Size fractionation of graphene oxide sheets assisted by circular flow and their graphene aerogels with size-dependent adsorption

Abstract

Lateral dimensions of graphene oxide (GO) sheets are acknowledged to critically influence the properties of macroscopic GO-based materials. However, restricted by current size fractionation technologies, it is still a great challenge to obtain GO sheets in a relatively narrow size distribution by a simple method. Here, we introduce a facile size fractionation of GO sheets assisted by circular flow. With the help of hydrogen gas (H₂) bubbles generated from a metallic replacement reaction, continuous turbulence forms a circular flow in a special tubular container to make it possible to fractionate different sized GO sheets. By using this process, crude GO (CGO) sheets can be facilely separated into three size ranges, namely LGO (>20 μm), MGO (2–20 μm) and SGO (<2 μm). Another advantage is that the separation approach can simultaneously remove residual graphite oxides. Furthermore, graphene aerogels made of LGO precursors (LGA) show better absorptivity than those of CGO, MGO and SGO precursors.

---

1. Introduction

Owing to its unique properties, graphene has enjoyed significant attention for application in many fields such as graphene membranes, graphene fibers and graphene aerogels.^1–6 But regular graphene is not enough to satisfy the application requirements, so it is necessary to obtain a suitable graphene in order to build materials possessing better properties. In recent years, lateral dimensions of graphene sheets are also acknowledged to be critical in influencing the performances of graphene materials.^7–9 For instance, large-size graphene sheets (>20 μm) are promising candidates for adsorbents^10,11 and conducting electrodes,¹² while small-size graphene sheets (<2 μm) mean with great favor for biomedical and optoelectronic functions as biosensors¹³ and solar cells.¹⁴ Medium-size graphene sheets (2–20 μm) also exhibit their outstanding mechanical performance in binder-free or composite fibers.^15,16

Graphene sheets have usually been prepared by reduction of graphene oxide (GO) sheets,¹⁷ while GO is a typical chemically exfoliated graphene yielded by use of strong oxidizing agents, whose basal planes and edges are decorated with epoxy, hydroxyl and carboxyl groups.¹⁸ Hence, GO sheets could easily be ruptured into small pieces with a wide size distribution when induced along the aligned epoxy groups by external stimulation.¹⁹ However, the narrow size distribution and corresponding mean size of GO sheets are of significance for the microstructures and properties of the GO-based materials for specific applications such as graphene aerogels as stated above. Moreover, residues graphite oxides are inevitably produced during oxidation and exfoliation of natural flake graphite but are difficult to be removed. Worse yet, the procedures in ingredients purification could aggravate structural defects growth in the basal planes of GO sheets, thus further breaking their pristine lateral dimensions randomly.^19–21 For these reasons, it is necessary to separate GO sheets into different narrow size distributions and to remove the impurities in a suitable method.

Fortunately, a series of work has been reported about size fractionation.^22–28 Chen et al.²³ presented the synthesis of GO sheets which could directly control size of GO by controlling the oxidation and exfoliation procedure. However, the mean size of obtained GO sheets was only less than 1 μm, which would greatly limit their applications. Additionally, Shi et al.²⁵ reported a size fractionation method based on pH-assisted sedimentation to separate GO sheets into two portions. But GO sheets as a kind of buffering agents was difficult to adjust pH values of the GO dispersion while dialysis process as a purification step was also costly and time-consuming. Kim et al.²⁶ found liquid crystal size selection of large-size GO flakes. In the report, large-size GO flakes tended to cause a high-density nematic phase in a stable environment, while small flakes spontaneously remained in a low-density isotropic phase for 1–2 weeks. So far, most fractionation processes have been used to separate crude GO sheets into merely two size distributions, so the application prospects of the GO potions seem to be limited. In recent years, Shi et al.²⁷ have successfully separated GO sheets into three size distributions via filtration using track-etched membranes. This method has the merits of additive-free, fast, and scalable, which is promising for GO size fractionation. The minor issue of their method is the special or maybe relatively expensive track-etched membranes. Zhang et al.²⁸ further separated GO sheets into four size distributions by the polar solvent-selective natural deposition method, which is good at size fractionation in a large scale, but the usage of large amount of organic solvents and relatively wide size distributions are the drawbacks.

There are disadvantages for most fractionation methods as mentioned above. In addition, huge workload has to be taken to remove residual graphite oxides during purification steps for the size fractionation, which bedevils many researchers in the field. To find a suitable purification approach for the size fractionation is still a challenge.

