Bottom-Up Formation of Robust Gold Carbide
March 16, 2015 – Gold and carbon hardly mix, and a crystalline gold carbide had never been made. In heating experiments up to 1300 K on graphene decorated with gold, researchers from Ulm and Vienna, led by SALVE director Ute Kaiser, watched one form atom by atom with 80 kV aberration-corrected HRTEM: gold atoms gathered into squares, layers and small cuboids with a far larger spacing than in pure gold. Electron energy loss spectroscopy and density functional theory identify the material as gold carbide in the zinc-blende structure, metastable but surprisingly robust.
Gold and carbon hardly mix: bulk gold dissolves almost no carbon, and a crystalline gold carbide had never been made. Now SALVE researchers in Ulm, with colleagues from the University of Vienna, have watched one form atom by atom. In heating experiments up to 1300 K on graphene decorated with gold nanoislands in an FEI Titan at 80 kV, using aberration-corrected high-resolution transmission electron microscopy (AC-HRTEM), they found clusters, layers and small cuboids of gold atoms arranged in squares, but with a far larger spacing than in pure gold. Electron energy loss spectroscopy (EELS) and density functional theory (DFT) identify the new material as gold carbide, AuC, in the zinc-blende structure. It is metastable, but surprisingly robust. Single silicon atoms in graphene imaged with a Nion UltraSTEM at 60 kV by other groups serve as validation of the EELS result.
A carbide that should not exist
The existence of a gold carbide was first reported in 1900, when Mathews & Watters described the highly explosive gold acetylide Au2C2[1]. Today, a whole research field, organogold chemistry, deals with compounds that contain carbon–gold bonds[2]. Inorganic crystalline gold–carbon compounds, however, had not been found experimentally[3]. This is not surprising: bulk gold has almost no solubility for carbon under equilibrium conditions, and the only crystalline gold carbide proposed so far turned out to be extremely unstable[4]. Small gold carbide clusters can be made, but they are electrically charged and contain only a few atoms[3].
The successful formation of robust gold carbide was made possible by the 2D material graphene. “We used graphene as the substrate for the gold atoms and as an in-situ heating element by applying a voltage. Furthermore, both graphene and the hydrocarbons on graphene could have served as a supplier of carbon for the chemical reaction,” said Prof. Ute Kaiser, director of the SALVE project, who led the study. “In addition, the electron beam itself affects the interaction of carbon and gold. The combination of these special features of the experiment enabled us to synthesize a new gold–carbon compound through a bottom-up process, atom by atom, in situ in the electron microscope for the first time.”
Graphene as substrate, heater and carbon source
Work with carbon nanostructures in the TEM has benefited enormously from aberration correction[5], since it gives atomic resolution at low electron energies, at which low-dimensional carbon materials remain much more stable. Thanks to its mechanical stability, graphene can serve as a sample support in electron microscopy[6], and because of its high thermal conductivity, graphene is stable at least from room temperature up to 2000 K[7].
In-situ experiments with 2D materials, however, place high demands on sample preparation. The aim was to image heated Au and C atoms with atomic resolution in real time. A mechanically exfoliated graphene flake was transferred onto a Si/SiN membrane with open windows and gold contacts, which gave an electrically contacted, completely free-standing few-layer graphene sheet in a TEM-compatible geometry[7]. Gold nanoislands had been deposited on it beforehand by thermal evaporation. A current through the graphene heated the sample, and the melting of the gold particles gave an estimate of the local temperature. How the microscope was set up is described in the recipe box at the end of the article.
Gold atoms in unusual squares
Gold atoms diffuse on the surface of gold particles already at room temperature[8],[9], which makes the particles change shape on heating. Above a certain temperature, almost spherical particles become liquid droplets and start to evaporate[7],[10],[11]; here this happened between 800 K and 1300 K, depending on particle size (3 to 20 nm). All observations described below were made just below the melting point of the gold particles. The diffusion barrier for gold atoms on graphene is extremely low, theoretically estimated at only 0.05 eV[12],[13], far too fast to be followed at exposure times of 0.5 to 1 s per image. Single gold atoms are therefore only visible when they are trapped at grain boundaries, edge reconstructions or vacancies[14], in line with DFT calculations[13],[15],[16]. The electron beam adds radiation-induced diffusion, as discussed by Urban & Seeger[17] and Banhart[18].
