Synthesis of Highly Oriented Graphite Films with a Low Wrinkle Density and Near-Millimeter-Scale Lateral Grains

ABSTRACT: Principal defects found in graphite films include grain boundaries and wrinkles. These defects are well known to have detrimental effects on properties such as thermal and electrical conductivities as well as mechanical properties. With a two-fold objective of synthesizing graphite films with large single crystal grains and no wrinkles, we have developed a relatively low-temperature (<1150 °C) process that involves the precipitation of graphite films from nickel− carbon solid solutions followed by in-situ etching of the nickel foil substrates with anhydrous chlorine gas to obtain near wrinkle-f ree continuous graphite films. The size (La) distribution of lateral grains (or single crystal regions) in these highly oriented graphite films has been found to be asymmetric with the largest grains about 0.9 mm in diameter, as determined by electron backscatter diffraction (EBSD). Thus, the growth of large-area graphite films with grains much larger than those reported for highly oriented pyrolytic graphite (HOPG, ∼20 to 30 μm), and with a low density of wrinkles, as in the case of HOPG, but prepared at much lower temperatures, is reported here.


■ INTRODUCTION
Methods such as the molten flux route (>1500°C) 1−5 or the metal−carbon solid solution route 6−17 have been explored to synthesize graphite crystals in the bulk form or as thin films. However, the graphite films generated are often heavily wrinkled, a defect that, along with grain boundaries, dislocations, and vacancies, is largely detrimental to their properties, including thermal and electrical conductivities and mechanical properties. Moreover, although some published studies provide selected area diffraction patterns from transmission electron microscope (TEM) analyses, the analyses are limited to regions ≲1 μm 2 , and none of these studies report the grain sizes of these graphite materials. Another route to synthesizing graphite involves the thermal cracking of hydrocarbons at very high temperatures (>2000°C ) with applied pressure. This process tends to yield nonwrinkled "highly oriented pyrolytic graphite" (HOPG) as products, but the grain sizes are typically limited to ∼20 to 30 μm (for grade A HOPG). 18 Another variant of the molten flux method to prepare graphite is the kish-method, where graphite flakes are obtained by the precipitation of C from molten steel. Although these flakes may contain single crystals up to ∼1 mm in size, they contain iron impurities that must be removed by additional purification. 4 In this work, we have explored the synthesis of graphite films with minimal wrinkles and large grains by precipitation from Ni−C solid solutions in the temperature range of 1000−1150°C . The Ni−C solid solution may be considered as a model system to study and control the precipitation of graphite films for the following reasons: (1) the Ni{111} surface is lattice matched to the graphene layers, (2) Ni does not undergo additional reactions (such as phase changes or carbide formation) at our temperatures of interest that could otherwise complicate the process, and (3) C dissolution and segregation and precipitation from Ni have been previously studied in detail by in situ techniques such as LEED and Auger electron spectroscopy by several surface science groups. 19−23 Methane was selected as the C precursor of choice as it is a thermally stable hydrocarbon that can only be catalytically decomposed in H 2 -rich environments at temperatures <1200°C: this approach minimizes "soot" or amorphous carbon formation from thermally mediated reactions. Our typical graphite growth process involved (1) exposure of the Ni foils to CH 4 gas at T dissolution to form Ni−C solutions followed by (2) precipitation of the graphite films at a lower T precipitation under vacuum and (3) a final cooling step to room temperature (RT) (Process A, Figure 1). Films generated using this process were heavily wrinkled (or buckled), where wrinkles are defined as vertical deviations from a flat surface ( Figure S2). Such films are referred to as Type A films in this manuscript ( Figure 2 and Figures S3 and S4). The major cause of the formation of wrinkles in thin films synthesized at high temperatures (>800°C) is the thermal stress generated due to the mismatch in the coefficients of thermal expansion (CTEs) between the thin film and the growth substrate (other factors such as intrinsic stresses generated during growth and extrinsic stresses due to external factors may also cause wrinkling). For example, when cooled from a growth temperature of 1000°C to RT, the lattice parameter "a" of face-centered cubic (FCC) Ni changes from 3.58 to 3.52 Å (i.e., ∼1.7% contraction), 24 while for graphite, the change in the in-plane lattice constant is negligible ( Figure S5). 25,26 This unequal contraction introduces biaxial compressive stresses in the film that are released by forming a network of delamination buckles in the film.
Wrinkling may distort the local crystal symmetry and thus introduce bond strain and crystallographic defects in the graphite film. At a more severe level, wrinkling may even lead to bond breakage in graphite films. Wrinkling has been observed to degrade several material properties in graphene 27−29 or to enhance its chemical reactivity. 30,31 Strategies to prevent wrinkling in single-layer graphene include growth on copper substrates with a specific crystallographic orientation (Cu(111)), 32,33 switching to lower CTE substrates like metallic platinum 34 or nonmetallic germanium, 35 or growth at lower temperatures (Cu/Ni(111) at 750°C). 36 However, to the best of our knowledge, no attempts on eliminating wrinkling in graphite or multilayer graphene films have been reported in the literature.

