9694
Langmuir 2006, 22, 9694-9703
Graphitic Structures by Design Jonathan Phillips,*,†,‡ Toshi Shiina,† Martin Nemer,§ and Kelvin Lester‡ Los Alamos National Laboratory, MS-E549, Los Alamos, New Mexico 87545, Sandia National Laboratories, 4100 National Parks Highway, Carlsbad, New Mexico 88220, and Department of Mechanical Engineering, UniVersity of New Mexico, Albuquerque, New Mexico 87131 ReceiVed April 5, 2006. In Final Form: August 10, 2006 It is shown that self-supporting graphitic structures of specific shape can be grown in a variety of forms, from nanoscale to macroscale, on metal templates, in a fuel-rich mixture of ethylene and oxygen at temperatures between 750 and 900 K. The evidence presented suggests graphite can be grown in any shape created from catalytic metals (e.g., Ni) under the proper conditions of temperature and gas composition. Structures produced include macroscale bodies, centimeters in dimension, composed of micrometer-scale graphite elements such as graphite “foam” and regular graphite “lattices”. Nanoscale hollow graphite spheres were also produced. The production rate in the apparatus employed was roughly shown to be 1 layer/s and was steady with time over several hours. The process of producing self-supporting bodies generally produces hollow graphite structures, as the underlying metal template must be removed by acid following the completion of graphite growth. The process is believed to be possible only in an environment, such as combustion, in which a high concentration of particular radical species is present in the vicinity of the template surface. The following process is postulated: (i) a single layer of graphite is formed from gas-phase radicals by the catalytic action of the metal template, (ii) additional graphite growth is “autocatalytic” and occurs via the decomposition of radicals on the surface and the incorporation of “free” carbon atoms, or other radical fragments, into “edge sites” on the graphite surface.
Introduction In this paper we describe a novel method to grow graphite in precisely designed forms, rapidly and with simple equipment. Potentially, this “technology” can be adapted for a host of applications. Indeed, highly ordered graphite has properties that have led to its use in computers, life sciences, microelectromechanical devices (“MEMS”), coatings, lubrication, metals forming, and nuclear, aerospace, and specialty military applications. Macroscopic graphite either is found naturally or is produced by variations on the basic recipe of baking for extended periods precursor carbons, such as petroleum coke, at temperatures in excess of 2500 K, a temperature so high that energy is the most significant cost.1,2 Still, each new method appears to produce a type of graphite with unique morphology, and hence new applications, such as the recent proposal that graphitic foams may have unique heat-transfer capability.3 However, for complex shapes, in addition to the high temperatures required, specialized processing is required. Probably the best example, one leading to the production of designed shapes on the nanometer scale, is the two-step process pioneered by Zakhidov and co-workers.4,5 In one example, amorphous carbon is deposited on voids in nanoscale opal structures via the thermal decomposition of * To whom correspondence should be addressed. † Los Alamos National Laboratory. ‡ University of New Mexico. § Sandia National Laboratories. (1) Oya, A.; Marsh, H. Phenomena of Catalytic Graphitization. J. Mater. Sci. 1982, 17 (2), 309. (2) Brooks, J. D.; Taylor, G. H. The formation of some graphitizing carbons. In Chemistry and Physics of Carbon; Walker, P., Ed.; Marcel Dekker: New York, 1968; Vol. 4, pp 243-286. (3) Klett, J.; Hardy, R.; Romine, E.; Walls, C.; Burchell, T. High-thermalconductivity, mesophase-pitch-derived carbon foams: Effect of precursor on structure and properties. Carbon 2000, 38 (7), 953. (4) Zakhidov, A. A.; Baughman, R. H.; Iqbal, Z.; Cui, C. X.; Khayrullin, I.; Dantas, S. O.; Marti, I.; Ralchenko, V. G. Carbon structures with three-dimensional periodicity at optical wavelengths. Science 1998, 282 (5390), 897. (5) Zakhidov, A. A.; Khayrullin, II.; Baughman, R. H.; Iqbal, Z.; Yoshino, K.; Kawagishi, Y.; Tatsuhara, S. CVD synthesis of carbon-based metallic photonic crystals. Nanostruct. Mater. 1999, 12 (5-8), 1089-1095.