The AC-HRTEM images gave unique insights into the continuous interaction of Au and C. When adsorbed amorphous carbon crystallized under heating[10], many gold atoms were trapped in it (Fig. 1a). In many cases they formed small clusters with square arrangements. In contrast to pure gold, these clusters had a considerably larger distance between neighbouring atoms, 3.35 Å (Fig. 1a, ii). Only one rare example matched the distance in bulk gold, 2.88 Å (Fig. 1a, i)[19]. Light elements between the heavy gold atoms are not visible under these imaging conditions, so the kind and composition of the unknown particles could not be read from the images. The scientists therefore named the structure “Au-X” for the time being; from the experiments it was only clear that X had to be an element with a low atomic number.
Regular single layers of Au-X with the 3.35 Å spacing grew only up to about 30 gold atoms; larger clusters formed two or more layers (Fig. 1b, c). In some cases, the switching between single and double layers could be followed in real time (Fig. 1b). A side view came from a cluster on the curled edge of a few-layer graphene sheet (Fig. 1d): there, the first and second Au-X layers are 2.35 Å apart, and with the second-layer atoms above the gaps of the first, as in the plan view of Fig. 1c, this again gives a gold–gold spacing of 3.35 Å.
Cuboids that eat carbon
Under continued heating and irradiation, the clusters grew into cuboids of 1–2 nm with sharp edges and constant thickness (Fig. 2a). Most of them sat close to ordinary gold nanoparticles, and the largest ones grew directly on them, preferably on their (111) facets (Fig. 2b, Video 1). Growing Au-X crystals did not just collect gold atoms but seemingly “ate” partly crystallized carbon layers and tiny fullerenes to gain more material (Video 2). At the edges of holes in the few-layer graphene, tiny crystals grew and dissolved atom by atom (Video 3), and an amorphous droplet of gold and carbon crystallized into a small cube (Video 4). The new material was fairly stable under the electron beam but still reorganized under it: when the dose rate was raised from 5 × 105 to 3 × 106 electrons per nm2 and second, larger objects broke up into several smaller ones.
Video 1: AC-HRTEM image sequence at 80 kV of gold on heated graphene, one frame every 2 s: Au-X crystals growing on the facets of a gold particle, played at 20 frames per second (Movie M1a of the paper).
Video 2: AC-HRTEM image sequence at 80 kV of gold on heated graphene, one frame every 2 s: a small AuC cube forms on the upper left facet of a gold particle, a second one on the lower right facet; both seemingly eat fullerenes (M1b).
Video 3: AC-HRTEM image sequence at 80 kV of gold on heated graphene, one frame every 2 s: tiny AuC crystals grow and decompose at the edges of holes in few-layer graphene at 3 × 106 e/nm² per second (M2).
Video 4: AC-HRTEM image sequence at 80 kV of gold on heated graphene, one frame every 2 s: an amorphous droplet of gold and carbon crystallizes, and new crystals nucleate at partly crystallized carbon layers (M3).
From the top and side views and the transitions between single and multiple layers, the gold part of the lattice is clear: the gold atoms form a face-centred cubic (fcc) lattice with a lattice parameter of √2 · (3.35 ± 0.1) Å, measured in the Fourier transform of Fig. 2a with the graphene lattice as calibration. Bulk gold is also fcc, but with √2 · 2.88 Å. The larger spacing suggests a chemical compound in which a second element sits in the gaps of the gold lattice. Three crystal structures are compatible with the observed fcc gold lattice: zinc blende (ZnS), rock salt (NaCl) and fluorite (CaF2) (Fig. 3). Image simulations show why the second element cannot be seen, as described in the box below, which also holds the simulated images (Fig. 4).