■ RESULTS AND DISCUSSION
Our approach for preventing wrinkle formation in graphite films has been to remove physical contact between the Ni foil and the graphite films immediately after growth at 1000°C and then cool the resulting free-standing graphite films to RT. This essentially involved chemical etching of the Ni foil(s) at T precipitation or 1000°C. An ideal reagent for this purpose should have the following characteristics: it should (1) be stable at 1000°C, (2) have sufficient vapor pressure for easy introduction into a reactor under dynamic gas flow conditions, (3) selectively react with the Ni foil(s), (4) form reaction products with sufficient vapor pressures that can be removed without contaminating the graphite films, (5) not introduce defects into the graphite films, and finally, (6) not interfere with the graphitization process of the dissolved C in the Ni foil(s).
Commercial etchants and aqueous solutions of nonoxidizing and oxidizing acids, such as HCl, HNO 3 , H 2 SO 4 , or H 3 PO 4 , that are commonly used for etching Ni at temperatures <200°C cannot be used for high-temperature etching inside a reactor. Oxidizing reagents such as supercritical water or oxygen gas will also completely or partially oxidize the graphite film (and possibly also generate CH 4 , CO, or CO 2 ) while converting the Ni foil to nickel oxide powder. Similarly, sulfurcontaining etchants such as H 2 S acid vapor are known to produce stable S-doped graphene at temperatures above ∼600°C . 37 Again, anhydrous HCl or other nonoxidizing acid vapors cannot be used because HCl will produce H 2 gas, both by thermal dissociation (∼0.73% at 1000°C) 38 and upon reaction with Ni, which may etch the graphene layers. Moreover, as the remaining dissolved C in the Ni foil graphitizes as the foil is etched (discussed in detail later), the presence of H 2 may complicate this process by creating vacancies or sp 3 carbon. Indeed, H 2 −Cl 2 mixtures have been used to form nanodiamond particles or sp 3 carbons that are not wanted in our process. 39,40 It has been shown that the carbon−halogen bonds in halographenes 41 and graphite fluoride 42 are unstable at temperatures above 400−700°C. For example, the C−Cl bond is not stable at high temperatures and pristine graphene can be recovered from chlorinated graphene by heating to temperatures >500°C. 43,44 Indeed, one method of purifying natural and kish graphites industrially is by exposing them to anhydrous Cl 2 gas at high temperatures. 4 Moreover, the synthesis of carbide-derived carbons (CDCs) has been explored with the reaction of carbides MC x (M = Si, Al, Ti, Zr, V, Nb, Ta, etc.) with Cl 2 −Ar under dynamic flow and ambient pressure at ∼900 to 1000°C to form graphitic C and Figure 1. Details of the production of Type A and B films: (i) overall process, (ii) schematic of the process inside the reactor, and, (iii) plots of variation in C solubility with temperature in polycrystalline Ni showing the amounts of C precipitated as graphite during steps 3 (due to a reduction in temperature), and, 4 (due to Ni etching) of the process (ST = step). MCl x . 45,46 The other halogens react similarly with MC x and are all potentially suitable agents for etching Ni at high temperatures without damaging the graphite film. 46,47 Because anhydrous Cl 2 gas has a higher reactivity than Br 2 and I 2 , and because they are respectively a liquid and a solid that would require additional optimization of appropriate bubbler-type set-ups, we selected Cl 2 for the etching process. Fluorine was not selected because of its severe safety hazards.
In order to remove wrinkles, after the graphite films had formed at 1000°C (T precipitation ), anhydrous Cl 2 gas was introduced for ∼3 h to etch away the Ni foil sandwiched between the two graphite films that had grown on its two sides (Process B, Figure 1). The temperature for graphite growth and subsequent Ni etching was selected to be 1000°C to ensure that the NiCl 2 produced did not contaminate the graphite films because NiCl 2 has a high enough vapor pressure to be removed from the reaction zone under dynamic vacuum.
It is important to note that for the graphite growth runs in which the Ni was etched (process B), the Ni foils were placed in boats with covers to prevent the resulting free-standing graphite films from flying away under this dynamic vacuum ( Figure S6). As the Ni etching progressed, an orange anhydrous NiCl 2 powder was deposited on the cooler surfaces of the reactor ( Figure S7). When the graphite growth and Ni etching were completed, the Cl 2 gas was turned off, and the films were cooled to RT after a 5 min wait period at 1000°C. This wait time was necessary to pump out all the Cl 2 gas before cooling in order to eliminate any possible chance of C−Cl bond formation at lower temperatures (<500°C). At this point, these graphite films no longer had a Ni substrate in physical contact and hence did not experience the ∼1.7% biaxial compressive stress they otherwise would have had. At RT, the films were recovered from the furnace and floated in ethanol or acetone. However, the two graphite films that had grown on the two sides of the Ni foil had collapsed onto each other (with gaps of ∼80 to 100 μm) after the Ni etching process. The films were then separated by gentle mechanical agitation with liquid N 2 (see the Supporting Information for details) and tested for the presence of any metallic Ni(0) residue by bringing a magnet close to them. If the films were not attracted to the magnet, they were tentatively considered free of Ni(0) ( Figure S8). The residual Ni concentration was later measured on select films by inductively coupled plasmaoptical emission spectroscopy (ICP-OES). The results given in Table S2 show that the Ni content was quite small, ∼0 to 1%. The films were then transferred onto silicon wafers and air dried for further characterization. These films are referred to as Type B films in this manuscript.
