First Observation of Self-Interstitials in Graphene

January 1, 2015 – Extra atoms built into a complete crystal lattice had never been imaged at atomic resolution. Researchers of Ute Kaiser's group at Ulm University have now done it: with an ordinary carbon coater they landed carbon atoms on graphene at about 1 eV, and 80 kV aberration-corrected HRTEM showed all three self-interstitial dimers predicted by theory, as well as larger aggregates and dislocation dipoles that bulge out of the sheet and are removed atom by atom by the electron beam.

Every real crystal holds defects, and they decide much of how it behaves. Atoms missing from the lattice are easy to see in two-dimensional materials and easy to make in an electron microscope, but their opposite, extra atoms squeezed into a complete lattice, had never been imaged at atomic resolution, neither in a three-dimensional nor in a two-dimensional material. Researchers of the Electron Microscopy of Materials Science group at Ulm University, led by Ute Kaiser, have now done it in graphene[1]. With an ordinary carbon coater, normally used to deposit conductive carbon films on non-conducting samples for electron microscopy, they landed extra carbon on single-layer graphene at energies of about 1 eV. Low-voltage aberration-corrected high-resolution transmission electron microscopy (AC-HRTEM) at 80 kV then showed all three self-interstitial dimer structures predicted by theory[2],[3], and in addition larger aggregates of extra atoms and dislocation dipoles. An independent study from the University of Oxford, with a JEOL JEM-2200MCO HRTEM also operated at 80 kV, confirms how dislocations buckle the sheet and how they vanish under the beam; it is compared in a validation box below.

Why defects matter so much for the properties of a material is explained in the box below.

▸ Background: why crystal defects matter

Challenges in creating and imaging implanted self-interstitials
Two-dimensional crystals have opened a completely new view of defects, for a simple reason: in a 2D material, no three-dimensional bulk hides the defects, since the surface is essentially all there is[1]. This has made atomically resolved studies possible of defects produced by ion[4] or electron irradiation[5], of grain boundaries[6], of dislocations[7],[8] and of foreign atoms that replace lattice atoms or are adsorbed on the lattice[9]. The implanted self-interstitial, an extra atom of the material's own kind built into its lattice, was missing from this list.

Besides AC-HRTEM, aberration-corrected scanning transmission electron microscopy (STEM) and scanning tunnelling microscopy (STM) can resolve the exact atomic structure of 2D materials[10]. A single carbon adatom sitting on top of a carbon–carbon bond of graphene, the so-called bridge position, had already been detected[11]. Structures resembling the interstitial dimers in graphene had been seen in incompletely crystallized hexagonal 2D silica, by STM and by AC-HRTEM[12],[13], and in graphene after extreme electron doses in a TEM[13],[14]. In those cases, however, the structures did not result from adding atoms: a large number of atoms had been removed around the defects and the whole lattice had been reordered, so the interstitial structures appeared at random among many other defects[1]. Isolated self-interstitial dimers had never been reported in untreated graphene, despite extensive atomic-scale studies of both exfoliated and chemically vapour-deposited (CVD) samples.

Several obstacles explain why. Missing atoms are easy to make in a TEM: at 80 kV and above, a single knock-on collision between an electron and a carbon atom is enough to remove the atom from graphene[5]. Extra atoms, by contrast, need an external source of new carbon. They can also be knocked out by the beam themselves, and more easily than lattice atoms, because the displacement threshold of atoms at defect sites tends to be lower than in pristine graphene[15]; the electron dose on the sample therefore has to be kept low. Finally, squeezing extra atoms into any crystal costs much energy because of the strain it causes[1].

Soft landing on graphene
Here graphene has an advantage: it can bend out of its plane, and because its bending modulus is small[16], extra atoms cost relatively little energy[1]. To find out whether carbon can be implanted at all, the team ran molecular dynamics (MD) simulations of single C atoms and C2 dimers landing on graphene. Below about 30 eV per atom, a large fraction of them is expected to bond to the sheet. A carbon coater delivers far less: a graphitic filament is heated to its sublimation temperature by a current pulse, and the emitted C atoms and C2, C3 and larger molecules land on the sample at thermal energies, typically below 1 eV per atom. With the coater run at the shortest settings, one to four pulses of 100–300 ms, submonolayer coverage was reached.

Single carbon adatoms are predicted to be mobile at room temperature[2] and are therefore not expected to remain where they landed[1]. Isolated dark spots seen after a short deposition matched the contrast of single carbon atoms, but they were far too stable; they are more likely immobile adsorbates such as CH3 molecules[17]. When two mobile adatoms meet, however, they can form a highly stable self-interstitial dimer[3], and such dimers are easy to identify: they are built into the sp2-bonded lattice without dangling bonds[2], and their pattern of pentagons, hexagons and heptagons is clearly visible in the images (Fig. 1).

