Carbon Glass
February 11, 2014 – How does a crystal turn into a glass? Researchers from Vienna, Helsinki and Ulm, among them SALVE director Ute Kaiser, followed it atom by atom in graphene with 100 kV aberration-corrected HRTEM, introducing disorder with the electron beam and extracting every atom from the images. The change comes suddenly and early: first a vitreous network separates small crystallites, then it takes over until a random network remains. Both competing theories of glass thus describe the same material at different degrees of disorder. A 60 kV STEM study from Vienna on ion-amorphized graphene confirms the crystallinity.
What does a glass look like at the atomic level, and how does a crystal turn into one? Researchers from the University of Vienna, the University of Helsinki and Ulm University, among them former SALVE member Jannik C. Meyer and SALVE director Ute Kaiser, have followed this transformation atom by atom in graphene[1]. Using 100 kV aberration-corrected high-resolution transmission electron microscopy (AC-HRTEM) on an FEI Titan, they introduced disorder step by step with the electron beam and extracted the positions of all atoms automatically from the images. The crystal turns into a glass suddenly and surprisingly early: first a vitreous network separates small crystallites, as the crystallite theory of glass describes, and then this network takes over until a random network remains, as Zachariasen proposed. A later study with a Nion UltraSTEM at 60 kV in Vienna, on graphene amorphized with an ion beam, serves as validation.
Two theories of glass
For more than 80 years, ideas about the structure of glass have been shaped by the intuition of Zachariasen and his illustrations of a random network[2],[3]. With the participation of SALVE scientists, earlier studies confirmed experimentally that vitreous regions in 2D silica come very close to these illustrations[4],[5] (see also the SALVE news article about 2D silica). However, crystalline areas were found as well, and with small samples and limited statistics it could not be decided whether they are an intrinsic part of the glass or a separate phase. The direct study of atomic coordinates in a glass has therefore remained mostly in the realm of theoretical models and computer simulations[6],[7]. The crystalline regions gave credibility to the crystallite theory[2], which posits that a glass is a disordered arrangement of small crystalline particles separated by a disordered network. In 2010, a fraction of up to 50 % crystallites was found in amorphous sputtered silicon[8], which supported but did not settle this question. In a single carbon layer, flat clusters of pentagons, hexagons and heptagons between small crystalline islands had already been imaged in reduced graphene oxide[9]. An answer would require watching a glass form at the atomic level, which the traditional way of making a glass, quenching a liquid[10], renders impossible.
Disorder in a solid, atom by atom
The team found a way around this problem. Instead of melting and quenching, they introduced disorder into graphene with the electron beam of the microscope, so that the material remained a solid throughout[1]. Being two-dimensional, graphene allows every atom to be seen directly. The electron energy of 100 keV lies slightly above the damage threshold of graphene, so the transformation is slow enough to follow every change. Disorder enters in two ways, by the removal of atoms[11] and by bond rotations[12], which occur at random places and times and form non-hexagonal carbon rings. As shown earlier, graphene stays flat even with pentagons and heptagons[13]. The image sequence itself had already been recorded for an earlier study, which followed graphene from point defects to a two-dimensional amorphous carbon[14]; its amorphous areas had not been analysed until now.
At first, small isolated defects appear in the lattice (Fig. 1a). Under continued irradiation they grow into a vitreous network that divides the original crystal into nanometre-sized crystallites (Fig. 1b), and in the final stage the crystallites vanish into the growing network (Fig. 1c). Holes also form, but they are caused by chemical effects driven by the beam[11] and are not counted as part of the disordered structure. The structure stays two-dimensional throughout, providing a direct view of every stage.
Fourier transforms of the images measure the change across the whole field of view (Fig. 1d–f). The distance between the peaks stays constant, while their width changes. The radial width, a measure of the crystallite size, rises only slightly while the defects are isolated, then increases quickly once they merge into the vitreous network, and finally saturates. The azimuthal spread grows almost linearly with the dose up to the fully disordered case, a spread of 60°. Because the effect of the beam depends on the voltage, the team uses the density deficit, the fraction of atoms removed, as a transferable measure of disorder, which can be determined directly in the experiment[11].