The surfaces of the Type B graphite films were examined by optical microscopy ( Figure S9) and SEM ( Figure 2 and Figure  S10) and showed a huge reduction in the total number of wrinkles. The films had lost their corrugated appearance, and the larger wrinkles were completely absent. The few remaining wrinkles were possibly caused by mechanical stresses during handling of these free-standing submicrometer-thick films. AFM imaging of the graphite surfaces showed that such wrinkles were ∼5 to 20 nm high and ∼0.1 to 0.2 μm wide. AFM imaging also showed that the films were "wavy", as might be expected from such free-standing films. Unsurprisingly, the film had occasional tears or holes, which were most likely a result of mechanical damage during the separation of the two films ( Figure 2 and Figure S11).
The etching rate of Ni with Cl 2 was observed to be important for two main reasons. First, etching conditions that were "too harsh" mechanically damaged the films, forming, for example, torn and crumpled films when the Cl 2 flow rate or the reactor pressure were too high ( Figure S12). Second, at 1000°C , the Ni foil not only has graphite films on both surfaces but also about 1.3 at. % C in the bulk (Figure 1). This dissolved C graphitizes during the Ni etching step (neither the dissolved C nor the precipitated graphite are etched by Cl 2 at this temperature, as explained earlier). This suggests that the Ni etching rate may be an important parameter to consider to better control the C precipitation during Ni removal and the final structure of the graphite films. Hence, in order to form large graphite grains, the etching must be carried out at an optimized rate so that these C atoms are added to existing grains instead of randomly nucleating smaller new ones. Although we were able to optimize the etching rates so that minimally damaged films with large grains were produced, we were not able to independently evaluate the effects of rapid etching on the grain size, because this forms crumpled graphite films that could not be characterized by the diffraction techniques to be discussed later.
Graphite growth in process B thus occurs in two stages: (1) by nonisothermal precipitation as a result of a reduction in temperature from T dissolution to T precipitation (step 3, Figure 1), and (2) by isothermal precipitation at T precipitation as the amount of Ni is reduced by chemical etching (thus continuously saturating the Ni with C; step 4, Figure 1). No obvious difference in quality was observed between the graphite precipitated in steps 3 and 4 (see Supporting Information for details). As a consequence of this Ni etching step, Type B films were found to be much thicker than the films grown without Ni etching, as performed for Type A films. Thus, Type B films have an average thickness in the range of ∼750 to 800 nm, while Type A films have an average thickness of ∼250 nm ( Figure S13), values that approximately match the predicted thickness from C solubilities (Predicted: Type B, 805 nm; Type A, 265 nm; Table S1).
The initial characterization of the chemical nature of the assynthesized films included electron energy loss spectroscopy (EELS), X-ray photoelectron spectroscopy (XPS) analysis, and Raman spectroscopy. EELS and XPS analyses showed that the film contained only sp 2 carbon and that no additional functionalization was present ( Figures S14 and S32). Raman mapping was performed on several samples to obtain information on the film quality from the I D /I G ratio, and it ranged from ∼0.04 to 0.09 ( Figure S15, Table S3-i), indicating that they were high quality with mostly sp 2 -bonded C atoms.
In order to determine the stacking order of the graphene layers, Raman mapping of the 2D peak was performed, and the stacking order was evaluated based on the method of Cancado et al. (see the Supporting Information for details). 48 For a Grade A HOPG sample with almost complete AB stacking, the metric obtained by this method was ∼1.0, whereas for both Type A and Type B films, it varied from ∼0.84 to 1.0 (Table  S3-ii). However, our films did show regions, where the AB stacking order is obviously broken ( Figure S17), but at this time, we are not certain what causes such disorder and hypothesize that such regions may be due to defects such as stacking faults or grain boundaries. These Raman results are consistent with the XRD data (discussed later) from which an "average" d 0002 spacing of 3.35 Å was obtained for all samples, which also implies that the layers are mostly AB stacked.
The crystal dimensions in both the "a" (L a or lateral dimensions) and "c" directions (L c ) must be determined to define the size of a single crystal or grain in our films, and electron backscatter diffraction (EBSD) has been used to determine the orientation and lateral grain size of the films. EBSD is a common technique for the determination of orientation and grain size in metallic and ceramic samples that utilizes Kikuchi patterns produced by backscattering of diffusely scattered electrons inside the sample for phase identification and determination of the crystal orientation. EBSD analysis has been used for characterizing graphitic regions in iron 49,50 and has been reported for graphite samples in two conference proceedings. 51,52 To the best of our knowledge, there has been no earlier work on the use of EBSD to obtain a detailed and rigorous evaluation of the grain orientation and size in graphite.
In order to verify that EBSD can be used to determine the graphite grain orientation and size, EBSD was first performed on flakes of natural graphite transferred to silicon wafers. In EBSD, the software utilizes a user-defined sample co-ordinate system and phase to match the Kikuchi bands of the sample with reference patterns and determines crystal orientation. 53 Thus, one relatively stringent requirement is that the sample surface must be completely flat and perpendicular to the normal direction (ND) of the sample co-ordinate system at all points. Hence, all samples were pressed with a roller a few times before measurement, which may have introduced additional wrinkles. Because graphite is a soft and layered material, no other chemo-mechanical polishing that is common for the preparation for metallic and ceramic samples for such studies was performed as it could have severely damaged the surface.