The second element is carbon
To identify the second element, local EELS was recorded from a single Au-X cuboid on graphene with a beam of about 5 nm diameter, with several energy windows covering the absorption edges of all candidate elements: silicon, nitrogen and carbon, which are present in the sample, and carbon, oxygen, silicon and potassium, which may come with contamination (Fig. 5). The spectra show carbon (K edge at 284 eV) and gold (M5 edge at 2206 eV). Silicon, nitrogen and oxygen can be excluded, since their edges are absent. The paper names the missing silicon signal the Si-K edge at 99 eV; it is the Si-L edge, as labelled in the paper’s own figure (Fig. 5c). A weak signal near 540 eV, only 0.1 % above background, most probably comes from gold (N2,3 edge); as oxygen it would correspond to a handful of atoms at most. Detecting carbon on graphene may seem trivial, but since all other reasonable elements are excluded, carbon must be the only partner of gold.
Density functional theory points to zinc blende
To confirm this interpretation, the lattice parameters of all candidate compounds in the three structures were calculated by DFT. Only the small elements C, N and O gave a reasonable lattice match; the values for silicon deviated by 11 to 25 % depending on the structure and could thus be excluded, in line with the EELS result. The smallest deviation is shown by the NaCl structure of AuC, AuN and AuO. However, DFT with the generalized gradient approximation (GGA) systematically overestimates lattice parameters[20],[21]: for gold, the DFT value is 2.4 % larger than the experimental one. Taking this systematic deviation into account, the best agreement is obtained for AuC in the ZnS structure. All values and the comparison with the experiment are given in the box below, together with the DFT surface energies (Fig. 6).
Metastable, but robust
The AuC formed in this bottom-up process is metastable, but surprisingly robust. The cuboids survived exposure to air and heating of the whole sample to 200 °C on a hotplate. In the vacuum of the microscope, they were at least as stable as the gold nanoparticles: while a neighbouring gold particle larger than 10 nm was already evaporating, which indicates at least 1300 K, the cuboids kept their size and crystal structure. Only on further heating did they decompose, directly, without a liquid phase in between (Video 5). The cuboids were first found by chance in three independent heating experiments, and their growth by Joule heating up to 1000 K under continuous irradiation was then reproduced in two further experiments.
Video 5: AC-HRTEM image sequence at 80 kV of gold on heated graphene, one frame every 2 s: at rising temperature the AuC crystals finally decompose, without a liquid phase in between (M4).
DFT predicts that AuC is metallic, like the stable tungsten carbide WC. The electronic structure of a small AuC cluster on graphene is almost the same as that of the isolated cluster, so AuC interacts only weakly with graphene and keeps its properties on it. The calculated bulk modulus of AuC is 141 GPa, smaller than that of bulk gold (180 GPa); for comparison, WC with 338 GPa is slightly stiffer than tungsten (310 GPa). Like other transition metal carbides, the new material may be attractive for applications.
Challenges named by the authors
Light elements between the heavy gold atoms are invisible in the HRTEM images, so the composition could only be inferred from EELS and DFT (Fig. 4). The temperature was not measured directly but estimated from the melting of the gold particles, whose melting point depends on their size. The weak signal near 540 eV fits gold, but energetically also the oxygen K edge. DFT with GGA systematically overestimates lattice parameters, and the identification of the ZnS structure relies on correcting for this deviation. All calculated compounds are endothermic: the material forms only under heat and electron irradiation together, and the carbon is assumed to come from hydrocarbon deposits or defects on graphene rather than from the graphene lattice itself.
Outlook
In-situ AC-HRTEM, with graphene as substrate, heater and carbon source at once, made it possible to watch the formation of a new gold–carbon compound atom by atom: from single gold atoms via square clusters, single and double layers to cuboids of AuC in the zinc-blende structure. The crystals are stable under the electron beam, in air and up to about 1000 K in vacuum. They are predicted to be metallic, with a bulk modulus of 141 GPa; other properties of this new material remain to be explored.
Resource:
Westenfelder, B., Biskupek, J., Meyer, J. C., Kurasch, S., Lin, X., Scholz, F., Groß, A., Kaiser, U. (2015).
Bottom-up formation of robust gold carbide.
