smaller, and focusing at the top surface is better defined since it coincides with a contrast maximum of the GaAs matrix lattice planes. Dopant atoms are detected in Figure 4 as intensity peaks against the background signal from the GaAs matrix. The horizontally and vertically striped lattice images of GaAs depict the intentional tilts of the nanowire relative to the beam direction. The intense dots are restricted to lattice plane projections as expected for substitutional atoms in contrast to
Figure 2. Calculation of the SE due to a Au atom within a GaAs lattice. Image (a) shows a ball and stick model and a correspondingly simulated HAADF image of the doped GaAs lattice projected along the electron beam direction including a 3° sample tilt. (b) Schematic illustration for the quantification of the SE, which is calculated as the intensity peak ratio between the gold containing column and the adjacent pure Ga columns. (c) The SE due to a single Au atom at different locations within a 9 nm GaAs wire (24 layers of Ga) in the tilted case compared to the zone-axis orientation. The probe focus was set to the top surface of the wire.
significantly smaller compared to the tilted object case as shown in Figure 2c. In both cases, the signal strongly depends on the position of the Au atom within the column, revealing once again the oversimplified relation of the signal intensity to Z1.6−1.8. In the off-axis configuration, the probe is less attracted by the crystal lattice compared to the zone-axis orientation. Consequently the HAADF intensity is much more dependent on the distance between the focal plane and the incorporated Au atom, while there is only a weak dependence on the position of the Au atom within the nanowire. An important parameter for the detection of a single Au atom is therefore the location of the focal plane of the electron probe relative to the location of the incorporated Au atom. Figure 3 presents four different profiles of the SE, corresponding to focusing the beam at the top of the GaAs nanowire with a focus value of Z = 0 nm as well as at depths of 2, 4.5, and 9 nm. When the dopant atom is situated close to the top of the nanowire and the beam is focused to this plane, the SE can reach 70% in the ideal case. In contrast, when 2354
dx.doi.org/10.1021/nl300314k | Nano Lett. 2012, 12, 2352−2356
Nano Letters
Letter
Figure 4. Experimental HAADF images (a,c) and SE line profiles (b,d) taken along lattice planes of a GaAs nanowire in a projection slightly off a zone axis. The circles in the HAADF images mark intense peaks with a SE well above the experimental noise level. In each image, three of these peaks were selected and displayed enlarged in the false color insets to the right. For the peaks marked by 1 (red circle), intensity line profiles along lattice planes are presented in (b,d). The red curve in the profiles is the SE along a lattice plane containing an intense peak, which is attributed to a single Au atom. In comparison, the black curves show the SE along a neighboring lattice plane containing no Au atom. The estimated experimental noise levels are denoted by the gray bands.
range of about 3 nm as an upper limit. In experiment, the detection depth range may be even smaller. Therefore the lower limit of the doping level is estimated by 10 ppm or 1017 cm−3 for a 10−15 nm thick nanowire. Such a doping level is close to the maximum solubility of Au in GaAs, which can reach values in the order of 10 16−18 cm −3 . Therefore, the incorporation of the gold atoms within the GaAs nanowires is high, possibly even reaching supersaturation, as reported previously for InAs nanowires.8 Hence GaAs nanowires synthesized by the VLS method do not provide the required impurity level for ballistic transport studies, and a self-catalytic process may be a favorable synthetic route.12 It is still unclear whether the position of the Au atom is substitutional or interstitial. Simulations show that the visibility of interstitial Au atom is also enhanced in the tilted configuration. The intensity of a single interstitial atom is comparable to a whole tilted Ga column (Supporting Information, section 7). No intense peaks between the atomic rows are observed in the experimental images. Image simulations show sufficient sensitivity for interstitial Au atoms, but we could not conclude whether this points to an absence of interstitial atoms in the probed nanowires or to a lower experimental signal-to-noise ratio than in the calculation.