Orientation mapping involves the determination of the crystal orientation (or index) of all pixels in a scan and then delineating grain boundaries based on the mis-orientation between these pixels. Hence, the first step after collecting the scan from a natural graphite flake was to evaluate the quality of the Kikuchi patterns obtained using an image quality (IQ) map (Figure S18-i) followed by the evaluation of the confidence index (CI) map that shows the CI values associated with each pixel. The CI values represent the acceptability of the indexing results and range from 0 to 1 ( Figure S18-ii). Only orientations with CI values >0.1 were considered for subsequent analysis after processing (Figure S18-iii). Finally, the indexing success rate (ISR) was determined, where ISR is simply the ratio of the points with CI values >0.1 to the total number of scanned points. As shown in Figure S18, there is good correlation between the IQ and CI maps, and the ISR is ∼99.6%, indicating that this is a good scan. The details of this process have been provided in the Supporting Information.
The EBSD data may be represented by pole figures (PFs), inverse pole figures (IPFs), IPF maps, and kernel average misorientation (KAM) maps. The PFs show plots of specified crystal directions or poles of the different unit cells of the grains in the sample with respect to the sample coordinate system. On the other hand, IPFs show plots of a sample direction (such as the ND) with respect to specific crystal directions of the differently oriented unit cells in the sample. Thus, two key pieces of information are obtained simultaneously: the number or set of points reflects the number of grains in the sample, whereas their locations indicate the location of specific crystal directions of the different grains (i.e., single crystal regions) with respect to the sample co-ordinate system. In both the PFs and IPFs, the points have been weighted by the associated grain sizes for easy identification of a particularly large grain and multiple smaller grains, as is often the case. Figure S19-I-i shows the plots of the PFs for the ⟨0001⟩ and ⟨101̅ 1⟩ crystal directions of the different grains in the scan for natural graphite. It contains two large points (for the (0001) PF) or sets of points (for the (101̅ 1) PF) and several tiny (or sets of tiny) points, which implies that the scanned region contains two large grains. The IPF for the natural graphite scan ( Figure S19-I-ii) shows the plots of the sample ND with respect to the crystal directions of the different grains. It contains points close to the ⟨0001⟩ direction, implying that the ⟨0001⟩ directions of the different grains are essentially parallel to the sample ND. In other words, the {0001} planes of all grains in the film are parallel to the sample surface. The IPF also shows two coincident large points, implying that two large grains are present. Finally, the IPF map is color coded to represent the orientation of the ND of each pixel in the scan, which for this scan is parallel to the ⟨0001⟩ directions of all associated grains ( Figure S19-III). The last step of the EBSD scan analysis involves delineating the grain boundaries and calculating the grain size (see Supporting Information). The grain diameters for regions 1 and 2 have been calculated from their areas to be ∼500 and ∼450 μm, respectively. Figure S19-IV shows the KAM map from the natural graphite scan comparing the mis-orientation between neighboring pixels: the mis-orientation between the first nearest neighbors is generally <2°.
Several Type B films have been analyzed by EBSD. However, perhaps because of an uneven surface topography, the ISR values were somewhat lower. Smaller scan areas (≤125 × 200 μm) tended to have larger ISR values (∼99%), but the grain sizes were artificially limited by the areas scanned ( Figure  3 and Figures S20 and S21 and Table S4). To determine the "true" grain sizes, EBSD scans over an area of about 1.5 × 1.5 mm 2 were performed. Figure 4 shows the IPF and PF plots for two Type B films with the {0001} orientation and the largest grains about 0.9 mm in diameter. The (0001) PFs for both films show a single large point, and the (101̅ 1) PFs show one set of large points, confirming that these regions are single crystals. The yellow regions in the IPF map and the circled points in the PFs and the IPFs are from the exposed silicon wafer where the films have holes or tears ( Figure S27). A large area scan from a Type A-wrinkled film is shown in Figure S23 and shows that the largest grain diameter is about 0.8 mm. The KAM maps and details of all large area scans are provided in Figure S22 and Table S5.
A probable reason for the lower ISRs can be obtained by comparing the maps in Figure S24, where the SEM image (i), the "raw" IQ map (ii) and the IQ (iii) and the CI (iv) maps from postprocessing are shown. From these figures, it is obvious that nonflat regions have lower IQ values that result in low CI values and poor indexing. In this example, the nonflat regions are the ones where the film has torn materials deposited on top of it, which perhaps obscure the Kikuchi patterns from the film lying beneath it. However, even without material deposition on the top and a sharp Kikuchi pattern, the local surface curvature (i.e., any deviation from the user defined sample co-ordinate system) may result in a failure by the software to index the pixel with an acceptable CI value and lead to an overall lower ISR for the scan. Figure 5 shows a comparison between similar area scans from a Type B film, a natural graphite flake, and commercial grade A HOPG. It is concluded from the unique grain color maps that grain sizes in grade A HOPG are much smaller (6− 26 μm) than in the natural or Type B graphite samples. Again, the quasi ring-like appearance of the (101̅ 1) PF of grade A HOPG versus the single set of large points for natural graphite or the Type B graphite film clearly demonstrates that the grains in HOPG are much smaller.