Sci. Rep. 5, 8891.
https://doi.org/10.1038/srep08891
Download this news article as PDF – for reading on paper, taking notes or filing.
-
Mathews, J. A., & Watters, L. L. (1900). The Carbide of Gold. Journal of the American Chemical Society, 22: 108-111. https://doi.org/10.1021/ja02040a010
-
Parish, R. V. (1997). Organogold Chemistry: I Structure and Synthesis. Gold Bulletin, 30: 3-12. https://doi.org/10.1007/BF03214751
-
Cohen, Y., Bernshtein, V., Armon, E., Bekkerman, A., & Kolodney, E. (2011). Formation and emission of gold and silver carbide cluster ions in a single C60− surface impact at keV energies: Experiment and calculations. The Journal of Chemical Physics, 134: 124701. https://doi.org/10.1063/1.3561317
-
Okamoto, H., & Massalski, T. B. (1984). The Au-C (Gold-Carbon) System. Bulletin of Alloy Phase Diagrams, 5: 378-379. https://doi.org/10.1007/BF02872953
-
Haider, M., Uhlemann, S., Schwan, E., Rose, H., Kabius, B., & Urban, K. (1998). Electron Microscopy Image Enhanced. Nature, 392: 768-769. https://doi.org/10.1038/33823
-
Ross, F. M. (2007). In Situ Transmission Electron Microscopy. In Science of Microscopy, 445-534. Springer New York. https://doi.org/10.1007/978-0-387-49762-4_6
-
Westenfelder, B., Meyer, J. C., Biskupek, J., Algara-Siller, G., Lechner, L. G., Kusterer, J., Kaiser, U., Krill III, C. E., Kohn, E. & Scholz, F. (2011). Graphene-Based Sample Supports for In Situ High-Resolution TEM Electrical Investigations. Journal of Physics D, 44: 055502. https://doi.org/10.1088/0022-3727/44/5/055502
-
Bovin, J. O., Wallenberg, R., & Smith, D. J. (1985). Imaging of Atomic Clouds Outside the Surfaces of Gold Crystals by Electron Microscopy. Nature, 317: 47-49. https://doi.org/10.1038/317047a0
-
Iijima, S., & Ichihashi, T. (1985). Motion of Surface Atoms on Small Gold Particles Revealed by HREM With Real-Time VTR System. Japanese Journal of Applied Physics, 24: L125. https://doi.org/10.1143/JJAP.24.L125
-
Westenfelder, B., Meyer, J. C., Biskupek, J., Kurasch, S., Scholz, F., Krill III, C. E., & Kaiser, U. (2011). Transformations of Carbon Adsorbates on Graphene Substrates Under Extreme Heat. Nano letters, 11: 5123-5127. https://doi.org/10.1021/nl203224z
-
Barnard, A. S., Young, N. P., Kirkland, A. I., Van Huis, M. A., & Xu, H. (2009). Nanogold: A Quantitative Phase Map. ACS nano, 3: 1431-1436. https://doi.org/10.1021/nn900220k
-
Jensen, P., Blase, X., & Ordejón, P. (2004). First Principles Study of Gold Adsorption and Diffusion on Graphite. Surface Science, 564: 173-178. https://doi.org/10.1016/j.susc.2004.06.188
-
Hardcastle, T. P., Seabourne, C. R., Zan, R., Brydson, R. M. D., Bangert, U., Ramasse, Q. M., Novoselov, K. S., & Scott, A. J. (2013). Mobile metal adatoms on single layer, bilayer, and trilayer graphene: An ab initio DFT study with van der Waals corrections correlated with electron microscopy data. Physical Review B, 87: 195430. https://doi.org/10.1103/PhysRevB.87.195430
-
Cretu, O., Krasheninnikov, A. V., Rodríguez-Manzo, J. A., Sun, L., Nieminen, R. M., & Banhart, F. (2010). Migration and Localization of Metal Atoms on Strained Graphene. Physical review letters, 105: 196102. https://doi.org/10.1103/PhysRevLett.105.196102
-