From the image to the atoms
In modern aberration-corrected TEM instruments and for very thin samples, the images can be interpreted directly as the projected atomic structure. So far, however, the atomic configurations had been taken from the images by hand, which is laborious and error-prone, and earlier work was therefore limited to small isolated defects[1]. “We have for the first time analyzed larger image areas with the assistance of computers,” says Jannik Meyer. The automatic method, and how well its structures reproduce the images, is described in the box below, which also holds the comparison of an image with its simulation (Fig. 2).
Three regimes
Figure 3 shows atomic structures derived from the images for the three regimes[1]. At a density deficit of 0.3 %, the lattice is crystalline with a few isolated defects (Fig. 3a). With increasing dose, the defects grow and merge, remarkably in such a way that isolated crystallites remain, separated by a vitreous network (Fig. 3b, 7.1 %). This structure somewhat resembles polycrystalline graphene with nanometre-sized grains[16],[17], but the vitreous network is much wider than typical grain boundaries in graphene[18]–[20] and covers a much larger part of the area. Under further irradiation, the vitreous areas grow at the expense of the crystallites, until almost no extended hexagonal areas remain and the structure is well described by the random network theory (Fig. 3c, 16.7 %).
Order at the shortest scale
Unlike vitreous oxides such as silica, which are built from tetrahedral units of several atoms, the building block here is the carbon atom. The shortest range of disorder therefore lies in how neighbouring atoms are connected[2]. To quantify it, the scientists computed the distribution of interatomic distances, the radial distribution function g(r) (Fig. 4a–c). Deviations from the ideal distances are clear close to the transition from crystal to glass (Fig. 4b, structure in Fig. 4e), but the long-range order disappears completely only once the crystallite structure has formed (Fig. 4c, structure in Figs. 4f and 3b). At this stage, g(r) is very similar to that of the 2D silica glass[4]. The bond-angle distribution (Fig. 4d) shows a growing spread around the ideal 120° and a second peak near 108°, the angle of pentagons. A peak of the heptagons near 129° does not appear, because local curvature lets the larger rings keep angles close to 120°.
Rings and crystallinity
Ring statistics show how the topology changes (Fig. 5a–d). The spread grows step by step, yet hexagons still dominate even at the highest degree of disorder (Fig. 5d), similar to a theoretical random network[21][1]. On a log-normal probability plot, the data approach the literature values with increasing disorder (Fig. 5e). The main change during the experiment is the falling share of hexagons, while the ratios between the other polygons stay nearly constant: pentagons to heptagons, for example, remain at 1.14 ± 0.02 throughout. Knowing this ratio allows the structural changes that an electron beam makes in graphene to be predicted[22]. How carbon glass compares with the 2D silica glass is summarized in the box below.
This share of hexagons among all rings is the crystallinity C. For pristine graphene C = 1, for the vitreous 2D silica C ≈ 0.4[4]. The data set shows for the first time the transition between these two extremes (Fig. 5f)[1]. As long as the disorder consists of defects in the original lattice, C falls slowly and linearly. As soon as the defects merge into the vitreous network, it falls much faster and approaches a constant ratio of about one hexagon to one non-hexagon, C ≈ 0.51. The kink appears at the same point as in the Fourier analysis (Fig. 1e) and supports the view that it marks the actual transition from crystalline graphene to the glassy phase. The final points lie almost at the constant offset of an exponential fit, so the maximum disorder was reached: further irradiation no longer changes the statistics. Glasses with any crystallinity between 0.5 and 0.9 can thus be made at will.
Order at the longest scale
Finally, the scientists looked at density variations in the final structure on different length scales (Fig. 5g, h)[1]. They divided it into square areas of 0.2 to 2.0 nm width w and calculated the spread of the atom density as a function of w. As expected, the spread becomes smaller for larger areas, the structure looks smoother on a larger scale. The data follow a 1/w law, which is exactly what completely random atomic coordinates, or white noise, would give. This indicates the absence of long-range density fluctuations, or hyperuniformity[23].
Several years later, the Vienna group amorphized graphene with an ion beam instead of electrons and imaged it with a different microscope. The validation box below compares the two microscopes and shows both carbon glasses side by side (Fig. 6).
Challenges named by the authors
Disorder made by an electron beam differs fundamentally from the traditional route of quenching a melt[1]. Holes appear during irradiation, caused by chemical effects of the beam, and adsorbates cover parts of the sheet; both were excluded from the analysis, as were the edges of each structure. The automatic extraction has an estimated error below 5 % for non-hexagons. The interatomic potential used for the relaxation affects only the absolute distances, not the conclusions. The radial distribution was limited to intermediate structures, because holes and other complications increase during the experiment. A weak ring-shaped feature in some images is a distortion from the CCD camera.