XRD experiments that involved coupled ω-2θ scans with a 0-D detector were performed to probe the crystal structure in the "c" direction and showed that the films have a preferred orientation because only the {0002} family of peaks was detected. The d 0002 values calculated from these scans were all clustered around 3.354 Å (Table S7-i), which implies that the films are, on average, AB stacked (a higher d 0002 value would imply turbostratic stacking). 54 This result does not contradict the Raman spectroscopic analysis discussed earlier showing occasional deviations from AB stacking. This is because Raman spectroscopy has a much higher resolution (spot size of beam ∼700 to 800 nm at 100× magnification; information depth ∼30 nm), whereas XRD provides an average value for the entire film (spot size of beam ∼1 mm; information depth ∼entire film).
Mosaic spread is a parameter that is commonly used for HOPG samples to determine crystallite (or single crystal) alignment in the "c" direction and a lower value of this metric signifies greater crystallite alignment (see the Supporting Information for details). The mosaic spread values obtained were in the range of ∼0.1 to 0.2°for Type B films and ∼0.5°f or Type A films. This indicates that the crystallite alignment in Type B films is at least as good as in Grade A HOPG, keeping in mind that these films are not perfectly flat which may artificially increase the mosaic spread value.
High-resolution TEM (HR-TEM) was also used to study and compare Type A and Type B graphite films ( Figure 6). Electron transparent Type A films imaged in the top-down mode (plan view TEM) showed a heavily wrinkled appearance with 3−4 wrinkles often emerging from common junction (v) grain size chart with the largest grain highlighted in gray. Grain boundaries are represented in white and black areas represent pixels with CI < 0.1 that have not been included in the IPF maps. Yellow regions in the IPF maps or the points circled in yellow on the IPFs and PFs are from the bare silicon wafer. The points in the PFs and IPFs have been weighted by grain size. ND refers to the normal or (001) direction of the sample coordinate system.
Chemistry of Materials pubs.acs.org/cm Article points. High-magnification images of these junction regions revealed "gaps" where the graphene layers had physically separated, possibly due to shear stress. Selected area electron diffraction (SAED) of such junctions was then done to explore whether wrinkling had locally modified the orientation. SAED showed no change in orientation although additional diffraction spots were present, possibly from the folded over part in the wrinkle. On the other hand, cross-sectional TEM imaging performed on thicker and thinner Type A films revealed two types of large angular bends (>100°) in the {0002} lattice fringes of the graphene layers. These include ∼130 to 160°conformal bends adjacent to the Ni substrate and ∼130 to 150°nanobuckles within the graphene layers of the film. Nanobuckles may show variations in the d 0002 lattice fringe spacing and may have adjacent physical gaps. Both nanobuckles and conformal bends may have neighboring regions of lower crystallinity. The details of the TEM characterization of Type A-wrinkled films and the possible implication of buckling of the graphene layers on the growth process and film properties will be reported elsewhere. For Type B films, the plan-view HR-TEM image on a film thinned to electron transparency by mechanical exfoliation showed the region to be a single crystal with AB stacking. Figure 6-viii,ix show the cross-section of a Type B film and respectively a high-resolution image and a diffraction pattern, proving that it is a highly crystalline material. Again, the sharpness of the diffraction spots indicates that this is a singlecrystal region.
In order to further understand the role of the Ni substrate on graphite grain size, attempts were made to correlate the grain structure of the Ni substrate with that of the graphite film. It was found that considerable grain growth had occurred in the Ni foil after the H 2 pretreatment step (step 1, Figure 1) and that the foil contained grains of different orientations that could be as large as ∼850 μm ( Figure S29). The fact that the largest graphite grains were also ∼800 to 900 μm in size suggests that a relationship may exist between the grain sizes of the graphite film and the Ni substrate. Again, the graphite grains were always found to have the {0001} orientation. Thus, the graphene layers of the films grow parallel to the Ni substrate irrespective of the orientation of the Ni grains. This has been previously observed by Johansson et al. while exploring the orientations of their graphite films and the underlying polycrystalline Ni substrate by electron diffraction in the TEM. 55 Additional details on the role of the substrate on graphite film growth will be reported elsewhere.
One of the major improvements expected from the removal of wrinkles in graphite films is a significant increase in their Young's modulus/ stiffness. Hence, removing wrinkles from an area fraction = 0.1, from films that have wrinkle amplitude/film thickness ratios ranging from 1 to 10, would increase their respective Young's modulus by a factor of 0.6−60. 56 However, the removal of wrinkles is not expected to significantly influence the film strength and the potential applications for wrinkle-free graphite films are where "high-modulus carbon films" are required (similar to high-modulus carbon fibers that are obtained by the removal of the misalignment between the constituent crystalline carbon ribbons).
On the other hand, removal of grain boundaries is expected to increase the film strength and stiffness. 57,58 An increase in grain size has been observed in silico as capable of increasing the fracture strength of polycrystalline graphene, following an inverse Hall−Petch-like relation and demonstrating that the grain boundary junctions are the weakest links. From this result, a ∼2 times increase in strength may be expected with an increase in grain size d in a graphite film of a size L, when the  Chemistry of Materials pubs.acs.org/cm Article L/d ratio changes from 10,000 to 10. Accordingly, the potential applications for both wrinkle and grain boundaryfree graphite films are where "high-modulus and high-strength carbon films" are required. A detailed discussion can be found in the Supporting Information.

