Chapter 1: Carbon-based Nanomaterials in Analytical Chemistry

Nanotechnology, a scientific area that studies and exploits the new properties of materials at the nanoscale (10⁻⁹ m), is one of the most promising research fields at the present time. Within this field, carbon nanomaterials are some of the most used tools because of their unique properties (high mechanical strength, high conductivity, attractive optical properties, chemical versatility, etc.). Taking advantage of these interesting features, a myriad of new electronic devices, sensors, batteries, and composites, have been developed covering the majority of scientific areas (e.g. biology, engineering, physics, chemistry, and medicine). In fact, there is extensive literature about this subject and it continues to be a hot topic. A quick search of the past five years (2013–2018) in the Web of Science database with the words “Fullerene”, “Carbon nanotube” and “Graphene”, yielded more than 12 000, 32 000 and 98 000 articles published, respectively. These results show that graphene has received more attention than carbon nanotubes in the scientific community, despite being the youngest member of the carbon-based nanomaterials family. Furthermore, the search also suggests that interest in fullerenes may be decaying,¹  as there are fewer new applications being published in the analytical chemistry field.²  Considering these trends, this chapter is mostly focused on applications involving carbon nanotubes (CNT) and graphene (G). Specifically, this chapter aims to provide a general overview of the structure, properties and synthetic routes employed to obtain both carbon nanotubes and graphene. Along the same lines, we will describe the most important aspects linked to the techniques used for their physicochemical characterization. Additionally, a specific bibliography will be recommended for readers who are interested in looking more deeply into any of the previous aspects. Finally, the potential of carbon-based nanomaterials for the development of analytical applications will be discussed.

Due to the contributions by material engineers, physicists, and chemists, a large number of complex carbon nanostructures can be grouped under the term “carbon-based nanomaterial”.^3,4  In this chapter, we want to simplify this aspect so we will only take into account carbon-based nanomaterials generally used in analytical chemistry. Consequently, materials have been classified as (see Figure 1.1):

• 0-D: All dimensions are at the nanoscale (dimensionless). Graphene quantum dots.

• 1-D: Two dimensions at the nanoscale. Carbon nanotubes and graphene nanoribbons.

• 2-D: One dimension at the nanoscale. Graphene and few-layer graphene (<10 layers).

• 3-D: No dimension at the nanoscale. Graphite.

While all these nanomaterials include a significant fraction of sp² hybridized carbon atoms, they also feature different shapes, depending on how the hexagonal lattice is arranged (with the exception of the capping ends of CNTs, in which carbon atoms form pentagons or activated functional groups). One important characteristic is that despite being formed by the same atoms, these different shapes confer unique and distinguishable properties that will be further discussed.

It is also important to note that a number of other materials containing carbon-based nanomaterials have been produced. Among those, it is worth mentioning carbon materials developed by pyrolysis and by most hydrothermal approaches.^4,5  These materials are considered to be outside the scope of this chapter.

Strictly speaking, the term “graphene” refers to a material composed of a single layer of carbon with partially filled sp²-orbitals above and below the plane of the sheet. With that being said, this term has been used in the literature to refer to materials comprising several layers (“few-layer” graphene). This categorization has been made as the electronic and mechanical properties of few-layer graphene (<10 layers) are distinct from those of graphite.^6–9  Monolayer graphene exists in a crumpled form with no stacking of sheets¹⁰  and its edges can be described as having armchair or zigzag motifs (Figure 1.2). Few-layer graphene can have a number of stacking arrangements, including ABAB (Bernal stacking), ABCABC (rhombohedral stacking), and less commonly, AAA. Furthermore, few-layer graphene with no discernible stacking order has also been reported and is termed “turbostratic”.¹¹ 

These materials feature extraordinary properties, such as: a large surface area (theoretically, 2630 m² g⁻¹ for a single layer,¹²  about double that of single-walled carbon nanotubes); optical transparency (absorbing ≈2.3% of white light);¹³  high mechanical strength and high elasticity (Young's modulus ≈1100 GPa, stronger than steel);^14,15  and excellent thermal conductivity (5×10³ Wm⁻¹ K⁻¹).¹⁶ 