Here, we introduce a facile three-potion size separation method for crude GO (CGO) sheets assisted by circular flow, simultaneously integrating with removing the impurities residual graphite oxides. By this separation methods, CGO sheets are separated into three portions with relatively narrow size distributions: large sized GO (LGO), medium sized GO (MGO) and small sized GO (SGO). Based on these different sized GO precursors, graphene aerogels are prepared and the adsorption capacity of the aerogels are investigated.

2. Experimental section

2.1 Materials

Natural flake graphite (50 meshes) was purchased from Qingdao Jinrilai graphite. Co., Ltd (China). Concentrated sulfuric acid (H₂SO₄, 98%), hydrochloric acid (HCl, 35%), sodium nitrate (NaNO₃), hydrogen peroxide (H₂O₂, 10%) and potassium permanganate (KMnO₄) were all analytical reagents and supplied by Sinopharm Chemical Reagent Co., Ltd (China).

2.2 Preparation of CGO

According to Hummers' method,²⁹ CGO sheets were fabricated by oxidation of flake graphite (3.0 g) with KMnO₄ (9 g) and sodium nitrate (1.5 g) in concentrated H₂SO₄ (70 mL) at 0 °C for 8 h, at 38 °C for 30 min and 105 °C for 20 min continually, followed by dilution (H₂O, 270 mL) and reduction (H₂O₂, 2 mL). As a result, the solution containing CGO sheets and small-grained graphite oxides was obtained. FTIR spectra (Fig. S1†) indicate the existence of oxygenated functional groups such as carboxyl groups, hydroxyl and epoxide groups in CGO and in the following LGO, MGO and SGO sheets.

2.3 Size fractionation of CGO

Fig. 1a is schematic illustration of separation device for size fractionation of CGO. A glass tube was used as a main body of the fractionation device. A silicon sheet worked as an anode in the top of the device, while a metal needle tubing as cathode at the bottom connecting syringe pump for collecting different potions of GO sheets. A nylon sieve was equipped above the cone groove to prevent GO sheets from falling into the groove. Electric current was applied to both sides of the separation device. The above-mentioned solution containing CGO and graphite oxides was added into the tube for size fractionation after it was regulated with HCl and deionized water to keep CGO concentration (C_CGO) and hydrogen ions (H⁺) concentration (C_H-ion) in certain ranges. The related parameters of the device and the fractionation are shown in Table S1.†

Fig. 1 (a) Schematic illustration of separation device for size fractionation of CGO sheets; (b) digital images of CGO separation results after 4 h with C_CGO and C_H-ion at 0.5 mg mL⁻¹ and 0.1 mol L⁻¹ respectively.

As shown in Fig. 1a and b, LGO sheets were mainly agglomerated and precipitated at the bottom of the device. SGO sheets were slowly collected near the anode in the top. Circulating MGO sheets were continually kept in the circular flow in the middle of the tube. LGO sheets at the bottom were first pumped out by the syringe pump through the cathode tube, then the MGO and lastly SGO sheets. Meanwhile, the main impurities in the process, residual graphite oxides, gradually precipitated into the cone groove at the bottom of the tube and were removed.

2.4 Preparation of graphene aerogels

The pH value of GO aqueous dispersions (2 mg mL⁻¹) was adjusted to 3.5 measured by a pH meter (Sartorius, PB-10). Then, the different sized GO dispersions were added into sealed vessels, heated at 180 °C for 6 h and freeze-dried for 24 h to obtain final graphene aerogels. Graphene aerogels made of LGO, MGO, SGO and CGO precursors were respectively marked as LGA, MGA, SGA and CGA.

2.5 Characterization

Scanning electron microscopy (SEM) images was taken on a field-emission SEM system (Hitachi, S-4800). The X-ray diffraction (XRD) patterns were obtained on a XRD diffractometer (Panalytical, X'Per-Pro MPD) with a Cu-Kα radiation source. The composition of GO sheets was recorded by X-ray photoelectron energy spectra (XPS) spectrometer (Kratos, Axis Ultra HAS). Thermal gravimetric analysis (TGA) of aerogels with different sized GO precursors was carried out under nitrogen flow (TA, Discovery TGA). Structural defects of different sized GO sheets were obtained using a Raman microscope (Horiba, HR800) with a 514 nm laser source. Ultraviolet visible (UV-vis) spectra of different sized GO sheets were measured by UV-vis spectrometer (Varian, CARY50). Fourier-transform infrared (FTIR) spectra of different sized GO sheets were measured by FTIR spectrometer (Thermo, Nicolet6700). Resistances of the GO films prepared by casting in petri dishes and vacuum drying at 50 °C were measured in four-probe technique (JG, ST2263). Elementary composition of GO samples was determined by an elemental analyzer (Vario, EL III). Nitrogen adsorption–desorption isotherms of graphene aerogels were measured by surface area and porosity analyzer (Micromeritics, ASAP2020) at 77 K. The surface areas here refer in particular to BET surface areas given by the analyzer according to the theory of Brunauer, Emmett, and Teller.