■ CONCLUSIONS
We have shown that we can grow graphite films with few wrinkles by growing them on Ni foils at high temperatures and then removing the Ni substrates by chemical etching to eliminate issues caused by the CTE differences between the films and the foils. Moreover, by optimizing the process parameters, we have grown continuous graphite films with near-millimeter-scale lateral grains. We believe that this is the first report of an attempt to reduce wrinkling in thin films of a layered material, with single-layer graphene being a notable exception. This is also the first rigorous attempt to unambiguously determine grain sizes in graphite by EBSD, which has been used extensively for this purpose. Initial calculations on mechanical properties predict a significant improvement in axial stiffness and strength in these films attributed to a reduction in the number of wrinkles and an increase in grain size. However, intrinsic stresses from the growth process and external stresses generated after the growth process that may cause other defects including fine wrinkles have not been eliminated or studied. The current technical limitations of the overall process include the formation of holes or tears in the films and the inability to obtain pristine thin films of graphite or multilayer graphene, both due to the difficulty of handling such free-standing films in the reactor without mechanical damage. Such limitations may be removed with appropriate equipment and experimental design modifications, such as by using more suitable sample holders. In the future, we hope to see such films generated so that the expected property improvements discussed above can be studied and evaluated.

■ EXPERIMENTAL SECTION
In a typical process, the reactor was loaded with ∼1 cm 2 pieces of Ni foil (thickness: 100 μm; purity: 99.994%, metals basis) in quartz boats covered with alumina lids. The furnace temperature was slowly increased to 1150°C in 1 h 15 min (∼15°C/min) under an Ar/H 2 flow at 100 Torr followed by a 15 min hold at 1150°C (step 1). This was followed by a CH 4 exposure step at a pressure of <0.1 Torr (step 2). Then, all gases were turned off and the temperature was quickly reduced to 1000°C (at 10°C/min) and held at 1000°C for 1 h under dynamic vacuum to precipitate the graphite films (step 3). This quick temperature reduction was to minimize graphite precipitation during this ramp down process and ensure that most of the film is generated at 1000°C. In one set of experiments (Process A, Figure  1), the samples were then fast cooled to RT by switching off and sliding the furnace ∼20 cm away from the samples and turning on a table fan. Finally, the Ni foils with graphite films on both sides were recovered. Selected films were delaminated using the electrochemical delamination method 59 with ∼1 M NaOH after (spin)coating them with polymeric PMMA (polymethyl methacrylate) support layers. These graphite-PMMA films were transferred to silicon wafers or other substrates of choice followed by PMMA removal using acetone and H 2 /Ar treatments at 400°C.
In a second set of experiments (Process B, Figure 1), after the formation of the graphite films at 1000°C, ∼10 to 20 sccm of anhydrous Cl 2 gas was introduced into the reactor for ∼3 h (step 4). At the end of this process, the Cl 2 gas was turned off, and after a 5 min wait time, the furnace was turned off and cooled to RT (step 5). At the end of this process, only unsupported graphite films were recovered. The graphite films were then collected and floated in ethanol or acetone. At this point, each recovered film was composed of two graphite films ("collapsed" films) that had originally grown on the two sides of each Ni foil and collapsed onto each other during the Ni etching. Liquid N 2 was added slowly to the ethanol where the films were floating, and these two graphite films were separated by gentle mechanical agitation. The films were then transferred to substrates of choice for further characterization.
Additional experimental details including characterization methods can be found in the Supporting Information.
Details of synthetic and characterization methods and mechanical property calculations; and profilometry data, XPS spectra, EELS spectra, Raman spectra, XRD data, and additional optical microscopy, SEM and AFM images, and EBSD scans (PDF) 1 was used as received.

