Reconstructed Ribbon Edges in Thermally ... - ACS Publications

Sep 27, 2012 - Reconstructed Ribbon Edges in Thermally Reduced Graphene. Nanoribbons. Muge Acik,. †. Javier Carretero-González,. ‡. Elizabeth ...
1 downloads 0 Views 4MB Size
Article pubs.acs.org/JPCC

Reconstructed Ribbon Edges in Thermally Reduced Graphene Nanoribbons Muge Acik,† Javier Carretero-González,‡ Elizabeth Castillo-Martínez,‡ Duncan M. Rogers,‡ R. Guzman,† Ray H. Baughman,‡ and Yves J. Chabal*,† †

Department of Materials Science and Engineering and ‡The Alan G. MacDiarmid NanoTech Institute, University of Texas at Dallas, Richardson, Texas 75080, United States S Supporting Information *

ABSTRACT: Graphene oxide nanoribbons (GONRs) with a high aspect ratio and high gravimetric density of their edges relative to those of graphene flakes are promising platforms for graphene-based devices. Since the edge chemistry of GONRs determines their final electronic and transport properties, it is important to understand the interactions of oxygen with the edges of the ribbons. Although oxidative unzipping of carbon nanotubes has been studied by Dai’s1 and Tour’s2 groups for GONR production, the role of oxygen concentration, the nature of edge oxygen groups, and their effect on the ribbon-edge geometry is still not well understood. We have therefore studied thermal annealing of GONRs, obtained by unzipping few-walled or multi-walled carbon nanotubes, focusing on the reduction process. For this purpose, in situ infrared absorption spectroscopy is used to monitor the edge reconstruction during the thermal reduction process. The ribbon edges of reduced graphene nanoribbons (rGNRs), initially functionalized with carboxyls, are found to convert to edge carbonyls during annealing at high temperatures (∼850 °C). The formation of these highly stable carbonyl species therefore leads to edge reconstruction of rGNRs after high-temperature anneals. The concentrations of initial hydroxyl, edge carbonyl, and carboxyl are found to be key factors that determine the resulting oxygen concentration in rGNRs after annealing. Although the initial concentration of these oxygen groups introduced during unzipping is associated with the concentration of oxidant (KMnO4 or H3PO4), the resulting amount of total oxygen in rGNRs upon thermal reduction is found to be independent of the wall thickness of the starting carbon nanotubes (i.e., number of original unzipped layers of GONRs).



INTRODUCTION Graphene,3 a 2D single atomic layer of sp2 carbon, has emerged as an exciting new material exhibiting unique properties in comparison to conventional allotropes of carbon. Development of chemical and physical synthesis methods for large-scale graphene production has focused on controlling its chemistry and structure.4 New chemical methods5 have been developed to produce graphene-based materials in flake,6 ribbon,7 or fiber8 morphologies. In particular, graphene oxide nanoribbons (GONRs) produced from oxidatively unzipped carbon nanotubes (CNTs)9,10 offer an alternative to ribbon formation with some control of the width (i.e., production of narrow semiconductor nanoribbons with a width below 10 nm).2 Apart from the conventional syntheses or growth techniques,11 an alternative route to producing graphene nanoribbons (GNRs) is the unzipping of carbon nanotubes via oxidation to produce GONRs. A subsequent thermal annealing process then yields reduced graphene nanoribbons (rGNRs) that are potentially useful for graphene nanoribbon-based devices,12 specifically for use in optoelectronics or spintronics applications.13 © 2012 American Chemical Society

Although various approaches for manufacturing GNRs have been presented, oxidative chemical unzipping of CNTs using sulfuric acid (H2SO4) and potassium permanganate (KMnO4) developed by Tour’s group2 offers a unique way for bulk production of tiny GONRs. Chemical unzipping of the singlewalled carbon nanotubes (SWCNTs)14,15 and multi-walled carbon nanotubes (MWCNTs)16,17 results in several types of oxygen functionalities localized mainly at the ribbon edges or located along the plane of the carbon nanoribbon. Moreover, an alternative method based on phosphoric acid (H3PO4) further improves the quality of graphene layers after reduction as described by Higginbotham et al.18 Although several theoretical studies have been presented,19,20 a solid mechanism has not yet been verified by experimental studies that fully describes the oxygen removal mechanism at the ribbon edges. In general, a combination of chemical unzipping followed by a thermal treatment to restore the sp2 hybridization in Received: March 30, 2012 Revised: September 5, 2012 Published: September 27, 2012 24006

dx.doi.org/10.1021/jp303035m | J. Phys. Chem. C 2012, 116, 24006−24015

The Journal of Physical Chemistry C

Article

rGNRs21,22 is believed to enhance the electrical properties.23,24 In all these approaches, the role of edge oxygen species in GONRs during oxidative unzipping and their removal during thermal annealing to produce rGNRs still need to be experimentally elucidated. In this study, we initially examine the chemical unzipping process starting with either MWCNTs or few-walled carbon nanotubes (FWCNTs). The initial edge oxygen interactions in GONRs and their stepwise removal during thermal annealing for rGNRs are monitored by in situ infrared (IR) spectroscopy. In particular, we study the dependence of the arrangement and chemical nature of the oxygen groups in GONRs during unzipping of MWCNTs on the KMnO4 concentration and the presence of H3PO4. On the basis of these spectroscopic results, we establish that the concentration of KMnO4 used for oxidative unzipping determines the initial concentration of total carbonyl and carboxyl at the ribbon edges upon unzipping. Thereby, we demonstrate that the structure of GONRs and their chemistry are controlled by the oxidative reaction conditions used during their bulk synthesis. The initial amounts of each oxygen group in addition to the presence of edge hydroxyls are found to determine the resulting edge oxygen concentration after thermal annealing. We also show that the resulting amount of oxygen in rGNRs does not depend on the wall thickness of starting CNTs (i.e., number of unzipped layers). The conductivity can therefore be restored easily with good control of the initial amount and the type of oxygen groups after unzipping, in part because the remaining oxygen is preferentially located at the edges upon thermal annealing.

