doi: 10.1038/nature06016 SUPPLEMENTARY INFORMATION Preparation and characterization of graphene oxide paper Dmitriy A. Dikin, 1 Sasha Stankovich, 1 Eric J. Zimney, 1 Richard D. Piner, 1 Geoffrey H. B. Dommett, 1 Guennadi Evmenenko, 2 SonBinh T. Nguyen, 3 and Rodney S. Ruoff 1 1 Department of Mechanical Engineering, 2 Department of Physics and Astronomy, 3 Department of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, Illinois 60208, USA SI-1. Proposed mechanism for the formation of graphene oxide paper We propose the following mechanism for the formation of graphene oxide paper by vacuum filtration. In the initial stages of the filtration the graphene oxide sheets are forced rapidly onto the surface of the Anodisc membrane by the water flow and randomly assembled (folded, crumpled, and wrinkled) on the surface of the membrane. As such, the layers of graphene oxide paper assembled at this stage are not particularly well-packed and arranged. After a short time the filter becomes clogged due to the deposition of the sheets, and the water flow slows down considerably. During the subsequent period of slow filtering (evaporation of water is also occurring), the concentration of graphene oxide sheets in the suspension rises, resulting in a significant increase in the sheet-to-sheet interactions. During this stage, the sheets are more likely to be aligned on top of each other in the ever-growing deposit and are probably also smoothed out by the water flow. The whole filtration process takes from about 12 hours to 2 days, depending on the amount of colloidal dispersion of graphite oxide used for achieving the desired film thickness. Toward the end, one can visually observe the formation of a mat with a gradually decreasing thickness. We suggest that the unique properties of the slowly flowing water in the confined galleries 1-5, together with the electrostatic and van der Waals attractive forces a between the very large aspect ratio compliant sheets of graphene oxide, are largely responsible for their sequential deposition into the observed macroscopic structure. The top layer of the resulting graphene oxide paper is not as dense and ordered as its core, perhaps due to the lack of water flow from the top of the suspension and the possibility of more water leaving the structure at the top surface by evaporation. Upon drying, the van der Waals attractions between the sheets squeezed and pushed out the remaining water molecules, leaving only those that are immobilized a The latter are significant at a distance of a few nanometers between the surfaces of these hydrophilic sheets (See Ref.6. Raviv, U., Perkin, S., Laurat, P. & Klein, J. Fluidity of water confined down to subnanometer films. Langmuir 20, 5322-5332 (2004).) www.nature.com/nature 1
by the formation of hydrogen bonds with donor and acceptor sites on the neighboring graphene oxide sheets. After drying, if water is poured on top of the graphene oxide paper, the paper swells sufficiently to allow the water to seep through, and then returns back to the dry state. A piece of graphene oxide paper left in water for several hours does not disperse and maintains its shape (in contrast, graphite oxide powder samples disperse immediately), but disintegrates easily if it is handled while still wet. However, moderate ultrasound agitation of a wetted graphene oxide paper readily re-disperses the graphene oxide sheets into colloidal dispersions. This behavior strongly suggests that no covalent bonding was present between the individual graphene oxide sheets in the graphene oxide paper. If a wetted graphene oxide paper is left to dry, it will regain its mechanical integrity and can again be handled without failure. If a drop of water is placed on the graphene oxide surface, the water uptake is very slow and only localized seepage of water into graphene oxide paper occurs. SI-2. Summary of the tensile tests of the graphene oxide paper samples As mentioned in the Methods section, the samples for mechanical testing were used as prepared without further modification. We were curious