High-Quality Reduced Graphene Oxide by CVD ... - ACS Publications

6 downloads 18 Views 3MB Size Report
Jan 20, 2016 - graphene flakes or CVD graphene and call for improved oxidation and ..... quality at the edges was lower compared to center regions of.

Article pubs.acs.org/JPCC

High-Quality Reduced Graphene Oxide by CVD-Assisted Annealing Stefan Grimm,†,‡ Manuel Schweiger,†,‡ Siegfried Eigler,§ and Jana Zaumseil*,† †

Institute for Physical Chemistry, Universität Heidelberg , D-69120 Heidelberg, Germany Department of Materials Science and Engineering, Friedrich-Alexander-Universität Erlangen-Nürnberg, Martensstraße 7, D-91058 Erlangen, Germany § Department of Chemistry and Pharmacy, Institute of Advanced Materials and Processes, Friedrich-Alexander-Universität Erlangen-Nürnberg, D-90762 Fürth, Germany ‡

S Supporting Information *

ABSTRACT: Graphene oxide is a promising solution-processable precursor for the mass production of graphene thin films. However, during the wet chemical oxidation and reduction process toward reduced graphene oxide (rGO) a large number of defects are created. Although it is possible to synthesize rGO with an average defect distance of 3−4 nm, the performance is still limited. Here we demonstrate the partial restoration of the graphene basal plane of rGO by annealing in Ar/H2/isopropanol flow. Detailed statistical Raman analysis over large areas corroborates that the mean defect distance increases from initially 2−3 to 10−12 nm after CVD annealing. Some areas even reach defect distances of up to 18 nm. However, residual manganese impurities from the oxidation process lead to undesired carbon nanotube growth on the substrate under these conditions and had to be removed before the deposition of graphene oxide flakes on the substrate. The observed defect reduction during CVD annealing indicates that the lattice defects in rGO are mostly decorated vacancies that can be healed by addition of carbon under suitable conditions.



INTRODUCTION The synthesis of graphene oxide (GO), for example by Hummer’s method,1 is easily scalable as shown by Lee et al.,2 which makes GO a promising precursor for the mass production of graphene-based devices. Yet, the creation of defects in the carbon lattice that persist after the reduction to reduced graphene oxide (rGO) is inevitable during this wet chemical process. These defects lead to vastly inferior electrical characteristics of rGO compared to mechanically exfoliated graphene flakes or CVD graphene and call for improved oxidation and reduction methods. By strict temperature control of the oxidation process, the final defect density in rGO can be lowered substantially as shown by Eigler et al.3 This type of graphene oxide could also be termed as oxo-functionalized graphene (oxo-G) since it is functionalized with hydroxyl groups and epoxy groups, next to few structural defects with concentrations below 1%. The low density of structural defects enables large flakes as well as a better reconstruction of the graphene lattice during reduction. Consequently, reduction routes that were introduced by Pei et al. and Eigler et al. have led to highly conductive rGO flakes, which indicate a largely restored graphene lattice with decorated vacancies as defects.4,5 However, the average defect distance LD of these materials is still only 3−5 nm. Decreasing the defect density further is however crucial for applications of rGO, especially in electronic graphene devices. In order to achieve this goal, the defect density of a sample must also be quantified reliably. Raman microscopy is the most versatile tool for carbon allotrope characterization,6,7 combining © 2016 American Chemical Society

chemical sensitivity with submicrometer spatial resolution even over large areas as is necessary for thin films of rGO flakes. Four characteristic features of the Raman spectrum of defective graphene are used for analysis: the defect-related D-mode (∼1350 cm−1), the G-mode (∼1590 cm−1), the D′-mode (∼1620 cm−1), and 2D-mode (∼2690 cm−1). The 2D-mode arises from resonant two phonon intervalley phonon scattering and its presence and a narrow full width at half-maximum (fwhm(2D)) together with the absence of a D-peak indicate a defect-free sp2-hybridized carbon lattice. The D-peak only occurs in graphene with Raman-active defects, and thus the intensity ratio of the D-mode and G-mode peak (ID/IG) is a measure of the defect density, as reviewed by Ferrari et al.8 Any disturbance of the honeycomb lattice of graphene can be considered a defect. Numerous different defect types (e.g., grain boundaries or migrated vacancy defects, Stone−Wales defects, single and double vacancies, adatoms with a low migration energy, etc.) have been predicted and found in graphene. Most of these defects exhibit large energies of formation and should not be present at thermal equilibrium, as reviewed by Banhart et al.9 In graphene oxide, however, due to the harsh chemical functionalization of graphite, functional groups like hydroxyl or carboxylic acid moieties can be found at flake edges or at vacancy defect sites even after chemical reduction. These Received: November 27, 2015 Revised: January 17, 2016 Published: January 20, 2016 3036