Outlook
The study shows that the two competing theories of glass, the crystallite theory and the random network model, describe the same material, each for a specific range of disorder, at least in two dimensions[1]. “Aberration-corrected electron microscopy is virtually unparalleled for imaging defective 2D materials with atomic resolution. In modern aberration-corrected TEM instruments and for very thin samples, one can interpret the obtained images directly as the projected atomic structure,” says Ute Kaiser. “The new investigation of ‘2D carbon glass’ is a further key component in the continued improvement of our understanding of vitreous structures.”
Resource:
Eder, F. R., Kotakoski, J., Kaiser, U., Meyer, J. C. (2014).
A journey from order to disorder — atom by atom transformation from graphene to a 2D carbon glass.
Sci. Rep. 4, 4060.
https://doi.org/10.1038/srep04060
Download this news article as PDF – for reading on paper, taking notes or filing.
-
Eder, F. R., Kotakoski, J., Kaiser, U., & Meyer, J. C. (2014). A Journey From Order to Disorder — Atom by Atom Transformation From Graphene to a 2D Carbon Glass. Scientific Reports, 4, 4060. https://doi.org/10.1038/srep04060
-
Wright, A. C. (1994). Neutron Scattering From Vitreous Silica. V. The Structure of Vitreous Silica: What Have We Learned From 60 Years of Diffraction Studies? Journal of Non-Crystalline Solids, 179, 84–115. https://doi.org/10.1016/0022-3093(94)90687-4
-
Wright, A. C., & Thorpe, M. F. (2013). Eighty Years of Random Networks. Physica Status Solidi (b), 250(5), 931–936. https://doi.org/10.1002/pssb.201248500
-
Huang, P. Y., Kurasch, S., Srivastava, A., Skakalova, V., Kotakoski, J., Krasheninnikov, A. V., Hovden, R., Mao, Q., Meyer, J. C., Smet, J., Muller, D. A., & Kaiser, U. (2012). Direct Imaging of a Two-Dimensional Silica Glass on Graphene. Nano Letters, 12(2), 1081–1086. https://doi.org/10.1021/nl204423x
-
Huang, P. Y., Kurasch, S., Alden, J. S., Shekhawat, A., Alemi, A. A., McEuen, P. L., Sethna, J. P., Kaiser, U., & Muller, D. A. (2013). Imaging Atomic Rearrangements in Two-Dimensional Silica Glass: Watching Silica’s Dance. Science, 342(6155), 224–227. https://doi.org/10.1126/science.1242248
-
Kumar, A., Wilson, M., & Thorpe, M. F. (2012). Amorphous Graphene: A Realization of Zachariasen’s Glass. Journal of Physics: Condensed Matter, 24(48), 485003. https://doi.org/10.1088/0953-8984/24/48/485003
-
Sheng, H. W., Luo, W. K., Alamgir, F. M., Bai, J. M., & Ma, E. (2006). Atomic Packing and Short-to-Medium-Range Order in Metallic Glasses. Nature, 439, 419–425. https://doi.org/10.1038/nature04421
-
Gibson, J. M., Treacy, M. M. J., Sun, T., & Zaluzec, N. J. (2010). Substantial Crystalline Topology in Amorphous Silicon. Physical Review Letters, 105(12), 125504. https://doi.org/10.1103/PhysRevLett.105.125504
-
SALVE news (2010). Reduced Graphene Oxide. salve-project.de, March 3, 2010. https://salve-project.de/newspress/reduced-graphene-oxide.html
-
Turnbull, D., & Cohen, M. H. (1958). Concerning Reconstructive Transformation and Formation of Glass. Journal of Chemical Physics, 29(5), 1049–1054. https://doi.org/10.1063/1.1744654
-
Meyer, J. C., Eder, F., Kurasch, S., Skakalova, V., Kotakoski, J., Park, H. J., Roth, S., Chuvilin, A., Eyhusen, S., Benner, G., Krasheninnikov, A. V., & Kaiser, U. (2012). Accurate Measurement of Electron Beam Induced Displacement Cross Sections for Single-Layer Graphene. Physical Review Letters, 108(19), 196102. https://doi.org/10.1103/PhysRevLett.108.196102