to see if the skin layers would have an effect on the mechanical properties of samples with different thicknesses. As the thickness of the skin layers remains constant from sample to sample (see above), its effect on the mechanical properties would manifest more strongly in thin samples. (For example, it is only ~6 % of the total cross section when the paper-like material is 5 µm thick and ~1% for a 25-µm-thick sample.) Surprisingly, our tensile test data (Table SI-2-1) for a range of samples with thicknesses between 2.5 to 25 µm, suggest that the skin layers have little effect on the mechanical properties of our samples. Thinner samples such as 10-1 (2.5-µm-thick) have almost the same Young s modulus as 2-2 (25-µm-thick). Residual stress that results from the fabrication process in paper-like materials is often manifested by visible curling or defects. Such stress, if any, does not appear to be significant in our materials over the range of examined paper thicknesses (1 µm to 25 µm). Visible curling is not observed. In addition, SEM examination shows no evidence of either morphological inhomogeneity (thick/thin stacks) or void defects. Finally, fracture analysis of all samples always resulted in sharp, straight breakage lines, indicating very uniform mechanical properties across the section. The following table displays the complete list of the graphene oxide samples which were successfully tested via static mechanical testing in a uniaxial in-plane tensile load-to-fracture configuration. The first digit in the sample number indicates the graphene oxide membrane from which the strip was cut. Thus, samples 6-1 thru 6-5 were five strips derived from the same piece of graphene oxide paper and have similar thicknesses. R in the sample number indicates that the broken fragment of the initial sample was reloaded for the second tensile test. In particular for sample # 10-1, it was possible to reload the fragments twice. www.nature.com/nature 2
Table SI-2-1. Complete results of the tensile test. Sample # t (µm) w ± 0.05 (mm) L ± 0.001 (mm) E (GPa) σ (MPa) ε (%) W (kj/m 3 ) 1-1 22±1 5.65 19.033 22±3 39 0.22 2-1-R 25±1 5.2 9.137 19±3 15 0.1 2-2 25±1 5.2 12.210 26±2 32 0.26 2-2-R 25±1 5.2 8.990 25±2 67 0.8 348 2-3 25±1 4.64 14.071 29±3 32 0.1 3-1 23±1 4.6 23.346 15±2 28 0.22 3-1 R 23±1 4.6 12.593 18±2 64 0.45 155 4-1 4.9±0.2 5.90 18.115 37±3 105 0.53 352 4-2 4.9±0.2 4.40 17.618 34±2 70.6 0.38 168 5-1 5.2±0.2 5.05 19.291 31±4 28.9 0.1 5-1-R 5.2±0.2 5.05 15.951 35±2 121 0.47 325 5-2 5.2±0.2 5.03 18.714 39±2 112 0.37 224 5-2-R 5.2±0.2 5.03 27.661 34±4 85 0.29 6-3 5.5±0.2 5.50 28.387 36±2 118 0.42 286 6-3-R 5.5±0.2 5.50 16.695 19±3 15.8 0.15 6-4 5.5±0.2 5.45 24.608 41±2 52.8 0.13 6-4 R 1 5.5±0.2 5.45 14.175 42±2 32...48 0.1 1.28 6-5 5.5±0.2 5.9 25.435 36±2 91 0.34 173 8-2 4.8±0.2 5.90 21.155 37±3 133 0.46 338 8-2 R 4.8±0.2 5.90 14.216 33 80 68 0.23 10-1 2.5±0.2 4.1 23.533 30±3 62.9 0.22 10-1-R1 2.5±0.2 4.1 19.152 31±2 83 0.32 129 10-1-R2 2.5±0.2 4.1 9.742 34±2 82 0.29 10-3 2.5±0.2 4.5 26.592 30±3 55 0.21 10-3-R 2.5±0.2 4.5 14.765 30±1 112 0.4 226 12-1 11±0.5 6.0 17.542 28±2 109 0.48 270 12-1 R 11±0.5 6.0 13.103 29±3 42 0.16 12-2 11±0.5 5.8 25.093 28±2 92 0.46 230 12-2 R 11±0.5 5.8 11.645 29±2 104 0.51 312 12-3 2 11±0.5 5.85 22.600 27-32 97 0.4 12-4 3 11±0.5 5.6 20.831 17-25 43 0.2 t, w, L are the thickness, width, and length, of the samples, respectively. E is Young s modulus, determined by fitting the stress-strain plot in the elastic regime with a straight line. σ is engineering stress at fracture, referred to in the manuscript as the stress and computed using the sample width and thickness of the fracture surface. ε is engineering strain at fracture, referred to in the manuscript as the strain and computed from the instantaneous length of the sample between the clamps. W is the work of extension to fracture, the amount of energy absorbed to fracture, calculated by taking the integral beneath the stress-strain curve. The values shown above in the table are for those samples that went through the elastic regime (usually with the strain value above 0.3%). 1 Sample shows slip-stick behavior as described in the manuscript. 2 Tensile tests were carried out in the temperature range between 20 and 150 0 C. 3 Tensile tests were carried out at temperatures of 40, 90, and 120 0 C; results are described in the manuscript. www.nature.com/nature 3
