SEARCH

SEARCH BY CITATION

Keywords:

  • carbon;
  • graphene;
  • nanoribbons;
  • nanotechnology;
  • synthetic methods

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Physical Properties and Applications of GNRs
  5. 3. Methods of GNR Synthesis
  6. 4. Summary and Perspective
  7. Acknowledgements

Graphene, the thinnest two-dimensional material in nature, has abundant distinctive properties, such as ultrahigh carrier mobility, superior thermal conductivity, very high surface-to-volume ratio, anomalous quantum Hall effect, and so on. Laterally confined, thin, and long strips of graphene, namely, graphene nanoribbons (GNRs), can open the bandgap in the semimetal and give it the potential to replace silicon in future electronics. Great efforts are devoted to achieving high-quality GNRs with narrow widths and smooth edges. This minireview reports the latest progress in experimental and theoretical studies on GNR synthesis. Different methods of GNR synthesis—unzipping of carbon nanotubes (CNTs), cutting of graphene, and the direct synthesis of GNRs—are discussed, and their advantages and disadvantages are compared in detail. Current challenges and the prospects in this rapidly developing field are also addressed.


1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Physical Properties and Applications of GNRs
  5. 3. Methods of GNR Synthesis
  6. 4. Summary and Perspective
  7. Acknowledgements

Graphene, a novel allotrope of carbon with a single atomic layer, has unique electronic properties, as first reported by Geim, Novoselov, and co-workers in 2004.1 Soon after, great attention was devoted to the exploring of the fundamentals and applications of this unique two-dimensional (2D) material. Fundamentally, owing to two nonequivalent atomic sublattices, the band structure of graphene has two Dirac conical points per Brillouin zone. The conduction band and valence band touch each other at each Dirac point, and an unusual linear energy dispersion emerges in the Dirac cone.2 The linear energy dispersion at the Dirac point leads to massless carrier behavior3 and extraordinary quantum properties, such as the anomalous quantum Hall effect and absence of localization.47 On the other hand, the single-atom-thick crystal has many novel electronic, physical and chemical properties:

  • 1
    exceptionally high carrier mobility at room temperature
    (>200 000 cm2 V−1 s−1);810
  • 2
    superior thermal conductivity (3000–5000 W m−1 K−1), which is as good as those of diamond and carbon nanotubes;11, 12
  • 3
    extremely high modulus (≈1 TPa) and tensile strength
    (≈100 GPa);13 and
  • 4
    high transparency to incident light over a broad wavelength range (97.7 %).14, 15

These intriguing properties of graphene have inspired exploration of its applications in electronic, photonic devices, sensors and as strong materials.

Because of its superior electronic, thermal, and mechanical properties, as silicon technology approaches its quantum limit, graphene is broadly considered to be the most promising candidate to replace Si in future electronic devices. However, as a semimetal, the zero bandgap of graphene normally gives graphene-based field effect transistors (FET) a low on/off ratio and hinders the dream of graphene microelectronics. Great efforts have been dedicated to opening a bandgap in graphene′s density of states (DOS). Among the most used techniques, cutting graphene into narrow nanoribbons to open the bandgap through the lateral quantum confinement effect is one of the most promising means of achieving this target (Figure 1 a).16, 17 More specifically, if the width of the GNR is less than 10 nm, the gap would be large enough to synthesize graphene FETs with on/off ratios of about 107 at room temperature.18, 19

thumbnail image

Figure 1. a) Graphene and graphene nanoribbons with armchair (transverse strip) and zigzag edges (longitudinal strip). The ribbon width is denoted by the number of dimer lines or zigzag chains. b) Variation of bandgap as a function of width for AC-GNRs [Reprinted with permission from ref. 16, Copyright (2006) by the American Physical Society]. c) Y-type tri-wing armchair-edge graphene nanoribbon and relationship between bandgap and wing width [Reprinted with permission from ref. 50, Copyright (2010) American Chemical Society].