Figure 1. Infrared absorbance spectra (400−3800 cm‑1) of (a) unzipped MWCNTs (GO nanoribbons, solid red line) in transmission and only MWCNTs (10 layers, solid black line), i.e., prior to unzipping. The presence of carbonyls/carboxyls (1650−1750 cm−1), hydroxyls with a contribution from carboxyls (∼3000−3600 cm−1), ethers (1000−1330 cm−1), and sp2 C (1550−1600 cm−1) is shown upon unzipping via oxidation using 850 wt. % KMnO4 in (a). A broad infrared absorption is present in (b) due to the effect of physisorbed trapped H2O. None of these absorbance spectra involve baseline corrections.

the characteristic sp2-C (CC, 1550−1600 cm−1) becomes IR active, although there is also a weak infrared absorption of edge carbonyls that overlaps with the sp2-C absorption. We note that this mode is IR inactive (e.g., undetectable) in pristine MWCNTs (Figure 1b) because the system’s symmetry is not perturbed by oxygen near the CC bonds (i.e., the sp2 conjugation is not activated). Similar oxygen groups are observed in GONRs unzipped from FWCNTs, as shown in Figure S1 (Supporting Information). To summarize, GONRs have similar edge oxygen groups whether they are unzipped from MWCNTs (Figure S1) or FWCNTs (Figure S2, Supporting Information). The broad-band, almost linear absorption observed in the ∼400−3800 cm−1 region for MWCNTs arises from IR radiation scattering that most likely arises from the formation of highly inhomogeneous clumps of conductive CNTs, as has been previously observed during the initial growth of Co films using atomic layer deposition (ALD).26 Stepwise Oxidation and Spectroscopic Evolution of Unzipping. To determine the initial chemical structure of GONRs, i.e., types of oxygen and their relative concentration during the chemical unzipping process, aliquots from the reaction mixture were taken and further characterized stepwise. The evolution of the unzipping process is based on two reaction variables: unzipping time and temperature during oxidation. As the oxidation reaction evolves, the samples are taken from the reaction flask with a long-needle syringe equipped and quenched in deionized (DI) water. Each experimental step (addition of oxidant, dependence on the oxidation time and temperature) is then monitored using IRAS (Figure 2i), Raman spectroscopy (Figure 2ii), and X-ray diffraction (Figure S3, Supporting Information). The degree of oxidation is determined by calculating the peak areas of IR absorption peaks corresponding to the oxygen concentration per oxygen species. The dependence of the oxidation efficiency of the unzipping process on the oxidant concentration is also examined using integrated peak areas.



RESULTS AND DISCUSSION Unzipped MWCNTs and FWCNTs Using KMnO4 in H2SO4. Unzipped MWCNTs vs Pristine MWCNTs. Infrared absorption spectroscopy (IRAS) is used in transmission mode to monitor the unzipping process of MWCNTs, forming GONRs upon oxidation using KMnO4. Herein, we first determine the presence of edge oxygen groups that are formed as a result of oxidation. The interpretation is provided by comparing experimental with simulated vibrational frequencies and by recording the absolute infrared intensity of each mode.25 We also study the effect of the oxygen concentration on the assynthesized GONRs to understand the reduction of GNRs upon annealing. Figure 1a shows the IR absorbance spectrum of the oxygen functional groups (especially out-of-plane hydroxyls and epoxides and in-plane edge carboxyls, carbonyls, and ethers) of unzipped MWCNTs after oxidation using 850 wt. % KMnO4. In comparison, these oxygen groups are absent in pristine MWCNTs (Figure 1b), i.e., prior to oxidation (i.e., before unzipping). Once unzipped, both edge carboxyls (COOH) and edge carbonyls (CO) form (∼1650−1750 cm−1). To determine these oxygen species, we compare the vibrational modes of each oxygen group calculated from the first principles calculations by density functional theory (DFT) simulations with the experimental findings of our previous study.25 On the basis of these calculations, the broad peak at ∼3000−3600 cm−1 can be associated with carboxyls (a C−OH stretch) at the ribbon edges and hydroxyls (possibly adsorbed on the ribbon surface, including defective sites, and at the ribbon edges). At lower frequencies, an additional weaker infrared absorption of hydroxyls (∼1040−1120 cm−1) is present that overlap modes of C−O species (∼1000−1330 cm−1). Once GONRs are formed via unzipping of MWCNTs, 24007

dx.doi.org/10.1021/jp303035m | J. Phys. Chem. C 2012, 116, 24006−24015

The Journal of Physical Chemistry C

Article

Figure 2. Normalized infrared absorbance spectra (i, left) and Raman spectra (ii, right) of GONRs from unzipped MWCNTs. Each spectrum in (i) represents a step in the oxidation process of MWCNTs to unzip them. Both temperature and time dependences are given for (a)−(f) in (i) and (a)−(g) in (ii). Absorbance spectra in (i) are normalized to the peak intensity of sp2-C (1575 cm−1) as a reference data point. Raman spectra in (ii) are normalized to the maximum G band (∼1600 cm−1) data points. (iii) AFM and TEM images for both unzipped MWCNTs and FWCNTs upon oxidation. (a) AFM image (1 μm × 1 μm) of GONRs. (b) AFM image (200 nm × 200 nm) taken at the same area of GONRs. The inset of (b) is a cross section across a GONR taken along the dashed line in the image. (c) TEM image of unzipped MWCNTs using 850 wt. % KMnO4 in H2SO4 (scale bar 100 nm) with a higher resolution image in (d), scale bar 10 nm. (e) TEM image of unzipped FWCNTs (scale bar 20 nm) with a higher magnification image in (f), scale bar 5 nm. (g) TEM image of unzipping MWCNTs using 850 wt. % KMnO4 in H2SO4/H3PO4 (9:1) (scale bar 100 nm) with a higher magnification image in (h), scale bar 5 nm.

necessary to remove impurities since similar IR spectra (Figure S4d vs Figure S4c) are obtained after minor annealing in the furnace. The infrared absorbance spectrum given in Figure 2i-a is the reference spectrum before introduction of any oxidant agent. The unzipping process is then initiated upon addition of KMnO4 (850 wt. %) shown in Figure 2i-b. Although the intensity of the CC stretch mode at 1575 cm−1 depends on the nature and concentration of neighboring oxygen functionalities, by normalizing the data to the sp2-C absorption at ∼1575 cm−1 involving weak CO contributions at ∼1600 cm−1, we see that carbonyls appear at ∼1650−1750 cm−1 (yellow region) at room temperature. In addition, epoxides at 1280−1330 cm−1 (green region) are formed. These spectral results confirm the fact that the oxidation attacks both the defective sites on the ribbon surfaces forming epoxides and also the ribbon edges decorated either with epoxides or with carbonyls. As the oxidation continues to unzip MWCNTs, the infrared absorption of epoxides is reduced since they are opened to form dicarbonyls (see the absorption increase showing formation of carbonyls at 1650−1750 cm−1 in green