Characterization
AFM was performed on a Bruker Dimension Icon system. SEM imaging was performed on a Verios 460 scanning electron microscope and EBSD scans were obtained using a Hikari camera from Ametek attached to this SEM. TEM images and EELS spectra were recorded on a probe corrected FEI Titan G2 60-300 instrument with an electron monochromator at 80 kV. XRD data was obtained on a Rigaku SmartLab powder X-ray diffractometer. Confocal Raman spectroscopy was performed on a WiTEC instrument using a 532 nm laser and 600 g/mm or 1800 g/mm gratings. XPS measurements were acquired on a Thermo Scientific ESCALAB 250 Xi instrument. ICP-OES data was acquired on a Varian 700-ES instrument.

Sample Preparation for EBSD
Flakes were extracted from natural graphite crystals or HOPG pieces by pressing them to sticky notes followed by removal of the adhesive sticky notes with HPLC grade acetone. The free-floating pieces were then transferred to silicon wafers and dried in an oven at 80-100 °C. All the natural graphite or HOPG flakes thus obtained and Type A and Type B graphite films on silicon wafers were pressed a few times using a Kolami 330 S roller prior to performing EBSD scans.

Sample Preparation for Cross-section TEM
Regions of interest (ROI) from selected graphite films on Si wafers or Ni substrates were first coated with a Ti film and then with a black marker pen outside the SEM chamber. Once inside the SEM chamber (Helios 450HP) the ROI was coated with amorphous carbon and then sectioned with a focused ion beam (FIB) of gallium ions at 30 kV, and at an angle of ±2° to the surface normal. After transferring this wedge to a half grid, the final milling was also done with a gallium FIB sequentially at 30 kV, 5 kV and 2 kV and without tilting. This procedure has been adopted from a previous study by Colby et. al. to minimize ion beam damage to the graphite film. 1 Figure S1. Figure representing a cross-section sample prepared from a graphite film for TEM imaging.  Figure S2 depicts what has been defined as the height of a wrinkle and how it has been measured using AFM. Generally, AFM analysis is performed on the side that was away from the Ni substrate during growth (outer side, Figure S3). However, select films have been flipped over and AFM has been performed on the side that was adjacent to the Ni surface (inner side). They show similar wrinkling patterns ( Figure S4).    (ii) Raman spectrum of anhydrous NiCl 2 . 6