Download figure to PowerPoint

To achieve large-scale production of high-quality GNRs with narrow widths, numerous fabrication strategies, including both top-down and bottom-up schemes, have been proposed, such as lithographic patterning followed by plasma etching of graphene,20, 21 sonochemical breaking of chemically derived graphene,18 metal-catalyzed22, 23 or oxidation cutting of graphene,24 direct chemical vapor deposition synthesis (CVD),25 chemical synthesis,26, 27 and unzipping of carbon nanotubes.28, 29

In this Minireview, we focus on recent progress in the synthesis of GNRs by different techniques, especially longitudinal unzipping of CNTs. The mechanical, electronic, and magnetic properties and edge reconstruction of GNRs are briefly summarized as well. The advantages and disadvantages of the methods of GNR synthesis, current challenges, and future perspectives are also discussed.

2. Physical Properties and Applications of GNRs

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Physical Properties and Applications of GNRs
  5. 3. Methods of GNR Synthesis
  6. 4. Summary and Perspective
  7. Acknowledgements

When a 2D graphene sheet is cut into 1D GNRs, depending on the orientation of cutting, GNRs with various edge structures can be obtained, namely, armchair- or zigzag-edged GNRs (denoted AC-GNRs or ZZ-GNRs, respectively; Figure 1 a) and chiral ones. The bandgap of a GNR is essentially governed by the ribbon width and the edge configuration.16, 17 Besides, the carrier mobility and magnetic properties of edge atoms also depend on the specific edge configuration.

By employing quantum approaches, Louie and co-workers demonstrated that all GNRs with homogeneous AC or ZZ edges have energy gaps which decrease nearly linearly with increasing GNR width.16 The bandgaps of AC-GNRs oscillate with a periodicity of three, and thus the AC-GNRs can be classified into three subgroups with bandgap hierarchies of ΔN=3p+1N=3pN=3p+2 (where Δ denotes the bandgap, N denotes the width of the GNR, and p is an integer (Figure 1 a,b).16, 17 In contrast to AC-GNRs, whose ground state is nonmagnetic, ZZ-GNRs have an antiferromagnetic (AFM) ground state. Each zigzag edge of a ZZ-GNR is ferromagnetic (FM), but the two edges are antiferromagnetically coupled.16, 30, 31 For most chiral GNRs, the bandgap oscillations vanish quickly as a function of the chiral angle.17 These studies show that careful design of both the width and the edge configuration of the GNR is crucial for achieving high-performance GNRs.

A tunable bandgap is highly desirable for designing and fabricating electronic devices. The presence of edges and the finite widths of GNRs provides additional degrees of freedom to control the electronic properties of GNRs, for example, by tailoring the ribbon width8, 18 or edge functionalization.32 Moreover, applying external electric fields,30 doping,3335 functionalization,36, 37 and edge modification38, 39 can even turn ZZ-GNRs into half-metallic ferromagnets, in which the DOS of one spin orientation is metallic and that of another spin orientation is semiconducting. Therefore, application of ZZ-GNRs in spintronics, such as spin filters or spin logic gates, can be anticipated.

The predicated bandgap opening and the relationship between gap and ribbon width were confirmed experimentally.18, 19, 40 The on/off ratio of sub-10 nm GNR FETs reached 107.18, 19 In agreement with theoretical predictions, low-temperature (≈1.6 K) electrical transport measurements on GNR-based FETs showed that the bandgap is indeed inversely proportional to the width of the GNR.20 However, no obvious dependence of the electronic transport properties of the GNRs on orientation was observed experimentally.20 Some theoretical simulations on nonideal GNRs revealed that this inconsistency is mainly attributable to the enhanced localization induced by the edge roughness or reconstruction in real GNR-based devices.4144 This demonstrates the importance of controlling edge smoothness in the fabrication of high-performance GNR devices. Besides, experimental studies also indicated that impurities, doping, and gate material exert important influences on the transport performance of GNR FETs.20, 21, 45

Despite intriguing electronic, magnetic, and optical properties, GNRs are not mechanically robust against in-plane compression and can be easily twisted or buckled because of the very short persistence length.4649 To overcome this shortcoming, new 1D Y-type graphene nanostructures, namely, tri-wing graphene (TWG) nanoribbons (Figure 1 c) were designed.50 This not only significantly improves the mechanical and thermal stability of graphene but also results in tunable electronic and magnetic properties.51 For example, each wing of TWG has independent electronic properties, and zigzag TWGs are stable ferromagnets with large magnetic moment.50