DOI: 10.1021/acs.jpcc.5b11598 J. Phys. Chem. C 2016, 120, 3036−3041

Article

The Journal of Physical Chemistry C

Figure 1. (A) Schematic illustration of the reduction of graphene oxide and single vacancy defect healing during CVD treatment. (B) Scanning electron micrograph of rGO flakes on Si/SiO2. Raman maps of the same area (100 × 200 μm) showing ID/IG ratio of rGO before (C) and after CVD annealing (E). CVD annealing conditions: 750 °C, 990 sccm Ar, 5 sccm H2, 5 sccm IPA, 10 min. Raman maps showing fwhm(2D) of rGO before (D) and after CVD annealing (F). Scale bar in (C−F) is 20 μm. Arrows indicate the positions (1, 2, 3) of single Raman spectra of monolayered rGO.

could be decorated and stabilized with hydroxyl moieties, withstanding chemical reduction conditions. By annealing in the presence of hydrogen and a carbon source, reduction of those decorated defects and the insertion of carbon into the basal plane of rGO are possible and would increase the size of intact sp2-domains. Starting with GO prepared by Hummer’s method, Cheng et al. achieved the highest quality rGO so far with a defect distance of 16.6 nm using methane as carbon source in a remote plasma enhanced CVD system.14 Here, we apply CVD-assisted annealing in argon/hydrogen and isopropanol (IPA) atmosphere to rGO flakes of already high quality and show further restoration of the sp2-carbon lattice. Because of the highly inhomogeneous nature of rGO thin films, that is, flakes with different sizes and thus edge contributions, and various defect densities, we use large area Raman mapping and statistical analysis to quantify the density of defects reliably as shown by Englert et al.17 We determine the defect density over a large number of monolayer rGO flakes before and after CVD annealing. Statistical evaluation and cross-correlation of the ID/IG ratio with the width of the 2Dmode enable extraction of a mean defect distance and also show the change of rGO quality depending on process parameters. Annealing of rGO under hydrogen and IPA flow improves the quality of rGO substantially and increases the average defect distance from 3 nm to up to 13 nm. However, we also find that small amounts of residual manganese from the oxidation procedure lead to unwanted nanotube growth under these conditions and had to be removed before GO deposition.

groups tend to stabilize dangling bonds and inhibit the selfreorganization of the honeycomb lattice. Sato et al.7 and Cançado et al.6 distinguished two main stages of defective graphene: stage 1, where the honeycomb lattice is largely intact and the mean defect distance LD is larger than 3 nm, i.e., sp2-hybridized carbon domains contain at least 300 atoms. The ID/IG ratio increases with defect density and the 2D-peak broadens. In stage 2, the graphene is dominated by defects, LD is shorter than 3 nm, and the intact sp2-domains between defects are small. The G-mode intensity is also low because it directly depends on the size of the intact honeycomb lattice. Thus, in stage 2 the ID/IG ratio actually decreases with further increasing defect density. The fwhm(2D) increases and the D′- and G-mode merge into one broad feature as observed for GO.6,7,10,11 The transition between stage 1 and stage 2 graphene occurs at a maximum of ID/IG ∼ 3.8 (for a Raman excitation wavelength of 532 nm). By combined analysis of the ID/IG ratio and the fwhm(2D), a reliable extraction of the defect distance for stage 1 graphene is possible by using the equation introduced by Cançado et al.6 LD =

