www.sciencemag.org/cgi/content/full/337/6096/825/DC1 Supplementary Materials for Long-Range Ordered Carbon Clusters: A Crystalline Material with Amorphous Building Blocks Lin Wang,* Bingbing Liu, Hui Li, Wenge Yang, Yang Ding, Stanislav V. Sinogeikin, Yue Meng, Zhenxian Liu, Xiao Cheng Zeng, Wendy L. Mao *To whom correspondence should be addressed. E-mail: lwang@ciw.edu Published 17 August 2012, Science 337, 825 (2012) DOI: 10.1126/science.1220522 This PDF file includes: Materials and Methods SupplementaryText Figs. S1 to S11 References Materials and Methods The starting material was synthesized by a simple evaporation method (18). The in situ high-pressure XRD was conducted at the High Pressure Collaborative Access Team (HPCAT) and GSECARS sectors of the Advanced Photon Source (APS), Argonne National Laboratory (ANL). Focused monochromatic X-rays with beam sizes of 7 µm FWHM were utilized for angle-dispersive XRD experiments. The wavelengths of the Xrays are 0.4151 and 0.5166 วบ for in-situ high pressure measurements and recovered samples, respectively. The diffraction patterns were collected on an MAR345 image plate. Pressure was determined from the equation of state of Au which was added as an internal standard. Samples recovered from different pressures were also characterized using a Raman spectrometer (Renishaw inVia, UK) with 514.5 nm excitation laser. The inelastic X-ray scattering (IXS) experiments were performed at station 16ID-D of the HPCAT at APS of ANL. Incident monochromatic X-rays were focused to 10 μm by 40 μm (vertical by horizontal) (FWHM) at the sample location. A 50 μm pinhole was used to cut the tail of the X-ray beam at 50 mm upstream from the focus point. A panoramic type diamond anvil cell with a culet size of 300 μm was used to generate high pressure for the sample. A composite c-BN gasket assembly was used for the IXS measurements to increase the gap size between two diamond anvils and isolate the sample from the surface of the diamonds. A drilled Be gasket served as the outer part of the gasket to hold the c-BN insert. The gap size is ~80 μm at 42 GPa. The X-ray beam was directed into the gasket and was aligned parallel to the gap between the diamond anvils. The beam was at least 20 μm away from the surface of each anvil. The IXS signal was collected by scanning the incident x-ray energy relative to the fixed energy of 9.885 keV which was set for each of the 17 spherically bent Si (555) analyzers, which were mounted vertically in a Rowland circle to refocus the IXS signal to the Si detector (Amptek R®) in a back scattering geometry. The resolution for the measurements was 1 eV. The infrared absorption spectrum for recovered sample was performed at station U2A at National Synchrotron Light Source (NSLS) of Brookhaven National Laboratory (BNL). Type II diamond anvils were used for the measurement. The background was carefully subtracted by taking a background at the area without sample. Supplementary Text Detailed Information of Simulations 1.Comparison of Lattice Constants from Experiment and Quantum Molecular Dynamics Simulation for the Purpose of Calibration. First, we used a classic molecular dynamics (MD) simulation to produce a thermally equilibrated hexagonal-closed packing (hcp) crystalline system of mixed C60 and m-xylene (C8H10) (based on atomic coordination published in ref. 23) for subsequent quantum MD simulations. The ratio of C60 to m-xylene molecules is 1:1. A unit cell of the hcp crystal contains six C60 molecules and six m-xylene molecules. The m-xylene molecules are filled in the cavity region among the close-packed C60 molecules. The supercell of the simulation system has 8 unit cells, which contains 3264 C atoms and 480 H atoms. The consistent valence force field (CVFF) implemented in the Material Studio 2 4.4 package is employed. The MD is performed in an isothermal-isobaric (NPT) ensemble simulation (using the Discover program in