Evidence for a New Two‐Dimensional C4H‐Type Polymer Based on Hydrogenated Graphene

A new polymer with C4H stoichiometry based on graphene is synthesized in situ using template-induced polymerization of self-organizing hydrogen adsorbates on graphene. The polymerization is observed "live" on the surface of graphene by photoemission spectroscopy. Photoemission spectroscopy allows for an accurate determination of the carbon/hydrogen stoichiometry, an aspect that is extremely important for understanding functionalized graphene. Carbon–hydrogen bonds are fundamental in chemistry and can be found as classical and non-classical bond configurations, yielding a manifold of molecular species and hydrogenated forms of carbon. Modifications of the CxHy stoichiometry result in different molecular systems with a significant change in structure and geometry as well as entirely different physical and chemical properties. Linking the individual CxHy units with covalent bonds, i.e., polymerization, allows for adopting this wide spectrum of properties to a 1D to 3D bulk material.1 Polymerization can be performed as a bottom up approach (chemically linking together individual CxHy units) or by functionalization of a template using self-organizing adsorbates. Graphene2, 3 is an ideal template system for the latter type of experiment because it can be grown in large areas by chemical vapor deposition on metals4, 5 and by Si sublimation on SiC6 and can be easily functionalized covalently.7-12 Fully hydrogenated graphene, also known as graphane,13 has carbon–hydrogen bonds directed upwards and downwards in the sp3 type carbon atom framework. Electronically, graphene is a zero-gap semiconductor and graphane is an insulator with an energy gap of 3.5 eV.13 The situation for hydrogenation of epitaxial graphene is different because substrate interactions prevent direct formation of graphane type structures and thus reduce the highest achievable hydrogen uptake. Partially hydrogenated graphene7, 8 has peculiar electronic9, 12, 14-17 and magnetic18, 19 properties that are strongly affected by the chemisorption pattern.20 For hydrogenated graphene, theory predicts electron localization,21 a peculiar midgap state,22, 23 strong excitonic effects,24 and high temperature superconductivity in hole-doped graphane25 but it is not yet clear whether any of these effects have actually been observed experimentally. The hydrogen-to-carbon ratio is crucial and so far it was estimated using scanning tunneling microscopy and electron energy loss spectroscopy.26-28 Therefore most transport and optical spectroscopy experiments to date are carried out without any knowledge of the H/C ratio (stoichiometry). For hydrogenated bilayer graphene the stability of a para-type chemisorption (the H atom pairs are located on opposite edges of the hexagon) was inferred on the basis of geometry optimizations.29 Much less is known about experiments and simulations of the hydrogenation kinetics of a quasi-free-standing graphene monolayer and on the existence of stable phases with certain chemisorption patterns and their electronic properties. In this work we provide the experimental proof for the existence of a new and stable C4H phase. We also provide strong evidence for a para-type chemisorption pattern for this phase. This new phase is characterized by the presence of aromatic rings arranged in a 2 × 2 super structure, which enhances its chemical stability. As a spectroscopic tool we employ photoemission, which allows for consistent determination of the hydrogen quantity chemisorbed on graphene. We further unravel the elementary processes involved in the surface reactions by using chemisorption models and non-equilibrium quantum chemical molecular dynamics (QM/MD) simulations. The photoemission experiments are corroborated by scanning tunneling microscopy (STM) studies of hydrogen chemisorption on graphene confirming the dosage dependence of the H-coverage. STM provides further insight into the hydrogen arrangement, which lacks long-range order for small H/C ratios. Pristine graphene samples were prepared in situ under ultrahigh vacuum (UHV) conditions by chemical vapor deposition (CVD) on Ni(111) thin films.30, 31 This growth method works for a relatively low temperature window from ≈500–620 °C31 as compared to growth by carbon precipitation, which uses 900 °C to dissolve carbon into the nickel.32 We thus conclude that these "soft" CVD conditions dissolve less carbon into the nickel film. The absence of dissolved carbon into the nickel is corroborated by the appearance of a single peak in the core-level photoemission spectra of carbon as we will show later. After CVD, we performed Au intercalation into the graphene/Ni interface in order to render the graphene layer quasi-free-standing.9, 33, 34 Hydrogenation of graphene/Au was performed by exposing graphene to a beam of atomic H that was produced by cracking H2 at ≈3000 K in a W capillary.9 The