All three predicted dimers
The samples contained the so-called inverse Stone–Thrower–Wales defect (Fig. 1a)[2] and both of its other two forms, which result from rotating one or two carbon–carbon bonds (Fig. 1b, c)[3]. The theoretical predictions of these defects in isolated form are thus confirmed, and these are the first atomic-resolution images of implanted self-interstitials in any material[1]. In the two rotated forms, it is impossible to say which two atoms are the extra ones. They can also be read as minuscule grain boundary loops with a density surplus[18], since a chain of pentagons and heptagons encloses one or two hexagons. In a 2D material, the line between adatoms and interstitials becomes fuzzy; the team calls single atoms sitting on top of the lattice adatoms, and atoms incorporated through perfect sp2 bonding self-interstitials[1].

Relaxed with a valence force field model in a cell of 20 002 atoms, all three dimers bulge strongly out of the plane, by 1.9–2.1 Å, over long distances; their formation energies are 5.8, 6.2 and 7.1 eV[1]. Under the beam, dimers switched back and forth between the forms of Fig. 1b and c, but never to or from the inverse Stone–Thrower–Wales form, and all of them eventually disappeared when the beam knocked out the extra atoms, after doses of the order of 109 electrons per nm2. How the images compare with image simulations, which also show why the raw images were corrected for a residual lens aberration, is shown in the box below (Fig. 2).

▸ Validation: HRTEM image vs. image simulation

Extra atoms cluster
Larger aggregates of extra atoms were found in the lattice as well, with four and with ten extra atoms (Fig. 3)[1]. Their models again bulge strongly out of the plane. The formation energies are 11.4 and 14.8 eV, or 5.7 and 3.0 eV per interstitial dimer, lower than for an isolated dimer: it pays for the extra atoms to agglomerate. These are not necessarily the best configurations for each number of atoms, but those actually observed. Contamination often stuck to the interstitial defects and was removed only after prolonged exposure to the beam, which revealed the sp2-bonded structures underneath. Molecules adsorbed on graphene thus seem to be drawn to the interstitials, which is promising for functionalizing graphene.

As with aggregates of vacancies[8], extra atoms can also rearrange the lattice into a dislocation dipole, with the extra atoms filling an additional lattice row between the two dislocation cores (Fig. 4)[1]. A first, rather disordered aggregate changed constantly during imaging, and dark high-contrast spots hinted at, for example, sp3-bonded carbon atoms, so its model is only an approximation; even so, it holds a density surplus. After a dose of 1.4 × 109 e/nm2, the atoms had arranged into a dipole with cores five lattice rows apart, which corresponds to ten extra atoms and a formation energy of 10.82 eV, or 2.16 eV per dimer. The defect then shed its surplus atoms step by step, shrank to a single interstitial dimer and finally vanished after a total dose of 6 × 109 e/nm2, leaving pristine graphene behind. How this compares with the missing-atom case is summarised in the box below.

▸ Comparison: missing atoms vs. extra atoms in graphene

A second graphene layer
There is a limit to how many interstitials can be packed into the sheet[1]. When the deposited carbon approached about one monolayer, the extra atoms no longer went into the graphene lattice but segregated into a new layer. High-magnification images showed that this layer is nanocrystalline graphene, seen as Moiré patterns on top of the first layer (Fig. 5a). Once the contribution of the first layer is removed by Fourier filtering[19], grains a few nanometres wide become visible, separated by continuous chains of pentagonal and heptagonal rings (Fig. 5b). The new layer was rather non-uniform, but it became more uniform during imaging, as the electron impacts helped the atoms over configuration barriers, much like annealing at high temperature. By careful tuning of the deposition, however, a density of extra carbon low enough for isolated point defects can be reached.

Theory put to the test
The experiment checks earlier calculations point by point, and not every prediction holds. The box below sets the observations against theory.

▸ Validation: experiment vs. theory

Dislocations in a second microscope
At the same time, dislocations in graphene were studied at the University of Oxford with another image-corrected microscope at the same voltage, a JEOL JEM-2200MCO HRTEM at 80 kV[20],[21]. The validation box below compares the two microscopes and shows the dislocations of both studies side by side.

▸ Validation: FEI Titan at 80 kV vs. JEOL JEM-2200MCO at 80 kV

How the samples were made and imaged is described in the box below.

▸ Recipe: implantation and imaging

Challenges named by the authors
The formation of the dimers could not be watched: all defects were already in place when the first images were taken, and the beam may have helped to turn less ordered structures into the final stable ones[1]. The approach gives no spatial control over where atoms land; precise engineering of graphene with interstitials will need more refined methods. Single carbon adatoms could not be identified, and the mobility of the larger carbon molecules that land on the sheet is unknown, since no predictions exist and simulating it was beyond the study. The beam that images the extra atoms also removes them, so the dose has to be kept low, and the model of the disordered aggregate in Fig. 4a is only an approximation, because the structure kept changing and sp3-bonded atoms are hard to interpret in a projected image.

