Article pubs.acs.org/JACS
Sodide and Organic Halides Effect Covalent Functionalization of Single-Layer and Bilayer Graphene Mandakini Biswal,† Xu Zhang,*,† David Schilter,† Tae Kyung Lee,⊥ Dae Yeon Hwang,⊥ Manav Saxena,† Sun Hwa Lee,† Shanshan Chen,§ Sang Kyu Kwak,†,⊥ Christopher W. Bielawski,†,#,∇ Wolfgang S. Bacsa,∥ and Rodney S. Ruoff*,†,#,⊗ †
Center for Multidimensional Carbon Materials (CMCM), Institute for Basic Science (IBS), Ulsan 44919, Republic of Korea School of Energy and Chemical Engineering, Ulsan National Institute of Science and Technology (UNIST), Ulsan 44919, Republic of Korea § Department of Physics, Renmin University of China, Beijing, 100872, P. R. China # Department of Chemistry, UNIST, Ulsan 44919, Republic of Korea ∇ Department of Energy Engineering, Ulsan National Institute of Science and Technology (UNIST), Ulsan 44919, Republic of Korea ∥ CEMES-CNRS and University of Toulouse, 29 rue Jeanne Marvig, 31055 Toulouse, France ⊗ School of Materials Science and Engineering, Ulsan National Institute of Science and Technology (UNIST), Ulsan 44919, Republic of Korea ⊥
S Supporting Information *
ABSTRACT: The covalent functionalization of single and bilayer graphene on SiO2 (300 nm)/Si was effected through sequential treatment with the alkalide reductant [K(15-crown-5)2]Na and electrophilic aryl or alkyl halides, of which the iodides proved to be the most reactive. The condensation reactions proceeded at room temperature and afforded the corresponding aryl- or alkyl-appended graphenes. For each sample, Raman and X-ray photoelectron spectroscopies were used to evaluate the degrees and uniformities of functionalization. Statistical analyses of the Raman data revealed that the introduction of the organic moieties was accompanied by sp3-rehybridization of the basal plane atoms. When bilayers consisting of 13C and 12C layers were treated, both the top and bottom sheets were decorated with organic groups. The reaction was followed using Raman spectroscopy, and the mechanism was studied by theoretical calculations. Indicative of its structure and reactivity, 4-pyridyl-decorated single-layer graphene was readily benzylated and appears to be an ideal platform to develop functional materials.
■
a lower chemical potential.10 Indeed, the covalent functionalization of BLG has proven to be relatively challenging when compared to SLG, a phenomenon that has been attributed to charge fluctuations and/or van der Waals forces between the two layers.11−13 The reactivity of graphene on Si with a superficial SiO2 layer is further enhanced by intrinsic substrate roughness and impurities that cause graphene to exhibit electron−hole charge fluctuations that affect its charge doping and Fermi level.11,12 The binding energy of such twodimensional materials toward addends is strongly influenced by van der Waals forces between individual sheets and/or the substrate.11 When on SiO2/Si, BLG exhibits stronger van der Waals forces than SLG, with the former exhibiting a lower binding affinity for adatoms and thus being more reluctant to undergo functionalization. In the case of hydrogenation,
INTRODUCTION The functionalization of graphene not only has a profound effect on its physical and chemical properties, but also represents an important step in the development of new twodimensional materials. In this context, two prominent targets are graphane and diamane, which refer to fully hydrogenated single-layer and bilayer graphene, respectively.1−3 Theoretical investigations have indicated that such sp3-hybridized carbon materials should exhibit high stabilities, with such predictions spurring experimentalists to devise routes toward these elusive species.1−3 In general, the functionalization of the graphene basal plane entails its treatment with either gas- or solutionphase reagents, and may afford a variety of products featuring covalently bonded atoms or organic moieties.4−8 While such reactivity has been explored with single-layer graphene (SLG), the functionalization toolkit for bilayer graphene (BLG) is comparatively small. When atop SiO2/Si, the electronic structures of SLG and BLG differ,9 with the latter exhibiting © XXXX American Chemical Society
Received: January 27, 2017
A
DOI: 10.1021/jacs.7b00932 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society
Figure 1. Covalent functionalization protocol. (a) Schematic illustration of graphene functionalization using alkalide solution and organic reagents. (b) Dark blue solution prepared by mixing NaK and 15-crown-5 in THF.