Focusing on the electronic properties, graphene is considered a zero-gap semiconductor because it presents no energy difference between the conduction and valence bands, so it might be considered a semiconductor or a metal.^17,18  Furthermore, under certain conditions it can be an excellent conductor of electricity with values of charge carrier mobility higher than 200 000 cm² V⁻¹ s⁻¹, reported for freely suspended graphene.¹⁹  This high conductivity was attributed to the fact that charge carriers mimic relativistic particles and are more easily and naturally described starting with the Dirac equation rather than the Schrödinger equation. The interaction between electrons and the periodic potential of graphene's honeycomb lattice gives rise to new quasi-particles with an effective speed of light v_F≈10⁶ m s⁻¹. These quasiparticles, called massless Dirac fermions, can be seen as electrons that have lost their rest mass m₀ or as neutrinos that acquired the electron charge e.¹⁰  As a consequence of this, electrons can travel distances in the order of micrometers without scattering, a phenomenon called “ballistic transport”.²⁰ 

Inside the graphene “family”, there are two components that are worth considering individually because of their prominent importance in analytical chemistry applications: graphene nanoribbons and graphene quantum dots.

Cutting graphene sheets into narrow strips yields to the formation of graphene nanoribbons (GNRs); a material possessing a large aspect ratio. In GNRs, the boundaries gain prominence, exposing non-three coordinated atoms at the edges.^3,21  One of most interesting properties is that GNRs display a finite band gap when their width is less than 10 nm and their electronic behavior changes from semiconductors to semimetals as their width is increased.^21,22  Thus, producing GNRs with defined widths and, in consequence, specific electronic behavior, constitutes a great challenge that many chemists and materials scientists are attempting to deal with.²¹ 

If graphene is cut into small pieces with dimensions of a few nanometers (2–20 nm), graphene quantum dots (GQDs) can be obtained. As reported with GNRs, the smaller the size, the higher the quantum confinement and edge effects, particularly once their dimensions fall below the 10 nm threshold.²³  This means that whereas graphene sheets have a band gap of zero width, which limits their usefulness in electronic and optoelectronic applications, GQDs have nonzero band gaps.⁴ 

The main reason why GQDs are gaining popularity is that they can be synthesized from almost an endless number of organic precursors (sugar, proteins, metabolites, etc.) and they exhibit strong photoluminescence (PL), which shifts in parallel with changes in the band gap size.^24,25  Furthermore, it has been demonstrated that chemical functionalization of GQDs affects their band gaps and PL.²⁶  Their PL spectra feature an intense emission band between 400 and 600 nm.²⁷  Additionally, GQDs are being investigated to develop low-toxicity, eco-friendly alternatives that have the desirable performance characteristics of conventional (CdSe) quantum dots, which are toxic.²⁸  It is interesting to note that there is a broad family of carbon quantum dots (they do not have a graphene structure) with excellent photoluminescence properties, but this topic is out of the scope of this chapter. Interested readers are encouraged to read the following reviews.^29–31 

Carbon nanotubes (CNT) gained popularity after the report by Iijima in 1991.³²  CNTs can be broadly classified in two main groups: Single-walled carbon nanotubes (SWCNTs) and multi-walled carbon nanotubes (MWCNTs). SWCNTs consist of a single graphene layer wrapped into a cylindrical tube and multi-walled nanotubes (MWCNTs) are made of several graphene layers, which are concentrically nested like the rings of a tree trunk.

Despite the structural similarity of both types, they possess a very different electronic behaviour. MWCNTs are always metallic but SWCNTs may be either metallic or semiconducting, depending on their chirality (defined by the direction in which the hypothetical graphene layer is rolled to form the nanotube). The nanotube chirality can be defined in terms of a chiral vector C_n (C_n=m · a₁+n · a₂), which also determines the tube diameter d.^33,34  This vector marks the direction of wrapping a graphene layer, in which a lattice point (m, n) is superimposed with an origin defined as (0, 0) (see Figure 1.3). Indeed, the diameter of a carbon nanotube can be expressed as:

This chiral vector defines the chiral angle as follows:³⁵ 

• Zigzag (θ=0, and m=0 or n=0)

• Armchair (θ=±30°, and m=n)

• Chiral (θ≠0, ±30°, and m, n values that are different from previous cases)

All armchair SWCNTs are metals; zigzag and chiral with m−n=3k, where k is a nonzero integer, are semiconductors with a tiny band gap (they display metallic behaviour); and all the others are semiconductors with a band gap that inversely depends on the nanotube diameter.^35,37 