−1 4.3·103 ⎛ ID ⎞ ⎜ ⎟ E L 4 ⎝ IG ⎠

where EL is the laser energy (here 2.33 eV). The presence of a 2D-mode peak (fwhm ∼70 cm−1), an observable D′-mode, narrow D-mode (fwhm ∼45 cm−1), and G-mode generally indicates high quality rGO, e.g., as produced by the method by Eigler et al.,3,12,13 and justifies the application of this equation for defect density calculation. One approach toward reducing the defect density in rGO even further is chemical vapor deposition (CVD) assisted annealing with a carbon source. Such treatment with, for example, CH4, ethylene, or ethanol was previously found to result in the reduction and significant lattice restoration of GO,14−16 and one may expect similar enhancement for rGO flakes that are already of good quality. Figure 1A shows a simplified structure of GO with one single vacancy defect in addition to epoxy, hydroxyl, and acidic moieties due to the wet oxidation of graphite. After reduction most of these moieties are removed, but the vacancy remains. This vacancy defect



EXPERIMENTAL METHODS Sample Preparation. Graphene oxide dispersions in water/methanol (3:1 vol %) were prepared according to the modified Hummer’s method using concentrated sulfuric acid and potassium permanganate. The hydrolysis was carried out under strict temperature control following the procedure by Eigler et al.3 GO films were transferred onto oxygen plasma treated Si/SiO2(300 nm) substrates by the Langmuir−Blodgett (LB) technique and subsequently reduced by a mixture of hydriodic acid (HI) and trifluoroacetic acid (TFA) for 10 min at 80 °C. The obtained rGO films were annealed at 80 °C for 30 min to prevent delamination. 3037

DOI: 10.1021/acs.jpcc.5b11598 J. Phys. Chem. C 2016, 120, 3036−3041

Article

The Journal of Physical Chemistry C

and fwhm(2D) to 59 ± 9 cm−1 for the exact same sample area (see Figures 1E and 1F, respectively). For direct spectral comparison three large rGO flakes in Figure 1B were picked (arrows) and labeled as 1, 2, 3 before and as 1′, 2′, 3′ after CVD-assisted annealing. The corresponding spectra are shown in Figure 2A in comparison

CVD Annealing. rGO films on Si/SiO2 were placed in a 1 inch quartz tube. The tube was flushed for 10 min with 1000 sccm argon and heated to the target temperature under argon flow. After the system reached the target temperature, all gas flows (Ar, H2) were adjusted (see Table S1) and kept for 10 min before cooling down to room temperature under pure argon flow. The carbon source was introduced to the system by argon bubbling through cooled IPA.18 For annealing under completely oxygen and water-free conditions the quartz tube with the samples was evacuated to 5 mbar, baked at approximately 75 °C for 30 min, and cooled down under argon flow. This procedure (vacuum−bake−purge process) was repeated twice before CVD annealing was started. Characterization. Raman spectra were recorded with a Renishaw inViaReflex confocal Raman microscope with a 532 nm Nd:YAG diode laser, a 50× objective (NA = 0.75), and 2400 lines/mm grating. A line lens was used to collect Raman spectra (StreamLine mode) over a large area within a short time frame and with minimal thermal stress to the sample. The spectrometer was calibrated to silicon (520.6 cm−1). For statistical analysis, large areas of 100 × 200 μm were scanned with lateral increments of 1.3 μm (horizontal and vertical). Approximately 11 000 spectra were collected, filtered, and linear background corrected, and the D-, G-, D′-, and 2D-peaks were fitted by a self-written Python software. Lorentzian peak fits were optimized utilizing the Levenberg−Marquardt algorithm. Spectra showing multilayered, graphitic, or uncoated substrate regions were discarded by using the G-band intensity as the main argument. Thus, the final statistics mainly show monolayer rGO (∼90%). Atomic force micrographs (AFM) were recorded in tapping mode with a Veeco NanoMan VS atomic force microscope. Scanning electron micrographs were obtained with a Carl Zeiss Auriga field-emission scanning electron microscope (SEM, 1 kV acceleration voltage).