Material Studio 4.4). The simulation time is 2 ns with a time step of 1.0 fs. Next, the equilibrium structure (as shown in Fig. S4) is used as the initial crystalline structure for the Born-Oppenheim molecular dynamics (BOMD) simulation. The initial pressure of the system is computed from BOMD simulation in the constant-volume and constant-temperature ensemble (NVT). The density function theory (DFT) in the PerdewBurke-Ernzerhof (25) scheme is chosen with unrestricted wave-function for the BOMD simulation. The core electrons of carbon are described by the Goedecker-Teter-Hutter (26) norm-conserving pseudopotential. The basis sets are a combination of the polarized double-ζ quality Gaussian basis and a plane-wave basis set (with an energy cutoff of 280 Ry). The total simulation time for equilibration is 2.0 ps with the time step of 1.0 fs. All the BOMD simulations are performed using the QUICKSTEP program in CP2K software package (27,28). The supercell contains six C60 and six m-xylene molecules (408 C atoms and 60 H atoms). The BOMD simulation shows that the initial system (a = 23.8 Å) is in slight compression state with a compression pressure of 1.85 GPa. The lattice constant (a = 23.8 Å) of the hcp C60*m-xylene system is slightly larger than experimental lattice constant a = 23.06 Å at the low pressure 1.7 GPa. The computed lattice constant is slightly larger due to the neglect of dispersion interaction in the BOMD simulation. However, we expect that at high-pressure (strong compression) states, the dispersion interaction becomes negligible. To prove this point, we performed two additional BOMD simulations (~1 ps each) to examine the hcp crystalline structure of C60*m-xylene system at ~5 and ~10 GPa, respectively (NPT simulations; see Fig. S5). The computed average lattice constant of a is 23.25 and 22.54 Å, respectively, whereas the experimental values of a = 22.60 and 22.20 Å at 5.6 and 10.8 GPa, respectively. Hence, with increasing the pressure, the computed lattice constant from BOMD simulation becomes closer to the experimental lattice constant. This benchmark calibration suggests that the simulated compression in the pressure range of 20 – 40 GPa is expected to mimic, semi-quantitatively, the experimental compression over the same pressure range. 2. Simulation of Compression and Decompression using a Similar Close-Packed System To demonstrate that the experimentally observed long-range ordered amorphous carbon cluster is a generic phenomenon, we performed independent BOMD simulations using the similar close-packed (face-centered cubic, fcc) system as a test model system. Note that both hcp and fcc systems are close-packed whereas the a fcc unit-cell demands a much smaller number of atoms (see below), which is feasible with our computational resources. Our BOMD simulations indicate that many physical/chemical behaviors observed in the experiment can be qualitatively captured (or even semi-quantitatively reproduced in terms of transition pressure; see below), thereby suggesting the appearance of long-range ordered amorphous carbon cluster is indeed a general phenomenon for the compressed C60*m-xylene system. Specifically, the spin-unrestricted DFT method is used in the BOMD simulation to account for the bond breaking and forming process under high pressures. Here, the supercell contains four C60 and four m-xylene molecules (272 C atoms and 40 H atoms). 