As reported for graphene, electronic transport in metallic SWCNTs and MWCNTs also occurs ballistically along nanotube lengths, enabling them to carry high currents with hardly any heating.^35,38,39  Reported resistivity values range from 1.2×10⁻⁴ to 5.1×10⁻⁶ ohm cm for MWCNTs,^40,41  and from 0.34×10⁻⁴ to 1.0×10⁻⁴ ohm cm for metallic SWCNTs.⁴²  The thermal conductivity at room temperature for an individual MWCNT⁴³  (>3000 W m⁻¹ K⁻¹), which is greater than basal plane of graphite (2000 W m⁻¹ K⁻¹) but lower than graphene (5000 W m⁻¹ K⁻¹).

Concerning the stiffness of the CNTs, the Young's modulus for an individual SWCNT nanotube is ≈0.64 TPa,⁴⁴  but higher values have been reported for MWCNT (1.28±0.5 TPa)⁴⁵  and graphene. With respect to the surface area of the MWCNT, ≈10–20 m² g⁻¹ values were determined by BET techniques, which is higher than that of graphite but is small compared to activated porous carbons. This value for the SWCNT is expected to be an order of magnitude higher.⁴⁶ 

A survey of the synthetic methods for graphene and CNTs is considered to be outside the scope of this introductory chapter. However, a brief summary of the most employed synthesis methods and their influence on carbon-based nanomaterials properties will be provided in this section. For the sake of a better understanding, graphene and CNTs will be individually dealt with, although there are synthesis methods in common.

Synthetic methods for graphene (2D), graphene nanoribbons (1D), and graphene quantum dots (0D) will be discussed separately because there are noticeable differences between them and they often employ different raw materials. Firstly, synthesis methods for graphene 2D will be considered, which can be divided into two groups: Top-down and bottom-up methods. Top-down methods consist in separating the stacked layers of graphite to yield single graphene sheets, whereas bottom-up methods consist in synthesizing graphene “atom by atom” from other carbon containing sources.

Mechanical exfoliation of fresh graphite using adhesive tape to cleave the layers apart (“Scotch tape” method), was the first technique employed to prepare graphene. This was successfully done by Novoselov et al. in 2004, who reported the preparation of isolated graphene layers from monocrystalline graphite.⁴⁷  The layers prepared by this approach are of high quality, but the method is slow and labour intensive and so is not considered to be suitable for large-scale production.^11,48  Since that year, a number of groups have attempted to develop exfoliation strategies that can be scaled up for large-scale and continuous graphene production. Generally, two strategies are employed for graphite exfoliation: Electrochemical and solvent-based.

Electrochemical exfoliation consists in applying a voltage between two electrodes, at least one of them being made of graphite. The exfoliated material from the graphite sacrificial electrode is collected in the electrolyte solution. For example, G. Wang et al. synthesized graphene applying a constant potential (5 V) between two graphite rods using a poly(sodium-4-styrenesulfonate) solution as electrolyte.⁴⁹  Surfactants, such as poly(sodium-4-styrenesulfonate), prevent re-agglomeration of the graphene, by stabilizing the individual layers in the aqueous solutions.¹¹  One drawback of this strategy is that surfactants can be difficult to remove from the graphene,⁴⁹  and it was demonstrated that their presence can also affect the electrochemical properties of the material.⁵⁰  In another work, sulfuric acid was used as an electrolyte instead of surfactant solutions.⁵¹  To reduce the oxidation of graphene, KOH was added to the H₂SO₄ solution to lower the acidity of the electrolyte solution. The electrochemical exfoliation process was carried out by applying a DC bias on the graphite electrode from −10 to +10 V. Sulfuric acid was found to be a good electrolyte for graphite exfoliation, which is thought to be due to intercalation of SO₄²⁻ ions.⁵¹ 

Solvent-based exfoliation is carried out by sonication of natural flake graphite in an adequate solvent (N,N-dimethyl formamide DMF, N-methyl-pyrrolidone NMP).^52,53  The best solvent with regard to the percentage of monolayer graphene dispersed was found to be NMP, while the solvent that gave the highest absolute concentration (mono- and few-layer graphene) was found to be cyclopentanone, with a solubility of 0.008±0.001 mg mL⁻¹.¹¹  The main limitations of this strategy are the high cost and high boiling point of these solvents, which makes removal difficult when forming films or coatings from the solution.⁴⁸  Aqueous surfactant solutions have also been employed for graphene exfoliation by sonication,^54,55  which has the benefit of avoiding expensive and often harmful solvents. However, as aforementioned, surfactants are difficult to remove from graphene and affect its electrochemical properties. It is also worth mentioning that deep eutectic solvents have been also used as green alternatives for the exfoliation of graphite.⁵⁶ 

The most commonly employed method for obtaining graphene is exfoliation and reduction of graphite oxide. Graphite oxide is produced by the oxidation of graphite using concentrated acids and strong oxidants, in Straudenmaier, Brodie, or Hummers’ methods.