RESULTS AND DISCUSSION The optimized oxidation and reduction method by Eigler et al.3 produces rGO flakes with lateral dimensions of up to 30 μm as shown in the scanning electron micrograph in Figure 1B and heights between 1.3 and 1.5 nm (see atomic force micrograph in Figure S1) after Langmuir−Blodgett (LB) transfer. A large area of such flakes (corresponding to the SEM image) was analyzed by Raman microscopy. The presence of multi- and monolayers as well as clean substrate regions were determined via the G-mode intensity histogram of the corresponding Raman map (see Figure S2). Monolayered flakes of rGO covered over 80% of the examined surface, whereas 9.5% of the surface was covered with graphitic or multilayered material. Areas with multilayered rGO and bare substrate were rejected from further analysis and appear white in all Raman maps. The quality of the pristine monolayer rGO is quantified in terms of ID/IG and the fwhm of the 2D mode as shown in Figures 1C and 1D as color-coded Raman maps. For the pristine rGO flakes average ID/IG values of 2.4 ± 0.3 and 70 ± 13 cm−1 for fwhm(2D) were extracted. These correspond to a mean defect distance of 2−3 nm. As schematically shown in Figure 1A, one might expect an improvement of the rGO quality after annealing in a reducing atmosphere with a carbon source. Indeed, after annealing this sample at 750 °C in 990 sccm argon, 5 sccm H2, and 5 sccm argon/IPA, we find a significant improvement of the rGO quality as indicated by the decrease of ID/IG ratio to 1.4 ± 0.4

Figure 2. (A) Raman spectra of average GO, average monolayer rGO before CVD annealing, and single spectra of positions 1′, 2′, 3′ after CVD annealing (750 °C; 990 sccm Ar, 5 sccm H2, 5 sccm IPA, 10 min). (B) Statistical analysis of Raman maps in Figure 1. The ID/IG ratio and fwhm(2D) of spectra from positions (1, 2, 3, circles) before and after (1′, 2′, 3′, squares) CVD annealing are marked.

to average GO and nonannealed rGO. We calculated the defect distances LD for these three points before annealing as 1.8, 2.1, and 2.0 nm, respectively. After CVD annealing the defect distances at these points drastically increased to 12.0, 13.5, and 17.8 nm. These values correspond to sp2 domains that contain approximately 4500, 6000, and 10 500 carbon atoms, respectively. The last value corresponds to an area with only 0.01% of lattice defects. The substantial improvement of rGO quality at these selected points indicates that the remaining defects after wet-chemical reduction of GO are mostly functionalized carbon vacancies with −OH or −COOH moieties that could be healed by low-temperature CVD annealing as illustrated in Figure 1A. Because of the highly inhomogeneous defect density distribution in a thin film of many rGO flakes with a variation of sizes, statistical analysis of all spectra is necessary for an estimation of the average rGO quality as shown in Figure 2B. A 3038