3. Methods of GNR Synthesis

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Physical Properties and Applications of GNRs
  5. 3. Methods of GNR Synthesis
  6. 4. Summary and Perspective
  7. Acknowledgements

Synthesis of high-quality GNRs with smooth edges and narrow and well-defined widths is crucial and a prerequisite for the development of GNR-based electronic and spintronic applications. Many methods, including lithographic patterning followed by plasma etching, sonochemical breaking of chemically derived graphene, metal-catalyzed or oxidation cutting of graphene, CVD, direct chemical synthesis, and unzipping of carbon nanotubes, have been developed. Details of experimental studies on the synthesis of GNRs are summarized in Table 1.

Table 1. Summary of recent experimental methods for GNR synthesis.[a]
MethodTEnvironmentYieldWidth [nm]Edge qualityBandgapFET on/off ratioRefs.
  1. [a] N/A: not available.

Plasma etching of CNTsN/ACNTs partially embedded in polymer filmmedium (20 %)10–20highsmall≈1028
Chemical attack on CNTs≈330 KH2SO4/KMnO4 solutionnearly 100 %100–500lowmetallicN/A29, 52
Intercalation and exfoliation of CNTs≈300–500 Ksolutionhigh (60 %)100–250lowN/AN/A53, 54
Metal-catalyzed cutting of CNTs≈1100 KSi substratelow (5 %)15–40highN/AN/A61, 62
Sonochemical unzipping of CNTsN/Asolutionlow (2 %)10–30high10–15 meVN/A55, 56
Laser irradiation of CNTslaser energy (≈200–350 mJ)substratehigh (60 %)60-160N/AN/AN/A57
Electrochemical unzipping of CNTsN/Asolutiongood70–100highN/AN/A58
Hydrogen treatment and annealing of CNTs∼670–820 Ksubstrate with Fe catalystN/AN/AN/AN/AN/A63
Unzipping functionalized CNTs by STM tipsN/AsubstrateN/AN/AN/AN/AN/A64
Electrical unwrapping of CNTs by TEM≈3000 KsubstrateN/A∼45lowN/AN/A65
Patterning and etching of grapheneN/ASi/SiO2 substrateshigh6–100low0.1–0.5 eV1.7–16020, 21, 76, 77
Sonochemical breaking of grapheneN/Asolutionlow<10–50high>0.1 eV (<10 nm)>105 (<10 nm)18, 19, 78
Metal-catalyzed cutting of graphene1200–1300 KSi/SiO2 substratesvery low≈10–15highN/AN/A22, 23, 7983
Oxidation cutting of graphene300–350 Ksolution0    24, 84
CVD1000–1700 Ktemplated substratehigh20–300low≈0 V10–10425, 85, 86
Chemical synthesis500–700 Ksolution or Au substrateN/A0.18–0.25 nmhigh≈1.6 eVN/A26, 27

3.1. Experimental Studies on Cutting CNTs into GNRs

In principle, a single-walled carbon nanotube (SWCNT) can be viewed as a folded or zipped GNR (Figure 2 a). It is thus natural to seek the reverse process, that is, unzipping SWCNTs to synthesize GNRs. Many attempts have been dedicated to synthesizing GNRs experimentally, including plasma etching,28 chemical attack,29, 52 intercalation and exfoliation,53, 54 sonochemical unzipping,55, 56 laser irradiation,57 electrochemical unzipping,58 catalytic cutting under microwave radiation59, 60 or by transition metal particles (e.g. Co, Ni, or Fe),61, 62 hydrogen treatment,63 in situ STM manipulation64 and electrical unwrapping of CNTs.65

thumbnail image

Figure 2. a) SWNT and graphene nanoribbon (GNR). Different techniques for unzipping of CNTs: b) plasma etching of partially embedded CNTs;28 c) longitudinal cutting of CNTs by chemical attack;29 d) intercalation of alkali metal atoms followed by exfoliation of CNTs;53 and e) metal particle catalyzed cutting of CNTs.61 f) Final graphene nanoribbon.