The exfoliation of graphite oxide yields graphene oxide layers (GO), which must be reduced to produce graphene. This material is generally termed “reduced graphene oxide” (rGO) or “chemically modified graphene” (CMG) rather than “graphene” because the structure and properties are not exactly the same, due to the high levels of defects induced in the harsh oxidation processes used to produce graphite oxide. The main advantage of this method is that graphite oxide is exfoliated more readily than graphite, using thermal treatments or via sonication in water, and the graphene oxide produced can then be reduced using either thermal, electrochemical, or chemical methods.⁵⁷ 

Chemical reduction is the most common method to reduce GO and the typical reducing agents are hydrazine,⁵⁸  NaBH₄⁵⁹  and hydroquinone.⁶⁰  These reagents showed some disadvantages such as toxic waste production and being harmful to the environment.⁴⁸  Thus, environmentally-friendly and highly effective reducing agents are needed to substitute the conventional methods to reduce GO.⁶¹ 

Graphene oxide is electrically insulating due to the oxygen-containing functional groups disrupting the sp² hybridization of the graphene layers (see Figure 1.5),⁶²  so efficient reduction is mandatory to restore the desirable electrical properties.¹¹  One of the challenges is to avoid the aggregation and precipitation of rGO after its reduction because it becomes less hydrophilic.⁵⁸ 

However, there is another strategy to reduce GO that has been employed for developing compact on-chip energy storage devices (microscale supercapacitors). This methodology consists in writing graphene patterns onto GO films by laser reduction.^63,64  Later, Tour's group improved this methodology by replacing GO films by a polyimide (PI) film so that production costs are lower.⁶⁵  PI film is converted into porous graphene by a CO₂ infrared laser (see Figure 1.6c and d). This laser-induced graphene (LIG) can be readily written into various geometries by using computer-controlled laser scribing (see Figure 1.6a and b).

Graphite intercalation compounds (GICs) are complex materials having a formula CX_m where the ion Xⁿ⁺ or Xⁿ⁻ is intercalated between the oppositely-charged graphite layers. Among GICs, alkali metal GICs are the most employed for graphene synthesis in combination with a solvent-assisted or thermal exfoliation.

For solvent-assisted exfoliation, GICs are sonicated in solution to aid exfoliation, although spontaneous exfoliation of potassium GICs in NMP has been published.⁶⁶  Furthermore, GICs can be produced “in situ” by electrochemical methods. For example, Wang et al. developed a method in which a negative graphite electrode was electrochemically charged and expanded in an electrolyte of Li salts and organic solvents (DMF) under high current density, and then exfoliated efficiently into few-layer graphene sheets with the aid of sonication (see Figure 1.7).⁶⁷  By this method, the conversion efficiency (few-layer graphene) was higher than 70%. Using a similar strategy but replacing the organic solvent with water, the conversion efficiency was improved (≈80% few-layer graphene) due to lithium and water reacting to form hydrogen gas, which is thought to further aid exfoliation.⁶⁸ 

With respect to thermal exfoliation, it is well known that expanded graphite (EG) is prepared by rapid heating of GIC, resulting in the abrupt ejection or decomposition of guest molecules and the subsequent huge unidirectional expansion of the initial platelets. Using this methodology, mono- to few-layer graphene has been reported for expanded graphite ground in ethanol,⁶⁹  or sonicated in NMP.⁷⁰ 

CVD is the most popular method for large-scale production of mono- or few-layer graphene films. Unlike CNTs, in which catalyst metal nanoparticles are used, the synthesis of graphene by CVD employs a catalytic metal foil on which a carbon source decomposes into carbon atoms. In this way, these atoms are deposited on the foil taking the same shape. After CVD, the metal substrate is etched to detach the graphene layers so that it can be transferred to a new substrate. This etching step is one of the handicaps for the use of metals in large-scale CVD growth of graphene because of the high cost and availability of these metal foils.¹¹ 