The amount of oxidant is optimized as 800−850 wt. % KMnO4 vs wt. % pristine MWCNTs on the basis of the complete disappearance of diffraction resulting from interwall spacing in X-ray diffraction patterns (Figure S3, Supporting Information). The spectral information in IR data is normalized with respect to the sp2-C absorption at 1575 cm−1 for comparison. Each sample is taken from the same batch. First, bare MWCNTs (i.e., prior to oxidation, Figure S4a, Supporting Information) are only soaked in H2SO4 for 12 h (Figure S4b). Hydroxyls (∼3300 cm−1) and ethers (1000−1330 cm−1) are present together with the sp2-C infrared absorption (IR active) at ∼1575 cm−1. Since the material is not washed after each, there is some contamination from H2SO4 as evidenced by a band at 1000−1330 cm−1 associated with C−O. This is also the case for unzipped MWCNTs (Figure S4b), with weak C−O and C−OH contributions. These impurities are washed away with several rinses (at least three) in fresh ultrapure DI water; i.e., the small features disappear as shown in Figure S4c). A minor anneal at 150 °C in Ar flow for 2 h in the furnace also preserves the oxygen groups in these GONRs despite some water removal. Therefore, an additional annealing step is not 24008

dx.doi.org/10.1021/jp303035m | J. Phys. Chem. C 2012, 116, 24006−24015

The Journal of Physical Chemistry C

Article

also shown in Figure 2iii-e with a high-magnification image in Figure 2iii-f. GONRs obtained from unzipped MWCNTs are noticeably wider than those produced from unzipped FWCNTs (note the scales of the TEM images) as expected from their initial diameter. A TEM image of MWCNTs unzipped using KMnO4 in a H3PO4/H2SO4 mixture (Figure 2iii-g, highmagnification image in Figure 2iii-h) is also included for comparison. Overall, evidence from these microscopic images indicates that the “H3PO4-modified” unzipping process results in smaller aspect ratio GONRs.9 Spectroscopic Evidence for Edge Oxygen Interactions and Edge Reconstruction. Transmission infrared absorbance spectra of GONRs recorded at room temperature provide spectral evidence for the presence of functional groups such as sp2-C and ethers. Hydroxyls, carboxyls, carbonyls, and epoxides are difficult to distinguish because they have modes overlapping in frequency in the regions 3000−3600, 1650−1750, and 1000−1330 cm−1, respectively. Thermal annealing studies are useful to identify each of these oxygen species and to understand their chemical nature because their thermal evolution is specific. The evolution during thermal reduction from 60 to 225 °C is shown for GONRs obtained from unzipped MWCNTs (Figure S5i, Supporting Information) and FWCNTs (Figure S5ii). All data are collected at 60 °C in a vacuum, including the reference spectrum. The differential spectra of both unzipped MWCNTs and FWCNTs have a similar trend involving oxygen group removal or formation of new oxygen species. After annealing at 125 °C (c−g), a negative feature at ∼1700−1750 cm−1 corresponds to the disappearance of carbonyls. The negative feature nearby at ∼1650−1700 cm−1 has contributions from both carboxyls and carbonyls. A negative feature at ∼3000−3600 cm−1 (a−g) is associated with the loss of corresponding carboxyls (C−OH stretch) and some hydroxyl groups, also shown in Figure 2i. As soon as MWCNTs are soaked in H2SO4, hydroxyls form (at ∼3000−3600 cm−1 in Figure 2i-a). Therefore, the C−OH stretch mode at ∼3000−3600 cm−1 has contributions from carboxyls and hydroxyls, and possibly water, identified by its scissor mode at ∼1620 cm−1. The loss of epoxides (∼1330 cm−1 band) in Figure S5 (Supporting Information) is either weak or absent. Therefore, we use the absorbance spectrum in Figure 2i-f to confirm the presence of a weak epoxide contribution. An earlier study30 of thermal reduction of GO flakes (GOFs) showed that CO2 production is an indication that defects are formed during annealing. The same is expected with the formation of defective sites on the ribbon surfaces, formed during the unzipping process by oxidation. The absence of a CO2 peak (2300 cm−1) during annealing at 60−225 °C (Figure S5) and a small increase in the D′ band in the Raman spectra confirm that defect formation is minor during the thermal reduction of GNRs. In addition, it has been shown that, at intermediate stages of thermal annealing, there is formation of carbonyls at the etch holes as a result of reactions between defects and trapped water.30 In the case of both unzipped MWCNTs (Figure S5i-b−d) and FWCNTs (Figure S5ii-b−d), a strong infrared absorption is also present at ∼1000−1600 cm−1 associated with carbonyls after annealing at 100−150 °C. Since there is no CO2 production, such carbonyl formation corresponds to the conversion of edge carboxyls to edge carbonyls rather than intermediate carbonyl species related to defect formation on the basal plane. This is in contrast to what was observed for thermally reduced GOFs.30

regions, Figures 2i-b−f). A complete unzipping is reached after oxidation at 55 °C for 1 h and then at 70 °C for 10 min (Figure 2i-f) as was confirmed by XRD patterns (Figure S3, Supporting Information). Figure 2i-f therefore indicates that few epoxides remain unzipped as shown with a weak absorbance at ∼1280− 1330 cm−1. In this case, IR spectroscopy is more useful to analyze epoxides and to distinguish carboxyls from carbonyls at the ribbon edges than NMR and X-ray photoelectron spectroscopy (XPS) studies. Normalized Raman intensity for each sample is also shown in Figure 2ii. A sharp 2D band appears at ∼2700 cm−1 after MWCNTs are soaked in H2SO4 (Figure 2ii-a). In this case, the normalization is performed with respect to the G band intensity at 1600 cm−1. Thereafter, a characteristic triplet 2D band appears right after a complete unzipping process is achieved (Figure 2ii-b−g). As the oxidation takes place with time and increasing reaction temperature, the Raman intensity of the 2D band (specifically the band at ∼2700 cm−1) is reduced (Figure 2ii-g). The relative increase in intensity of the D′ peak at ∼1300 cm−1 indicates the formation of more defects than on CNTs once the GONR edges are decorated with the edge oxygen groups. The ratio ID′/IG is calculated from the Raman intensities of each peak and found to be 0.80, 1.03, 1.15, 1.21, 1.30, 1.32, and 1.34, respectively, in Figure.2ii-a−g. These results indicate that GONRs produced by oxidative unzipping of CNTs are highly defective nanoribbons, similar to graphene flakes chemically produced from graphite,27 approximately 5−6 times more defective compared to GNRs generated by other processes such as nanotomy-based unzipping as shown by Berry et al.28 It should also be noted that the defect concentration increases as the oxidation proceeds with unzipping (Figure 2ii-a−g). The unzipping process via oxidation is also monitored by Xray diffraction (XRD) (Figure S3, Supporting Information), collected for bare CNTs soaked in H2SO4 and GONRs after each processing step (Figure S3). After 15 min of heating at 55 °C in the presence of the oxidant in the reaction mixture, the reflection corresponding to the interwall distance, i.e., 2θ ≈ 26° in MWCNTs, is slightly shifted to lower angles, corresponding to a larger interplanar distance consistent with unzipping initiation. Nevertheless, this shift remains very small for the following 45 min at 55 °C. On the other hand, as soon as the reaction mixture is heated to 70 °C for 1 min, this 2θ reflection is shifted more and almost disappears after 10 min at 70 °C. At that point, a high-intensity reflection appears (centered at 2θ ≈ 9.6°). This interlayer distance corresponds to the basal plane distance between GO layers (c axis), which confirms a complete unzipping of the MWCNTs to produce stacks of piled GONRs. The morphology of these GONRs unzipped from MWCNTs is shown in Figure 2iii, where the atomic force microscopy (AFM) images (Figure 2iii-a) reveal isotropic and random arrangement of these GONRs (magnified image in Figure 2iiib). The inset of Figure 2iii-b shows a cross section across a nanoribbon (dashed line) that is 40.0 nm wide with a 7.5 nm upward bend at the edges. The upward bend is a remnant of the curvature of the pristine MWCNT and has been observed in other unzipped MWCNTs.29 The measured nanoribbon width is an upper limit because of the finite width of the AFM tip. Transmission electron microscopy (TEM) images of these GONRs are also given in Figures 2iii-c−h. For GONRs from MWCNTs unzipped using KMnO4 in H2SO4 (Figure 2iii-c), these GONRs are randomly oriented as shown in the magnified image (10 nm scale) in Figure 2iii-d. Unzipping of FWCNTs is 24009