Download figure to PowerPoint

3.1.1. Plasma Etching

In 2009, Dai and co-workers unzipped multiwalled nanotubes (MWCNTs) to GNRs by Ar plasma etching.28 The MWCNTs were first embedded in a poly(methyl methacrylate), PMMA, layer, and then the PMMA/MWNT film was peeled off in KOH solution and a narrow strip of the outmost side wall of partially embedded MWCNTs was exposed to a 10 W Ar plasma. The top side walls of MWCNTs were etched faster than other areas protected by PMMA, and this led to longitudinal unzipping of the MWCNTs (Figure 2 b). This route can be used to produce single-, bi-, and multilayer GNRs and GNRs with inner CNT cores, depending on the diameter, the number of walls of the MWCNT, and the etching time. The synthesized GNRs normally have smooth edges, narrow width distributions (10–20 nm), and semiconducting behavior.28

3.1.2. Chemical Attack

Tour and co-workers synthesized GNRs by lengthwise cutting of MWCNTs by a simple, efficient, and scalable oxidation method.29 The MWCNTs were suspended in concentrated sulfuric acid and treated with KMnO4 to break the C[BOND]C bonds along the axial direction of the CNTs (Figure 2 c). The synthesized GNRs are up to 4 μm long with widths of 100–500 nm and thicknesses of 1–30 graphene layers. The electrical conductivity of these GNRs is poor, owing to edge attachment of many oxygen-containing chemical groups, and most of the GNRs show metallic behavior.29

3.1.3. Intercalation and Exfoliation

Vega-Cantu and co-workers synthesized GNRs by longitudinally unzipping MWCNTs through intercalation of lithium and ammonia followed by exfoliation53 (Figure 2 d). The final products consist of multilayered GNRs, partially opened MWCNTs, and graphene flakes. Although this method is not very promising for high-yield GNR synthesis, it provides a new possibility to realize longitudinal unzipping of CNTs.

3.1.4. Metal-Catalyzed Cutting

Terrones and co-workers and Srivastava and co-workers both exploited transition metal particles (e.g. Co, Ni, or Cu) as chemical scissors to cut MWCNTs.61, 62 In the cutting procedure, the metal particles serve as catalysts to break H[BOND]H and C[BOND]C bonds and as solvents for etched C atoms. These processes do not involve any aggressive chemical treatment, and thus smooth graphitic edges can be easily achieved (Figure 2 e).

3.2. Theoretical Studies on Unzipping CNTs to GNRs

Quite a few ab initio calculations have been dedicated to understand the unzipping mechanism of CNTs at the atomic level. Seminario et al.66 explored oxidation unzipping of armchair SWCNTs and found that unzipping started with potassium permanganate attacking, stretching, and breaking of one of the C[BOND]C bonds perpendicular to the CNT axis. The resulting defect weakens the neighboring parallel C[BOND]C bonds longitudinally, making them more easily attacked. Initial attack took place more easily in the middle than at the ends of the CNTs. Zhao et al. considered oxidative longitudinal unzipping of SWCNTs with different diameters and chiralities.67 Their calculations showed that the highly strained C[BOND]C bond that has a maximum angle to the tube axis breaks first. Then, subsequent breakage of the adjacent C[BOND]C bond parallel to the broken one is barrierless, and gives rise to continuous unzipping of CNTs to GNRs.

Recently, Guo and co-workers demonstrated that oxygen atoms are favorably adsorbed on small-diameter CNT walls to form unzipped C[BOND]O[BOND]C epoxy chains along a direction with minimum angle to the tube axis (Figure 3).6870 They further considered that the highly curved sidewalls of deformed armchair SWCNTs are the energetically most favorable areas to adsorb O atoms to form the unzipped C[BOND]O[BOND]C epoxy chains and further oxidation would break the epoxy groups into carbonyl groups, leading to formation of bilayer ZZ-GNRs.71

thumbnail image

Figure 3. Theoretical studies on mechanisms in oxidation cutting of CNTs. a) Three possible directions for oxygen atoms adsorbed on the SWCNT wall.68 b) Energetically favorable directions of formed epoxy lines on armchair, chiral, and zigzag CNTs.