DOI: 10.1021/acs.jpcc.5b11598 J. Phys. Chem. C 2016, 120, 3036−3041

Article

The Journal of Physical Chemistry C clear overall shift toward lower ID/IG ratios, and a reduction of the fwhm(2D) can be observed. The average defect distance after annealing can be calculated as 10−11 nm. The number of rGO spectra with a fwhm(2D) < 50 cm−1 nearly triples after annealing from 5.8% (pristine) to 16.4%, thus indicating larger intact sp2-carbon lattices. Note that bilayers of graphene show lower reactivities than monolayers and edge regions exhibit higher reactivities.19 After CVD annealing the increase of rGO quality at the edges was lower compared to center regions of rGO flakes. To exclude misleading data of rGO edges and multilayered material, we extracted the Raman data from interior positions of monolayers only (see Figure S3). The decreased defect density suggests that, indeed, the sp2carbon lattice of rGO is repaired by annealing with an additional carbon source. Interestingly, the same treatment of GO flakes before reduction with HI/TFA does not lead to the same improvement of the carbon lattice (see Figure S4). Although rGO is formed and a 2D peak appears, the average ID/IG ratio is about 1.0 ± 0.2, and fwhm(2D) is very broad with about 100 cm−1. These values place this rGO in the defectdominated stage 2. It suggests that under these reducing conditions the defects within the GO are not efficiently removed, but lattice rearrangement and loss of carbon take place. Higher temperatures, plasma, or catalytic surfaces are probably necessary to enable direct restoration of the sp2 lattice starting from GO flakes. Annealing of rGO with hydrogen but without IPA leads to less improvement than with IPA (see Table S2 and Figure S5) with a mean defect distance of 8−9 nm. The obtained defect density is higher than without an additional carbon source. Thus, we conclude that under CVD conditions with IPA significant carbon insertion into the honeycomb lattice of rGO takes place after reducing decorated vacancy defect sites by hydrogen. Although these defect distances are encouraging further improvement is desirable. Reckinger et al. showed that residual moisture and oxygen negatively influence the CVD growth of graphene on copper.20 Also, simulations and experimental data suggest that metallic residue, Si/SiOx nanoparticles, oxygen, or water can etch graphene.21−26 Therefore, we examined the rGO flakes after annealing by SEM (see Figure 3A) and found a large number of holes that indicate some loss of carbon even under CVD conditions with a carbon source, probably more defective areas. In order to exclude any residual oxygen or water (e.g., adsorbed on the quartz tube walls) during the CVD annealing, we refined our process and used a vacuum−bake−purge procedure before annealing that eliminated such effects. Even without isopropanol this procedure immediately reduced the ID/IG ratio to 1.4 ± 0.3 and resulted in defect distances of 10−11 nm (see Table S2, Figure S5). The percentage of spectra showing a fwhm(2D) < 50 cm−1 increased to 19.5%. With IPA as a carbon source during annealing we again found slightly larger defect distances (11− 12 nm, see Figure 3B), but we also observed significant growth of single-walled carbon nanotube (SWNT) (see Figure 4A and Figures S6 and S7). Raman spectra showed radial breathing modes (RBM) at wavenumbers between 150 and 300 cm−1, which indicate SWNTs with diameters of 0.9−1.7 nm. For the statistical analysis of this sample, the Raman maps were filtered so that SWNTs were excluded, and only the rGO flakes were taken into account. Clearly the appearance of SWNTs is surprising and unwanted; hence, a number of control experiments were performed to identify the source of the SWNT growth.

Figure 3. (A) Scanning electron micrograph of rGO flakes on Si/SiO2 after CVD annealing. Red areas indicate etched graphene, and dark gray areas correspond to double layers. (B) Statistical Raman analysis of 11 000 spectra of pristine rGO flakes (black), CVD-treated rGO (red, 750 °C, 990 sccm Ar, 5 sccm H2, 5 sccm IPA), and CVD-treated rGO with a vacuum-baked quartz tube (blue).

Particles of 3−5 nm diameter appeared after annealing even at lower temperatures (see Figure S8). Chua et al. recently showed that GO produced with permanganate as the oxidant contains large amounts of manganese (>5000 ppm), which is known as a catalyst for the growth of carbon nanotubes.27,28 We suspect that these metallic residues from the oxidation process form nanoparticles that catalyze SWNT growth during the CVD annealing step if etching of SWNTs, and rGO is suppressed by the exclusion of water from the system. To investigate the origin of these particles further, the supernatant of a precipitated GO dispersion (ultracentrifuged at 250000g for 8 h) was drop-cast on a cleaned substrate and annealed at 750 °C with 990 sccm Ar, 5 sccm H2, and 5 sccm IPA. In this case no particles or SWNTs were found on the surface. Thus, we assume that the manganese impurities are directly attached to the GO flakes. This notion was corroborated by complete oxygen plasma etching (10 min, 100 W) of rGO flakes on an LB-coated Si/SiO2 substrate and annealing under the same conditions afterward. Nanoparticles as well as SWNTs were found to be evenly distributed on the surface as shown by atomic force microscopy (see Figure S9). To exclude contaminations originating from the CVD system, control experiments with extensively cleaned substrates showed no SWNT growth or amorphous carbon after annealing under the same CVD conditions. Next, we attempted to remove the manganese from the GO flakes on Si/SiO2 prior to reduction with HI/TFA and annealing by EDTA chelation (immersion in 125 mM EDTA solution, pH ∼ 9 for 2 h) and HCl etching (immersion in 0.5 M HCl for 2 h). However, SWNTs still grew on these samples (see Figure 4B). Probably, manganese contaminations are 3039