dx.doi.org/10.1021/jp303035m | J. Phys. Chem. C 2012, 116, 24006−24015

The Journal of Physical Chemistry C

Article

Using “edge” to indicate the ribbon edges and “basal plane” to indicate the ribbon surfaces, we note that, once edge carboxyls are transformed to edge carbonyls at intermediate temperatures, condensation reactions take place assisted by water removal. This process stabilizes the edges with carbonyl termination (Figure 3). In both of these reduced samples

(unzipped MWCNTs and FWCNTs), carboxyls and carbonyls start decomposing as the temperature is raised to 225 °C (Figures S6 and S7, respectively, Supporting Information). The peak at 1750 cm−1 is no longer observed after annealing at 600 °C (Figures S6f and S7f for unzipped MWCNTs and FWCNTs, respectively). The C−OH stretch peak (∼3000− 3600 cm−1) also disappears after annealing at 150 °C (Figure S5, Supporting Information), which indicates the removal of both edge carboxyls and hydroxyls. Therefore, edge carbonyls are the only oxygen species remaining in these rGNRs, mostly at the ribbon edges, with possibly minor contributions from defective sites of the ribbon surfaces). Therefore, the negative peaks at 1500−1700 cm−1 (below the dashed baseline) only correspond to the decomposition of edge carbonyls. As recently shown by DFT simulations,25 dicarbonyls (1,2- and 1,3benzoquinones), ketones (isolated carbonyls), or 2-pyranones are likely to exist at atomically straight ribbon edges of reduced GNRs. Dependence of rGNR Production on Initial Oxidation. To study the role of initial oxidation on the resulting oxygen concentration, we examine the infrared transmission absorbance spectra (obtained by referencing to the clean bare Si/ SiO2) for GONRs unzipped from MWCNTs using either 800 or 850 wt. % KMnO4 (Figure 4i,ii). The amount of oxidant (KMnO4) during unzipping determines the initial epoxide concentration. For instance, if the amount of KMnO4 used for an oxidation unzipping process is increased from 800 to 850 wt.

Figure 3. Schematic representation of edge carbonyl formation from carboxyls during thermal annealing. The oxygen groups are colorcoded: blue stands for carboxyls, yellow for hydroxyls, red for carbonyls, gray for epoxides, and purple for ethers.

Figure 4. Infrared absorbance spectra of reduced GONRs unzipped from MWCNTs using (i) 800 wt. % KMnO4 and (ii) 850 wt. % KMnO4 in H2SO4. Note the different scales and different temperatures for signal disappearance. Normalized total integrated infrared absorbance vs annealing temperatures for GONRs (iii) unzipped from MWCNTs (a) and FWCNTs (b) using 850 wt. % KMnO4 and (iv) unzipped from MWCNTs using 800 wt. % KMnO4. 24010

dx.doi.org/10.1021/jp303035m | J. Phys. Chem. C 2012, 116, 24006−24015

The Journal of Physical Chemistry C

Article

Table 1. Summary of the Initial Amount of Oxygen Groups Prior to the Thermal Annealing and Remaining Amount of Oxygen Groups upon Reduction oxygen species

FWCNTs,a 850 wt. % KMnO4

MWCNTs,a 850 wt. % KMnO4

MWCNTs,a 800 wt. % KMnO4

MWCNTs,a H3PO4

GO flakes, H3PO4

CO + C−O C−OH (H2O) total oxygen (atom %)b CO/total (%)b C−OH/total (%) remaining O (atom %)

2.8 6.2 54.1 5 12 25

17 18 154.4 11 12 20

11.1 9.5 91.7 12 38 10

2.8 4.7 23.8 12 20 55

2.8 11.2 32.3 9 35 50

a

Unzipped FWCNTs or MWCNTs resulting in formation of GNRs through oxidation. bStandard deviations are in the range of 1−3 atom %.

MWCNTs using 800 wt. % KMnO4, no such increase in the infrared absorbance was detected upon annealing to 250 °C (Figure 4iv). This experimental finding indicates that oxidation assisted by an 800 wt. % oxidant concentration (instead of 850 wt. %) does not generate defects and there is a negligible amount of edge carbonyl formation at intermediate-temperature anneals (Figure 4iv). For the case of 850 wt. % KMnO4, there is still remaining oxygen (∼20 and 25 atom % of the initial total oxygen concentration for GONRs unzipped from MWCNTs and FWCNTs, respectively) after annealing at 850 °C. However, the peak at 800 cm−1 remains weak, suggesting that there is little oxygen aggregation at the edges.31 In the case of MWCNTs unzipped with 800 wt. % oxidant, the oxygen concentration remaining after annealing at 750 °C is ∼10 atom % (less than for the case of 850 wt. % oxidant). Upon annealing to 850 °C, however, the infrared intensity is enhanced by ∼50%, characteristic of agglomerated edge ethers. These results therefore highlight the fact that the oxidant concentration controls the type and degree of oxidation at the edges of GONRs as well as the edge termination (e.g., agglomerated ethers). The oxygen amount calculated from the integrated peak areas shows that the initial edge carbonyl, edge carboxyl, and hydroxyl concentrations determine the remaining oxygen concentration after high-temperature anneals. Table 1 summarizes both the initial oxygen concentration per total initial oxygen concentration at room temperature prior to annealing and the remaining amount of oxygen after annealing at 850 °C. Unzipped MWCNTs using 850 wt. % KMnO4 exhibit a large initial edge carbonyl concentration (roughly ∼11 atom % of the ∼154 atom % total initial amount of oxygen functionalities). In contrast, unzipped FWCNTs using 850 wt. % KMnO4 only have ∼5 atom % edge carbonyls of the total initial oxygen concentration (∼54 atom %). Therefore, the edge carbonyl concentration in unzipped MWCNTs using 850 wt. % KMnO4 is relatively lower than that in unzipped FWCNTs using 850 wt. % KMnO4 (Table 1). However, the total initial hydroxyl concentrations in these GONRs are similar (∼12 atom %). Once the hydroxyl groups decompose, they are likely to form hydroxyl radicals during annealing, which then transform edge carboxyls into edge carbonyls. Therefore, the initial amount of carboxyls is also a key factor to determine the resulting edge carbonyl concentration in reduced GONRs. For instance, once MWCNTs are unzipped using 800 wt. % KMnO4, there is a negligibly small carbonyl formation upon annealing to 250 °C (Figure 4iv) with an initial total hydroxyl concentration of ∼38 atom %. This results in a low oxygen concentration (∼10 atom %) after annealing at 750 °C relative to the total initial amount of oxygen. Since the defect concentration is small in GONRs derived from unzipped MWCNTs using 800 wt. % KMnO4, all

