Do defects enhance fluorination of graphene?

Abstract

Controlling the functionalization of graphene is essential for many applications. Here, we probed the reactivity of monolayer and bilayer graphene samples with intentionally prepared specific numbers of defects. Fluorination was used as a model reaction. We demonstrate that the reactivity of the single-layer graphene is not significantly affected by oxygen plasma-generated defects. On the other hand in the case of graphene bilayers, a decrease in the reactivity was observed for a small number of new defects, while an increase in reactivity was found for a larger number of defects. The long-term stability test of the fluorinated samples showed that minor changes in the sample occur during the first week after fluorination.

---

Introduction

The fluorinated graphene is the thinnest insulator,¹ which has great potential for application in nanoelectronics and spintronics.^2–5 However, the formation of this material from single-layer graphene is still not well explored. Incorporation of fluorine atoms into graphene causes local structural modifications, which can lead to a change from the semi-metallic material – graphene to the insulating material – fluorographene.^2,6,7 In other words, the structural transformation leads to opening of the band gap. In contrast to conventional crystalline materials with a fixed band gap, it has been shown that the bilayer graphene band gap is in principle tunable.⁸ Much effort has been given to developing approaches to control the band gap-opening process to tune the final material properties, in particular in the case of bilayer graphene.^8–10 Fluorination of graphene has been intensely investigated as a promising pathway to this goal. The course of the fluorination reaction has been studied in detail with the aid of Raman spectroscopy and isotopically labelled bilayer graphene of both Bernal and turbostratic stacking of the layers.¹¹ Nevertheless, full control of the fluorination has not been achieved yet, as several external parameters, such as the presence of defects^12,13 and samples containing both bilayers and monolayers¹¹ of graphene, essentially influence the reaction. As shown recently,¹¹ the fluorination process is more efficient for turbostratic than for Bernal bilayer (AB 2-LG) graphene under the same conditions. In turbostratic graphene (T 2-LG) the fluorination rate is similar in both layers, while in AB 2-LG-stacked graphene the top layer becomes much more fluorinated than the bottom layer. In addition, the fluorination rate is suggested to increase with the degree of fluorination, which is rationalized by the increased number of defects.¹⁴ Although the crucial role of defects in graphene functionalization has been suggested,¹⁵ no systematic study has been performed yet.

Defects are commonly found in real graphene samples and being important reactive sites, they should contribute to the susceptibility of graphene to any functionalization.^16,17 The defects in the graphene structure can occur either as sites possessing sp³ hybridization instead of the common sp² hybridization or in the form of vacant spots in the two-dimensional carbon network.¹⁸

There are several approaches to defects' investigation in graphene layers. A straightforward method is the direct observation via AC-HRTEM or STM microscopy,^19–21 but both are demanding on equipment and the specific preparation of the sample for imaging makes its further processing complicated. In addition, DFT computational methods are used for assistance in the structure considerations.²² The number of defects in graphene can be indirectly observed via Raman spectroscopy as well. This is an easily executable, convenient method that does not need any special adjustments of the sample; all of this makes it a method of first choice for defects' investigations. One of the most important features in the Raman spectrum to be observed to evaluate defects is the D band.^23,24 Together with its second overtone 2D and the G band, corresponding to the primary in-plane vibrational mode, these modes typical for graphene carry information about the defects as well.²⁵ In many cases it is more convenient to evaluate the A_(D)/A_(G) or A_(D)/A_(2D) ratios; both increase with the disorder. In the case of a large number of defects, broadening of Raman bands of graphene is also observed.²³ A simple relation for quantification of defects in single-layer graphene is given by the equation:²⁶ n_D = 7.3 × 10⁹ × (I_D/I_G) × E_L,⁴ where n_D is the number of defects, I_D and I_G the intensity of the D and G bands, respectively, and E_L is the laser excitation energy. This approach gives correct results for monolayer graphene (1-LG). However in the case of graphene bilayer the situation is more complex. Depending on stacking order, T 2-LG or AB 2-LG, one needs to include a possible scattering of phonons in the non-defect layer by the next defective layer.²⁷

In the presented work, we intentionally prepared graphene samples with a defined number of defects using oxygen plasma. The defective graphene samples were subsequently fluorinated and the degree of fluorination was analysed based on the Raman spectroscopy and XPS data. These experiments allowed us to unravel the influence of the number of defects on the fluorination of graphene. Moreover, the change in the fluorinated samples over time was investigated, which allowed us to address the influence of the initial defects on the stability of the final product – fluorinated graphene.