Download figure to PowerPoint

Other theoretical models, such as unzipping of hydroxyl group saturated and deformed nanotubes under chemical attack, annealing hydrogenated CNTs, and hydrogen loading at open CNT ends to unzip CNTs into GNRs, were also proposed.7274 Some consensuses can be found in these theoretical models, that is, the closer to the perpendicular direction of the CNT axis, the easier the attack on C[BOND]C bonds of CNTs; the smaller the CNT chiral angle (from θ=30° for AC-CNTs to θ=0° for ZZ-CNTs), the harder the unzipping of CNTs; the smaller the CNT diameter, the easier the unzipping of CNTs.

Recently, we proposed a unique method of metal-catalyzed unzipping of SWCNTs to synthesize narrow GNRs in H2 gas.75 Our ab initio calculations showed that the energy barrier for unzipping a (5,5) SWCNT in H2 gas is as high as 3.11 eV (Figure 4 a), which means that such a reaction can not happen under normal conditions. However, the barrier could be drastically reduced to 1.16 eV with the assistance of a single Cu atom (Figure 4 b). Therefore, such metal-catalyzed unzipping of SWCNTs could be implemented at moderate temperature (≈200–300 °C). This method is feasible for most transition metals, for example, Mn, Fe, Co, Ni, Pd, and Pt, and SWCNTs with different diameters, lengths, and chiralities. The transition metal atom plays two important roles in the unzipping procedure: 1) lowering the dissociation barrier of the H2 molecule (Figure 4 b) and 2) catalyzing C[BOND]C bond breakage. Besides, H termination of edge C atoms along the cutting channel is crucial for driving the movement of the catalyst atom forward. Eventually, with the assistance of a single metal atom, the SWCNT can be catalytically cut into a long and narrow ZZ-GNR (Figure 4 c). This method opens a door to synthesizing high-quality GNRs at low temperature, and if it were achieved experimentally, it would be major breakthrough for electronic/spintronic device fabrication.

thumbnail image

Figure 4. Optimized reaction path of metal-catalyzed cutting of CNTs from first-principles calculations. The reaction barriers of unzipping of (5,5) SWCNT without (a) and with (b) Cu atom in H2 gas. Inset of (b) shows the reaction barrier of Cu-catalyzed dissociation of the H2 molecule. c) Partially unzipped SWCNT (adapted form ref. 75).

Download figure to PowerPoint

3.3. Other Methods for Synthesis of GNRs

3.3.1. Lithographic Patterning and Plasma Etching of Graphene

Kim and co-workers synthesized GNRs by employing the electron-beam lithography and etching technique.20 Bulk graphene flakes were deposited onto Si/SiO2 substrates and patterned, and then GNRs with various widths were obtained after removing the unprotected graphene area by oxygen-plasma treatment (Figure 5 a). Electronic measurements confirmed the theoretical predications that the GNR bandgap is inversely proportional to the ribbon width. However, the fact that dependence of the bandgap on crystallographic direction was not observed indicates that this method can not control the GNR edge with atomic precision. Avouris et al. also carried out similar experiments to fabricate GNRs21 with edge roughnesses mostly around 1–3 nm and widths down to 10–15 nm. By employing silicon nanowires as etching masks in the lithography process, Huang et al. produced GNRs with controllable widths down to 6 nm. An 8 nm GNR FET showed a dramatically improved on/off ratio of about 160.76 Other lithography methods like scan tunneling microscopy (STM) lithography were also successfully used to produce GNRs with well-defined armchair edges (Figure 5 b).77 Nevertheless, the widths of GNRs obtained by this technique were scattered over a broad range, and most of the edges were extremely rough. Hence, fabricated devices showed unpredictable performance.

thumbnail image

Figure 5. Schematic of other nanoribbon synthesis techniques. a) Lithographic patterning and plasma etching of graphene.20 b) STM lithography.77 c) Metal-catalyzed cutting of graphene sheets.22 d) Chemical synthesis of GNRs from hydrocarbon molecules [Reprinted by permission from Macmillan Publishers Ltd: Nature (ref. 27), copyright (2010)].