%, the initial epoxide concentration in GONRs increases (green dashed line showing the C−O−C contribution at ∼1330 cm−1; note the scale difference for comparison, Figure 4i-a vs Figure 4ii-a). For reduced GONRs unzipped from MWCNTs using 800 wt.% KMnO4, the absence of a C−OH peak at ∼3000−3600 cm−1 after annealing at 200 °C (Figure 4i-c) indicates the removal of hydroxyls or edge carboxyls. The decrease of the infrared band at ∼1650−1850 cm−1 and the increase of the sp2C band (∼1575 cm−1) are both consistent with the decomposition of edge carboxyls and edge carbonyls. After annealing to 300 °C, carboxyls are consumed, as evidenced by the disappearance of the C−OH stretch mode at ∼3000−3600 cm−1 (Figure 4ii-d). In contrast, there are still remaining edge carboxyl groups in the samples oxidized using 850 wt. % KMnO4, even after annealing at 300 °C (Figure 4i-c). This is due to the initial large total epoxide concentration in GONRs produced by 850 wt. % KMnO4. In the case of MWCNTs oxidized with 850 wt. % KMnO4, all carbonyls are removed after annealing at 500 °C (Figure 4ii-e), whereas they remain for MWCNTs oxidized with 800 wt. % KMnO4. After annealing at 750 °C, all edge carbonyls decompose and the only remaining groups are in the form of edge ethers (Figure 4i-f), as evidenced by the appearance of a sharp mode at 800 cm−1 upon annealing at 850 °C (Figure 4i-g,ii-g; Figure S8, Supporting Information). This mode is associated with edge ether agglomeration of at least 10 C−O−C groups, as shown in a previous study.31 A similar sharp mode at 800 cm−1 is also present for unzipped FWCNTs after annealing at 850 °C (Figure S9, Supporting Information). While in both materials edge ethers are present after annealing at 850 °C (Figure 4ii-g), the peak intensity of the 800 cm−1 mode is weaker in MWCNTs unzipped with 850 wt. % KMnO4 than in the case of MWCNTs oxidized by 800 wt. % KMnO4. This phenomenon is associated with the presence of a large initial edge carbonyl concentration in GONRs oxidized using 850 wt. % KMnO4 in comparison to the GONRs oxidized using 800 wt. % KMnO4. The oxygen concentration, calculated from the total integrated area of each infrared absorbance feature and normalized to the initial total oxygen concentration (integration of the whole spectrum), is shown in Figure 4iii,iv for GONRs unzipped from FWCNTs or MWCNTs using 850 wt. % KMnO4 and from MWCNTs using 800 wt. % KMnO4 for comparison. For both MWCNTs (Figure 4iii-a) and FWCNTs (Figure 4iii-b) unzipped using 850 wt. % KMnO4, the plots indicate a total increase in the infrared absorbance by about 10% after annealing at temperatures between 60 and 250 °C. This finding indicates carbonyl formation at defective etch holes as a result of reacting trapped water with dangling bonds.30 However, in the case of GONRs unzipped from 24011

dx.doi.org/10.1021/jp303035m | J. Phys. Chem. C 2012, 116, 24006−24015

The Journal of Physical Chemistry C

Article

Figure 5. Transmission infrared differential spectra as a function of infrared frequencies at 800−3750 cm−1. Thermally reduced (i) GONRs and (ii) GOFs, initially oxidized with 850 wt. % KMnO4 in H2SO4/H3PO4 (9:1) mixtures. (iii) Normalized total integrated infrared absorbance as a function of annealing temperature for samples initially oxidized with 850 wt. % KMnO4 in H2SO4/H3PO4 (9:1) mixtures: (a) GONRs unzipped from MWCNTs and (b) GOFs.

release CO2 after annealing at 100 °C (Figure 5i-a). This experimental observation is important to explain the high concentration of defects in reduced GOFs (Figure 5ii-a). After annealing at 200 °C, there is a higher concentration of carbonyls in reduced GOFs (Figure 5ii-b) than in reduced GONRs (Figure 5i-b). Figure 5iii shows that the normalized integrated infrared absorbance relative to the total initial oxygen concentration at room temperature increases upon annealing from 60 to 250 °C and is larger in reduced GOFs (Figure 5iiib) than in GONRs (Figure 5iii-a). For comparison, the edge carbonyl concentration increase is larger than ∼10% in the case of unzipped MWCNTs and FWCNTs without H3PO4 (Figure 5iii). Edge carbonyl formation is therefore a result of the transformation from edge carboxyls. Water removal, evidenced by a negative peak at ∼1620 cm−1 in the differential spectra upon annealing at 100 °C, suggests that water reacts at the ribbon edges. Therefore, there is a substantial amount of water that is removed at the initial stages of thermal annealing, which however does not lead to a large etch hole carbonyl concentration as is the case for reduced GOFs oxidized with KMnO4. After annealing to 300 °C, a continuous removal of edge carbonyls occurs for both reduced GONRs (Figure 5i-c−f) and GOFs (Figure 5ii-c−f) oxidized with H3PO4 in the reaction mixture. After annealing at 850 °C, there is ∼50 atom % more remaining oxygen concentration in reduced GOFs than in