Experimental

Graphene synthesis

Graphene samples containing single and bilayer grains were obtained via chemical vapour deposition (CVD) of methane (CH₄) on a Cu foil.²⁸ Briefly, the Cu foil was placed in the centre of a furnace, which was then heated to 1273 K and annealed for 20 min in a flow of H₂ (50 standard cubic centimetres per minute). Subsequently, CH₄ was introduced for 45 min. Afterwards, the samples were annealed in H₂ for 5 min and then cooled to room temperature, maintaining the H₂ flow. The pressure was kept at 0.35 Torr during the whole process. The graphene layers were transferred onto SiO₂/Si using poly(methyl methacrylate) (PMMA) polymer as described elsewhere.²⁹ The SiO₂/Si substrate was then cut into four parts, from now on referred to as S1, S2, S3 and S4.

Formation of defects

The graphene samples were placed inside the chamber of an oxygen plasma etcher (PICO, Diener Plasma-Surface Technology, Germany). The chamber was first pumped out and then filled with O₂. When the set pressure was reached (0.90 mbar), the plasma was turned on (350 W) for 2, 4 or 10 s (samples S2, S3 or S4, respectively). Afterwards, the plasma chamber was purged. Sample S1 was not subjected to any plasma treatment and was used as a control sample.

Fluorination

The samples were fluorinated in a homebuilt chamber using XeF₂ crystals.¹¹ The chamber containing the graphene samples S2–S4 was first evacuated to 10⁻⁵ Torr. Then, the samples were exposed to XeF₂ for 10 min.

Raman spectroscopy

Raman maps containing 625 spectra (25 × 25) were acquired just after the graphene samples were produced (raw), after the samples were treated in the oxygen plasma (defected samples), and after the samples were subjected to fluorination (fluorinated samples). In addition, Raman maps were also acquired at different periods of time after fluorination (1–40 days). The Raman data were obtained by measuring the same graphene grains in each sample and analysing the same area in the different steps (pristine samples, after formation of defects, fluorinated samples) to ensure a reliable comparison of the results. The Raman maps were obtained using a WiTec Alpha300, equipped with a 532 nm laser, a 50× objective, a 600 g mm⁻¹ grating and a 105 μm diameter fibre.

XPS

The XP spectra of the samples were measured using a modified ESCA3 MkII electron spectrometer equipped with a hemispherical electron analyser operated in a fixed transmission mode. Al Kα radiation was used for electron excitation. The binding energy scale was calibrated using the Au 4f_7/2 (84.0 eV) and Cu 2p_3/2 (932.6 eV) photoemission lines. The pressure in the XPS analysis chamber during spectra acquisition was 3 × 10⁻⁹ mbar. The spectra were collected at a detection angle of 35° with respect to the macroscopic surface plane. Survey scan spectrum and high resolution spectra of F 1s, C 1s and Si 2p photoelectrons were measured. The accuracy of the measured binding energies was ±0.1 eV. The spectra were curve fitted after subtraction of Shirley background³⁰ using the Gaussian–Lorentzian line shape and nonlinear least-squares algorithms. Quantification of the elemental concentrations was accomplished by correcting the photoelectron peak intensities for their cross-sections³¹ and for the analyser transmission function. The typical error of the quantitative analysis by XPS is ≈10%.