donors,34 and the latter property is exploited here in a one-pot, two-step methodology for covalent functionalization of SLG and BLG. The procedure begins with the mechanical exfoliation of graphite onto SiO2 (300 nm)/Si, a process known to afford SLG and BLG regions of very high crystallinity. Treatment with alkalide triggers reduction of the graphenes to the corresponding graphenides, whose exposure to organic halides then yields single and bilayer products with organic surface addends. Described below is the synthesis and characterization of these new two-dimensional materials, as well as their further elaboration.
theoretical calculations indicate that the total binding energy (i.e., the formation energy of each C−H bond) associated with the reduction of SLG is 2.9 eV whereas for BLG it is 0.9 eV.13 Functionalization of BLG is a unique problem with practical complications. It is not obvious whether one or both layers will react, and the outcome may depend on the interlayer stacking orientation. Moreover, the passage of reactants between the interfaces or through defects may lead to a variety of regiochemical outcomes. For all the challenges involved in the chemistry and manipulation of such thin materials, these experiments are arguably of greater fundamental value than those involving bulk materials. As will become clear, the use of SLG and BLG (including isotopically labeled samples) affords mechanistic insights not available from the study of graphite, for example. In terms of functionalization methodology, significant efforts have focused on treating graphenes with reactive solutions, including those containing radicals derived from diazonium salts or peroxides.14−20 Such methods are useful in decorating graphene with covalently bonded organic addends, although the extent of these reactions is generally much lower for BLG relative to SLG on SiO2/Si.16,17,20 An exception occurs when an external voltage is applied, where the chemical potential of BLG is tunable such that its reactivity is comparable to that of SLG.21,22 However, such experiments are less practical as they require the deposition of electrodes if the graphene is supported by a dielectric such as SiO2/Si. Another method for graphene functionalization involves use of reductive alkali metal solutions or emulsions as sources of solvated electrons.23−25 The reduction of graphene to graphenide is typically followed by treatment with electrophiles such as organic halides. Among the most widely used reductants are the Birch-type solutions, prepared by dissolving an alkali metal such as Li in NH3 at low temperature (below −33 °C).26−28 A more convenient procedure involves treatment with sodium− potassium (NaK) alloy emulsified in aprotic solvents such as tetrahydrofuran (THF). This method enables functionalization of graphite and graphene at room temperature,23 although the inhomogeneous reagents afford less control over the reactions. Employed in the present study is a sodide (Na−) compound, an example of an alkalide salt.29,30 Such compounds are readily prepared by dissolving alkali metals in aprotic solvents such as THF containing crown ethers or cryptands.31−33 Chelation of the macrocycle(s) to the more electropositive zerovalent metal can lower its oxidation potential enough to induce “disproportionation”. In the present case, equimolar NaK alloy dissolves in a THF solution of 15-crown-5 to give the dark blue alkalide [K(15-crown-5)2]Na, in which K and Na are in the +I and −I oxidation states, respectively.32 Sodide and other alkali anions are known to be strong bases and electron
■
RESULTS AND DISCUSSION Covalent Functionalization and Thermal Recovery of Graphene. The steps involved in the present graphene functionalization protocol are shown in Figure 1a. Addition of NaK to 15-crown-5 in THF affords a highly O2- and H2Osensitive dark blue salt (Figure 1b), in which the large K+ cations exist in the sandwich [K(15-crown-5)2]+ dissociated from the Na− counteranions.31,32 The dissolution of liquid NaK in THF/15-crown-5 is rapid, and this starting alloy is used in preference to slower-reacting solid Na or K. The mixed-metal salt [K(15-crown-5)2]Na efficiently reduces graphenes, and the resulting graphenides are susceptible to reaction with organic halides to afford materials decorated with the organic chains. The present work involved mechanical exfoliation of graphite onto SiO2 (300 nm)/Si, with a sample imaged by atomic force microscopy (AFM) featuring a thin flake of SLG and BLG adjacent to a thicker graphitic flake (Figure 2a). The graphene regions were identified according to their height profiles, with SLG and BLG having thicknesses of around 0.9 and 1.4 