the hydroxyls are preferentially consumed by transforming edge carboxyls into edge carbonyls rather than participating in the reactions with etch holes at moderate-temperature (∼200 °C) anneals. Effect of H3PO4 (Unzipped MWCNTs vs GO Flakes). XRD studies show that the interlayer distance of unzipped MWCNTs after oxidation in the presence of 10% H3PO4 is, on average, larger than after oxidation in KMnO4 with 98% H2SO4 (Figure S10, Supporting Information, red vs blue spectra). Similarly, Figure S11 (Supporting Information) indicates a complete unzipping of FWCNTs in the presence of H3PO4. Once GONRs are unzipped using H3PO4, the oxygen groups are similar to those found in GONRs unzipped without H3PO4 (i.e., only KMnO4, Figure 1a). A strong absorbance peak is observed at 1000−1200 cm−1, assigned to the presence of phosphate and hydrogen phosphate groups overlapping with ether contributions. GOFs produced from graphite powder in a H2 SO4 /H 3PO4 mixture 32 are characterized by infrared absorbance spectra similar to those of GONRs produced in a H2SO4/H3PO4 mixture (Figure S5ii vs Figure S5i, Supporting Information). However, the thermal reduction of these GONRs produced in H3PO4 (Figure 5i) results in a low defect concentration compared to the reduced GOFs oxidized with H3PO4 (Figure 5ii). This is consistent with the lack of CO2 formation (no feature at ∼2300 cm−1) in the case of GONRs unzipped with H3PO4, in constrast to reduced GOFs that 24012

dx.doi.org/10.1021/jp303035m | J. Phys. Chem. C 2012, 116, 24006−24015

The Journal of Physical Chemistry C

Article

the degree of functionalization33 or creating atomically straight edges.34 Therefore, once the chemistry of the ribbon edge is well understood, further studies with this knowledge can provide subsequent interesting edge designs via chemical edge functionalization.35 Edge reconstruction36 in rGNRs via edge carbonyl groups is also an alternative route to decrease the optical band gap while opening a band gap.37 Hopefully, the functionalization of these ribbon edges with edge carboxyl groups38 may make it possible to modulate the electronic structures and transport properties of rGNRs.

reduced GONRs. The differential spectra for reduced GONRs at 750 °C (Figure S13, Supporting Information) also show less decomposition of edge carbonyls (at ∼1500−1700 cm−1) than in reduced GOFs (Figure S14, Supporting Information). This is associated with the total initial hydroxyl concentration that is present in reduced GOFs (∼35 atom %), larger than in reduced GONRs (∼20 atom %) (Table 1). Therefore, addition of H3PO4 is shown to promote the reduction of structural defects during thermal annealing more than for GONRs produced by using only KMnO4 and H2SO4. However, the decrease of the initial defect formation results in a large concentration of remaining oxygen functionalities, even after annealing at high temperatures. This is associated with the stabilization of oxygen groups in the presence of H3PO4, which remain stable in the graphene plane. Defects therefore help decomposition of the basal plane oxygen groups, tearing apart GONRs and GOFs, leading to a further oxygen removal during annealing. This is an important factor for distortions of the basal plane occurring in the defective sites of the ribbon surfaces.



EXPERIMENTAL METHOD GONR Preparation (Unzipping Process). GONRs are produced by chemical unzipping of vertically aligned MWCNTs or FWCNTs, grown by chemical vapor deposition39 (CVD) by using an extension of the method developed by Tour’s group.2 We also follow an unzipping processes for the production of graphene nanoribbons2 and low defect content nanoribbons by adding H3PO4 to the reaction mixture.18 In general, in a 50 mL Erlenmeyer flask, 35 mg (1 wt equiv) of carbon nanotubes (MWCNTs or FWCNTs) or graphite flakes is dispersed in 35 mL of concentrated H2SO4 (ACS grade, 95− 98 wt. %, Merck) and stirred overnight. Then KMnO4 (ACS grade, Merck) (amount ranging between 5 and 8.5 wt equiv) is added to this mixture, the reaction is progressibly heated first to 55 °C for approximately 1 h and finally to 70 °C, and then heating is stopped after KMnO4 has been consumed. For the low defect content nanoribbon production, H3PO4 (ACS grade, ≥85 wt. % in H2O, Sigma-Aldrich) is added to the mixture of carbon nanotubes and sulfuric acid (final H2SO4:H3PO4 ratio 9:1, v/v) and stirred for 15 min before the addition of 0.34 g (8.5 wt equiv) of KMnO4 (ACS grade, Merck) . This mixture is then heated at 65 °C for 1 h. For both processes, either using H3PO4 or not, the protocol followed after reaction is the same. The reaction flask is placed in an ice bath (∼200 mL) containing 3 mL of H2O2 (ACS grade, 30% in water, Fisher). The mixture is filtered over a poly(tetrafluoroethylene) (PTFE) membrane with a 0.45 μm pore size (Sartorius Stedim Biotech) and then successively washed and filtered twice with 200 mL of 30% HCl (ACS grade, 36.5−38%, Fisher) and 200 mL of deionized water (HPLC grade, Fisher). The resulting graphene oxide nanoribbons are first redispersed in 40 mL of deionized water (HPLC grade, Fisher) by stirring overnight, then placed in SnakeSkin dialysis tubing (10K MWCO, Pierce), and dialyzed for 10 days to remove the traces of impurities from the inorganic acids and the salt employed in the oxidizing mixture. X-ray Diffraction Analysis. Powder XRD is used to follow the unzipping process and assess its completion. In general, samples are prepared by drop-casting from the dialyzed graphene oxide nanoribbons. For the conversion study of the unzipping process, samples are prepared by taking a drop from the Erlenmeyer flask containing the reaction mixture after different reaction times at 55 °C or further at 70 °C and placing it on top of a silicon substrate. Data are collected in a Rigaku Ultima III diffractometer using Cu Kα radiation. Atomic Force Microscopy Analysis. AFM samples are fabricated by vacuum filtration of a colloidal suspension of graphene oxide nanoribbons in deionized water after being dialized. AFM is performed with a Veeco Multimode atomic force microscope, model MMAFMLN, coupled with a Nanoscope IV for digital imaging.



CONCLUSIONS This study examines the role that the choice of oxidant plays in affecting the oxygen concentration at room temperature in chemically unzipped GONRs. Initially, the edges are functionalized by edge carboxyls, carbonyls (in a minor amount), hydroxyls, and epoxides (possibly on the ribbon surfaces in addition to the ribbon edges). The use of a slightly less than optimal concentration of oxidant (i.e., 800 wt. % KMnO4 instead of 850 wt. %) decreases the total carbonyl concentration at the ribbon edges. The initial amounts of edge carboxyls and carbonyls also determine the resulting total edge oxygen concentration upon annealing. Thermal annealing promotes the transformation of edge carboxyls to edge carbonyls. At high-temperature anneals, the remaining oxygen is only in the form of carbonyls and ethers at the edges of the rGNRs. The amount of these remaining edge groups also determines the degree of infrared absorption once annealed at 850 °C, which is critical to short-range arrangement of ether groups. This experimental finding is consistent with ether edge termination upon removal of carboxyls and carbonyls at the ribbon edges. This edge termination (i.e., geometry) is very similar to what was found in GO flakes after annealing at high temperatures,34 as recently confirmed by atomistic field emission studies with HRTEM (Figure 2iii-a−h). In contrast, an increase of the oxidant concentration (i.e., 850 wt. % KMnO4) does not enhance the infrared absorption at 800 cm−1, although there is almost 2 times more oxygen remaining in the annealed samples at 850 °C than with those unzipped with 800 wt. % KMnO4 concentration. We also show that the final amount of oxygen in rGNRs does not depend on the wall thickness of CNTs (i.e., number of layers, unzipped FWCNTs vs MWCNTs). Indeed, it is found that the use of H3PO4 decreases defect formation in GONRs upon initial unzipping compared with GOFs. Moreover, the defect concentration is shown to determine the degree of reduction. The reduction efficiency of both GONRs and GOFs diminishes when initially decorated with few defects. Therefore, a low defect concentration promotes reduction of the remaining oxygen groups upon annealing to 850 °C. Overall, these findings can help guide the scientific community to develop controlled ways for edge termination of GONRs by annealing. Both GONRs and rGNRs are ideal ribbon systems for either tuning the transport gap by varying 24013