Somani et al. were the first group to synthesize few-layer graphene films using CVD. Ni was used as the catalytic foil and camphor as the carbon source.⁷¹  Apart from nickel, graphene growth has been demonstrated on a wide range of metals, including group 8–10 transition metals (Fe, Ru, Co, Rh, Ir, Pd, Pt, Cu, Au) and a number of alloys (Co–Ni, Au–Ni, Ni–Mo, stainless steel).⁷  Among them, copper^72–74  and nickel^75–77  are the most studied, but the growth process features are different. Whilst the low solubility of carbon in copper (<0.001 at%) helps make this growth process self-limiting,⁷³  the growth on nickel is more difficult to control due to medium-high carbon solubility (>0.1 at%), although it has the advantage of not requiring ultra-high vacuum (UHV) conditions,^75,76  as used in the majority of syntheses of graphene on copper.

Briefly, the most commonly-reported growth mechanism of graphene is as follows. Catalytic metals are used to facilitate the diffusion of carbon into the metal thin film at high temperature, followed by precipitation of carbon out from the metal thin film to the metal surface when cooling. During the CVD, the metal substrate is placed in a furnace at a reaction temperature below 1000 °C and low pressure or UHV conditions with a diluted hydrocarbon gas or other carbon sources.^{72,73,75–77}  The thickness and quality of the graphene layers can be controlled by optimizing the reaction parameters such as the cooling rate, concentration of carbon precursor, reaction time, and reaction temperature. Besides that, the type of carbon precursor also affects the formation of graphene.⁴⁸  Most of the carbon sources used were purified chemicals that could be expensive for mass production. Additionally, greener and more environmentally-friendly synthesis methods must be developed utilizing natural, renewable, and cheaper waste materials.⁷⁸  In this sense, there is an interesting example in which graphene grows directly on the back side of a Cu foil using low- or negatively-valued raw carbon-containing materials (cookies, chocolate, grass, plastics, roaches, and dog faeces).⁷⁴ 

Graphene was also synthesized without a metal catalyst by using microwave plasma-enhanced CVD.^79,80  This technique allows one to work at atmospheric pressure and it is capable of continuously producing graphene.

Graphene can be produced on a SiC surface by annealing the SiC surface using UHV. During the heating of the SiC substrate (>1000 °C) under UHV, silicon atoms preferentially sublimate from the substrate. With the removal of silicon atoms, the arrangement of carbon atoms will take place to form graphene layers.⁸¹  The number of graphene layers is affected by the annealing time and sublimation temperature.^82–84  Moreover, graphene layers grow at different patterns on the C-face surface and Si-face surface of the SiC. For the Si-face, graphene grows in a single orientation (rotated 30° with respect to SiC) and exhibits regular Bernal stacking, but for the C-face, graphene exhibits rotational stacking, where each of the rotations forms a commensurate structure with either SiC or the underlying graphene layer.^85,86  Consequently, the electronic and physical properties of graphene are affected. It is worth mentioning that SiC is commercially available but expensive so it limits its use in commercial applications.¹¹ 

Graphene nanoribbons can be synthesized by unzipping SWCNTs or MWCNTs using different strategies. The graphene nanoribbon width is ruled by the diameter of the carbon nanotubes. These strategies can be classified in: (1) intercalation-exfoliation of MWCNTs, involving treatments in liquid NH₃ and Li, and subsequent exfoliation using HCl and heat treatments;⁸⁷  (2) chemical reaction with acids that start to break carbon–carbon bonds (e.g. H₂SO₄ and KMnO₄ as oxidizing agents);⁸⁸  (3) catalytic approach, in which metal nanoparticles “cut” the nanotube longitudinally;⁸⁹  (4) the electrical method, by passing an electric current through a nanotube;⁹⁰  and (5) physicochemical method by embedding the tubes in a polymer matrix followed by Ar plasma treatment.⁹¹ 

There are other methods for the synthesis of graphene nanoribbons but they are out of the limits of this chapter. If readers want to deepen their knowledge of this topic, we recommend the following reviews.^3,4 

As with graphene, GQD synthesis methods can be classified in two groups: Top-down and bottom-up methods.