3. Results and discussion

In an aqueous media, GO sheets are negatively charged because of the ionization of their carboxylic acid and phenolic hydroxyl groups mostly at their edges.³⁰ It suggests that formation of a stable GO dispersion is due to electrostatic repulsion against van der Waals interaction, aggregation and residual π-conjugated domain among GO sheets. It is known that the edge-to-area ratio of a GO sheet decreases with the increase of its lateral dimension. Thus, large GO flakes would favor aggregation and precipitation as the rising of C_H-ion and the concentration of multivalent cations in the aqueous solutions, compared with their small counterparts. According to aggregation kinetics of GO sheets in aqueous solutions, the GO dispersion could provide appropriate environment for cross-linking of GO sheets in bridging edges of GO sheets through chelating carboxylate groups (edge-to-edge/face mode) with existence of multivalent cations and high C_H-ion.³¹ Gao's studies also suggested that GO aggregates at low C_H-ion might dominantly exhibit a GO–water–GO sandwich-like structure (face-to-face mode) in the form of complexes consisting of parallel GO sheets and a large amount of water molecules.³²

According to above statements and our experimental phenomena, the schematic separation model is designed as shown in Fig. 2. By adjusting C_H-ion to high value in the experimental range, high C_H-ion prompt residual graphite oxides to rapidly precipitate because of their low charge-to-mass ratio. With the presence of multivalent cations and high C_H-ion, carboxylic groups in LGO sheets are rapidly protonated and multivalent cations link LGO sheets through interacting with the functional groups on LGO surfaces, particularly at the edges. H⁺ and other cations gradually gather around the cathode when external electric field is applied. Then, LGO sheets are also soon aggregated near the bottom of the tube in the edge-to-edge/face mode (Fig. 2b), and most of the LGO sheets precipitate in a relatively short time period at the beginning of the fractionation.

Fig. 2 (a) CGO separation mechanism assisted by circular flow; (b) proposed face-to-face mode forming aggregation by hydrogen bonding network and proposed edge-to-edge/face mode forming aggregation and precipitation by multivalent cations.

H₂ bubbles are mainly generated from metallic replacement reaction around the cathode, moving near the side wall of the tube and causing continuous turbulence in the dispersion, which leads to a circular flow in the device. Simultaneously, the other small part of H₂ bubbles are also generated by electrochemical reaction in the cathode. Then, the circular flow would prevent the sheets in the flow from aggregation especially in the middle part of the dispersion. SGO sheets have a higher charge-to-mass ratio than large ones at the same C_H-ion due to high-density carboxyl groups at the edges. Thus, SGO sheets, benefitting from the circular flow in the CGO dispersion, are attracted to the anode in the top of the device because of many ionized carboxylic acid groups charged negatively. The face-to-face mode (Fig. 2a) is the dominant interaction pattern of SGO aggregation with electrostatic repulsion among SGO sheets. MGO sheets have to stay in the circular flow because of low charge-mass ratio without interrupting the LGO aggregation at the bottom and the impurities at the groove of the device.

There are characterized peaks at 2θ = 10.8°, 11.3°, 10.9° and 9.9° for CGO, LGO, MGO and SGO samples respectively in the XRD diffraction pattern as shown in Fig. S2.† Particularly, the spectrum of CGO has another two small peaks at 21.2° and 26.5°, which indicates the existence of graphite flakes, whereas the spectra of LGO, MGO and SGO have no the peaks, suggesting the impurity-removing is effective in the fractionation process.