3 900,000 CPU hours granted by DOE Oak Ridge National Laboratory Computing Facility were spent on this series of simulations. (2.1) Initial Crystalline Structure Again, we use the classical MD to acquire a thermally equilibrated fcc system in the NPT ensemble (290 K, 1 atm). Based on the final configuration obtained from the classical MD simulation, a BOMD simulation in the NVT ensemble is performed to compute the initial pressure at 290 K, which is about 0.4 GPa. Thus, the fcc system is also in the slight compression state as in the case of hcp system. (2.2) BOMD Simulation of Instant Compression and Decompression (Fig. S6, Fig. S7) (2.3) Temperature Profile in the Course of Compression and Decompression Simulations As shown in Fig. S8, in the BOMD simulations of instant compression, the temperature of the system increases typically very fast within the initial 0.5 ps, especially at high pressures. As a result, some chemical bonds of m-xylene molecules appear to be broken (see Fig. S6). This bond-breaking is not observed in the cold-compression experiment because in experiments the pressure is normally increased in a step-by-step fashion (a few GPa per step). (2.4) Vibrational Spectra of the System after Decompression Fourier transformation of the velocity auto-correlation functions obtained from the BOMD simulation gives rise to the vibration spectra at zero-pressure state after decompression from the six high-pressure states, as shown in Fig. S9. It can be seen that the characteristic peaks of C60 molecule in the range of 1200-1600 cm-1 of the vibration spectra are quite high if decompressed from high-pressure state at 20 or 30 GPa. These peaks become much weaker if decompressed from high-pressure state at or greater than 32 GPa. This behavior is consistent with that shown in the experimental Raman spectra, in that C60 molecules are still severely broken even after decompressed to zero-pressure state from the high-pressure states (with pressure > 32 GPa). (2.5) Pair Distribution Functions The pair distribution functions (PDFs) of carbon atoms at six high-pressure states are computed from the BOMD simulations (Fig. S10(A)). As shown in Fig. S10(A), the first peak (1.4 Å) and the second peak (2.5 Å) correspond to C-C bond length and C-C-C length with a bent angle of 120º, respectively, indicating that the 6-member rings are well retained even at very high pressures. The third and fourth peaks are not significant in PDFs at pressure higher than 32 GPa, implying that the C60 molecules are crushed. Moreover, the PDFs at zero-pressure states (decompressed from the six high-pressure states) are computed (Fig. S10(B)). By comparing Figure S10(A) and (B), it can be seen that the compressed C60 system is fairly elastic when compressed to a pressure lower than 30 GPa. However, when the system is compressed beyond 32 GPa, the deformation of C60 molecules is irreversible, evidenced by high similarity between PDFs corresponding to the same high-pressure state. From the enlarged plots of PDFs from 1.0 to 2.0 Å (Fig. S10(C) and (D)), it can be seen that the C-C length shifts slightly to the right with increasing the pressure, suggesting a transition of sp2 carbon to sp3 carbon at higher pressures. In addition, the radial distribution functions (RDFs) with respective to the center of C60 molecules are shown in Fig. S10(E) and (F). The single sharp peak in C60-centered 4 RDFs (at compression state) further proves the compression up to 30 GPa is quite reversible. The position is exactly the same as the radius of fullerene (3.6Å). Significant deformation of C60 molecules can be seen at the pressure higher than 32 GPa, as sharp peaks corresponding to the C60 shell in the RDFs become more broad and lower. The RDFs corresponding to 40 GPa exhibits very different features from the others, indicating that the symmetric structure of C60 molecules is largely destroyed due to the high pressure, forming the amorphous carbon. 5 Fig. S1. IXS spectra of the solvated C60 sample at high pressure. The π* and σ* peaks correspond to 1s – π* and 1s – σ* transitions respectively. The resolution for the measurements was 1 eV. The X-ray beam was directed into the gasket and was aligned parallel to the gap between the diamond anvils. The beam was at least 20 μm away from the surface of each anvil, which ensures no signal from diamond anvils could be counted. The IXS signal was collected by scanning the incident x-ray energy relative to the fixed energy of 9.885 keV. 6 Fig. S2. IR spectrum of the OACC recovered from 32.8 GPa. The absorption peaks from the solvent are labled. No absorption from C60 is observed in the whole spectrum. This suggests that C60 cages are collapsed, but the solvent molecules are well preserved. 