cracking efficiency of the H2 molecule at these parameters is close to unity. The hydrogen partial pressures during hydrogenation were in the range of 1 × 10−9 mbar to 1 × 10−7 mbar. The hydrogenation time was varied from a few seconds to 10 mins. During in situ hydrogenation, the sample was kept at room temperature for all microscopy and spectroscopy experiments. STM experiments were conducted using a commercial Omicron VT Multiscan STM, with all measurements carried out under UHV conditions at room temperature. W tips prepared by electrochemical etching immediately before introducing them into the UHV chamber were used for imaging and the STM images were processed using WSxM.35 Typical scanning conditions used with the sample were 1 nA for the tunneling current at a positive 0.02 V sample bias. STM allows for direct imaging of the C atoms in the graphene layers and the hydrogen coverage on the graphene surface.12, 36 Furthermore, we performed X-ray photoelectron spectroscopy (XPS) of the C 1s core-levels using synchrotron radiation. These measurements were carried out at the BESSY II synchrotron (Berlin, Germany) using the German–Russian beamline with the HIRES end station. STM and XPS experiments were performed using the same parameters for the hydrogenation procedure, therefore the morphology and electronic properties can be related to each other for a given hydrogenation time. In order to study the hydrogenation kinetics, a phenomenological kinetic chemisorption model calculation and QM/MD simulations of the hydrogenation process were performed. We have fitted the H adsorption and desorption probabilities to the experimental data using a kinetic model. Quantum chemical MD simulations are based on an approximate density functional theory (DFT) method calculating the spin-polarized self-consistent-charge density-functional tight-binding potential.37-39 Therefore they provide an a priori description of the hydrogenation kinetics and the information about the preferred chemisorption patterns. Further details on the MD simulations can be found in the Supporting Information. Finally, we calculated the electronic band structure and the density of states of C4H using DFT. Figure 1a–e shows STM images of pristine and hydrogenated graphene at increasing hydrogen exposure times, as indicated on each panel. The atomically resolved image recorded on pristine graphene is shown in Figure 1a and a larger scale image highlighting the characteristic Moiré super lattice with a period of 2.6 nm, roughly consistent with the super cell of Au on the Ni(111) surface9 is presented in Figure 1b. The Moiré pattern appears as a result of lattice mismatch when superimposing the graphene and the Au lattices. These STM images are important as they prove that no Au islands remain on the surface after the intercalation process. This is crucial since Au islands might block the H chemisorption and lead to a wrong determination of the hydrogen storage capacity of graphene. With proceeding hydrogenation, bright protrusions appear on the graphene surface, initially spot-like features commensurate with low hydrogen exposure (Figure 1c), followed by the formation of filamentary structures at intermediate hydrogen exposure (Figure 1c,d). At all outlined exposure times the hydrogen chemisorption sites appear to be distributed randomly on the scan sizes investigated in this study. This random chemisorption pattern for low H coverages has important implications for the mechanism of bandgap opening in the electronic structure of hydrogenated graphene. From an electronic point of view, hydrogen adsorption introduces a local sp3 character, reducing the number of π electrons and therefore the carrier concentration at the Fermi level in graphene.40 For the complete surface passivation after increasing the hydrogen exposure to 600 s (Figure 1e), the hydrogenated graphene layer has the lowest carrier concentration. As we will show by photoemission, this hydrogenation time corresponds to an H/C ratio of 25%. The simulations of the hydrogenation procedure predict that para-type chemisorption is the dominant pattern for this exposure time. Such a C4H compound with a para-type H chemisorption is a wide bandgap insulator. The insulating behavior and the coexistence of various orientations in the chemisorption make it difficult to retain the good spatial resolution that we showed for low H/C ratios. STM therefore proves that all Au is intercalated in between the graphene and nickel film and it provides access to the surface morphology changes of graphene during hydrogenation. In particular, it proves that H atoms chemisorb and there is no preferred pattern for low exposures. A quantitative determination of the H/C atomic ratio and chemisorption pattern remains difficult at the scan scales presented in this local study. We therefore employ photoemission as described in the next section, which is capable of probing larger areas on a