Outlook
With a simple carbon coater and 80 kV AC-HRTEM, a crystal was pushed into a state of supersaturation and studied atom by atom for the first time[1]. All self-interstitial structures predicted for graphene were confirmed, and larger aggregates and dislocation dipoles turned out to be favourable arrangements for the extra atoms. All of them buckle strongly out of the plane, and as the authors write, "such blister-like structures can be expected to have higher reactivity than pristine graphene, which can be advantageous for functionalization of graphene." Defects with surplus carbon atoms are also predicted to have unusual electronic and magnetic properties[2]. The study shows that such structures can be made, and it opens a route to tailoring graphene and other 2D materials by implanting self-interstitials into their lattice.

Resource:
Lehtinen, O., Vats, N., Algara-Siller, G., Knyrim, P., Kaiser, U. (2015).
Implantation and atomic-scale investigation of self-interstitials in graphene.
Nano Lett. 15(1), 235–241.
https://doi.org/10.1021/nl503453u

Update – September 25, 2018

The news article of January 1, 2015 ends above. Results published after it that test its findings are listed below.

Bulges measured in 3D
A projected image cannot show how far a defect rises out of the sheet; in the 2015 study the bulges came from force-field calculations[1]. In 2018, a group at the University of Vienna measured such bulges directly[22]. With a Nion UltraSTEM 100 at 60 kV, they recorded two MAADF images of the same defect, one at normal incidence and one with the sample tilted by about 20° (Fig. 7a, b). An algorithm then moved the atoms of a 3D model and adjusted the imaging parameters until simulated images matched both experimental images. Tested on simulated data with realistic noise, the method placed the atoms out of the plane to within 0.18 Å on average[22].

The first defect reconstructed this way holds two extra carbon atoms: a rotated grain with an ad-dimer built into the lattice, a close relative of the interstitial structures seen in Ulm (Fig. 7c, d)[22]. In the measured structure the sheet rises by about 4 Å at the defect, and the two atoms furthest from the plane are those of the ad-dimer. An energy-minimized model of the same defect has the same shape. The single dimers in Ulm were computed to rise by 1.9–2.1 Å[1]; the Vienna defect is larger. A flower defect without extra atoms, in contrast, stayed flat in the measurement and in the calculation alike. The Vienna paper names the Ulm study among those in which such out-of-plane distortions had only been inferred from HRTEM images[22].

Grain boundaries were measured in the same way. Boundaries with small misorientation, made of nearly isolated dislocation cores, lift the sheet by up to about 4 Å, with a kink of almost 40° at 2.6°, while boundaries near 30° stay almost flat[22]. This supports the picture of the Ulm dislocation dipole bulging out of the plane (Fig. 4)[1] and the hillocks measured with GPA in Oxford[20]. None of the authors of the Vienna paper took part in the Ulm study.

Download this news article as PDF – for reading on paper, taking notes or filing.

References
  1. Lehtinen, O., Vats, N., Algara-Siller, G., Knyrim, P., & Kaiser, U. (2015). Implantation and Atomic-Scale Investigation of Self-Interstitials in Graphene. Nano Letters, 15(1), 235–241. https://doi.org/10.1021/nl503453u

  2. Lusk, M. T., Wu, D. T., & Carr, L. D. (2010). Graphene Nanoengineering and the Inverse Stone-Thrower-Wales Defect. Physical Review B, 81(15), 155444. https://doi.org/10.1103/PhysRevB.81.155444

  3. Lusk, M. T., & Carr, L. D. (2008). Nanoengineering Defect Structures on Graphene. Physical Review Letters, 100(17), 175503. https://doi.org/10.1103/PhysRevLett.100.175503

  4. Standop, S., Lehtinen, O., Herbig, C., Lewes-Malandrakis, G., Craes, F., Kotakoski, J., Michely, T., Krasheninnikov, A. V., & Busse, C. (2013). Ion Impacts on graphene/Ir(111): Interface Channeling, Vacancy Funnels, and a Nanomesh. Nano Letters, 13(5), 1948–1955. https://doi.org/10.1021/nl304659n

  5. Meyer, J. C., Eder, F., Kurasch, S., Skákalová, 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

  6. Yazyev, O. V., & Louie, S. G. (2010). Topological Defects in Graphene: Dislocations and Grain Boundaries. Physical Review B, 81(19), 195420. https://doi.org/10.1103/PhysRevB.81.195420