propylene at 1100 K for 6 h.4 Raman spectroscopy suggests some of the carbon was converted to graphite following a heat treatment at 2300 K for several hours. A regular “inverse opal” array, a designed (partially) graphite structure with characteristic dimensions in the nanometer range, was formed in this manner. The potential value of new forms of graphite is even more clearly seen from the tremendous interest in “carbon” (in fact graphite) nanotubes.6 It is believed that these structures, properly grown and purified, will have unique strength, thermal characteristics, and electronic behaviors, and these properties will enable a host of new technologies from space elevators7 to molecular-scale logic circuits.8,9 Yet it is sometimes argued that nanotubes are simply an elegant nanoscale form of graphite, and some of the “special” properties such as fracture strength,10 mechanical stiffness, and thermal conductivity11 anticipated for nanotubes can be found in other graphitic structures as well. With the technology described below it will be possible to test the hypothesis that designed graphite structures can perform as well as nanotubes in many applications proposed for nanotubes. Some “tests” performed by others already support this hypothesis. For example, it is frequently suggested that nanotubes will be the key elements in “molecular-scale” logic circuits. Preliminary investigations have validated the feasibility of this concept, yet the necessary purification and then reorganization of nanotubes of the appropriate types into complex two-dimensional networks (6) Chen, C.-K.; Perry, L. W.; Xu, H.; Jiang, Y.; Phillips, J. Plasma torch production of macroscopic carbon nanotube structures. Carbon 2003, 41, 25552560. (7) Edwards, B. C. Design and deployment of a space elevator. Acta Astronaut. 2000, 47 (10), 735. (8) Baughman, R. H.; Zakhidov, A. A.; de Heer, W. A. Carbon nanotubes the route toward applications. Science 2002, 297 (5582), 787. (9) Huang, Y.; Duan, X. F.; Wei, Q. Q.; Lieber, C. M. Directed assembly of one-dimensional nanostructures into functional networks. Science 2001, 291 (5504), 630. (10) Dresselhaus, M. S.; Dresselhaus, G. Intercalation compounds of graphite. AdV. Phys. 2002, 51 (1), 1-186. (11) Stankovich, S.; Dikin, D. A.; Dommett, G. H. B.; Kohlhaas, K. M.; Zimney, E. J.; Stach, E. A.; Piner, R. D.; Nguyen, S. T.; Ruoff, R. S. Graphene-based composite materials. Nature 2006, 442 (7100), 282-286.
10.1021/la060915c CCC: $33.50 © 2006 American Chemical Society Published on Web 10/04/2006
Graphitic Structures by Design
Langmuir, Vol. 22, No. 23, 2006 9695
Figure 1. Quartz reactor system. To ensure consistency, the operating conditions, including the position of the sample holder (14 cm from either end), reduction time and temperature, deposition temperature (875 K), and gas flow rates, were kept the same in all cases, except as noted. The impact of variation of these parameters on graphite product properties is under investigation (e.g., see Figure 4).
remains unaccomplished.12 Other workers have found that flat graphite, of just a few atomic layers (“graphene”), can perform the same functions in logic circuits as nanotubes.13 Putting these two findings together suggests that designed graphitic structures, grown in place in the geometric configurations required for circuitry, using the approach described below, may be an easier avenue for producing nanoscale logic circuits than the use of nanotubes. Some of the demonstration structures already grown using the new technology described herein show that large graphitic structures can be rapidly and inexpensively grown in forms that in actual practice will be more practical, and perhaps stronger, than real nanotube structures. The novel technique introduced in this paper for growing designed graphitic structures is simple: partially combust hydrocarbons in the presence of appropriate catalytic metals (e.g., nickel, platinum, or palladium) that are preformed into (or coated onto) the desired shape (templates). Under the proper conditions of temperature and feed composition, graphite grows rapidly and follows the form of the template precisely. The mechanism of growth is not entirely proven, but an initial model that is completely consistent with all observations is presented. Experimental Section Reactor/Procedure. The reactor system used in this study for the production of the various graphitic structures shown is very simple. It consists of a 2 cm diameter quartz reactor inside a standard tube furnace. Also required are three mass flow controllers for the three input gases: an inert (UHP N2), a hydrocarbon (C2H4), and oxygen (UHP grade) (Figure 1). The process in all cases consisted of three steps: (i) reduction of the catalytic agent (nickel in all cases described herein) in a mixture consisting of nitrogen (95%) and hydrogen (about 5%) for 4 h at 625 K, (ii) flushing of the system with nitrogen while raising the temperature to the final temperature (875 K) for approximately 25 min, (iii) introduction of the “deposition mixture”, which consists of nitrogen (>90 vol %) and a fuel-rich (O2:C2H4 < 3) mixture of ethylene and oxygen. Microscopy. Both transmission (HRTEM, HR ) high resolution; JEOL 2010) and scanning (SEM; Hitachi S-800 and JEOL 5800LV) electron microscopies were used to characterize the templates and graphitic structures created in this study. The high-resolution transmission electron microscope is a 200 keV instrument equipped with a GATAN slow-scan camera and a DigitalMicrograph system. (12) Kong, J.; Soh, H. T.; Cassell, A. M.; Quate, C. F.; Dai, H. J. Synthesis of individual single-walled carbon nanotubes on patterned silicon wafers. Nature 1998, 395 (6705), 878. (13) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A. Electric field effect in atomically thin carbon films. Science 2004, 306 (5696), 666-669.
Microscopy was performed at the Department of Earth and Planetary Sciences at the University of New Mexico. Powdered samples were analyzed by X-ray diffraction (XRD) in the XRD Laboratory in the Department of Earth and Planetary Sciences at the University of New Mexico, using a Scintag Pad V diffractometer with DataScan 4 software (from MDI, Inc.) for system automation and data collection. Cu KR radiation (40 kV, 35 mA) was used with a Bicron scintillation detector (with a pyrolitic graphite curved crystal monochromator). Data were analyzed with Jade 6.5 software (from MDI, Inc.) using the ICDD (International Center for Diffraction Data) PDF2 database (rev. 2004) for phase identification.