Results and discussion

The graphene samples were prepared by the CVD method and then they were exposed to the plasma treatment for a specific period of time to create defects. Afterwards, the samples were fluorinated. Fig. 1a shows schematically the treatment sequence applied to graphene samples. The fluorinated samples were measured by high-resolution Raman mapping immediately after the fluorination and then again in defined time steps so that the changes in the samples over time could be followed. Note that all Raman maps were measured on the identified graphene grains, hence it is possible to follow the development of each spot of the Raman map after the treatment (formation of defects, fluorination).

Fig. 1 (a) Scheme summarising the experiments performed on the graphene layers. (b) Optical image of the analysed area of sample S2. (c) Raman map showing A_(D)/A_(G) of sample S2. (d) Raman map showing FWHM (2D) and the different types of graphene, monolayer, bilayer in Bernal stacking, and bilayer randomly stacked on sample S2.

To minimize the role of the preparation/transfer procedure, we used one large sample in which we identified four regions of the pristine sample that contained grains consisting of 1-LG, T 2-LG and AB 2-LG bilayers. An example of one of the regions is shown in Fig. 1b. Detailed Raman maps were measured in each of the regions and thus we could determine the structural perturbations (defects) through the ratio of the D and G band areas A_(D)/A_(G), see Fig. 1c, and to address separately the 1-LG, AB 2-LG and T 2-LG areas through the 2D band width (Fig. 1d). The pristine sample was cut into four parts, each containing one of the selected regions. Then, defects were introduced by oxygen plasma treatment for 0, 2, 4 or 10 s and the parts were denoted samples S1, S2, S3 or S4, respectively. Finally, the samples S1–S4 were simultaneously fluorinated.

At each step (pristine, after defect creation, after fluorination) the Raman maps of the samples S1–S4 were measured. Fig. 2a shows Raman spectra of graphene samples with different numbers of initial defects (S1 – no intentionally introduced defects, S3 and S4 – defects introduced by oxygen plasma treatment for 4 s and 10 s, respectively). Raman spectra are averaged over the 1-LG, T 2-LG and AB 2-LG areas. Note that samples S2 and S3 gave similar results, hence only S3 is shown for clarity.

Fig. 2 (a) Averaged Raman spectra of the D and G band region, of the graphene samples S1, S3 and S4. Black, blue and green spectra are assigned to raw, defected and fluorinated samples, respectively. (b) A_(D)/A_(G) data obtained for graphene samples S1–S4, in raw, defected and fluorinated states. Samples S1–S4 were subjected to 0, 2, 4 and 10 s of oxygen plasma treatment, respectively, and accordingly they underwent the identical fluorination process.

It was shown previously^32,33 that the D/G ratio correlates with the amount of defects created by oxygen plasma. It is very difficult to distinguish different types of defects in graphene. Nevertheless, it is very unlikely to have for example air-stable vacancy in graphene sheet. It was shown for example by Kaloni et al.³⁴ that vacancies in graphene react with oxygen. Consequently, it can be assumed that the defects prepared by oxygen plasma represent ‘typical’ defects found in typical graphene samples.

We note that the D band was observed also in the spectra of the pristine graphene samples, but it is significantly smaller than the D band observed in the spectra of the samples after the intentional creation of defects or after fluorination.

In Fig. 2b the number of structural perturbations (number of defects or degree of functionalization) is represented by the A_(D)/A_(G) ratio for the probed samples. Initially, the A_(D)/A_(G) ratio was found to be <0.4 (black bars) for all samples, increasing slightly after the defects were induced by oxygen plasma in samples S2 and S3, and increasing even more for sample S4, where significantly more defects were created by the longest exposure to the oxygen plasma (blue bars). It is noteworthy that for 1-LG we found in all cases a larger A_(D)/A_(G) ratio than for 2-LG.²⁷ In addition, note that the A_(D)/A_(G) ratio of AB 2-LG overestimates the number of defects, by up to a factor of 2 because the phonons in the bottom graphene layer are scattered by defects in the top graphene layer.¹¹

After the graphene samples were fluorinated, the A_(D)/A_(G) ratio increases significantly confirming a successful fluorination. Samples S1 and S4 after fluorination showed A_(D)/A_(G) ratios for 2-LG > 1, even reaching values above 2 for 1-LG, while in the case of the fluorinated graphene samples S2 and S3 the A_(D)/A_(G) > 1 only for 1-LG, while for T and AB 2-LG the A_(D)/A_(G) ratio remains below 1 and 0.5, respectively. As previously observed, both for defected and fluorinated graphene samples, 1-LG Raman spectra are more affected by the introduction of defects or functional groups than 2-LG.^11,27 The number of defects calculated and corrected using the approach in reference (ref. 27) is given in Table 1.