nm, respectively (Figure 2b). Such regions are unambiguously assigned from their Raman shifts and line widths.35 SLG on SiO2/Si gives rise to a sharp 2D peak at ∼2678 cm−1 with a full width at half-maximum (FWHM2D) less than 30 cm−1. In contrast, AB-stacked BLG has a blue-shifted, broader 2D peak (FWHM2D: ∼55 cm−1, 2D band shift: ∼10 cm−1).35 In the present case, the FWHM2D map (Figure 2c) allows one to distinguish SLG from BLG regions, providing a picture consistent with AFM measurements. Further, Raman spectra of graphenes are sensitive to the presence of defects, these giving rise to D and D′ bands otherwise unobserved on symmetry grounds.36 The intensity ratio of the D and G bands (ID/IG) can report on the defect density (see below), and it is noted that this quotient is small across SLG and BLG basal planes, being nontrivial only at their edges (Figure 2d). Indeed, the absence of D or D′ bands from the bulk (Figure 2e) is consistent with the material being essentially free of defects. The pristine graphene flakes were immersed in a solution of freshly prepared [K(15-crown-5)2]Na in THF under an B
DOI: 10.1021/jacs.7b00932 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society
Figure 2. Pyridyl functionalization and thermal recovery of mechanically exfoliated graphene. (a) AFM image of a mechanically exfoliated graphene flake on SiO2 (300 nm)/Si containing both SLG and BLG. (b) AFM height profile of SLG and BLG shown in (a). (c) Raman FWHM map of 2D band (FWHM2D) of SLG and BLG. (d) Raman intensity ratio (height) map of D and G bands (ID/IG) of SLG and BLG. (e) Raman spectra taken at the positions marked in (d). (f) ID/IG map of Py-SLG and Py-BLG. (g) Raman spectra taken at the positions marked in (f). (h) ID/IG map of graphene after annealing. (i) Raman spectra taken at the positions marked in (h). An excitation wavelength of 532 nm was used for all analyses.
indicated functionalization had taken place over the whole flake, with the ID/IG ratio being greatest at its edges (SI, Figure S2). The thermal stability of the 4-pyridyl-containing sample was investigated by annealing at 450 °C for 120 min (4 × 10−1 Torr Ar); the ID/IG map of the heat-treated sample is shown in Figure 2h. Heating Py-SLG/Py-BLG resulted in the disappearance of the D and D′ bands characteristic of functionalization (Figure 2i). This observation was consistent with a recovery of a graphene lattice, although some curling of the flake was observed after annealing. Although 4-pyridyl groups covalently attached to graphene can be removed at 450 °C, it is known that other types of defects such as basal plane vacancies or N or B dopant atoms cannot be removed at this temperature.37,38 This result, in addition to a series of X-ray photoelectron spectroscopic measurements (XPS, Figure S3), suggested to us that the pyridyl groups were not incorporated into the plane but instead were bound to basal C atoms. Thus, we concluded that the latter must then be the origin of the D and D′ peaks recorded for Py-SLG as well as Py-BLG. Statistical Analysis of Raman Spectra. Raman analysis enables quantification of defect density and degree of disorder
atmosphere of Ar. After 1 h, the mixture was treated with an organic halide dissolved in THF; the different halides employed are summarized in Table S1 (Supporting Information, SI). In the case of 4-iodopyridine, the reaction is rapid, with the characteristic blue color of the alkalide disappearing after only around 2 min. In contrast to the ID/IG map of the pristine flakes, that of the product (Figure 2f) reveals 4-pyridylfunctionalized SLG (Py-SLG) regions with ID/IG ratios typically exceeding 2 (∼5.4 on average, SI, Figure S1a), as well as functionalized BLG (Py-BLG) with ID/IG ≈ 1. Furthermore, the intensities of 2D peaks for Py-SLG and Py-BLG are less than those of the material before treatment (SI, Figure S1b, c). As discussed above, the emergence of a Raman D peak indicates the presence of defects (Figure 2g), which in Py-SLG and Py-BLG presumably take the form of sp3-hybridized basal C atoms bound to 4-pyridyl groups. Apart from the intense D peaks, D′ peaks appear at 1620 cm−1 for both Py-SLG and PyBLG, being less prominent for the latter. Since the BLG region is small for the flake in Figure 2, additional reactions were conducted on a larger AB-stacked BLG flake. Raman analysis C
DOI: 10.1021/jacs.7b00932 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society
Figure 3. Statistical analysis of Raman spectral parameters. (a-c) ID/IG (a), ID′/IG (b) and ID/ID′ (c) ratios from 100 Raman spectra of Py-SLG and Py-BLG. (d) Plots of ID/IG versus ID′/IG obtained from the 100 Raman spectra for both Py-SLG and Py-BLG.