Fig. 3 shows SEM images and corresponding histograms of size distributions of CGO, LGO, MGO and SGO sheets after size fractionation. Fig. S3† shows details for determination of the size of a single GO sheet. According to the monolayer structure of GO, the weight of a GO sheet is directly proportional to its area value. Hence, the weight percentage of GO size distribution corresponds to its area percentage. From the image of the CGO sheets, it can be found that the sizes of the CGO sheets are random. According to the size distribution histograms, the measured sizes of CGO sheets are widely distributed from 0 to above 100 μm. In contrast, the sizes of LGO, MGO and SGO sheets are all kept in relatively narrow distributions. Among them, SGO sheets own the narrowest size distribution, and the lateral dimensions of SGO sheets are almost under 2 μm. Meanwhile, the weight percentage of MGO dimensions between 2 and 20 μm is more than 93 wt%. The weight percentage of LGO dimensions over 20 μm is more than 82 wt%. These results indicate that the obvious effect of separating LGO, MGO, SGO sheets has been attained through the circulating flow process.

Fig. 3 SEM images (left) and corresponding histograms (right) of size distributions of CGO, LGO, MGO and SGO sheets prepared by spin-coating process on a silicon substrate.

In addition, the fractionation results can also be confirmed by the resistances of separated GO films. Fig. 4 shows the resistances of LGO, MGO and SGO films in the thickness direction of these films. The resistances of the different films have clearly different values. In comparison to the MGO and SGO films, LGO films have the lowest sheet resistance at the same thickness, for example, at the thickness of 40 μm, the sheet resistances of LGO, MGO and SGO are 1.21, 1.48 and 2.5 MΩ m⁻², respectively. It is clear that LGO has the lowest resistance because LGO is easy to overlap and to contact each other.

Fig. 4 Resistances of LGO, MGO and SGO films with different film thickness.

XPS measurements are performed to analyse composition of different GO specimens. As shown in Fig. 5a–c, three peaks or shoulders exist for each XPS spectrum. Through peak fitting for the raw XPS spectra, there are clearly three independent peaks at 284.6 eV, 286.6 eV and 288.3–289 eV for each GO specimen, which corresponds to C–C, C–O and C=O, respectively. The peak intensity ratio of intact carbon (C–C) to oxygenated carbon atoms (C–O and C=O) can reflect the percentage of the oxygenated functional groups, and the larger of the ratio the lower of the percentage. Obviously as shown in Fig. 5a–c, this value decreases in the following sequence: LGO > MGO > SGO, that is, the percentage of the oxygenated functional groups is in the following sequence: LGO < MGO < SGO. Element analysis shows that the contents of carbon (C) and oxygen (O) are different in CGO LGO, MGO and SGO sheets in Fig. 6. The values of C/O for CGO, LGO, MGO and SGO samples correspond to 0.86, 0.94, 0.83 and 0.72, respectively, which is in accordance with the XPS results mentioned above.

Fig. 5 C 1s XPS spectra of (a) LGO, (b) MGO and (c) SGO specimens prepared by vacuum freeze-drying; (d) UV-vis spectra of LGO, MGO and SGO aqueous dispersions at 0.05 mg mL⁻¹.

Fig. 6 Element analysis of CGO, LGO, MGO and SGO samples for the contents of C and O elements in pie charts.

Among structural defects, oxygenated functional groups, C=O and C–O are mainly located at the edges of GO sheets; thus the percentage of these two groups reflects the edge-to-area ratio of GO sheets. Actually, the relative percentage of C=O and C–O consistently declines with increasing the GO lateral dimension, which suggests that LGO flakes own fewer structural defects than small sized flakes. Raman spectra (Fig. S4†) show D peaks at 1350 cm⁻¹ and G peaks at 1590 cm⁻¹, demonstrating lattice distortions and structural defects of all GO specimens. I_D/I_G values of LGO, MGO, SGO and CGO specimens are 0.866, 1.013, 1.137 and 1.065, respectively, and that of LGO sheets is the lowest among these specimens, indicating fewest defects for the LGO sheet.

In Fig. 5d, UV-vis spectral results show that LGO, MGO and SGO dispersions all have an absorption peak at about 230 nm and a shoulder peak at around 305 nm, which agrees well with as reported for GO materials.¹⁷ The peak area corresponds to the degree of π → π* transitions (conjugation) of polyaromatic C=C groups, while the shoulder area corresponds to the degree of the n → π* electron transitions of C=O bonds. Compared with the peak absorptions of MGO and SGO sheets, LGO sheets have the most retention of aromatic rings in the basal plane, suggesting that LGO sheets which are obtained by this separation approach possess fewer defects in their microscopic structure.