DOI: 10.1021/acs.jpcc.5b11598 J. Phys. Chem. C 2016, 120, 3036−3041

Article

The Journal of Physical Chemistry C

Figure 4. False-color scanning electron micrographs of rGO flakes on Si/SiO2 after CVD treatment (750 °C, 990 sccm Ar, 5 sccm H2, 5 sccm IPA, 10 min) in a dried quartz tube. (A) Aqueous subphase during Langmuir−Blodgett transfer process. (B) Aqueous subphase during LB transfer and subsequent treatment with EDTA solution (2 h, 125 mM, pH ∼ 9). (C) HCl (∼0.5 M) subphase during LB transfer. Blue areas in SEM micrographs indicate monolayers of rGO. SWNTs are visible as bright lines in (A, B) on the insulating substrate but continue on the rGO flakes, although with much lower contrast.

located on both sides of the GO flakes. Thus, the removal of manganese from both sides of the GO flakes was attempted by exchanging the aqueous subphase during the LB transfer with HCl (∼0.5 M). The so coated samples were additionally immersed in 0.5 M HCl for 2 h prior to chemical reduction and annealing. This led to clean and smooth rGO surfaces after CVD annealing (see Figure 4C) without any formation of nanoparticles or growth of SWNT. Selected Raman spectra of monolayers of CVD-treated rGO are presented in Figure 5A comparing the pristine rGO with

defect distance of 10−11 nm. But this sample also showed SWNT networks, which are likely to be unwanted depending on the final application. For the HCl-subphase sample no nanotubes or etched areas were found after CVD annealing in Ar/H2/IPA flow in a dried tube and the rGO exhibited a mean ID/IG ratio of 2.0 ± 0.3 and a fwhm(2D) of 69 ± 8 cm−1 corresponding to a mean defect distance of 8−9 nm.



CONCLUSIONS We have shown that CVD-assisted annealing with Ar/H2/ isopropanol of already high-quality rGO flakes further decreases the defect density most likely through the reduction of decorated vacancy sites, subsequent insertion of carbon and thus repair of the sp2-carbon lattice. Other defects like Stone− Wales defects or line defects should still be present because their activation barrier is higher compared to decorated vacancy defects. Statistical Raman analysis was used to reliably quantify the quality of large areas of random rGO flakes, which confirmed an increase of the mean defect distance from 2−3 nm to 9−10 nm. Some flake areas even showed defect distances of up to 18 nm, which is well above previous reports for rGO. The significant defect reduction during CVD annealing indicates that the carbon lattice of chemically reduced rGO is largely intact, and the remaining vacancies can be filled by annealing under CVD conditions. We found that residual manganese from the oxidation process attached to the GO flakes results in carbon nanotube growth under water-free CVD conditions with a carbon source. Depending on the intended application the additional SWNT growth may be unwanted. The manganese residue can be removed by using an HCl subphase during the Langmuir−Blodgett transfer without too much impact on the final rGO quality. The presence of these impurities may also lead to uncontrolled surface reactivity of the rGO and highlights the shortcomings of the Hummer’s method for graphite oxidation. Alternative oxidation methods for the production of GO are thus highly desirable.

Figure 5. Comparison between CVD-annealed rGO samples with different subphases during LB transfer process of the GO. (A) Single, monolayer rGO Raman spectra of pristine rGO (black), CVDannealed rGO based on GO transferred by LB with aqueous subphase (red), and with HCl subphase (blue). (B) Statistical Raman analysis of rGO films in (A) over large areas.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.5b11598. Atomic force micrographs of GO and rGO flakes, filtering conditions of Raman maps, Raman analysis of CVD annealed GO, influence of different CVD parameters and Raman analysis after CVD processing

the CVD-annealed rGO obtained with water or HCl as the subphase for LB coating of the GO flakes. Although both samples showed a significant increase in quality, the sample with a pure water subphase reached a slightly better quality (see Figure 5B). The ID/IG ratio and fwhm(2D) were about 1.5 ± 0.5 and 59 ± 9 cm−1, respectively, corresponding to a mean 3040