an interstitial incorporation. The detection limit of the SE in the experimental images is determined essentially by the noise level, which depends on the electron dose and on the noise of the background intensity of the GaAs matrix. In addition, thin surface layers of gallium oxide and hydrocarbon contamination lower the SE due to beam broadening and contribute to the noise in the experiment. In order to improve the signal-to-noise ratio of the evaluation, an averaging along the lattice planes was applied to determine the background signal of the GaAs matrix. The line profiles containing the high intensity spots were normalized to the averaged background signal. By this way we obtain a plot of the SE along a lattice plane, where the intense image dots manifest as sharp peaks; see, e.g., in Figure 4b,d. The peak maximum reaches a SE of up to 25%, which is well beyond the experimental noise as marked by the gray bands. The bands reflect the uncertainty of twice the standard deviation of 6−8% related to signal and background noise. Taking this uncertainty as a detection threshold, we find significant intensity peaks above the background noise, which we associate with Au atoms. The calculated density of the Au atoms is between 0.025 and 0.1 per image area of 1 nm2. From the simulations in Figure 3, we approximate that the generation of a significant signal from the Au atoms is restricted to a depth 2355
dx.doi.org/10.1021/nl300314k | Nano Lett. 2012, 12, 2352−2356
Nano Letters
Letter
(7) Czaban, J. A.; Thompson, D. A.; LaPierre, R. R. GaAs Core-Shell Nanowires for Photovoltaic Applications. Nano Lett. 2008, 9 (1), 148− 154. (8) Perea, D. E.; Allen, J. E.; May, S. J.; Wessels, B. W.; Seidman, D. N.; Lauhon, L. J. Three-Dimensional Nanoscale Composition Mapping of Semiconductor Nanowires. Nano Lett. 2006, 6 (2), 181−185. (9) Perea, D. E.; Lensch, J. L.; May, S. J.; Wessels, B. W.; Lauhon, L. J. Composition analysis of single semiconductor nanowires using pulsed-laser atom probe tomography. Appl. Phys. A: Mater. Sci. Process. 2006, 85 (3), 271−275. (10) Allen, J. E.; Hemesath, E. R.; Perea, D. E.; Lensch-Falk, J. L.; LiZ., Y; Yin, F.; Gass, M. H.; Wang, P.; Bleloch, A. L.; Palmer, R. E.; Lauhon, L. J. High-resolution detection of Au catalyst atoms in Si nanowires. Nat. Nanotechnol. 2008, 3 (3), 168−173. (11) Shtrikman, H.; Popovitz-Biro, R.; Kretinin, A. V.; Kacman, P. GaAs and InAs Nanowires for Ballistic Transport. IEEE J. Sel. Top. Quantum Electron. 2010, 17 (4), 922−934. (12) Krogstrup, P.; Popovitz-Biro, R.; Johnson, E.; Madsen, M. H.; Nygård, J.; Shtrikman, H. Structural Phase Control in Self-Catalyzed Growth of GaAs Nanowires on Silicon (111). Nano Lett. 2010, 10 (11), 4475−4482. (13) Arbiol, J.; et al. Triple-twin domains in Mg doped GaN wurtzite nanowires: structural and electronic properties of this zinc-blende-like stacking. Nanotechnology 2009, 20 (14), 145704. (14) Perea, D. E.; Hemesath, E. R.; Schwalbach, E. J.; Lensch-Falk, J. L.; Voorhees, P. W.; Lauhon, L. J. Direct measurement of dopant distribution in an individual vapour-liquid-solid nanowire. Nat. Nanotechnol. 2009, 4 (5), 315−319. (15) Hemesath, E. R.; Schreiber, D. K.; Gulsoy, E. B.; Kisielowski, C. F.; Petford-Long, A. K.; Voorhees, P. W.; Lauhon, L. J. Catalyst Incorporation at Defects during Nanowire Growth. Nano Lett. 2011, 12 (1), 167−171. (16) Voyles, P. M.; Grazul, J. L.; Muller, D. A. Imaging individual atoms inside crystals with ADF-STEM. Ultramicroscopy 2003, 96 (3− 4), 251−273. (17) Voyles, P. M.; Muller, D. A.; Grazul, J. L.; Citrin, P. H.; Gossmann, H. J. L. Atomic-scale imaging of individual dopant atoms and clusters in highly n-type bulk Si. Nature 2002, 416 (6883), 826− 829. (18) Hartel, P.; Rose, H.; Dinges, C. Conditions and reasons for incoherent imaging in STEM. Ultramicroscopy 1996, 63 (2), 93−114. (19) van Benthem, K.; Lupini, A. R.; Kim, M.; Baik, H. S.; Doh, S.; Lee, J. H.; Oxley, M. P.; Findlay, S. D.; Allen, L. J.; Luck, J. T.; Pennycook, S. J. Three-dimensional imaging of individual hafnium atoms inside a semiconductor device. Appl. Phys. Lett. 2005, 87 (3), 034104. (20) Oh, S. H.; Benthem, K. v.; Molina, S. I.; Borisevich, A. Y.; Luo, W.; Werner, P.; Zakharov, N. D.; Kumar, D.; Pantelides, S. T.; Pennycook, S. J. Point Defect Configurations of Supersaturated Au Atoms Inside Si Nanowires. Nano Lett. 2008, 8 (4), 1016−1019. (21) Gemmell, D. S. Channeling and related effects in the motion of charged particles through crystals. Rev. Mod. Phys. 1974, 46 (1), 129− 227. (22) Shtrikman, H.; Popovitz-Biro, R.; Kretinin, A.; Houben, L.; Heiblum, M.; Bukała, M.; Galicka, M.; Buczko, R.; Kacman, P. Method for Suppression of Stacking Faults in Wurtzite III−V Nanowires. Nano Lett. 2009, 9 (4), 1506−1510. (23) Barthel, J. Dr. Probe - STEM multislice image calculation program; Ernst Ruska-Centre for Microscopy and Spectroscopy with Electrons: Jülich, Germany; www.er-c.org. (24) Loane, R. F.; Xu, P.; Silcox, J. Thermal vibrations in convergentbeam electron diffraction. Acta Crystallogr., Sect. A 1991, 47 (3), 267− 278. (25) Goodman, P.; Moodie, A. F. Numerical evaluations of N-beam wave functions in electron scattering by the multi-slice method. Acta Crystallogr., Sect. A 1974, 30 (2), 280−290.