Results Experiments were designed and carried out to meet the following objectives: (i) demonstrate conclusively that graphite can be grown on catalytic (nickel) templates and that the graphite so grown followed the structure of the template precisely (“template carbon”), (ii) determine the conditions under which template carbon will grow, (iii) demonstrate that a number of pure graphitic self-supporting structures can be formed by acid dissolution of the metal substrate, (iv) show that graphitic structures can be created using this novel approach on the nanoto macroscale, (v) demonstrate that the carbon structures found are in fact graphite, and (vi) demonstrate that in the absence of oxygen only filamentous carbon will grow. Example Structures. Figure 2 shows an example of graphite growth following the form of a nickel template. The “graphite” lattice, grown in the furnace following the procedure described in the Experimental Section, follows precisely the form of the original nickel template. The template (Figure 2a,b) used in this case was a 99.9% nickel screen with nominal size laminae of 11 µm and nominal interlamina dimensions of 40 µm obtained from Goodfellow Metals (Cambridge, U.K.). The particular structure shown (Figure 2c,d) required only 1 h of treatment under standard reactor deposition conditions (875 K, ethylene-to-oxygen ratio of 3:1). It is clear from the figure that even using the simple apparatus described it was possible to make macroscopic objects with dimensions of many centimeters very quickly. It is instructive to look at the crystals of graphite found on the surface of the object (Figure 2e). The final object that forms is composed of polycrystalline graphite, but clearly the crystals are on the order of micrometers or larger in dimension. The impact of changing growth conditions on this characteristic of the graphite is under further investigation. Further information about the crystals in the lattices is obtained using TEM (Figure 3). For example, at sufficiently high magnification (Figure 3a) one can clearly see the basal planes of graphite. The separation of these planes matches the known
9696 Langmuir, Vol. 22, No. 23, 2006
Phillips et al.
Figure 2. SEM study of a graphitic lattice: (a) a fresh uncoated 99.9% (metals basis) nickel screen with nominal size laminae of 11 µm and nominal interlamina dimensions of 40 µm, obtained from Goodfellow Metals, Cambridge, U.K.; (b) higher magnification of the nickel lattice; (c) screen after graphite deposition conditions for 1 h, followed by metal removal using 4:1 HCl/HNO3 (by volume) at room temperature for about 5 h [this left the hollow graphite only (determined by XRD)]; (d) same as (c) but higher magnification; (e) further magnification of (c) showing that the surface, at least, consists of a patchwork of graphite crystals.
value for graphite. The SAD pattern clearly is that of polycrystalline graphite. A finding relevant to the postulated model of graphite growth presented below is that the overall “character” of the graphite is impacted by precise position within the reactor. That is, as shown in Figure 4, two pieces of lattice graphite, grown simultaneously, can be significantly different, even if the only difference is that one was grown 7 cm further downstream than the other. In the example shown, one clearly is covered with
graphite, and the other is apparently free of deposits of any kind. Also note that the conditions employed (825 K) generally produced “rougher” graphite and some filaments, relative to the findings that at 875 K the graphite covered the template smoothly and was apparently totally free of filaments (e.g., Figure 2). Another graphitic object of macroscopic scale, but micrometer “short-range” dimension, is shown in Figure 5. This graphitic “foam” was created by treating a sample of nickel metal foam (Goodfellow Cambridge Ltd., Huntingdon, England; thickness
Graphitic Structures by Design
Langmuir, Vol. 22, No. 23, 2006 9697
Figure 4. Light microscope image of a treated lattice showing position sensitivity: (a) optical microscope image of a nickel screen after graphite deposition in the tube reactor shown in Figure 1 at position X ) -7 cm, placed 7 cm upstream in the quartz tube (Figure 1) from the center; (b) optical microscope image of the nickel screen after graphite deposition in the tube reactor shown in Figure 1 at the center of the quartz tube, X ) 0 cm. Both lattices were processed under the same conditions (550 °C, ethylene-to-oxygen ratio of 1:2) simultaneously for 1 h. Clearly, the position within the tube impacts the deposition rate.
Figure 3. TEM study of a graphite lattice: (a) TEM image of a graphitized and acid-treated screen showing the graphite basal planes; (b) SAD image of the same area as in (a) showing clearly that graphite is the only structure present.
1.6 mm, purity 95%, porosity 95%) in the reactor under deposition conditions similar to those employed to make the lattice structure (Figure 2). The selection of a foam template was made because of recent indications that graphitic foam has value as a heattransfer material.14 Graphitic structures can also be grown on the nanoscale, as shown in Figure 6. The template employed in this case was (14) Gaies, D.; Faber, K. T. Thermal properties of pitch-derived graphite foam. Carbon 2002, 40 (7), 1137-1140.