Raman Spectroscopy of Graphite Films and HOPG
A method by Cancado et al. has been used to determine the AB stacking order from the 2D peak. 7 Briefly, this involved fitting the 2D peaks with a 2D1 peak at ~2714 cm -1 , a 2D2 peak at ~2670 cm -1 and a 2DT peak (T: turbostratic peak) at ~2698 cm -1 (Figure S16). The AB stacking order over the scanned area was then calculated as the ratio of the area of the 2D2 peak to the sum of the areas of the 2D2 and 2DT peaks and compared to HOPG. For a Grade A HOPG sample with near complete AB stacking of the graphene layers, this 2D area ratio was ~1.0. Hence, scanned areas from the graphite films that had an area ratio equal or close to 1 were considered to be AB stacked.  Figure S16. An example of a 2D peak from a Type B graphite film being fitted with a 2D 1 , a 2D 2 and a 2D T peak. As the area under the 2D 2 peak is negligible, the graphite film at this point may be considered to be AB stacked. Figure S17. Examples of Raman spectra showing turbostratic character where, G1 denotes a 600 g/mm grating and G2 denotes a 1800 g/mm grating. The 2D peaks were recorded with the higher resolution G2 for clarity.

EBSD Analysis of Graphite Films
EBSD scans were typically performed using a 6.4 nA current and at 20 kV with the sample surface tilted by 70° and using a hexagonal sampling grid.
Step sizes for the maps ranged from 0.25 -25 µm.
The standard post-processing steps included confidence index (CI) standardization (where the low CI of a point was replaced by the higher CI value of a neighboring point with the same orientation) followed by the creation of a partition that excluded points with CI < 0.1. Finally, the grain tolerance angle and minimum size were defined before the different plots and charts were generated.
As explained by Demirel et. al., 8

Sample Details
Step size (µm)

ISR (%) Largest grain size Diameter (µm)
Area (mm 2 ) Type B (Fig. 4-I       to be determined by the Scherrer equation. [9][10][11] Mosaic spread may be determined by varying or 'rocking' the incident beam relative to the Bragg angle of interest.

H2 pre-treatment of Ni
The first part of all Type A and Type B processes included a H2 pre-treatment step (step 1, Figure   1). In addition to cleaning the Ni surface from surface oxides, considerable grain growth was observed, with larger average grain sizes with increasing temperature. Figure S28. Schematic of the H2 pre-treatment step reproduced to determine Ni grain size.

Graphite Films from 25 µm thick Ni foils
Graphite films with smaller grains were obtained under the same process conditions (Process B) with the largest grain sizes ~0.24 mm.

Mechanical property calculations
Effect of wrinkles: The increase in in-plane stiffness may be estimated using reference [13] for a film of thickness T from which wrinkles, having an amplitude A that is small compared to their mean separation distance and occupying an area fraction f, have been removed. We have accordingly found that E(f=0)/E(f)=1+k*f*(A/T) 2 , where k=6 is a theoretical value and E(f=0) and E(f) represent the Young's modulus of the wrinkle-free and wrinkled films respectively. 13  Accordingly, the potential applications for both wrinkle-free and grain boundary-free graphite films are where "high modulus and high strength carbon films" are required. Figure S33. Schematic depicting a possible route for the formation of Type B graphite film in steps 3 and 4 (as described in Figure 1).

Quality comparison of graphite formed in Steps 3 and 4 in Process B
The schematic above takes the following observations into account: (1) the etching started from the edges of the Ni foils and proceeded inward such as in Figure S8-(ii), and (2) a Ni foil with an area ~1 cm 2 formed 2 graphite films (on its top and bottom surfaces) with ~1 cm 2 lateral areas. The surface of the graphite film that was originally in contact with the Ni foil substrate has been referred to as the 'inner' surface and the opposite surface as the 'outer' surface. Thus, as per the above schematic, regions closer to the outer surface were formed during step 3, while regions closer to the inner surface were formed during step 4. Samples have been transferred to silicon wafers such that either the inner or the outer surface was exposed and thus, the surface studied. No obvious differences have been observed between the outer and the inner surfaces (from similarly synthesized different films) in terms of graphite quality, using multiple characterization techniques. Examples include: 1. EBSD scans: Figures 4-I and 4-II are, respectively, from the inner and outer surfaces of two different graphite films.  Figure S34) from HR-TEM imaging.

Figure S35
shows EELS spectra collected at ~20 nm intervals from a cross-section sample. For the C-K edge, the maximum π*/σ* intensity was observed close to the 'inner' surface of the film, implying that the highest quality graphite might be contained in this region. In any case, we note that the variation is relatively small, and the sample is "well graphitized" throughout the entire cross section. (Please note that variations in quality in cross-section TEM samples may also arise from exposure to ion or electron beams during sample preparation or imaging).        on which EBSD and other characterization would be possible. Our efforts as described above suggest to us that new methods (inventing new methods) would seem to be helpful for preparing "sections" of graphite film samples.