DOI: 10.1021/acs.jpcc.5b11598 J. Phys. Chem. C 2016, 120, 3036−3041

Article

The Journal of Physical Chemistry C



(15) López, V.; Sundaram, R. S.; Gómez-Navarro, C.; Olea, D.; Burghard, M.; Gómez-Herrero, J.; Zamora, F.; Kern, K. Chemical vapor deposition repair of graphene oxide: A route to highlyconductive graphene monolayers. Adv. Mater. 2009, 21 (46), 4683− 4686. (16) Su, C.-Y.; Xu, Y.; Zhang, W.; Zhao, J.; Liu, A.; Tang, X.; Tsai, C.H.; Huang, Y.; Li, L.-J. Highly efficient restoration of graphitic structure in graphene oxide using alcohol vapors. ACS Nano 2010, 4 (9), 5285−5292. (17) Englert, J. M.; Vecera, P.; Knirsch, K. C.; Schäfer, R. A.; Hauke, F.; Hirsch, A. Scanning-Raman-microscopy for the statistical analysis of covalently functionalized graphene. ACS Nano 2013, 7 (6), 5472− 5482. (18) Schweiger, M.; Schaudig, M.; Gannott, F.; Killian, M. S.; Bitzek, E.; Schmuki, P.; Zaumseil, J. Controlling the diameter of aligned single-walled carbon nanotubes on quartz via catalyst reduction time. Carbon 2015, 95, 452−459. (19) Sharma, R.; Baik, J. H.; Perera, C. J.; Strano, M. S. Anomalously large reactivity of single graphene layers and edges toward electron transfer chemistries. Nano Lett. 2010, 10 (2), 398−405. (20) Reckinger, N.; Felten, A.; Santos, C. N.; Hackens, B.; Colomer, J.-F. The influence of residual oxidizing impurities on the synthesis of graphene by atmospheric pressure chemical vapor deposition. Carbon 2013, 63, 84−91. (21) Bissett, M. A.; Tsuji, M.; Ago, H. Strain engineering the properties of graphene and other two-dimensional crystals. Phys. Chem. Chem. Phys. 2014, 16 (23), 11124−11138. (22) Cheng, G.; Calizo, I.; Hight Walker, A. R. Metal-catalyzed etching of graphene governed by metal−carbon interactions: A comparison of Fe and Cu. Carbon 2015, 81 (1), 678−687. (23) Choubak, S.; Levesque, P. L.; Gaufres, E.; Biron, M.; Desjardins, P.; Martel, R. Graphene CVD: Interplay between growth and etching on morphology and stacking by hydrogen and oxidizing impurities. J. Phys. Chem. C 2014, 118 (37), 21532−21540. (24) Gao, L.; Ren, W.; Liu, B.; Wu, Z.-S.; Jiang, C.; Cheng, H.-M. Crystallographic tailoring of graphene by nonmetal SiO x nanoparticles. J. Am. Chem. Soc. 2009, 131 (39), 13934−13936. (25) Ramasse, Q. M.; Zan, R.; Bangert, U.; Boukhvalov, D. W.; Son, Y.-W.; Novoselov, K. S. Direct experimental evidence of metalmediated etching of suspended graphene. ACS Nano 2012, 6 (5), 4063−4071. (26) Tsetseris, L.; Pantelides, S. T. Adsorbate-induced defect formation and annihilation on graphene and single-walled carbon nanotubes. J. Phys. Chem. B 2009, 113 (4), 941−944. (27) Chua, C. K.; Ambrosi, A.; Sofer, Z.; Macková, A.; Havránek, V.; Tomandl, I.; Pumera, M. Chemical preparation of graphene materials results in extensive unintentional doping with heteroatoms and metals. Chem. - Eur. J. 2014, 20 (48), 15760−15767. (28) Liu, B.; Ren, W.; Gao, L.; Li, S.; Liu, Q.; Jiang, C.; Cheng, H.-M. Manganese-catalyzed surface growth of single-walled carbon nanotubes with high efficiency. J. Phys. Chem. C 2008, 112 (49), 19231− 19235.

of rGO, SEM and AFM images of metal particles on rGO (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (J.Z.). Present Address

S.E.: Chemistry and Chemical Engineering, Chalmers University of Technology, SE-412 96 Göteborg, Sweden. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research was funded by the Deutsche Forschungsgemeinschaft (DFG) via the Collaborative Research Center “Synthetic Carbon Allotropes” (SFB 953). S.E. acknowledges funding by the DFG (EI938/3-1).