nickel nanoparticles obtained from Umicore Canada, Inc. The spacing between these planes was measured and was found to be 3 ( 0.1 Å, which bounds the value expected for graphite. The character of the graphite changes dramatically as a function of the conditions. In some cases, only “template graphite”, that is, graphite that appears to mimic the shape of the template, is found. In other cases, primarily filamentous graphite is found. Additional exploration of this phenomenon was undertaken using the nickel micrometer-scale particles as templates. This work revealed rough, but consistent, patterns as a function of the conditions. In particular, it appears that there is a window of temperatures, roughly from 750 to 850 K, over which the best quality graphite is formed for all oxygen:fuel ratios. At higher temperatures, the graphite clearly extends beyond the edges of the nickel and hence is called “template plus” graphite. It is also clear that at temperatures below about 750 K only filaments form and at temperatures below about 650 K very little if any carbon is deposited (Figure 7). Note that it may be that a different pattern of growth/morphology is observed on larger structures, although rough initial work indicates a similar pattern on the lattice templates. Examples of highly filamentous graphite formed on the nickel nanoparticles are shown in Figure 8. It is clear that the filaments grow in a manner apparently identical to that generally observed/ described in the literature. That is, each filament is associated with a metal particle, and it is generally believed that the metal particles are key to the growth process.15-19 In addition to the template graphite (Figure 6) formed under limited conditions and the graphitic filaments formed at relatively low temperatures, or in the absence of oxygen, at high
9698 Langmuir, Vol. 22, No. 23, 2006
Figure 5. Graphitized foam: (a) photograph of the macroscopic graphitized foam, approximately 0.75 in. on a side and 0.125 in. thick, after metal removal by acid treatment (light clearly passes through the material); (b) SEM image of the fresh uncoated nickel foam; (c) blowup of (b) (scale bar 100 µm); (d) SEM image of the nickel foam after deposition of graphite for around 1 h; (e) SEM image of the graphitized foam after acid removal of the nickel. Clearly the laminae are hollow, and some are broken, probably during handling.
Phillips et al.
Figure 6. Graphite nanoparticles grown on a Ni nanoparticle template: (a) TEM image of a single hollow graphite nanoparticle, grown in standard conditions, after Ni metal removal with acid; (b) TEM image of the edge section of the same particle, showing basal planes, with interplanar separation of the magnitude expected for graphite; (c) SAD image from a cluster of nanoparticles showing spacing on the order of 3 ( 0.1 Å.
Graphitic Structures by Design
Langmuir, Vol. 22, No. 23, 2006 9699
Figure 7. Map of the morphology as a function of the operating conditions. The morphology of the structures that form on nickel nanoparticles is clearly a function of the conditions employed.
Figure 8. TEM and SEM images of filaments growing from Ni nanoparticles: (a) TEM image clearly showing the character of filaments that grew on a 200 nm nickel particle at 625 K at a 1:1 ethylene:oxygen ratio; (b) SEM image showing apparently all nickel particles produced filaments. Contrast this figure with Figure 6.
temperatures in excess fuel graphitic carbon grew, but extended beyond the limits of the template. That is, the shape of this graphite, given the appellation template plus, was not faithful to the underlying template (Figure 9). This type of growth does not (15) Baker, R. T. K.; Harris, P. S. In Chemistry and Physics of Carbon; Walker, P. L., Thrower, P. A., Eds.; Marcel Dekker: New York, 1978; Vol. 14, p 83. (16) Baker, R. T. K.; Yates, D. J. C.; Dumesic, J. A. Coke Formation on Metal Surfaces; Baker, R. T. K., Ed.; ACS Symposium Series 202; American Chemical Society: Washington, DC, 1981; p 1. (17) Baker, R. T. K.; Barber, M. A.; Waite, R. J.; Harris, P. S.; Feates, F. S. Nucleation and Growth of Carbon Deposits from Nickel Catalyzed Decomposition of Acetylene. J. Catal. 1972, 26 (1), 51.
Figure 9. Template plus and template graphite: SEM images (scale bar 5 µm). (a) Template plus graphite (925 K, 2:1) does not appear to retain the original nearly spherical shape of the nickel template particles. In fact, platelets appear to be dominant. (b) In template graphite (725 K, 1:1) the original shape of the template is visible even after the graphite grows into “necks” that connect the particles.
appear to reflect the total amount of graphite that was produced as preliminary measurements show that the rates of growth of the two types of graphite are nearly identical. (18) Nielsen, J. R.; Trimm, D. L. Mechanisms of Carbon Formation on NickelContaining Catalysts. J. Catal. 1977, 48 (1-3), 155-165. (19) Boellaard, E.; Debokx, P. K.; Kock, P. A.; Geus, J. A. The Formation of Filamentous Carbon on Iron and Nickel-Catalysts: 3. Morphology. J. Catal. 1985, 96 (2), 481-490.
9700 Langmuir, Vol. 22, No. 23, 2006
Phillips et al.
polycrystalline graphite. Regarding the chemical analysis, Ni was approximately 350 ppm and sulfur, calcium, and potassium, if present, were at concentrations below the detection limits. XRD studies were performed on graphite grown on different template materials, and from these studies it is clear that the carbon in all cases is graphite. There is no evidence at all of turbostratic or amorphous carbon in the X-ray diffractogram.20-22 However, these studies also show that the size of the crystals formed is related to the size of the template material. The graphite fiber “grain size” (Figure 8) is on the order of angstroms as determined from an analysis of the extent of line broadening detected. In contrast, the XRD lines from the crushed graphite lattice (Figure 2) are not measurably broadened beyond instrument broadening, indicating that the graphite crystals are at least several hundred angstroms in size.
Discussion
Figure 10. XRD analysis: (a) graphite grown on a nickel lattice after nickel removal ( graphite peak at 2θ ) 23° is observed for all template forms); (b) graphite grown on 200 nm nickel powder at 625 K with a 1:1 hydrocarbon-to-oxygen ratio (filaments), Ni not removed. The XRD data clearly show that the only carbon structure present is graphite. Note: The acid removal of Ni generally left small signals from Ni, equivalent, on the basis of intensity analysis relative to a presumed unchanged graphite peak, to ∼1% of the original metal.