Table 1 Number of defects in analysed raw, defected and fluorinated samples, computed after references,^11,27 for 1-LG, AB 2-LG and T 2-LG (×10⁹ cm⁻²)

	S1			S2
	1-LG	AB 2-LG	T 2-LG	1-LG	AB 2-LG	T 2-LG
Pristine	41	1.8	11	33	2.2	9.6
Defected	—	—	—	54	4.1	15
Fluorinated	270	55	130	540	49	75

---

	S3			S4
	1-LG	AB 2-LG	T 2-LG	1-LG	AB 2-LG	T 2-LG
Pristine	38	1.9	6.6	31	2.5	8.1
Defected	69	4.6	9.7	210	8.5	31
Fluorinated	390	24	49	250	40	110

---

For fluorinated 1-LG, the value of the A_(D)/A_(G) ratio increases with the number of intentionally introduced defects and it is higher for all plasma-treated samples (before fluorination) as compared with the A_(D)/A_(G) of pristine 1-LG. However, only sample S2 exhibited a higher increase in the A_(D)/A_(G) ratio after fluorination than the plasma-untreated 1-LG sample. For samples S3 and S4 the increase in the A_(D)/A_(G) ratio after fluorination does not even reach the value obtained in the case of pristine graphene, as demonstrated by green bars in Fig. 2. A corresponding trend is also documented in Table 1 – the computed number of defects in fluorinated sample S2 is higher than for samples treated with oxygen plasma for a longer time, although the number of defects before fluorination simply grew with the plasma treatment duration. Consequently it is clear that the fluorination of 1-LG graphene is only slightly affected by defects. This can be understood by the relatively high reactivity of 1-LG graphene samples.

In the case of the moderately plasma-treated T and AB bilayer graphenes (samples S2 and S3) the value of the A_(D)/A_(G) ratio is lower after fluorination than that for the corresponding pristine T and AB bilayer graphenes (S1) after fluorination. In contrast, both AB and T regions of the strongly plasma-treated sample (S4) exhibit slightly higher A_(D)/A_(G) ratios than that of the corresponding areas of pristine graphene (S1) after fluorination. The number of defects in plasma-treated 2-LG AB samples before fluorination was comparable for samples S2 and S3, while for samples S4 the number of defects was almost three times higher (Table 1). This is consistent with a previous study on oxygen plasma-treated graphene.¹¹

The behaviour of T 2-LG is very similar to that of AB 2-LG. The only difference is that the T 2-LG is more susceptible to oxygen plasma and more defects are created compared with the number of defects in the AB 2-LG. This is then also reflected in the larger degree of fluorination of the T 2-LG as compared with the AB 2-LG, for the samples that were treated in the same way.