SI-3. Examples of stress-strain dependences measured for graphene oxide paper samples a b c Stress, MPa # 5-1 # 5-1-R # 5-2 # 5-2-R # 8-2 # 8-2-R d e f Stress, MPa # 6-3 # 6-3-R # 6-4 # 6-4-R Strain, % Figure SI-3-1. Examples of the tensile behavior for a few representative graphene oxide paper samples. a through e panels show results for 10 samples obtained from 3 different graphene oxide paper films (#5, 6, and 8). For easier comparison of the mechanical behavior of the samples, panel f shows a combined plot of all 10 measurements displayed in panels a through e. SI-4. Justifications for the comparison made in Figure 3 During the review process, a referee suggested that we include the following discussion for completeness and as guidance to the interested readers. Bucky paper has been used as a reinforcing component in polymer composites to achieve very high improvement in the mechanical properties. Wang et al. 7 have reported 15 GPa for an Epon/Bucky paper composite at 39 wt% loading of CNTs, and a modulus of 3.04 GPa was reported for an Epon/amino-functionalized SWCNTs composite at 0.5 wt% loading 8. When the CNTs in bucky paper are magnetically aligned prior to polymer addition, the modulus of the final nanocomposite can increase to 45 GPa along the direction of alignment and 15 GPa in the perpendicular direction 9. While Zhang et al. 10 had described MWNT fabric/sheet which has a density of ~0.5 g/cm 3 and specific strength of 465 MPa, this material was extensively drawn from a bed of vertically grown CNT and woven to enhance interfibril mechanical coupling www.nature.com/nature 4
along the filaments. Thus, it is not surprising that very high mechanical strength can be achieved in comparison to irregular laid-down individual fibrils that are obtained via filtration. Other works 11,12 where SWCNTs or MWCNTs were aligned, spun, twisted, and densified by forming woven threads and yarns 13 lie in this same category. As analyzed above, while there are other CNT-based materials that can claim higher moduli than our graphene oxide paper, they were either composites with polymers or extensively processed. As such, they cannot be appropriately compared to our simple, filtration-prepared, single-component materials. Hence, in comparing the tensile strengths and Young s modulus of our graphene oxide paper to other materials, we carefully select only those flexible, single-component materials that are either prepared by similar filtration stragies (vermiculite and bucky paper) or have similar morphologies (flexible graphite). SI-5. Bending of graphene oxide paper Bending experiments were performed inside of a Nova NanoSEM 600 by using our home-made testing stage with piezo-actuated linear nanopositioner (model ANPx100, made by AttoCube Systems, Germany). Samples for bending tests (approximately 2 x 20 mm 2 ) were cut from the same membranes as those for tensile tests. In all, 10 strips from membranes #1, 2, 5, 6, 10, and 12 (see Table SI-2-1) were prepared and tested. To perform the test, one end of a strip was clamped to a silicon substrate and carefully bent into a loop by hand. Both ends of the loop were then attached together using a piece of conductive double sided carbon adhesive tape (SPI Supplies Brand, www.2spi.com) to maintain the lower and upper portions parallel to one another. The sample-loaded silicon substrate was next afixed onto the stationary side of the testing stage. Another piece of silicon was attached to the movable side of the testing stage. The curved strip was then compressed between the two silicon substrates until it kinked or fractured. Experiments were performed in the displacement-controlled regime, with a typical step size of about 0.1 µm. A few steps of the experimental sequence are shown in Fig. SI-5-1. www.nature.com/nature 5