Clearly the rate of growth of the graphite is of interest. Analysis of the weight change of about a dozen samples of template graphite, produced over the whole range of conditions that produces template carbon (see Figure 7), suggests that the growth rate is about 1 graphite layer/s in the system employed and that it is fairly linear with time. Some tests were employed to determine if graphite grows on noncatalytic metals. It was found that no graphite forms on aluminum under conditions identical to those employed to make template graphite on nickel. On iron, some slight brown discoloration was noted, but no graphite was recovered following the acid removal of the metal. It is likely the brown color was simply iron oxide. Graphite. In addition to visual inspection (Figure 2) four different methods were employed to demonstrate that the material made using this novel procedure is in fact graphite: (i) measurement of interplanar spacing (Figures 3a and 6b), (ii) selected area diffraction (SAD) in the transmission electron microscope (Figures 3b and 6c), (iii) X-ray powder diffraction (Figure 10), and (iv) chemical analysis employing ICP (Galbraith Labs, TN). SAD was performed on structures produced from various templates, both before and after acid removal of the template, and in all cases clearly showed the existence of
All the experimental objectives outlined above were met. XRD, TEM-SAD, and TEM all clearly show that the structures produced on the catalytic (nickel) substrates in partial combustion environments were graphite. Thus, “bulk” graphite was produced quickly at far lower temperatures, using far simpler equipment, than in standard practice (more below). Clearly, graphite growth is a function of the substrate. For example, graphite grew very rapidly on Ni, but very slowly (if at all) on Fe and not at all on Al. Moreover, in partial combustion environments no discoloration was ever found on any ceramic or glass surface in the reactor. It was shown that simple acid bath treatments removed the nickel substrate, leaving pure, self-supporting graphite structures of sufficient strength to be handled without extensive damage. In the event that oxygen was carefully excluded from the reaction mixture, absolutely no template carbon formed. However, at sufficiently high temperatures (ca. >825 K) graphitic filaments grew even in the absence of oxygen. Interestingly, at any given temperature the rate of filament growth in the absence of oxygen was clearly slower than that of template graphite. The results of this work suggest that there is a viable alternative to the standard procedure for producing macroscopic graphitic objects. Clearly, macroscopic, self-supporting graphitic structures can be grown at far lower temperatures, much more quickly, using this technology than with conventional high-temperature/ long-soak methods. In fact, it is clear that the approach employed not only represents a unique method for making graphite, but can generate structures that cannot be made in any other fashion. For example, there is no method available for making either the macroscopic lattice structures or the graphite nanostructures generated with this new methodology. Critical Analysis of Literature Mechanisms. A critical analysis of the literature on the mechanism of graphite growth suggests both that there are common observations not readily explained by the existing models and that, in particular, many observations made in the present study cannot be readily explained with these models. However, models of diamond growth and soot growth can readily be extrapolated to graphite. These “extrapolated” models are completely consistent with the proposed mechanisms described below. At present there are two clearly unique mechanisms that can be used to create bulk graphite and another mechanism associated (20) Menendez, J. A.; Xia, B.; Phillips, J.; Radovic, L. R. On the modification and characterization of chemical surface properties of activated carbon: Microcalorimetric, electrochemical, and thermal desorption probes. Langmuir 1997, 13 (13), 3414-3421. (21) Franklin, R. E. Crystallite Growth in Graphitizing and Non-Graphitizing Carbons. Proc. R. Soc. London, Ser. A 1951, 209 (1097), 196. (22) Radovic, L. R.; Walker, P. L.; Jenkins, R. G. Importance of Carbon Active Sites in the Gasification of Coal Chars. Fuel 1983, 62 (7), 849-856.
Graphitic Structures by Design
with the formation of some graphitic fibers. The first, and commercially predominant, method for making bulk graphite is simply to heat high carbon content materials (e.g., petroleum coke) to temperatures well in excess of 2300 K and hold them at that temperature for prolonged periods. The postulated mechanism is that at sufficiently high temperatures diffusion processes are active enough to allow the carbon atoms/clusters to move to a low-energy configuration (graphite) from higher energy configurations (e.g., turbostratic carbon). At no time during the method used here was a temperature sufficient to activate this mechanism reached; hence, this mechanism plays no role in the formation of graphite in the present case. The second method for bulk graphite production is a catalytic process in which metal, generally in the form of small particles, is employed to convert solid carbon into graphite. The catalytic conversion of solid carbon to graphite has been thoroughly characterized. It is clearly a “localized” phenomenon taking place within a few atomic layers of the metal catalyst surface.1,17 It has never been demonstrated, and there is no reason to believe, that the method can be employed to make graphite in specific macroscopic shapes, such as those produced herein. The postulated mechanism for the catalytic mechanism is that the metal acts as a template for carbon atoms, such that as they diffuse to the metal surface they can organize into a low-energy form. As described in the literature this is at best a vague model. It does not provide fully satisfactory explanations for even the most elementary observations. For example, why is graphite formed several atomic layers from the metal surface? How does the catalytic nature of the metal extend beyond its surface? A simple extension of the catalytic model, consistent with the present paper, thermodynamics, the general theory of crystal growth, and earlier reports, is offered: Graphite is itself a catalyst for graphite formation. One application of this postulate