  7. Warner, J. H., Margine, E. R., Mukai, M., Robertson, A. W., Giustino, F., & Kirkland, A. I. (2012). Dislocation-Driven Deformations in Graphene. Science, 337(6091), 209–212. https://doi.org/10.1126/science.1217529

  8. 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

  9. Bangert, U., Pierce, W., Kepaptsoglou, D. M., Ramasse, Q., Zan, R., Gass, M. H., Van den Berg, J. A., Boothroyd, C. B., Amani, J., & Hofsäss, H. (2013). Ion Implantation of Graphene – Toward IC Compatible Technologies. Nano Letters, 13(10), 4902–4907. https://doi.org/10.1021/nl402812y

  10. N’Diaye, A. T., Coraux, J., Plasa, T. N., Busse, C., & Michely, T. (2008). Structure of Epitaxial Graphene on Ir(111). New Journal of Physics, 10(4), 043033. https://doi.org/10.1088/1367-2630/10/4/043033

  11. Meyer, J. C., Girit, C. O., Crommie, M. F., & Zettl, A. (2008). Imaging and Dynamics of Light Atoms and Molecules on Graphene. Nature, 454(7202), 319–322. https://doi.org/10.1038/nature07094

  12. Yang, B., Boscoboinik, J. A., Yu, X., Shaikhutdinov, S., & Freund, H.-J. (2013). Patterned Defect Structures Predicted for Graphene Are Observed on Single-Layer Silica Films. Nano Letters, 13(9), 4422–4427. https://doi.org/10.1021/nl402264k

  13. Björkman, T., Kurasch, S., Lehtinen, O., Kotakoski, J., Yazyev, O. V., Srivastava, A., Skákalová, V., Smet, J. H., Kaiser, U., & Krasheninnikov, A. V. (2013). Defects in Bilayer Silica and Graphene: Common Trends in Diverse Hexagonal Two-Dimensional Systems. Scientific Reports, 3, 3482. https://doi.org/10.1038/srep03482

  14. Robertson, A. W., He, K., Kirkland, A. I., & Warner, J. H. (2014). Inflating Graphene With Atomic Scale Blisters. Nano Letters, 14(2), 908–914. https://doi.org/10.1021/nl404266k

  15. Susi, T., Kotakoski, J., Arenal, R., Kurasch, S., Jiang, H., Skákalová, V., Stephan, O., Krasheninnikov, A. V., Kauppinen, E. I., Kaiser, U., & Meyer, J. C. (2012). Atomistic Description of Electron Beam Damage in Nitrogen-Doped Graphene and Single-Walled Carbon Nanotubes. ACS Nano, 6(10), 8837–8846. https://doi.org/10.1021/nn303944f

  16. Yakobson, B. I., Brabec, C. J., & Bernholc, J. (1996). Nanomechanics of Carbon Tubes: Instabilities Beyond Linear Response. Physical Review Letters, 76(14), 2511–2514. https://doi.org/10.1103/PhysRevLett.76.2511

  17. Erni, R., Rossell, M. D., Nguyen, M.-T., Blankenburg, S., Passerone, D., Hartel, P., Alem, N., Erickson, K., Gannett, W., & Zettl, A. (2010). Stability and Dynamics of Small Molecules Trapped on Graphene. Physical Review B, 82(16), 165443. https://doi.org/10.1103/PhysRevB.82.165443

  18. Cockayne, E., Rutter, G. M., Guisinger, N. P., Crain, J. N., First, P. N., & Stroscio, J. A. (2011). Grain Boundary Loops in Graphene. Physical Review B, 83(19), 195425. https://doi.org/10.1103/PhysRevB.83.195425

  19. Lee, Z., Jeon, K.-J., Dato, A., Erni, R., Richardson, T. J., Frenklach, M., & Radmilovic, V. (2009). Direct Imaging of Soft–Hard Interfaces Enabled by Graphene. Nano Letters, 9(9), 3365–3369. https://doi.org/10.1021/nl901664k

  20. Warner, J. H., Fan, Y., Robertson, A. W., He, K., Yoon, E., & Lee, G.-D. (2013). Rippling Graphene at the Nanoscale Through Dislocation Addition. Nano Letters, 13(10), 4937–4944. https://doi.org/10.1021/nl402902q

  21. Gong, C., He, K., Robertson, A. W., Yoon, E., Lee, G.-D., & Warner, J. H. (2015). Spatially Dependent Lattice Deformations for Dislocations at the Edges of Graphene. ACS Nano, 9(1), 656–662. https://doi.org/10.1021/nn505996c

  22. Hofer, C., Kramberger, C., Ahmadpour Monazam, M. R., Mangler, C., Mittelberger, A., Argentero, G., Kotakoski, J., & Meyer, J. C. (2018). Revealing the 3D Structure of Graphene Defects. 2D Materials, 5(4), 045029. https://doi.org/10.1088/2053-1583/aaded7