a c d 1 mm b e R f 0.5 mm 300 µm Figure SI-5-1. SEM images of a displacement sequence during the bending experiment on graphene oxide paper #10 (2.5-µm-thick). (e) A radius of curvature (R 105 µm) is shown for the bended sample just before it fractures. The scale bars for c-f are the same. SI-6. Thermal gravimetric analysis of graphene oxide paper The thermal gravimetric analysis (TGA) of a sample of graphene oxide paper under flowing N 2 gas was compared with that of the graphite oxide powder (Figure SI- 4-1). At relatively high ramp rate (at least 5 0 C/min) a rapid expansion of graphite oxide powder occurs at about 180 0 C, resulting in a partial physical loss of the material from the uncapped crucible used in the TGA instrument. A slower heating rate of 1 0 C/min was subsequently employed to prevent material loss. In these cases, one can observe two steps in the mass loss for both graphene oxide paper and graphite oxide powder. The first significant mass loss ( 15 wt%) occurs during sample heating from room temperature to 100 0 C. Above this temperature, the mass reading stabilizes until 160 0 C; between 160 and 190 0 C, the samples lose an additional 20 wt%. www.nature.com/nature 6
Figure SI-6-1. Normalized remaining mass of graphene oxide paper and powder as a function of temperature SI-7. Density measurements The density of graphene oxide paper was measured using the Archimedes method in water. The physical density of graphene oxide paper for relatively thick ( 10 µm) samples was 1.8 g/cm 3. Density measurements for thinner samples of graphene oxide paper were unreliable due to the small sample mass. The hydrophilicity of graphene oxide paper allows for the complete wetting of its surfaces immediately upon exposure to water and the in-water mass was constant for at least 15 minutes. References 1. Israelachvili, J. N. & McGuiggan, P. M. Forces between surfaces in liquids. Science 241, 795-800 (1988). 2. Klein, J. & Kumacheva, E. Confinement-induced phase-transitions in simple liquids. Science 269, 816-819 (1995). 3. Klein, J. & Kumacheva, E. Simple liquids confined to molecularly thin layers. I. Confinement-induced liquid-to-solid phase transitions. J. Chem. Phys. 108, 6996-7009 (1998). 4. Raviv, U., Laurat, P. & Klein, J. Fluidity of water confined to subnanometre films. Nature 413, 51-54 (2001). 5. Raviv, U., Giasson, S., Frey, J. & Klein, J. Viscosity of ultra-thin water films confined between hydrophobic or hydrophilic surfaces. J. Phys.: Condens. Matter 14, 9275-9283 (2002). 6. Raviv, U., Perkin, S., Laurat, P. & Klein, J. Fluidity of water confined down to subnanometer films. Langmuir 20, 5322-5332 (2004). 7. Wang, X. F., Chiang, M. Y. M. & Snyder, C. R. Monte-Carlo simulation for the fracture process and energy release rate of unidirectional carbon fiber-reinforced polymers at different temperatures. COMPOS PART A-APPL S 35, 1277-1284 (2004). 8. Wang, S. R., Liang, Z. Y., Liu, T., Wang, B. & Zhang, C. Effective aminofunctionalization of carbon nanotubes for reinforcing epoxy polymer composites. Nanotechnology 17, 1551-1557 (2006). www.nature.com/nature 7
9. Wang, B., Liang, Z., Shankar, R. & Zhang, C. Electrical resistivity and mechanical properties of magnetically aligned SWNT buckypapers and nanocomposites, in Proceedings of 49th International SAMPE Symposium and Exhibition (Long Beach, CA, USA, 2004). 10. Zhang, M. et al. Strong, transparent, multifunctional, carbon nanotube sheets. Science 309, 1215-1219 (2005). 11. Jiang, K. L., Li, Q. Q. & Fan, S. S. Nanotechnology: Spinning continuous carbon nanotube yarns - Carbon nanotubes weave their way into a range of imaginative macroscopic applications. Nature 419, 801-801 (2002). 12. Li, Y. L., Kinloch, I. A. & Windle, A. H. Direct spinning of carbon nanotube fibers from chemical vapor deposition synthesis. Science 304, 276-278 (2004). 13. Zhang, M., Atkinson, K. R. & Baughman, R. H. Multifunctional carbon nanotube yarns by downsizing an ancient technology. Science 306, 1358-1361 (2004). www.nature.com/nature 8