to the current proposed mechanisms of catalytic graphite formation is to the often observed phenomenon of the conversion of carbon to graphite at a distance of several atomic layers from a catalytic metal surface. Hence, the revised model of catalytic conversion of carbon to graphite is that it is a two-step process. First, the metal promotes the reorganization of carbon atoms in direct contact with its surface into graphite. Second, the newly formed graphite acts as a template for the conversion of “diffusing” carbon atoms/clusters into graphite as well. Clearly, this is a slow process, akin to sintering, and hence is arguably very slow below the Tamman temperature23 of carbon (approximately 1900 K), but it is a process that takes place at some rate at all temperatures. It is also true that thermodynamically this mechanism is sensible. It is often the case24 in solids that the structure with the higher (chemical potential) effective “vapor pressure” (e.g., amorphous or turbostatic carbon) gradually is converted to the form with the lower (chemical potential) vapor pressure (e.g., graphite) via species diffusion. It is argued below that this modified catalytic mechanism, that is, the catalytic conversion of carbon atoms into graphite at graphite surfaces, is a key aspect of the process observed in the present study. A third class of mechanisms of graphite growth are employed to explain the generation of filaments. There are many variations in this mechanism class, but all require that a gaseous hydrocarbon precursor decompose on a metal surface, leaving a carbon atom.15-19 The carbon atom then diffuses through or around the particle and attaches itself to a growing filament. It is argued that (23) Moulijn, J. A.; van Diepen, A. E.; Kapteijn, F. Catalyst deactivation: Is it predictable? What to do? Appl. Catal., A 2001, 212 (1-2), 3-16. (24) Phillips, J.; Tanski, J. Structure and kinetics of formation and decomposition of corrosion layers formed on lithium compounds exposed to atmospheric gases. Int. Mater. ReV. 2005, 50 (5), 265-286.
Langmuir, Vol. 22, No. 23, 2006 9701
this process is thermodynamically favored as the free energy (and effective carbon vapor pressure) of the “free atoms” on and/or in the metal is higher than that of the graphitic filament. A model very similar to that for filament growth is very similar to the most widely accepted theory of nanotube growth. To wit, in the so-called “root growth” mechanism6,25-27 for nanotubes, carbon atoms initially are adsorbed on a metal nanoparticle and then are transferred, by some mechanism, to the particlenanotube interface at which interface nanotube growth occurs. This mechanism will clearly lead to asymmetric growth around the catalytic metal particle and cannot explain those examples of symmetric growth, relative to the metal template observed in the present study. However, it clearly can explain those instances of filament formation observed in this study. The character of the graphite growth observed in the present work is not readily explained by any of the existing models of graphite growth. None of the models predict that under any circumstances layers of graphite will form symmetrically around a template directly from gas-phase precursors. Catalytic graphite growth, even the modified version posited above, requires a source of solid carbon. The only existing theory for conversion of a gaseous source of carbon precursors to graphite leads to asymmetric growth, filaments. Below we postulate a new mechanism for symmetric graphite growth directly from gasphase precursors that is consistent with observation, thermodynamics, combustion theory, and crystal growth mechanisms. Moreover, this proposed mechanism is actually an application of recent models for soot and diamond growth to graphite growth.28,29 Proposed Mechanism. It is postulated that metal template encapsulating graphite grows from radicals formed via homogeneous (partial) combustion in multiple steps. In the first step, radicals are formed in the gas phase due to a homogeneous partial oxidation reaction. Indeed, the mechanism for homogeneous oxidation of ethylene has long been studied.30 Modern mechanistic models indicate many different carbon-containing radicals exist during oxidation31,32 and that the particular population of radicals is a function of the gas composition, temperature, and reaction time. It is postulated, as a part of this model, that only over a limited temperature range are the correct radicals present in sufficient concentration to deposit graphite. This is not a novel concept. Indeed, it is well understood that in many combustion processes that soot will only form over a limited range of temperatures and fuel mixtures as only in those conditions are soot precursor radicals present.33,34 Similarly, it has been (25) Gavillet, J.; Loiseau, A.; Journet, C.; Willaime, F.; Ducastelle, F.; Charlier, J. C. Root-growth mechanism for single-wall carbon nanotubes. Phys. ReV. Lett. 2001, 87 (27I), 2755041-2755044. (26) Saito, Y.; Okuda, M.; Fujimoto, N.; Yoshikawa, T.; Tomita, M.; Hayashi, T. Single-wall carbon nanotubes growing radially from Ni fine particles formed by arc evaporation. Jpn. J. Appl. Phys., Part 1 1994, 33 (4A), L526-529. (27) Gavillet, J.; Loiseau, A.; Ducastelle, F.; Thair, S.; Bernier, P.; Stephan, O.; Thibault, J.; Charlier, J. C. Microscopic mechanisms for the catalyst assisted growth of single-wall carbon nanotubes. Carbon 2002, 40 (10), 1649-1663. (28) Frenklach, M.; Schuetz, C. A.; Ping, J. Migration mechanism of aromaticedge growth. Proc. Combust. Inst. 2005, 30 (1), 1389-1396. (29) Netto, A.; Frenklach, M. Kinetic Monte Carlo simulations of CVD diamond growth-Interlay among growth, etching, and migration. Diamond Relat. Mater. 2005, 14 (10), 1630-1646. (30) Lenher, S. J. Am. Chem. Soc. 1931, 53, 3752. (31) Westbrook, C. K.; Dryer, F. L.; Schung, K. P. Comprehensive Mechanism for the Pyrolysis and Oxidation of Ethylene. In 19th Symposium (International) on Combustion, Haifa, Israel; Combustion Institute: Pittsburgh, PA, 1982; pp 153-166. (32) Warnatz, J. Structure of Laminar Alkane, Alkene, and Acetylene Flames. In 18th Symposium (International) on Combustion, Waterloo, Ontario, Canada; Combustion Institute: Pittsburgh, PA, 1980; pp 369-384. (33) Frenklach, M.; Ramachandra, M. K.; Matula, R. A. In 20th Symposium (International) on Combustion; Combustion Institute: Pittsburgh, PA, 1984; p 871.