For bilayer graphene samples, it can be concluded that a smaller number of introduced defects (S2 and S3) by oxygen plasma results in a lower A_(D)/A_(G) ratio after fluorination. This result is in agreement with the stabilization of defects previously observed for defected graphene samples,²⁷ in which the lightly defected graphene samples showed a decrease in the D band width, increasing only when longer plasma treatments (t > 10 s) were applied. Defects are usually reactive sites, and thus it is generally assumed that a larger number of defects should facilitate the fluorination process. However, one has to consider also the stability of the defects. For 1-LG our experimental results suggest that for the samples oxidised to the first stage (1-LG, sample S2) we increase the reactivity of graphene but then the reactivity is reduced for the graphene oxidised to the second stage (1-LG sample S3) and finally it is and again increased for the graphene oxidised to the third stage (1-LG sample S4). We may assume that at the first stage new functional groups introduced by oxygen plasma will result in formation of charge inhomogeneities in graphene. These charge inhomogeneities are reactive sites but they can be occupied also by oxygen functional groups which are created during the further plasma treatment. If these sites are occupied, the reactivity of graphene should decrease. Introducing even more functional groups will lead probably to large deformation of graphene sheet and thus the graphene becomes more reactive if it is oxidised to the third stage. The 2-LG graphene is less reactive as compared to 1-LG. Hence after the points of charge inhomogeneities are ‘consumed’ it is more difficult to functionalize perfect parts of graphene. Therefore the reactivity drops more than in the case of 1-LG (both S2 and S3). When even more defects are finally created the structure becomes more deformed and higher reactivity is observed (S4).

The results of Raman spectroscopy investigations were further verified by X-ray photoelectron spectroscopy (XPS) and are summarized in Table 2. The F/C ratios were calculated assuming homogenous distribution of fluorine within the samples. It can be seen that the results are in qualitative agreement with Raman measurements. The fluorination procedure of pristine graphene provided material that possesses the least amount of fluorine, the defected samples S2 and S3 exhibited significantly higher amount of fluorine and the sample S4, the highest amount of fluorine. The spectrum of F 1s photoelectrons (Fig. 3a) indicated the presence of only one chemical state of fluorine atoms (binding energy of 687.5 ± 0.2 eV), which is in agreement with previous works.³⁵ The C 1s spectrum (Fig. 3b) showed a dominating contribution from sp²-hybridized carbon atoms (284.4 eV) with a less intense signal from sp³ hybridized atoms and weak components belonging to oxygen- and fluorine-containing functionalities.

Table 2 F/C atomic ratios obtained from XPS measurements

	S1	S2	S3	S4
F/C at ratio	0.017	0.031	0.030	0.037

---

Fig. 3 The XPS of the fluorinated graphene samples (S1–S4). (a) F 1s photoelectrons, (b) C 1s photoelectrons. The spectra in each plot are normalized to the same height.

In pristine graphene, it is expected that fluorine reacts directly with carbon network to form C–F bond. This reaction can be facilitated by presence of defects as the amount of local charge inhomogeneities will increase furthermore we may speculate that the exchange reaction of C–OH to form C–F bond takes place.³⁶

It was shown previously that fluorinated samples tend to release fluorine over time or at elevated temperature. Here, we probed the stability of the fluorinated samples over time.

Fig. 4 plots the A_(D)/A_(G) ratio for AB 2-LG, T 2-LG and 1-LG, as a function of time, in days. As shown in Fig. 2b and commented above, the initial A_(D)/A_(G) ratios for S1 and S4 in both 2-LG assemblies were higher than for samples S2 and S3, while in 1-LG, the two highest A_(D)/A_(G) ratios were found in S4 and S2, respectively. In other words the samples differed in the degree of fluorination and in the number and type of defects before the fluorination.

Fig. 4 Time dependence of A_(D)/A_(G) plotted for: (a) AB 2-LG, (b) T 2-LG and (c) 1-LG, in samples S1–S4.

In general, the behaviour of the samples is similar: there is slight development of the A_(D)/A_(G) ratio in the first few days but after about one week the A_(D)/A_(G) ratio is constant. In addition, the changes within the first week are not large. The A_(D)/A_(G) ratio of the 1-LG sample is slightly decreased immediately after fluorination and it is then increased to the same level (S2, S3) or even to a higher level (S1, S4) than observed immediately after fluorination.

In the case of the BLG, there is almost no change of the A_(D)/A_(G) ratio for the S2 and S3 samples even during the first week after fluorination. In the case of the S1 and S4 samples, there is a slight decrease of the A_(D)/A_(G) ratio, which is not followed by an increase.