Malola, S., Häkkinen, H., & Koskinen, P. (2009). Gold in Graphene: In-Plane Adsorption and Diffusion. Applied Physics Letters, 94: 043106. https://doi.org/10.1063/1.3075216
-
Tang, Y., Yang, Z., & Dai, X. (2011). Trapping of Metal Atoms in the Defects on Graphene. The Journal of chemical physics, 135: 224704. https://doi.org/10.1063/1.3666849
-
Urban, K., & Seeger, A. (1974). Radiation-Induced Diffusion of Point-Defects During Low-Temperature Electron Irradiation. Philosophical Magazine, 30: 1395-1418. https://doi.org/10.1080/14786437408207289
-
Banhart, F. (1999). Irradiation Effects in Carbon Nanostructures. Reports on Progress in Physics, 62: 1181. https://doi.org/10.1088/0034-4885/62/8/201
-
Wang, J., Wang, G., & Zhao, J. (2002). Density-functional study of Aun (n = 2–20) clusters: Lowest-energy structures and electronic properties. Physical Review B, 66: 035418. https://doi.org/10.1103/PhysRevB.66.035418
-
Lischka, M., & Groß, A. (2002). Hydrogen Adsorption on an Open Metal Surface: H2/Pd (210). Physical Review B, 65: 075420. https://doi.org/10.1103/PhysRevB.65.075420
-
Lischka, M., Mosch, C., & Groß, A. (2007). Tuning Catalytic Properties of Bimetallic Surfaces: Oxygen Adsorption on Pseudomorphic Pt/Ru Overlayers. Electrochimica acta, 52: 2219-2228. https://doi.org/10.1016/j.electacta.2006.03.113
-
Krishnamurthy, S., Montalti, M., Wardle, M. G., Shaw, M. J., Briddon, P. R., Svensson, K., Hunt, M. R. C. & Šiller, L. (2004). Nitrogen Ion Irradiation of Au (110): Photoemission Spectroscopy and Possible Crystal Structures of Gold Nitride. Physical Review B, 70: 045414. https://doi.org/10.1103/PhysRevB.70.045414
-
Maya, L., Paranthaman, M., Thundat, T., & Bauer, M. L. (1996). Gold Oxide as Precursor to Gold/Silica Nanocomposites. Journal of Vacuum Science & Technology B, 14: 15-21. https://doi.org/10.1116/1.589020
-
Biskupek, J., Hartel, P., Haider, M., & Kaiser, U. (2012). Effects of Residual Aberrations Explored on Single-Walled Carbon Nanotubes. Ultramicroscopy, 116: 1-7. https://doi.org/10.1016/j.ultramic.2012.03.008
-
Perdew, J. P., Burke, K., & Ernzerhof, M. (1996). Generalized Gradient Approximation Made Simple. Physical review letters, 77: 3865. https://doi.org/10.1103/PhysRevLett.77.3865
-
Kresse, G., & Furthmüller, J. (1996). Efficient Iterative Schemes for Ab Initio Total-Energy Calculations Using a Plane-Wave Basis Set. Physical review B, 54: 11169. https://doi.org/10.1103/PhysRevB.54.11169
-
Kresse, G., & Furthmüller, J. (1996). Efficiency of Ab-Initio Total Energy Calculations for Metals and Semiconductors Using a Plane-Wave Basis Set. Computational Materials Science, 6: 15-50. https://doi.org/10.1016/0927-0256(96)00008-0
-
Blöchl, P. E. (1994). Projector Augmented-Wave Method. Physical Review B, 50: 17953. https://doi.org/10.1103/PhysRevB.50.17953
-
Kresse, G., & Joubert, D. (1999). From Ultrasoft Pseudopotentials to the Projector Augmented-Wave Method. Physical Review B, 59: 1758. https://doi.org/10.1103/PhysRevB.59.1758
-
Birch, F. (1978). Finite Strain Isotherm and Velocities for Single‐Crystal and Polycrystalline NaCl at High Pressures and 300° K. Journal of Geophysical Research: Solid Earth, 83: 1257-1268. https://doi.org/10.1029/JB083iB03p01257