9702 Langmuir, Vol. 22, No. 23, 2006
demonstrated that catalytic etching during combustion reactions only occurs over a narrow temperature and composition range for the same reason. For example, the catalytic etching of platinum during ethylene oxidation only occurs in fuel-lean conditions over the temperature range from 775 to 925 K. These limits are believed to reflect the fact that etching requires a high concentration of the hydroperoxide radical, HO2.35-37 In the second step of graphite formation, homogeneously formed radicals collide and adsorb on solid surfaces found in the reactor. The third step is a function of the nature of the surface with which the radicals collide. Four rough categories of surfaces are postulated to produce distinct results: (i) “ceramic”, (ii) metal, (iii) metal catalyst, and (iv) graphite. In the event that the collision is with a “ceramic surface” (e.g., quartz reactor or alumina boat), no significant surface chemistry takes place. In particular, the radicals do not decompose to leave a nonvolatile carbon. Evidence in support of this is the finding in this and other studies17 that carbon is only deposited on metal surfaces. On some metal surfaces no carbon will be deposited (e.g., aluminum in this study), and on some metal surfaces some carbon will be deposited, but only on catalytic surfaces such as Ni, Pt, or Pd will graphite form. Indeed, it has long been known that certain metals are catalysts for the formation of graphite and that on noncatalytic metals carbon deposition is slow and graphite does not form.1 Thus, on catalytic surfaces, and only on catalytic metals, a layer of graphite is quickly generated from the decomposed radicals. The process as described to this point explains the formation of the first layer of graphite only. A change in process must occur to explain the second and subsequent layers. Specifically, we postulate that once a graphite layer is formed it acts as a self-catalyst for further growth. That is, we make a novel postulate regarding the process which occurs if the radical adsorbs on a graphitic surface: On graphitic surfaces, carbon-containing radicals will decompose, leaving carbon atoms to react with “dangling bonds” at the edge of the graphite sheets.20,38,39 Carbon atoms reacting at these sites will naturally assume the minimum free energy configuration; hence, the graphite lattice will be extended. The remaining species of the decomposed radicals will desorb as molecules, or as new radicals. Thus, for example, a methyl radical that decomposes on the graphite surface will release a hydrogen molecule, and a methylene radical will possibly release a hydrogen molecule and a hydrogen atom. The existence of free hydrogen atoms on graphite is well-known.40-43 H atoms thus created will diffuse across the surface, possibly to react and form H2 molecules. (34) Arefeva, E. F.; Tesner, P. A. Formation of Pyrocarbon from Methane and Acetylene at Temperatures of 1350-1500 degree C. Combust., Explos. Shock WaVes (Engl. Transl.) 1986, 22 (3), 326-330. (35) Wei, T. C.; Phillips, J. Thermal and Catalytic EtchingsMechanisms of metal catalyst reconstruction. AdV. Catal. 1996, 41, 359-421. (36) Dean, V. W.; Frenklach, M.; Phillips, J. Catalytic Etching of Platinum Foils and Thin-Films in Hydrogen Oxygen Mixtures. J. Phys. Chem. 1988, 92 (20), 5731-5738. (37) Chou, C. H.; Phillips, J. Platinum metal etching in a microwave oxygen plasma. J. Appl. Phys. 1990, 68 (5), 2415-2423. (38) Xia, B.; Phillips, J.; Chen, C. K.; Radovic, L. R.; Silva, I. F.; Menendez, J. A. Impact of pretreatments on the selectivity of carbon for NOx adsorption/ reduction. Energy Fuels 1999, 13 (4), 903-906. (39) Phillips, J.; Kelly, D. J.; Radovic, D. L.; Xie, D. F., Microcalorimetric study of the influence of surface chemistry on the adsorption of water by high surface area carbons. J. Phys. Chem. B 2000, 104 (34), 8170-8176. (40) Menendez, J. A.; Radovic, L. R.; Xia, B.; Phillips, J. Low-temperature generation of basic carbon surfaces by hydrogen spillover. J. Phys. Chem. 1996, 100 (43), 17243-17248. (41) Weigle, J. C.; Phillips, J. Novel Dual-Bed Reactors: Utilization of Hydrogen Spillover in Reactor Design. Langmuir 2004, 20 (4), 1189-1193. (42) Chang, H.; Phillips, J.; Heck, R. Catalytic synergism in physical mixtures. Langmuir 1996, 12 (11), 2756-2761. (43) Weigle, J. C.; Phillips, J. Modeling Hydrogen Spillover in Dual-Bed Catalytic Reactors. AIChE J. 2004, 50 (4), 821-828.
Phillips et al.