As an application of the fractionated GO, the graphene aerogels made of them have different thermal stability. TGA analysis of graphene aerogels (Fig. S5†) shows that LGA has largest weight loss between 50 and 150 °C among the graphene aerogels due to loss of absorbed surface water, whereas the major and steep weight loss stage is between 150 and 320 °C, which is likely to be attributed to the pyrolysis of unstable functional groups. Between 320 and 1000 °C, a slow mass loss is due to the removal of stable oxygenated groups, and the residual weight percentages at 1000 °C for CGA, LGA, MGA and SGA aerogels correspond to 31.8 wt%, 38.8 wt%, 35.9 wt% and 23.5 wt%, respectively. It is suggested that except good water adsorption for LGA, the precursor LGO used for preparing LGA has the fewest defects, which agrees with the statement above.

Graphene aerogels composed by these GO precursors not only have a large amount of active carbon atoms to react with strong oxidizing agents, but also own interconnected pores due to 3D porous architectures. In order to observe trapping capacity of different graphene aerogels, active carbon atoms are removed before adsorption test of strong oxidizing agents. Fig. 7 shows the adsorption capacity of LGA, MGA, SGA and CGA distinguished by the change of colour in potassium permanganate solution. The pink of the potassium permanganate solutions slowly fades by adsorption for all graphene aerogels. The extent of colour fading is different among the GO precursors, particularly, the colour of LGA fades most distinctly, indicating LGA is able to absorb most permanganate ions. To further verify the results mentioned above, nitrogen adsorption–desorption isotherms of CGA, LGA, MGA, SGA samples are shown in Fig. S6.† The BET surface areas (m² g⁻¹) corresponding to the data in Fig. S6† of different graphene aerogels decreases in the following sequence: LGA (86.4) > MGA (84.3) > SGA (56.3) > CGA (2.4). According to the values, the nitrogen adsorption–desorption isotherms again illustrate LGA have the strongest adsorption capacity, particularly in the low-pressure stage. As shown in Fig. 8, all the aerogels fill with pores, but their structure and morphology are quite different. It seems that LGA has larger sized pores than those of the other GO aerogels, the pores run through each other and the pore walls are particularly thin, leading to strong adsorbability for potassium permanganate. MGA also has many pores, but the pore walls are thicker than LGA and the pore dimensions are smaller. However, SGA seems to be fragile and have dense structure. Only in large magnification can the small pores be seen. Although CGA has some pores, the pore walls looks some thick and most pores are closed and dead-end. Obviously, the structure and morphology basically correspond to the adsorption capacity as shown in Fig. 7.

Fig. 7 Digital photographs of adsorbability for permanganate ions of LGA, MGA, SGA and CGA at equal weight in different standing time.

Fig. 8 (a) Digital photographs of LGA, MGA, SGA and CGA, and (b) corresponding SEM images of LGA, MGA, SGA and CGA in smaller magnification and (c) in larger magnification.

4. Conclusions

We present a facile method for size fractionation of CGO sheets assisted by circular flow. Notably, CGO sheets can be fractionated into three portions with relatively narrow size distributions, namely LGO (>20 μm), MGO (2–20 μm) and SGO (<2 μm), which simultaneously integrates the purification steps of removing residual graphite oxides. There are fewer oxygenated groups on the LGO surface than MGO, SGO and CGO, so LGO surface has fewer defects. In addition, the graphene aerogels composed of the large lateral dimension of GO precursors possess stronger adsorption capacity for strong oxidizing agents such as permanganate ions than the graphene aerogels made of the other sized GO precursors.

Acknowledgements

This research is supported by a Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions.