7 Fig. S3. XRD pattern of the decompressed material from 42 GPa after a mild heat treatment at 107°C for 8 h. The heat treatment was taken place in air. No sharp diffraction peaks were found in the pattern which suggests that OACC loses its long-range periodicity and transforms into an amorphous structure after the solvent is removed. 8 Fig. S4. A snapshot of the hexagonal close-packed (hcp) structure of C60*m-xylene crystal (with the ratio of C60 to m-xylene molecules being 1:1), obtained from the classical MD simulation at the ambient condition (298 K and 1atm). The green lines refer to the hexagonal symmetry of the system. The supercell has 8 unit cells, which contains 3264 C atoms and 480 H atoms. After 2 ns MD run, the system reaches thermal equilibrium, and the calculated lattice constant is a ~ 23.8 Å, compared to the experimental value (a = 23.63 Å) at 1 atm. Color code: C of C60 (pink), C of m-xylene (blue), and H (white). 9 Fig. S5. Snapshots of unit cell of the hcp crystal from BOMD simulation (after 1 ps) at the pressures of (A) 1.85, (B) 5, and (C) 10 GPa. The computed average lattice constant a = (A) 23.8, (B) 23.25 and (C) 22.54 Å. As a comparison, the experimental values of a are 23.06, 22.60, and 22.20 Å at 1.7, 5.6 and 10.8 GPa, respectively. At 10 GPa, the computed and measured lattice constants are very close to each other. 10 Fig. S6. BOMD simulations of instant compression processes of the (initial) fcc system from the initial pressure (0.4 GPa) to six different pressures: 20, 30, 32, 33, 35 and 40 GPa, respectively. The pressure-induced crystal-to-amorphous transition appears to occur in the pressure range of 32 to 35 GPa. The simulation time is 4 ps for each compression simulation. Polymerization of fullerene molecules can be observed for pressure 33 GPa or higher, due to the ultrahigh temperature of the system within the 1.5 ps of instant compression (see Fig. S8 below). 11 Fig. S7. BOMD simulations of instant decompression processes, from states of the system at six different high pressures to either the zero-pressure or 0.4 GPa-pressure state, respectively. The simulation time is 1or 2 ps for each decompression simulation. 12 Fig. S8. Temperature evolution in the BOMD simulations during the instant compression and decompression, respectively. 13 Fig. S9. Computed vibration spectra at zero-pressure state after decompressed from six different high-pressure states. The vibration spectra are computed form the Fourier transform of the velocity auto-correction functions of C atoms in the BOMD simulation. 14 Fig. S10. The pair distribution functions of the system at (A) compression states and (B) zero-pressure states after decompression; the enlarged plots of PDFs from 1.0 to 2.0 Å are shown in (C) and (D). The C60-centered radial distribution functions at (E) compression states and (F) decompression states. 15 Fig. S11. The relative shift of the strongest XRD peak (100) of the high pressure phase as a function of pressure. The peak shift of XRD of diamond (B=442GPa and B’=3.5) is also shown in the figure. From the comparison, OACC has a compressibility slightly lower than (or comparable to) which of diamond. 16 References 1. H. W. Kroto, J. R. Heath, S. C. O'Brien, R. F. Curl, R. E. Smalley, C60: Buckminsterfullerene. Nature 318, 162 (1985). doi:10.1038/318162a0 2. S. Iijima, Helical microtubules of graphitic carbon. Nature 354, 56 (1991). doi:10.1038/354056a0 3. K. S. Novoselov et al., Electric field effect in atomically thin carbon films. Science 306, 666 (2004). doi:10.1126/science.1102896 Medline 4. W. L. Mao et al., Bonding changes in compressed superhard graphite. Science 302, 425 (2003). doi:10.1126/science.1089713 Medline 5. Z. W. Wang et al., A quenchable superhard carbon phase synthesized by cold compression of carbon nanotubes. Proc. Natl. Acad. Sci. U.S.A. 101, 13699 (2004). doi:10.1073/pnas.0405877101 Medline 6. E. D. Miller, D. C. Nesting, J. V. Badding, Quenchable transparent phase of carbon. Chem. Mater. 