The model described above for autocatalytic growth of graphite in a gas environment containing carbon-containing radicals is only intended to be a “conceptual” model. There is insufficient data for postulating with certainty the precise chemistry of the process. Many alternatives can be envisioned. For example, it is possible hydrogen adsorbs or coadsorbs at the active site positions alongside carbon species. In fact, this is postulated to take place in models of diamond and soot growth. In these models an extra step is required to extract the hydrogen before the lattice can continue to grow by carbon addition.28,29 In the third step, due to the catalytic nature of the metal, these deposited carbon atoms are directed by the catalyst template to form a graphitic layer directly on the metal surface. Alternatively, it could be that the edge sites are also the only active sites for homogeneous radical decomposition. Thus, only at these sites will radicals decompose, leaving the carbon atoms so-produced in graphite sites, concomitantly causing the graphite lattice to “grow” by one additional atom. In any event, it is clear that this mechanism will only work in the presence of a metal known to catalyze the formation of the initial layer of graphite or in the presence of an existing graphite lattice. The ability of all metals to catalyze graphite formation has been thoroughly studied, and it is clear that nickel, cobalt, platinum, and palladium1 are the most active graphite catalysts. Application of the Mechanism to the Data. The above models can be employed to explain both filament formation and template graphite formation observed in this work. Observations made in this study indicate that filamentous carbon is the preferred form, even when oxygen is present, when the process is carried out below about 800 ( 25 K, and template graphite growth is totally dominant above 825 K, except when no oxygen is present. In the absence of oxygen filament growth is relatively rapid even above 800 K. The precise temperature below which filament growth dominates appears to be a function of the ethylene-to-oxygen ratio. Moreover, it appears that the filaments are associated with small nickel particles. Thus, for example, it was rare to observe filament growth on “lattice nickel” (see Figure 4). The welldescribed standard mechanism for graphite filaments found in the literature, summarized above, is sufficient to describe all observations. It is interesting to note that the process of filament growth does not require “radicals”, but it does require small particles of precisely the same metals known to catalyze graphite growth. Thus, it is reasonable to anticipate that at temperatures too low to create radicals filamentous carbon can form directly via the decomposition of ethylene molecules on the nickel surface. Indeed, it is generally understood that radicals only begin to form during ethylene oxidation at around 775 K.30,44 Thus, the dominance of filamentous growth below this temperature is consistent with many earlier observations. It is also clear that there is a maximum temperature for controlled template graphite growth. This is an observation requiring additional study, but it is possible that carbon atoms produced by radical decomposition above about 900 K are more mobile than those generated at low temperature. Possibly, these radicals diffuse farther before incorporation into the growing graphite lattice. Such a growth process could lead to more random shapes, rather than faithful reproduction of the template shape. The proposed mechanism is consistent not only with the observations made in the present study, but also with observations made in earlier studies as well. Using pure acetylene, three types (44) Nae Lih, W.; Phillips, J., Reaction-enhanced sintering of platinum thin films during ethylene oxidation. J. Appl. Phys. 1986, 59 (3), 769-779.
Graphitic Structures by Design
of carbon deposits have been obtained on nickel, amorphous (at temperature 825 K), and graphite platelets (temperature >1300 K).45 Filamentous material has also been obtained in a mixture of ethylene, carbon monoxide, and hydrogen on iron powder by cooking at 875 K for 3 h.46 It is also notable that in none of the earlier studies was sufficient coverage of the metal surface achieved for the creation of selfsupporting graphite structures by any process (e.g., acid removal of metal).
Summary A method for making designed graphite structures in a lowtemperature combustion environment is described. Specifically, it is shown that graphite structures can be grown that precisely mimic the shape of an underlying catalytic metal (i.e., Ni) template in fuel-rich ethylene/oxygen mixtures at between about 750 and (45) Saito, Y.; Yoshikawa, T.; Okuda, M.; Fujimoto, N.; Yamamuro, S.; Wakoh, K.; Sumiyama, K.; Suzuki, K.; Kasuya, A.; Nishina, Y. Cobalt particles wrapped in graphitic carbon prepared by an arc discharge method. J. Appl. Phys. 1994, 75 (1), 134-137. (46) Park, C. F.; Baker, R. T. K. F. Catalytic behavior of graphite nanofiber supported nickel particles. 3. The effect of chemical blocking on the performance of the system. J. Phys. Chem. B 1999, 103 (13), 2453-2459.
Langmuir, Vol. 22, No. 23, 2006 9703
900 K. It is also shown that the process is radical based simply from experiments that show the rate of deposition is a strong function of the position/residence time of the reactor/reacting gas mixture. These findings led to the suggestion that the mechanism of the process is one in which initially homogeneously formed carbon-containing radicals decompose on the metal template, which “organizes” the carbon into a graphite shape catalytically. Subsequently, the radical decomposition process is repeated on the graphite surface that now covers the metal template. Carbon atoms, created by the decomposition, are deposited at graphite edge sites, enlarging the size of the graphite. That is, graphite growth is autocatalytic in a gaseous environment containing a large number of carbon-containing radicals. It was demonstrated that the finding has possible technological value by showing that growth rates are rapid (1 layer/s) and by creating a host of unique graphitic shapes from “lattices” to “foam” to hollow nanospheres. Self-supporting, macroscopic, pure graphitic structures could be made by dissolving the metal templates away with acid. LA060915C