References

1. D. A. Dikin, S. Stankovich, E. J. Zimney, R. D. Piner, G. H. Dommett, G. Evmenenko, S. T. Nguyen and R. S. Ruoff, Nature, 2007, 448, 457–460.
2. K. S. Kim, Y. Zhao, H. Jang, S. Y. Lee, J. M. Kim, K. S. Kim, J. H. Ahn, P. Kim, J. Y. Choi and B. H. Hong, Nature, 2009, 457, 706–710.
3. Z. S. Wu, S. Yang, Y. Sun, K. Parvez, X. Feng and K. Mullen, J. Am. Chem. Soc., 2012, 134, 9082–9085.
4. H. P. Cong, X. C. Ren, P. Wang and S. H. Yu, Sci. Rep., 2012, 2, 613.
5. H. Hu, Z. Zhao, Y. Gogotsi and J. Qiu, Environ. Sci. Technol. Lett., 2014, 1, 214–220.
6. H. Hu, Z. Zhao, W. Wan, Y. Gogotsi and J. Qiu, Adv. Mater., 2013, 25, 2219–2223.
7. M. J. Deka, U. Baruah and D. Chowdhury, Mater. Chem. Phys., 2015, 163, 236–244.
8. S. Chatterjee, F. Nafezarefi, N. H. Tai, L. Schlagenhauf, F. A. Nüesch and B. T. T. Chu, Carbon, 2012, 50, 5380–5386.
9. S. Liu, M. Hu, T. H. Zeng, R. Wu, R. Jiang, J. Wei, L. Wang, J. Kong and Y. Chen, Langmuir, 2012, 28, 12364–12372.
10. S. Kabiri, D. N. H. Tran, T. Altalhi and D. Losic, Carbon, 2014, 80, 523–533.
11. X. Mi, G. Huang, W. Xie, W. Wang, Y. Liu and J. Gao, Carbon, 2012, 50, 4856–4864.
12. J. P. Zhao, S. F. Pei, W. C. Ren, L. B. Gao and H. M. Cheng, ACS Nano, 2010, 4, 5245–5252.
13. H. F. Dong, W. C. Gao, F. Yan, H. X. Ji and H. X. Ju, Anal. Chem., 2010, 82, 5511–5517.
14. C. X. Guo, H. B. Yang, Z. M. Sheng, Z. S. Lu, Q. L. Song and C. M. Li, Angew. Chem., 2010, 49, 3014–3017.
15. M. Z. Seyedin, J. M. Razal, P. C. Innis, R. Jalili and G. G. Wallace, Adv. Funct. Mater., 2015, 25, 94–104.
16. R. Jalili, S. H. Aboutalebi, D. Esrafilzadeh, R. L. Shepherd, J. Chen, S. Aminorroaya-Yamini, K. Konstantinov, A. I. Minett, J. M. Razal and G. G. Wallace, Adv. Funct. Mater., 2013, 23, 5345–5354.
17. D. C. Marcano, D. V. Kosynkin, J. M. Berlin, A. Sinitskii, Z. Z. Sun, A. Slesarev, L. B. Alemany, W. Lu and J. M. Tour, ACS Nano, 2010, 4, 4806–4814.
18. T. Szabo, O. Berkesi, P. Forgo, K. Josepovits, Y. Sanakis, D. Petridis and I. Dekany, Chem. Mater., 2006, 18, 2740–2749.
19. F. Banhart, J. Kotakoski and A. V. Krasheninnikov, ACS Nano, 2011, 5, 26–41.
20. S. Fujii and T. Enoki, J. Am. Chem. Soc., 2010, 132, 10034–10041.
21. Z. Y. Li, W. H. Zhang, Y. Luo, J. L. Yang and J. G. Hou, J. Am. Chem. Soc., 2009, 131, 6320–6321.
22. S. Y. Pan and I. A. Aksay, ACS Nano, 2011, 5, 4073–4083.
23. L. Zhang, J. Liang, Y. Huang, Y. Ma, Y. Wang and Y. Chen, Carbon, 2009, 47, 3365–3368.
24. X. Zhou and Z. Liu, Chem. Commun., 2010, 46, 2611–2613.
25. X. Wang, H. Bai and G. Shi, J. Am. Chem. Soc., 2011, 133, 6338–6342.
26. K. E. Lee, J. E. Kim, U. N. Maiti, J. Lim, J. O. Hwang, J. Shim, J. J. Oh, T. Yun and S. O. Kim, ACS Nano, 2014, 8, 9073–9080.
27. J. Chen, Y. Li, L. Huang, N. Jia, C. Li and G. Shi, Adv. Mater., 2015, 27, 3654–3660.
28. W. Zhang, X. Zou, H. Li, J. Hou, J. Zhao, J. Lan, B. Feng and S. Liu, RSC Adv., 2015, 5, 146–152.
29. W. S. Hummers and R. E. Offeman, J. Am. Chem. Soc., 1958, 80, 1339.
30. D. R. Dreyer, S. Park, C. W. Bielawski and R. S. Ruoff, Chem. Soc. Rev., 2010, 39, 228–240.
31. D. Deng, X. Jiang, L. Yang, X. Hou and C. Zheng, Anal. Chem., 2014, 86, 758–765.
32. L. Wu, L. Liu, B. Gao, R. Munoz-Carpena, M. Zhang, H. Chen, Z. Zhou and H. Wang, Langmuir, 2013, 29, 15174–15181.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental data, characterization details, size determination and circular-flow details of size fractionation. See DOI: 10.1039/c6ra16363g

---