9, 18 (1997). doi:10.1021/cm960288k 7. M. Wu, X. J. Wu, Y. Pei, Y. Wang, X. C. Zeng, Three-dimensional network model of carbon containing only sp2-carbon bonds and boron nitride analogues. Chem. Commun. 47, 4406 (2011). doi:10.1039/c0cc05738j 8. Y. Iwasa et al., New phases of c60 synthesized at high pressure. Science 264, 1570 (1994). doi:10.1126/science.264.5165.1570 Medline 9. C. S. Yoo, W. J. Nellis, Phase transformations in carbon fullerenes at high shock pressures. Science 254, 1489 (1991). doi:10.1126/science.254.5037.1489 Medline 10. M. N. Regueiro, P. Monceau, A. Rassat, P. Bernier, A. Zahab, Absence of a metallic phase at high pressures in C60. Nature 354, 289 (1991). doi:10.1038/354289a0 11. B. Sundqvist, Fullerenes under high pressures. Adv. Phys. 48, 1 (1999). doi:10.1080/000187399243464 12. F. Moshary et al.; de Vries MS, Gap reduction and the collapse of solid C60 to a new phase of carbon under pressure. Phys. Rev. Lett. 69, 466 (1992). doi:10.1103/PhysRevLett.69.466 Medline 13. L. Wang et al., Synthesis and high pressure induced amorphization of C60 nanosheets. Appl. Phys. Lett. 91, 103112 (2007). doi:10.1063/1.2768634 14. L. Marques et al., “Debye-Scherrer Ellipses” from 3D fullerene polymers: An anisotropic pressure memory signature. Science 283, 1720 (1999). doi:10.1126/science.283.5408.1720 Medline 15. M. Núñez-Regueiro, L. Marques, J. L. Hodeau, O. Béthoux, M. Perroux, Polymerized fullerite structures. Phys. Rev. Lett. 74, 278 (1995). doi:10.1103/PhysRevLett.74.278 Medline 16. M. Barrio et al., Solid-state studies of C60 solvates formed in the C60−BrCCl3 System. Chem. Mater. 15, 288 (2003). doi:10.1021/cm021284k 17 17. F. Michaud et al., Solid-state studies on a C60 solvate grown from 1,1,2-trichloroethane. Chem. Mater. 12, 3595 (2000). doi:10.1021/cm0011099 18. L. Wang et al., Synthesis of thin, rectangular C60 manorods using m-xylene as a shape controller. Adv. Mater. 18, 1883 (2006). doi:10.1002/adma.200502738 19. A. Talyzin, U. Jansson, C60 and C70 solvates studied by Raman spectroscopy. J. Phys. Chem. B 104, 5064 (2000). doi:10.1021/jp993658b 20. A. Graja, R. Swietlik, Temperature study of IR spectra of some C60 compounds. Synth. Met. 70, 1417 (1995). doi:10.1016/0379-6779(94)02903-C 21. R. Swietlik, P. Byszewski, E. Kowalska, Interactions of C60 with organic molecules in solvate crystals studied by infrared spectroscopy. Chem. Phys. Lett. 254, 73 (1996). doi:10.1016/0009-2614(96)00308-9 22. L. Wang et al., Highly enhanced luminescence from single-crystalline C60 ·1 m-xylene nanorods. Chem. Mater. 18, 4190 (2006). doi:10.1021/cm060997q 23. M. Ramm, P. Luger, D. Zobel, W. Duczek, J. C. A. Boeyens, Static disorder in hexagonal crystal structures of C60 at 100 K and 20 K. Cryst. Res. Technol. 31, 43 (1996). doi:10.1002/crat.2170310111 24. P. Dauber-Osguthorpe et al., Structure and energetics of ligand binding to proteins. Proteins 4, 31 (1988). doi:10.1002/prot.340040106 Medline 25. J. P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865 (1996). doi:10.1103/PhysRevLett.77.3865 Medline 26. C. Hartwigsen, S. Goedecker, J. Hutter, Relativistic separable dual-space Gaussian pseudopotentials from H to Rn. Phys. Rev. B 58, 3641 (1998). doi:10.1103/PhysRevB.58.3641 27. J. VandeVondele et al., Quickstep: Fast and accurate density functional calculations using a mixed Gaussian and plane waves approach. Comput. Phys. Commun. 167, 103 (2005). doi:10.1016/j.cpc.2004.12.014 28. G. Lippert, J. Hutter, M. Parrinello, A hybrid Gaussian and plane wave density functional scheme. Mol. Phys. 92, 477 (1997). doi:10.1080/002689797170220 18