The time-dependent measurements thus confirm higher stability of samples S2 and S3 even after fluorination. This is reasonable because the stable defects are ‘changed’ for fluorine only if the interaction of graphene with fluorine is strong. Samples S1 and S4 have more reactive sites, which can be changed by less strongly bonded fluorine. In the case of 1-LG with higher reactivity the vacancy is filled and disorder is increased (the A_(D)/A_(G) ratio increases) while in the case of the less reactive 2-LG the vacancy can be filled by reconstruction of the surface.

Conclusions

In the present study, we experimentally probed the influence of the number of defects on the fluorination reaction of graphene using Raman spectroscopy. The defects in graphene were created intentionally using oxygen plasma treatment. The samples were than fluorinated. To ensure reliability of the results, we measured Raman maps of the samples and compared the same selected areas in the samples after introducing defects and after fluorination. For 1-LG we did not observe significant changes in the fluorination rate with the different number of created defects. On the other hand, for 2-LG, we observed a lower degree of fluorination for the samples with low number of defects as compared with the as-prepared sample and the sample with a high number of defects.

The observed results were rationalized by two competitive processes. (1) It can be generally expected that the defects make graphene more reactive. (2) However, the reactivity of specific defects also plays a role, thus transforming less-stable parts of graphene (charge inhomogeneities) to more stable functional groups, reduces the graphene reactivity. However, at a certain stage graphene becomes deformed due to new functional groups and higher reactivity is observed.

Finally, we have also monitored the stability of the fluorinated samples over time showing that in all samples there is a slight development of the A_(D)/A_(G) ratio in the first few days, but after about one week the A_(D)/A_(G) ratio is constant.

Acknowledgements

The authors acknowledge the support of MSMT ERC-CZ project (LL 1301).