dx.doi.org/10.1021/jp303035m | J. Phys. Chem. C 2012, 116, 24006−24015

The Journal of Physical Chemistry C



Transmission Electron Microscopy Analysis. TEM is performed with an JEOL 2100 field emission gun (FEG) that is operated at 200 kV. A drop of the dialyzed graphene oxide nanoribbon solution is diluted in 2-propanol. A drop of this final suspension is evaporated on a copper grid covered with lacey carbon (SPI TEM grids). The solvent is completely evaporated by a thermal treatment at 150 °C in argon flow for 1 h to avoid the formation of amorphous material under the electron beam. Substrate Preparation and Film Deposition. Si(100) wafers with ∼6−7 nm of thermal oxide (Cz (Si grown by the Czochralski method), n-type, double side polished with a nominal resistivity of 1−20 Ω·cm) are cleaned by sequential rinsing in DI water (pure deionized water with a resistivity of 18.2 MΩ·cm and a total organic carbon (TOC) below 5 ppb obtained from a Millipore system), ethyl acetate, methanol, and DI water and immersion in piranha solution (98% H2SO4/30% H2O2, 2:1, v/v) at 90 °C for at least 30 min and rinsed with copious amounts of DI water. The dialyzed suspensions of GONRs are then deposited on one side of these clean SiO2/Si substrates (1.5 × 3.8 cm) by drop-casting. Transmission Fourier Transform Infrared Measurements. The infrared absorbance spectra are collected in direct transmission at normal incidence with 4 cm−1 resolution involving a deuterated triglycine sulfate (DTGS) detector. The measurements are performed at room temperature with 10 loops (1000 scans per data set) in nitrogen purge. In Situ Infrared Transmission Spectroscopy Measurements. FTIR measurements are performed in a transmission geometry (typically 500 scans per data set) using a DTGS detector with a mirror optical velocity of 0.6329 cm/s at 4 cm−1 resolution. The sample in the chamber is positioned so that the IR incident angle is close to the Brewster angle (70°) using direct transmission. Data collection per loop is performed at 60 °C after each annealing sequence. The annealing time during a sequence is 5 min at each temperature for a stepwise reduction, and the total annealing time for the overall experiment per sample is approximately 8 h. Differential spectra are referenced either to the reference temperature (60 °C) or to the spectrum collected at a previous annealing temperature. Each absorbance spectrum is referenced to the bare clean SiO2/Si substrate spectrum collected at 60 °C used as a reference room temperature. Thermal Annealing Process (Thermal Reduction). Thermal annealing of GONRs on SiO2/Si substrates is achieved by direct resistive heating of the Si substrates. The annealing chamber (a closed system) is placed in the main compartment of a N2-purged spectrometer (Thermo Scientific Nicolet 6700 FT-IR spectrometer with a KBr beam splitter, a DTGS detector, and constant nitrogen flow generated by a slight overpressure). The chamber is evacuated (10−3−10−4 Torr) to minimize the environmental effects. The sample is held by two tantalum clips to permit resistive heating in vacuum. Once the sample is in place, the measurements are in situ (no sample transfer needed). All temperature readings are monitored by a Eurotherm unit using type K thermocouples spot-welded to a Ta clip attached to the sample edge. Calibration with a pyrometer indicates that the thermocouple readings are systematically too low (∼20−50 °C) in the 500− 900 °C range. However, this is a systematic error so that the relative measurements are reproducible.

Article

ASSOCIATED CONTENT

S Supporting Information *

Transmission infrared absorbance spectra of unzipped GONRs from MWCNTs and FWCNTs before and after unzipping, XRD patterns of GONRs from MWCNTs using 850 wt. % KMnO4 in H2SO4, infrared transmission differential spectra of thermally reduced GONRs unzipped from MWCNTs and FWCNTs at 60−850 °C, XRD patterns of GONRs from MWCNTs and FWCNTs using 850 wt. % KMnO4 in H2SO4 with additional H3PO4, transmission infrared absorbance spectra of GOFs and unzipped MWCNTs (GONRs) at room temperature, and transmission infrared differential spectra of GOFs and unzipped GONRs (initially oxidized with H3PO4) after annealing at 60−850 °C. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions

M.A. performed all FTIR measurements of GONRs and GOFs and IR studies combined with the in situ thermal annealing process with the help of R.G. M.A. also worked in direct coordination with J.C.-G. and E.C.-M. for materials synthesis. J.C.-G. performed all oxidative unzipping processes. E.C.-M. ran all TEM and HRTEM measurements along with the sample deposition for the thermal annealing process. AFM images were collected by D.M.R. and analyzed with the help of J.C.-G. and E.C.-M. R.H.B. and Y.J.C. both supervised the work. M.A. wrote the manuscript together with J.C.-G. and E.C.-M. Both R.H.B. and Y.J.C. reviewed the manuscript, and all authors confirm the work for publication. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering, under Award DE-SC001951. We also acknowledge the support of Dr. Marcio Dias-Lima and Dr. Raquel Ovalle-Robles for providing multiwalled and few-walled carbon nanotubes, respectively.