REFERENCES

(1) Hummers, W. S.; Offeman, R. E. Preparation of graphitic oxide. J. Am. Chem. Soc. 1958, 80 (6), 1339−1339. (2) Lee, S.; Eom, S. H.; Chung, J. S.; Hur, S. H. Large-scale production of high-quality reduced graphene oxide. Chem. Eng. J. 2013, 233, 297−304. (3) Eigler, S.; Enzelberger-Heim, M.; Grimm, S.; Hofmann, P.; Kroener, W.; Geworski, A.; Dotzer, C.; Röckert, M.; Xiao, J.; Papp.; et al. Wet chemical synthesis of graphene. Adv. Mater. 2013, 25 (26), 3583−3587. (4) Eigler, S.; Grimm, S.; Enzelberger-Heim, M.; Müller, P.; Hirsch, A. Graphene oxide: efficiency of reducing agents. Chem. Commun. 2013, 49 (67), 7391−7393. (5) Pei, S.; Cheng, H.-M. The reduction of graphene oxide. Carbon 2012, 50 (9), 3210−3228. (6) Cançado, L. G.; Jorio, A.; Ferreira, E. H. M.; Stavale, F.; Achete, C. A.; Capaz, R. B.; Moutinho, M. V. O.; Lombardo, A.; Kulmala, T. S.; Ferrari, A. C. Quantifying defects in graphene via Raman spectroscopy at different excitation energies. Nano Lett. 2011, 11 (8), 3190−3196. (7) Sato, K.; Saito, R.; Oyama, Y.; Jiang, J.; Cançado, L. G.; Pimenta, M. A.; Jorio, A.; Samsonidze, G. G.; Dresselhaus, G.; Dresselhaus, M. S. D-band Raman intensity of graphitic materials as a function of laser energy and crystallite size. Chem. Phys. Lett. 2006, 427 (1−3), 117− 121. (8) Ferrari, A. C.; Basko, D. M. Raman spectroscopy as a versatile tool for studying the properties of graphene. Nat. Nanotechnol. 2013, 8, 235−246. (9) Banhart, F.; Kotakoski, J.; Krasheninnikov, A. V. Structural defects in graphene. ACS Nano 2011, 5 (1), 26−41. (10) Dresselhaus, M. S.; Jorio, A.; Souza Filho, A. G.; Saito, R. Defect characterization in graphene and carbon nanotubes using Raman spectroscopy. Philos. Trans. R. Soc., A 2010, 368 (1932), 5355−5377. (11) Jiang, J.; Pachter, R.; Mehmood, F.; Islam, A. E.; Maruyama, B.; Boeckl, J. J. A Raman spectroscopy signature for characterizing defective single-layer graphene: Defect-induced I(D)/I(D′) intensity ratio by theoretical analysis. Carbon 2015, 90, 53−62. (12) Eigler, S. Mechanistic insights into the reduction of graphene oxide addressing its surfaces. Phys. Chem. Chem. Phys. 2014, 16 (37), 19832−19835. (13) Eigler, S.; Hof, F.; Enzelberger-Heim, M.; Grimm, S.; Müller, P.; Hirsch, A. Statistical Raman microscopy and atomic force microscopy on heterogeneous graphene obtained after reduction of graphene oxide. J. Phys. Chem. C 2014, 118 (14), 7698−7704. (14) Cheng, M.; Yang, R.; Zhang, L.; Shi, Z.; Yang, W.; Wang, D.; Xie, G.; Shi, D.; Zhang, G. Restoration of graphene from graphene oxide by defect repair. Carbon 2012, 50 (7), 2581−2587. 3041

DOI: 10.1021/acs.jpcc.5b11598 J. Phys. Chem. C 2016, 120, 3036−3041