sample. a) STM topography images of pristine graphene/Au with atomic resolution. The images of the hydrogenation series (at 10−7 mbar partial pressure) on a larger scale are shown in (b–e) with the corresponding hydrogenation time. The hydrogen exposures in (b–e) were O L, 0.75 L, 3.0 L, and 45 Langmuir. The chemisorption of hydrogen changes the appearance of the surface and is clearly observable as bright spots. C 1s core-level spectra of graphene/Au at different hydrogen coverages up to maximum H/C ratio of about 25% (this ratio corresponds to C4H). All C 1s features were fitted with Doniach–Sunjic line shapes with components C1 and C2 representing unhydrogenated C atoms of graphene/Au. C1 has no neighboring C–H bond and C2 can have a neighbor C–H bond. The binding energies are given relative to the C1 component with values of 284.2 eV for pristine graphene/Au. From the integrated area of the C3 component, we directly determine the hydrogen coverage for every functionalization step according to Equation 1. C1, C2, and C3 denote the areas under the relevant constituents of the C 1s core-level spectrum shown in Figure 2. We find a maximum coverage of about 25%, which is in accordance to model calculations for single-sided hydrogenation.42 The peaks C1 and C2 are very close to each other and can exchange spectral weight in the fit. This does not affect the determination of η as the C3 peak is well separated and both C1 and C2 correspond to unhydrogenated C atoms. a) The H/C ratio determined by XPS (points) along with the calculated chemisorption curve from Equation 3. The lower scale is the hydrogenation time (normalized for 10−7 mbar partial pressure) and the upper scale the exposure in Langmuir. b–d) MD simulations: b) time dependence of the H/C ratio up to equilibrium concentration, c) ratio of H chemisorption, reflection, and H2 formation as a function of H/C stoichiometry, and d) the final snapshot (top and side view) of the simulated hydrogenation of graphene after 200 ps simulation time. In order to determine Pads it is necessary to estimate the hydrogen flux I on the sample. With the specifications of the turbomolecular pump (pumping speed) and the hydrogenation pressure we can estimate the H-molecule flux in the chamber using the ideal gas law. With a cracking efficiency of ≈1 and using the solid angles of H-source and sample, the H-atom flux on the sample is found to be Hatoms per second per C-atom. From Equation 4 and the above-mentioned relations between the probabilities we obtain on average , , and . The dominant role of H reflection is also found independently from MD simulations as is shown below. An atomistic description of the hydrogenation process yields a completely parameter-free solution of the H/C ratio. It furthermore provides information about the H chemisorption sites. This is in particular interesting because the electronic and magnetic properties of hydrogenated graphene depend sensitively on the chemisorption pattern. Their reliability and robustness can be verified by comparing the simulated maximum H/C ratio to the experimental one. The time dependence of the MD simulation is given in Figure 3b and the obtained maximum coverage of 25% agrees to the experiment. For the simulation we have used an incident energy corresponding to the experimental value of the kinetic energy, Ekin, of the H atoms at the cracking temperature T. The average kinetic energy of an H atom in the experiment is given by , where kB is the Boltzmann constant and T ≈ 3000 K. The simulation provides insight into the constituent processes, i.e., H reflection and H2 formation followed by desorption, which are depicted in Figure 3c. Clearly, the H reflection is the dominant process for this incident energy of 400 meV. QM/MD simulations provide further evidence for the self-organization of the H chemisorption pattern of the C4H-type graphene (final snapshots of the simulation). One example is shown in Figure 3d and others are shown in the Supporting Information. Here, all H atoms are arranged in a para-type chemisorption pattern and the resulting stoichiometry of this perfect para-type graphane is C4H. This pattern is dominant over other H arrangements. By averaging over all simulated trajectories, we can find for ortho:meta:para chemisorbed pairs of H atoms the ratio 0.14:0.24:0.62 (see Supporting Information). This distribution is also evident from the XPS spectrum shown in Figure 2 (a pure para-type C4H layer would have zero C1 component). From Figure 3d it is also obvious that hydrogenation leads to ripple formations due to distortions in the C- lattice. Finally, we have further investigated the electronic properties of this compound using DFT calculations. In Figure 4a,b we depict the electronic energy bands and the density of electronic states showing that C4H has a wide bandgap of 3.5 eV. In Figure 4c we show an image of the perfect C4H sheet with the unit cell that has 8 C atoms and 2 H