REFERENCES

(1) Li, X.; Wang, X.; Zhang, L.; Lee, S.; Dai, H. Science 2008, 319, 1229. (2) Kosynkin, D. V.; Higginbotham, A. L.; Sinitskii, A.; Lomeda, J. R.; Dimiev, A.; Price, B. K.; Tour, J. M. Nature 2009, 458, 872−877. (3) Geim, A. K.; Novoselov, K. S. Nat. Mater. 2007, 6, 183−191. (4) Li, X. S.; Cai, W. W.; An, J. H.; Kim, S.; Nah, J.; Yang, D. X.; Piner, R. D.; Velamakanni, A.; Jung, I.; Tutuc, E.; et al. Science 2009, 324, 1312−1314. (5) Park, S.; Ruoff, R. S. Nat. Nanotechnol. 2009, 4, 217−224. (6) Kwon, J.; Lee, S. H.; Park, K.; Seo, D.; Lee, J.; Kong, B.; Kang, K.; Jeon, S. Small 2011, 7, 864−868. (7) Fuhrer, M. S. Nat. Mater. 2010, 9, 611−612. (8) Li, X.; Zhao, T.; Wang, K.; Yang, Y.; Wei, J.; Kang, F.; Wu, D.; Zhu, H. Langmuir 2011, 27, 12164−12171. (9) Jiao, L.; Zhang, L.; Ding, L.; Liu, J.; Dai, H. Nano Res. 2010, 3, 387−394. (10) Terrones, M. ACS Nano 2010, 4, 1775−1781. (11) Yang, X.; Dou, X.; Rouhanipour, A.; Zhi, L.; Rader, H. J.; Mullen, K. J. Am. Chem. Soc. 2008, 130, 4216−4217. 24014

dx.doi.org/10.1021/jp303035m | J. Phys. Chem. C 2012, 116, 24006−24015

The Journal of Physical Chemistry C

Article

(12) Sinitskii, A.; Dimiev, A.; Kosynkin, D. V.; Tour, J. M. ACS Nano 2010, 4, 5405−5413. (13) Li, L.; Qin, R.; Li, H.; Yu, L.; Liu, Q.; Luo, G.; Gao, Z.; Lu, J. ACS Nano 2011, 5, 2601−2610. (14) Price, B. P.; Lomeda, J.; Tour, J. M. Chem. Mater. 2009, 21, 3917−3923. (15) Cataldo, F.; Compagnini, G.; Patanè, G.; Ursini, O.; Angelini, G.; Ribic, P. R.; Margaritondo, G.; Cricenti, A.; Palleschi, G.; Valentini, F. Carbon 2010, 48, 2596−2602. (16) Jiao, l.; Wang, X.; Diankov, G.; Wang, H.; Dai, H. Nat. Nanotechnol. 2010, 5, 321−325. (17) Zhang, Z.; Sun, Z.; Yao, J.; Kosynkin, D. V.; Tour, J. M. J. Am. Chem. Soc. 2009, 131, 13460−13463. (18) Higginbotham, A. L.; Kosynkin, D. V.; Sinitskii, A.; Sun, Z.; Tour, J. M. ACS Nano 2010, 4, 2059−2069. (19) Li, J.; Kudin, K. N.; McAllister, M. J.; Prud’homme, R. K.; Aksay, I. A.; Car, R. Phys. Rev. Lett. 2006, 96, 176101. (20) Rangel, N. L.; Sotelo, J. C.; Seminario, J. M. J. Chem. Phys. 2009, 131, 031105. (21) Shimizu, T.; Haruyama, J.; Marcano, D. C.; Kosynkin, D. V.; Tour, J. M.; Hirose, K.; Suenaga, K. Nat. Nanotechnol. 2011, 6, 45−50. (22) Sinitskii, A.; Fursina, A. A.; Kosynkin, D. V.; Higginbotham, A. L.; Natelson, D.; Tour, J. M. App. Phys. Lett. 2009, 95, 253108-1− 253108-3. (23) Kosynkin, D. V.; Lu, W.; Sinitskii, A.; Pera, G.; Sun, Z.; Tour, J. M. ACS Nano 2011, 5, 968−974. (24) Zhu, Y.; Lu, W.; Sun, Z.; Kosynkin, D. V.; Yao, J.; Tour, J. M. Chem. Mater. 2011, 23, 935−939. (25) Acik, M.; Lee, G.; Mattevi, C.; Pirkle, A.; Wallace, R. M.; Chhowalla, M.; Cho, K.; Chabal., Y. J. J. Phys. Chem. C 2011, 115, 19761. (26) Kwon, J.; Saly, M.; Halls, M. D.; Kanjolia, R. K.; Chabal, Y. J. Chem. Mater. 2012, 24, 1025. (27) Stankovich, S.; Dikin, D. A.; Piner, R. D.; Kohlhaas, K. A.; Kleinhammes, A.; Jia, Y.; Wu, Y.; Nguyen, S. T.; Ruoff, R. S. Carbon 2007, 45, 1558. (28) Mohanty, N; Moore, D.; Xu, Z.; Sreeprasad, T. S.; Nagaraja, A.; Rodriguez, A. A.; Berry, V. Nat. Commun. 2012, 3, 844. (29) Elias, A. L.; Botello-Mendez, A. R.; Meneses-Rodriguez, D.; Gonzalez, V. J.; Ramirez-Gonzalez, D.; Ci, L.; Munoz-Sandoval, E.; Ajayan, P. M.; Terrones, H.; Terrones, M. Nano Lett. 2010, 10, 366. (30) Acik, M.; Mattevi, C.; Gong, C.; Lee, G.; Cho, K.; Chhowalla, M.; Chabal, Y. J. ACS Nano 2010, 4, 5861. (31) Acik, M.; Lee, G.; Mattevi, C.; Chhowalla, M.; Cho, K.; Chabal, Y. J. Nat. Mater. 2010, 9, 840. (32) Marcano, D. C.; Kosynkin, D. V.; Berlin, J. M.; Sinitskii, A.; Sun, Z.; Slesarev, A.; Alemany, L. B.; Lu, W.; Tour, J. M. ACS Nano 2010, 4, 4806−4814. (33) Cresti, A.; Lopez-Bezanilla, A.; Ordejón, P.; Roche, S. ACS Nano 2011, 9271−9277. (34) Yamaguchi, H.; Murakami, K.; Eda, G.; Fujita, T.; Guan, P.; Wang, W.; Gong, C.; Boisse, J.; Miller, S.; Acik, M.; et al. ACS Nano 2011, 5, 4945−4952. (35) Lomeda, J. R.; Doyle, C. D.; Kosynkin, D. V.; Hwang, W.-H.; Tour, J. M. J. Am. Chem. Soc. 2008, 130, 16201−16206. (36) Dubois, S. M.; Lopez-Bezanilla, A.; Cresti, A.; Triozon, F.; Biel, B; Charlier, J; Roche, S. ACS Nano 2010, 4, 1971−1976. (37) Johari, P.; Shenoy, V. B. ACS Nano 2011, 5, 7640−7647. (38) Zhang, C. X.; He, C.; Yu, Z.; Zhang, K. W.; Sun, L.; Zhong, J. J. Phys. Chem. C 2011, 21893−21898. (39) Zhang, M.; Atkinson, K.; Baughman, R. H. Science 2004, 306, 1358.

24015

dx.doi.org/10.1021/jp303035m | J. Phys. Chem. C 2012, 116, 24006−24015