In conclusion, the detection of single foreign atoms within a III−V crystal lattice is feasible even in cases where the Z number ratio between the dopant and the host lattice is rather low, by avoiding strong channeling conditions. The volume of detection, where single Au atoms can be imaged, is only a fraction of the sample volume, close to the focal plane of the electron beam. The present work puts this technique on a firm foundation with quantitative modeling considering varying focal depths and positions of atoms. In GaAs nanowires, this technique allows us to identify single Au atoms within the lattice, proving that Au atoms are incorporated within the nanowire during the growth process. The high doping level in the wires encourages pursuing alternative, self-assisted synthetic routes for GaAs nanowires for ballistic transport. Since the presented method is general, our study lays the groundwork for other doped semiconductor nanostructures and for correlating dopant atom positions with their physical properties.
■
ASSOCIATED CONTENT
S Supporting Information *
Discussion regarding the precision and parameter dependence of the calculation of the SE, the detection of interstitial Au atoms, the observation of mobile surface Au atoms as a result of the interaction with the electron beam, and the analytical evidence for a native oxide layer in experiment. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors thank the DFG for the support of the subangstrom electron microscopy at the Ernst-Ruska Centre in Juelich. M.B. would like to acknowledge the support of the Sara Lee Schupf Postdoctoral Award Funded by the Clore Foundation and S. Donald Sussman, The Weizmann Institute of Science Program for Advancing Women in Science. H.S.’s work was partly supported by Israeli Science Foundation under grant 530/08 and Israeli Ministry of Science grant no. 2013028.
■
REFERENCES
(1) Gradecak, S.; Qian, F.; Li, Y.; Park, H.-G.; Lieber, C. M. GaN nanowire lasers with low lasing thresholds. Appl. Phys. Lett. 2005, 87 (17), 173111−3. (2) Bryllert, T.; Wernersson, L. E.; Lowgren, T.; Samuelson, L. Vertical wrap-gated nanowire transistors. Nanotechnology 2006, 17 (11), S227. (3) Heeres, E. C.; Bakkers, E. P. A. M.; Roest, A. L.; Kaiser, M.; Oosterkamp, T. H.; de Jonge, N. Electron Emission from Individual Indium Arsenide Semiconductor Nanowires. Nano Lett. 2007, 7 (2), 536−540. (4) Minot, E. D.; Kelkensberg, F.; van Kouwen, M.; van Dam, J. A.; Kouwenhoven, L. P.; Zwiller, V.; Borgström, M. T.; Wunnicke, O.; Verheijen, M. A.; Bakkers, E. P. A. M. Single Quantum Dot Nanowire LEDs. Nano Lett. 2007, 7 (2), 367−371. (5) Patolsky, F.; Timko, B. P.; Zheng, G. F.; Lieber, C. M. NanowireBased Nanoelectronic Devices in the Life Sciences. MRS Bull. 2007, 32, 142. (6) Wang, N.; Cai, Y.; Zhang, R. Q. Growth of nanowires. Mater. Sci. Eng., R 2008, 60 (1−6), 1−51. 2356
dx.doi.org/10.1021/nl300314k | Nano Lett. 2012, 12, 2352−2356