Download figure to PowerPoint

3.3.2 Sonochemical Breaking of Chemically Derived Graphene

A sonochemical method was developed by Dai and co-workers to fabricate GNRs with ultrasmooth edges and widths below 10 nm.18 The synthesized GNRs have variable widths ranging from about 50 to sub-10 nm and their edge roughness is well below ribbon width. FET devices fabricated from sub-10 nm GNRs show an extremely high room-temperature on/off ratios of Ion/Ioff≈107.18, 19 Recently, Cheng et al. further presented a facile strategy to synthesize GNRs in a high yield of about 5 wt % of the starting graphene sheets by sonochemically cutting chemically derived graphene sheets.78

3.3.3. Metal Nanoparticle Catalyzed Cutting of Graphene

Nanocutting of graphene is another promising method of GNR synthesis. In 2008, Strachan et al. used thermally activated Fe nanoparticles as chemical scissors to cut graphene sheet along a specific crystallographic direction (Figure 5 c).22 Parallel straight nanotrenches were observed. Long (>1 μm) crystallographic edges with widths as small as about 15 nm and lengths on the order of millimeters were synthesized. A similar synthesis route employing Ni nanoparticles was implemented by Ci et al.23 They found that the cutting direction could be controlled by means of the size of the metal particles, in principle allowing GNRs or pieces to be cut with discrete AC or ZZ edges.23 Many other catalysts were used to cut graphene into nanostructures with well-defined edges.7982 Although the catalyzed cutting of graphene could well control the edge structure and smoothness, the tracks of the nanoparticles often turn at 60 or 120° angles when they meet edges or defects, leading to a random width distribution of the GNRs (Figure 5 c). A recent report suggested that a combination of nanolithography and catalyzed cutting may provide a feasible way to control ribbon width and edge structure simultaneously.83

3.3.4. Chemical Oxidation Cutting of Graphene

The chemical oxidation method is another feasible route for graphene cutting in large quantities because of its low cost and feasibility. Aksay et al. employed oxidation and thermal expansion methods on graphite to produce graphene sheet.84 Graphite oxide (GO) was prepared by treating graphite with concentrated nitric and sulfuric acids, and gradually adding potassium chlorate to the mixture. Then the GO slurry was spray-dried (300 °C) and thermally exfoliated at high temperature (1050 °C) in a quartz tube. Fujii et al. used a scanning probe microscopic technique to cut oxidized graphene into nanosized islands.24 The cutting procedure was triggered by the local mechanical stress caused by a point contact between the preoxidized graphene sheet and the AFM probe, leading to rupture of the sheet.24 However, in most of these experiments, only graphene quantum dots were produced due to the high symmetry of the honeycomb lattice of graphene (Figure 6 b).

thumbnail image

Figure 6. a) Theoretically suggested linear alignment of epoxy chains during oxidation of graphene69 and possible unzipping mechanism of graphene oxide: from epoxy pairs to carbonyl pairs.70 b) Random unzipping of graphene into quantum dots without external tensile strain. c) Orientation-selective cutting of graphene into GNRs with external tensile strain.87

Download figure to PowerPoint

3.3.5. Bottom-Up Methods: Chemical Vapor Deposition and Chemical Synthesis

The processes mentioned above are all top-down methods. Some bottom-up strategies were also exploited to synthesis GNRs, such as CVD25, 85, 86 and chemical synthesis.26, 27 The great advantage of the CVD method is bulk production (grams per day) of GNRs.25 The synthesized ribbons (<20–30 μm in length) have widths of 20–300 nm and thicknesses of 2–40 layers. The on/off ratio and mobilities can reach 10 at low temperature (4 K) and 2700 cm2 V−1 s−1 at room temperature, respectively.86 At present, the template CVD method is viewed as the most promising way to achieve mass production of GNRs,85, 86 although the widths of GNRs should be narrowed and controlled in a small range in future development. On the other hand, self-assembly of planar graphene-like hydrocarbon molecules in one dimension led to successful synthesis of narrow GNRs with lengths of up to 12 nm (Figure 5 d).26, 27 The advantage of this approach is precise control of composition and structure.