atoms. It can be seen that isolated aromatic rings are separated from each other by H atoms. Aromaticity is not only of fundamental importance for the chemical stability of C4H but also regarding the synthesis of modified graphene-based systems from molecular precursors. The aromatic rings are arranged in a 2 × 2 super structure with respect to the graphene sheet. Interestingly, this super structure is also found in stage 1 graphite intercalation compounds, where the graphene lattice remains almost undistorted due to its ionic character in contrast to the present covalently bonded compound. The appearance of a bandgap is well known in hydrogenated graphene under the same conditions9, 43 and confirms our calculations. a) Calculated electronic energy band structure and b) density of states (DOS) (per eV per atom) for C4H. The DOS of graphene is also shown in (b) for comparison. c) Model of the C4H sheet with the isolated aromatic rings indicated. The vectors a and b define the unit cell of C4H and connect the aromatic rings. They define a 2 × 2 super structure with respect to pristine graphene. In conclusion, we have proven by XPS determination of the H/C stoichiometry that a new and stable C4H phase of hydrogenated graphene exists, which provides an upper limit to storage of H on graphene under the applied conditions. This quantitative determination of H/C ratios is complemented by an STM investigation of the surface morphology changes upon hydrogenation in which the presence of hydrogen is clearly visible as bright spots. For quantitatively low levels of H chemisorption where random bond formations predominate, these bright spots are distributed on the surface of graphene with no order or self assembly. The correlation of exposure time with the measured H/C ratio yields the hydrogenation kinetics for which we have developed the chemisorption model. In addition we have performed MD simulations and obtained an evolution of the H-coverage in excellent agreement to the experiment. Chemisorption can occur in principle at random positions. However, during prolonged H bombardment, a para-adsorption pattern emerges as a consequence of repeated surface reorganization due to associative H2 elimination. Since hydrogenation of graphene has been carried out on a substrate with a directed H beam, the stereochemistry implies a single-sided hydrogenation process, where all para-type positions are hydrogenated on the same surface site. This is different from a free standing graphene layer that is exposed to an H plasma where the hydrogenation can take place on either side.8 This observation compares well to partially fluorinated graphene of a gross stoichiometry C4F, where aromatic domains are also assumed to be present.10 The maximum coverage of 25% has to be compared to the maximum hydrogen to carbon atomic ratio of other forms of hydrogenated carbon such as nanotubes,44 graphite,45 or strongly interacting graphene on transition metal substrates.16 Our results agree very well with the nanotube and graphite cases. The differences in maximum hydrogen chemisorption of graphene on transition metal substrates may be explained by the corrugation of graphene on these substrates which would help the chemisorption process. We also report electronic structure calculations of C4H indicating that it is a wide bandgap semiconductor making it an attractive material for optoelectronics in the UV range. Moreover, the aromatic domains that are encircled by sp3 units may serve as a nanotemplate in a similar manner for smaller molecules comparable to the concave areas in the boron nitride nanomesh.46 The precise analysis of H/C stoichiometries by XPS is an important step towards functional materials based on hydrogenated graphene. Analysis of the C/H ratio by core-level photoemission will prove particularly useful in combination with other spectroscopic methods such as resonance Raman, optical absorption, and electron spin resonance. A.V.F., X.L., and A.G. acknowledge EU support through ELISA for the stay at BESSY. D.V. acknowledges DFG Grant No. VY 64/1-1. D.H., A.G. and M.K. acknowledge DFG Grant No. GR 3708/1-1. A.G. acknowledges an APART fellowship from the Austrian Academy of Sciences. Y.W. and S.I. acknowledge helpful discussions with Hiroaki Nakamura and Atsushi Ito of the National Institute of Fusion Science in Toki, Japan. The MD simulations were supported by a CREST (Core Research for Evolutional Science and Technology) grant in the Area of High Performance Computing for Multiscale and Multiphysics Phenomena from the Japan Science and Technology Agency (JST). Y.W. kindly acknowledges Hujun Qian for his assistance with the MD simulations. D.U. and A.F. acknowledge support from RFBR and SPbSU grants. A.G. and D.H. acknowledge R. Hbel and S. Leger for technical assistance. Detailed facts of importance to specialist readers are published as "Supporting Information". Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.