References

1. K. J. Jeon, Z. Lee, E. Pollak, L. Moreschini, A. Bostwick, C. M. Park, R. Mendelsberg, V. Radmilovic, R. Kostecki, T. J. Richardson and E. Rotenberg, ACS Nano, 2011, 5, 1042–1046.
2. K. K. Tadi, S. Pal and T. N. Narayanan, Sci. Rep., 2016, 6, 25221.
3. F. Withers, M. Dubois and A. K. Savchenko, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 073403.
4. W. Han, R. K. Kawakami, M. Gmitra and J. Fabian, Nat. Nanotechnol., 2014, 9, 794–807.
5. K.-I. Ho, C.-H. Huang, J.-H. Liao, W. Zhang, L.-J. Li, C.-S. Lai and C.-Y. Su, Sci. Rep., 2014, 4, 5893.
6. F. Withers, S. Russo, M. Dubois and M. F. Craciun, Nanoscale Res. Lett., 2011, 6, 526.
7. H. Y. Liu, Z. F. Hou, C. H. Hu, Y. Yang and Z. Z. Zhu, J. Phys. Chem. C, 2012, 116, 18193–18201.
8. Y. B. Zhang, T. T. Tang, C. Girit, Z. Hao, M. C. Martin, A. Zettl, M. F. Crommie, Y. R. Shen and F. Wang, Nature, 2009, 459, 820–823.
9. J. B. Oostinga, H. B. Heersche, X. L. Liu, A. F. Morpurgo and L. M. K. Vandersypen, Nat. Mater., 2008, 7, 151–157.
10. T. Ohta, A. Bostwick, T. Seyller, K. Horn and E. Rotenberg, Science, 2006, 313, 951–954.
11. J. Ek Weis, S. D. Costa, O. Frank, Z. Bastl and M. Kalbac, Chem.–Eur. J., 2014, 21, 1081.
12. J. Xiao, P. Meduri, H. H. Chen, Z. G. Wang, F. Gao, J. Z. Hu, J. Feng, M. Hu, S. Dai, S. Brown, J. L. Adcock, Z. Q. Deng, J. Liu, G. L. Graff, I. A. Aksay and J. G. Zhang, ChemSusChem, 2014, 7, 1295–1300.
13. B. Wang, J. J. Wang and J. Zhu, ACS Nano, 2014, 8, 1862–1870.
14. S. S. Lee, S. W. Jang, K. Park, E. C. Jang, J. Y. Kim, D. Neuhauser and S. Lee, J. Phys. Chem. C, 2013, 117, 5407–5415.
15. Z. M. Ao, Q. G. Jiang, S. Li, H. Liu, F. M. Peeters, S. Li and G. X. Wang, ACS Appl. Mater. Interfaces, 2015, 7, 19659–19665.
16. P. A. Denis and F. Iribarne, J. Phys. Chem. C, 2013, 117, 19048–19055.
17. L. J. Zhou, Z. F. Hou, L. M. Wu and Y. F. Zhang, J. Phys. Chem. C, 2014, 118, 28055–28062.
18. A. Zandiatashbar, G. H. Lee, S. J. An, S. Lee, N. Mathew, M. Terrones, T. Hayashi, C. R. Picu, J. Hone and N. Koratkar, Nat. Commun., 2014, 5, 3186.
19. O. Lehtinen, N. Vats, G. Algara-Siller, P. Knyrim and U. Kaiser, Nano Lett., 2015, 15, 235–241.
20. J. C. Meyer, C. Kisielowski, R. Erni, M. D. Rossell, M. F. Crommie and A. Zettl, Nano Lett., 2008, 8, 3582–3586.
21. F. Banhart, J. Kotakoski and A. V. Krasheninnikov, ACS Nano, 2011, 5, 26–41.
22. O. Leenaerts, H. Peelaers, A. D. Hernandez-Nieves, B. Partoens and F. M. Peeters, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 195436.
23. A. C. Ferrari and D. M. Basko, Nat. Nanotechnol., 2013, 8, 235–246.
24. A. Jorio, E. H. Martins Ferreira, L. G. Cancado, C. A. Achete and R. B. Capaz, in Physics and Applications of Graphene – Experiments, ed. S. Mikhailov, InTech, 2011, ch. 18, pp. 439–454.
25. A. Eckmann, A. Felten, A. Mishchenko, L. Britnell, R. Krupke, K. S. Novoselov and C. Casiraghi, Nano Lett., 2012, 12, 3925–3930.
26. L. G. Cancado, A. Jorio, E. H. M. Ferreira, F. Stavale, C. A. Achete, R. B. Capaz, M. V. O. Moutinho, A. Lombardo, T. S. Kulmala and A. C. Ferrari, Nano Lett., 2011, 11, 3190–3196.
27. S. D. Costa, J. Ek Weis, O. Frank and M. Kalbac, Carbon, 2016, 98, 592–598.
28. M. Kalbac, O. Frank and L. Kavan, Carbon, 2012, 50, 3682–3687.
29. A. Reina, X. T. Jia, J. Ho, D. Nezich, H. B. Son, V. Bulovic, M. S. Dresselhaus and J. Kong, Nano Lett., 2009, 9, 30–35.
30. D. A. Shirley, Phys. Rev. B: Solid State, 1972, 5, 4709.
31. J. H. Scofield, J. Electron Spectrosc. Relat. Phenom., 1976, 8, 129–137.
32. N. McEvoy, H. Nolan, N. A. Kumar, T. Hallam and G. S. Duesberg, Carbon, 2013, 54, 283–290.
33. A. Felten, A. Eckmann, J. J. Pireaux, R. Krupke and C. Casiraghi, Nanotechnology, 2013, 24, 355705.
34. T. P. Kaloni, Y. C. Cheng, R. Faccio and U. Schwingenschlogl, J. Mater. Chem., 2011, 21, 18284–18288.
35. P. W. Gong, Z. F. Wang, Z. J. Fan, W. Hong, Z. G. Yang, J. Q. Wang and S. R. Yang, Carbon, 2014, 72, 176–184.
36. P. W. Gong, J. Q. Wang, W. M. Sun, D. Wu, Z. F. Wang, Z. J. Fan, H. G. Wang, X. X. Han and S. R. Yang, Nanoscale, 2014, 6, 3316–3324.

---