3.3.6. Theory-Related Oxidation Cutting of Graphene

A few theoretical studies on the cutting mechanism and design of new methods were done at ab initio level. Li et al. proposed an unzipping mechanism of graphene oxide that the formed epoxy groups preferentially aligned in a line during oxidation of graphene.69 Li and Yang further revealed how oxygen-atom attack can transform a pair of epoxy groups into a a pair of carbonyl groups and rupture of graphene preloaded with epoxy chains (Figure 6 a).70

Due to the high symmetry of the honeycomb lattice of graphene, the cutting directions of graphene sheet are random during oxidation and lead to graphene quantum dots. Thus, producing graphene nanoribbons with smooth edges by means of oxygen attack is not possible. Recently, by employing DFT calculations, we proposed an effective way of cutting strained graphene into GNRs along a specific direction.87 Our studies demonstrated that the presence of uniaxial external strain not only guides alignment of O atoms along a ZZ direction that is closely perpendicular to the strain, but also significantly lowers the reaction barrier and enthalpy of reaction of graphene cutting along that direction. Moreover, the applied strain simultaneously increases the reaction barrier of cutting along other directions. Hence, orientation-selective cutting of graphene into GNRs on oxidation can be achieved (Figure 6 c).87 A potential experimental route to cut graphene into GNRs by applying external strain by stretching graphene on a polymer was proposed.

4. Summary and Perspective

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Physical Properties and Applications of GNRs
  5. 3. Methods of GNR Synthesis
  6. 4. Summary and Perspective
  7. Acknowledgements

We have reviewed recent progress in the synthesis of GNRs from both experimental and theoretical viewpoints. Among all these methods, synthesis of GNRs through longitudinal unzipping of CNTs is a very promising strategy to achieve mass production of high-quality GNRs, since large-scale fabrication of high-quality CNTs has been realized today. However, the diameter, number of concentric cylinders, and chirality of the CNTs will greatly affect the width, thickness, and edge formation of the synthesized GNRs. Thus, to synthesize GNRs with uniform widths and smooth edges, the synthesis of CNTs with narrow diameter distribution, fixed number of concentric cylinders, and, probably specific chirality is a prerequisite.

Besides unzipping of CNTs, lithographic patterning followed by plasma etching, sonochemical breaking, metal nanoparticle catalyzed cutting, chemical oxidation cutting of graphene, CVD, and chemical synthesis methods have also been developed to synthesize GNRs. Lithographic patterning followed by plasma etching can be used to synthesize GNRs, but the lack of ribbon-width control and high edge roughness may hinder its application in electronics. Sonochemical breaking of chemically derived graphene can be used to synthesize GNRs for fabrication of room-temperature FETs with high on/off ratio, but the yield of sub-10 nm GNRs with smooth edges is very low. Metal nanoparticle catalyzed cutting of graphene produces graphene nanostructure with well-defined armchair or zigzag edges, but changes in cutting direction lead to graphene quantum dots instead of GNRs. Chemical oxidation cutting of graphene has the advantage of simplicity and low cost, but the cutting direction is hard to control. CVD methods can achieve bulk production of GNRs, but the width, edge roughness, and thickness of the obtained GNRs are mostly not promising.

Future methods for GNR synthesis must address the following issues: 1) narrow ribbons, 2) well-defined and smooth ribbon edges, and 3) high-quality GNR wall formation. Among the known methods, only the unzipping of high-quality CNTs and cutting of high-quality graphene have the potential to achieve the three issues simultaneously. As most CNTs are currently unzipped large-diameter MWCNTs, developing methods of cutting small-diameter SWCNTs is pressing. For cutting of high-quality graphene, control of the cutting direction is crucial. One promising method of graphene cutting, namely, metal-catalyzed cutting, probably could achieve the final goal of GNR production, that is, the synthesis of narrow, smooth-edged GNRs with high-quality walls, if the motion of the catalyst could be controlled by means of an external field, strain, or interaction with the substrate.

Acknowledgements

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Physical Properties and Applications of GNRs
  5. 3. Methods of GNR Synthesis
  6. 4. Summary and Perspective
  7. Acknowledgements

This work is supported by NBRP (2010CB923401 and 2011CB302004), the NSF (21173040 and 11074035), SRFDP (20090092110025), and Peiyu Foundation of SEU.