SLG and Py-BLG. The ID′/IG histograms obtained for Py-SLG and Py-BLG were fitted to Gaussian curves, whose respective maxima at 0.61 and 0.20 are consistent with the higher reactivity displayed by SLG. A similar treatment of the ID/ID′ data recorded for Py-SLG and Py-BLG gave values of 9.1 and 3.9, respectively (Figure 3c). With intensity ratios in hand, the ID/IG ratio may be plotted against the ID′/IG ratio obtained for Py-SLG as well as Py-BLG (Figure 3d) such that the sample distributions become more apparent. This analysis is valuable in that the quotient ID/ID′ depends strongly on the type of defects in SLG.42,43 The typical ID/ID′ ratios arising from boundaries (3.5), vacancies (7), substitutional B atoms (9), and sp3hybridized basal sites (13; C−F, C−H, or C−O) in SLG are rather distinct.42,43 In the present work, the mode of ID/ID′ ratios for Py-SLG is greater than 7, indicating that the D bands do not arise from boundaries or vacancies. Rather, the data are consistent with “defects” in the present Py-SLG taking the form of sp3-hybridized C sites from covalent functionalization, a conclusion in agreement with that drawn from the annealing experiments. Indeed, while the ID/ID′ ratio is less than 13, the disparity may be accounted for in the difference between the possible groups that can be formed (i.e., C−pyridyl vs C−F, C−H, or C−O).42,43 For Py-BLG, the ID/ID′ ratio was found to be consistent with a report in which BLG, covalently functionalized using aryldiazonium salts, exhibited D and D′ bands with ID/ID′ ≈ 4.11 Again, this result indicates that the observed D and D′ bands result from the formation of sp3hybridized C centers bound to 4-pyridyl groups. Covalent Functionalization of Isotopically Labeled Graphene. The data described thus far do not address whether the present methodology results in functionalization of one or both of the two layers in BLG. Answering this question required the preparation of BLG with 12C and 13C layers, which
for carbon materials such as nanocrystalline graphite and graphene.39−41 In the case of SLG, the ID/IG ratio can allow calculation of the defect density nD (cm−2): nD(cm−2) =
(1.8 ± 0.5) × 1022 ⎛ ID ⎞ ⎜ ⎟ λL 4 ⎝ IG ⎠
(1)
where λL (nm) is the excitation wavelength.41 Statistical analyses of the ID/IG quotient obtained from 100 Raman spectra facilitate an estimation of the degree of functionalization in Py-SLG and Py-BLG (Figure 3a). The highest ID/IG ratios recorded for Py-SLG and Py-BLG were 6.4 and 2.1, respectively. Gaussian fits of the ID/IG histograms for Py-SLG and Py-BLG have maxima at 5.40 and 0.77, respectively, values that represent the most probable ID/IG quotients. The intense and sharp 2D signal arising from Py-SLG is consistent with a “stage I” classification, and the most probable ID/IG ratios can be substituted into eq 1 to give nD. The estimated defect density of (1.21 ± 0.34) × 1012 cm−2 corresponds to the surface density of pyridyl groups, assuming that side reactions including oxidation are negligible. In the case of Py-BLG, comparison is made to ID/IG values for Ar+-sputtered BLG of known defect density.36 The present Py-BLG has a defect density somewhat lower than 1012 cm−2, consistent with the lower reactivity of BLG toward alkalides. Despite this, the degree of BLG functionalization here is greater than that resulting from other wet-chemical routes including free radical addition using diazonium salts or peroxides, which afford SiO2/ Si-supported Py-BLG with ID/IG ratios of ∼0 or