-
Kotakoski, J., Meyer, J. C., Kurasch, S., Santos-Cottin, D., Kaiser, U., & Krasheninnikov, A. V. (2011). Stone-Wales-Type Transformations in Carbon Nanostructures Driven by Electron Irradiation. Physical Review B, 83(24), 245420. https://doi.org/10.1103/PhysRevB.83.245420
-
Crespi, V. H., Benedict, L. X., Cohen, M. L., & Louie, S. G. (1996). Prediction of a Pure-Carbon Planar Covalent Metal. Physical Review B, 53(20), R13303–R13305. https://doi.org/10.1103/PhysRevB.53.R13303
-
SALVE news (2011). From Graphene to 2D Amorphous Carbon. salve-project.de, March 9, 2011. https://salve-project.de/newspress/from-graphene-to-two-dimensional-amorphous-carbon.html
-
Lehtinen, O., Kurasch, S., Krasheninnikov, A. V., & Kaiser, U. (2013). Atomic Scale Study of the Life Cycle of a Dislocation in Graphene From Birth to Annihilation. Nature Communications, 4, 2098. https://doi.org/10.1038/ncomms3098
-
Westenfelder, B., Meyer, J. C., Biskupek, J., Kurasch, S., Scholz, F., Krill, C. E., & Kaiser, U. (2011). Transformations of Carbon Adsorbates on Graphene Substrates Under Extreme Heat. Nano Letters, 11(12), 5123–5127. https://doi.org/10.1021/nl203224z
-
Turchanin, A., Weber, D., Büenfeld, M., Kisielowski, C., Fistul, M. V., Efetov, K. B., Weimann, T., Stosch, R., Mayer, J., & Gölzhäuser, A. (2011). Conversion of Self-Assembled Monolayers Into Nanocrystalline Graphene: Structure and Electric Transport. ACS Nano, 5(5), 3896–3904. https://doi.org/10.1021/nn200297n
-
Huang, P. Y., Ruiz-Vargas, C. S., van der Zande, A. M., Whitney, W. S., Levendorf, M. P., Kevek, J. W., Garg, S., Alden, J. S., Hustedt, C. J., Zhu, Y., Park, J., McEuen, P. L., & Muller, D. A. (2011). Grains and Grain Boundaries in Single-Layer Graphene Atomic Patchwork Quilts. Nature, 469, 389–392. https://doi.org/10.1038/nature09718
-
Kim, K., Lee, Z., Regan, W., Kisielowski, C., Crommie, M. F., & Zettl, A. (2011). Grain Boundary Mapping in Polycrystalline Graphene. ACS Nano, 5(3), 2142–2146. https://doi.org/10.1021/nn1033423
-
Kurasch, S., Kotakoski, J., Lehtinen, O., Skákalová, V., Smet, J., Krill, C. E., Krasheninnikov, A. V., & Kaiser, U. (2012). Atom-by-Atom Observation of Grain Boundary Migration in Graphene. Nano Letters, 12(6), 3168–3173. https://doi.org/10.1021/nl301141g
-
Shackelford, J. F., & Brown, B. D. (1981). The Lognormal Distribution in the Random Network Structure. Journal of Non-Crystalline Solids, 44(2–3), 379–382. https://doi.org/10.1016/0022-3093(81)90040-5
-
Robertson, A. W., Allen, C. S., Wu, Y. A., He, K., Olivier, J., Neethling, J., Kirkland, A. I., & Warner, J. H. (2012). Spatial Control of Defect Creation in Graphene at the Nanoscale. Nature Communications, 3, 1144. https://doi.org/10.1038/ncomms2141
-
Torquato, S., & Stillinger, F. H. (2003). Local Density Fluctuations, Hyperuniformity, and Order Metrics. Physical Review E, 68(4), 041113. https://doi.org/10.1103/PhysRevE.68.041113
-
Kotakoski, J., Brand, C., Lilach, Y., Cheshnovsky, O., Mangler, C., Arndt, M., & Meyer, J. C. (2015). Toward Two-Dimensional All-Carbon Heterostructures via Ion Beam Patterning of Single-Layer Graphene. Nano Letters, 15(9), 5944–5949. https://doi.org/10.1021/acs.nanolett.5b02063