In the majority of top-down methods, graphene oxide (GO) is cut into small pieces by strong acid and oxidizing solutions,⁹²  and then the produced nanographene oxide (NGO) is thermally^93–95  or chemically⁹⁶  reduced to GQDs. Although most articles employ GQDs, NGO is photoluminescent in the visible and infrared regions.⁹²  In this work, a GO sample was activated with chloroacetic acid under strong, basic conditions, in order to activate epoxide and ester groups, and to convert hydroxyl groups to carboxylic acid moieties (GO-COOH). Then, polyethylene glycol (PEG) was grafted onto the COOH groups, obtaining (NGO-PEG) with high solubility and stability in salt and cellular solutions, which is desirable for biological applications. The ultra-small size of the NGO was caused by the sonication involved in both GO-COOH synthesis and pegylation steps (see Figure 1.8).⁹²  There are many articles about GQDs’ synthesis so we have selected some of them. For example, Pan et al. reported a hydrothermal route to cutting graphene sheets into GQDs with an average diameter of 9.6 nm, which exhibited blue luminescence.⁹³  Shen et al. prepared GQDs that were surface-passivated by polyethylene glycol (GQDs-PEG) by a one-pot hydrothermal reaction.⁹⁴  The collected GQDs-PEG possessed a diameter between 5 and 25 nm and a blue photoluminescence (464 nm). Interestingly, the photoluminescence properties of the GQDs-PEG were much better than those of the GQDs. Green photoluminescence GQDs were prepared by a one-step solvothermal method (using DMF as solvent).⁹⁵  GQDs were single or bi-layered with average diameters of 5.3 nm. Other interesting work about GQDs reduced by chemical reagents was carried out by Shen and coworkers.⁹⁶  GQDs were prepared by hydrazine hydrate reduction of GO and their surface was passivated by PEG. Strong blue photoluminescence was shown under 365 nm radiation and green fluorescence was observed under a 980 nm laser.

GQDs can be synthesized by electrochemical methods. For example, Li et al. prepared a graphene film by filtration, and then it was oxidised by performing a cyclic voltammetry (scan within ±3.0 V at a scan rate of 0.5 V s⁻¹) in 0.1 M PBS.⁹⁷  These GQDs had a uniform size of 3–5 nm and exhibited a green luminescence. Furthermore, the oxygen-containing groups on the surface of GQDs makes them soluble in aqueous media, facilitating further functionalization and various applications. Another top-down method was discovered by Gokus et al., demonstrating that strong photoluminescence could be induced in single-layer graphene using an oxygen plasma.⁹⁸  The process was as follows: graphene samples were produced by microcleavage of graphite on a silicon substrate, and then were exposed to oxygen : argon (1 : 2) radio-frequency plasma (0.04 mbar, 10 W) for increasing time (1–6 s). The O₂ plasma treatment of the single layer graphene caused the opening of a bandgap in terms of functionalization of its pristine lattice with oxygen atoms, generating the photoluminescent behaviour.⁹⁹ 

One of the most promising bottom-up approaches is the solution-phase chemical method using carbon precursors. GQDs have been synthesized by oxidative condensation of aryl groups.^100–104  Yan et al. produced large colloidal graphene quantum dots with a uniform and tunable size.¹⁰⁰  They consist of graphene moieties containing 168, 132, and 170 conjugated carbon atoms. The oxidation of polyphenylene dendritic precursors that were synthesized through stepwise solution chemistry led to fused graphene moieties. The stabilization of the resultant graphene is achieved by multiple 2′,4′,6′-triakyl phenyl groups covalently attached to the edges of the graphene moieties. The crowdedness on the edges of the graphene cores twists the substituted phenyl groups from the plane of the core, leading to alkyl chains closing the latter in all three dimensions. This results in reduced face-to-face interaction between the graphene, thus effectively increasing their solubility and stability.

Using a multistep methodology (see Figure 1.9), Liu et al. prepared multicolor GQDs with a uniform size of ≈60 nm diameter and 2–3 nm thickness by using unsubstituted hexa-peri-hexabenzocoronene (HBC) as the carbon precursor.¹⁰⁵  HBC is a large polycyclic aromatic hydrocarbon (PAH), which was synthesized from commercially available hexaphenylbenzene by cyclodehydrogenation. In the first step, HBC was pyrolyzed at a high temperature obtaining artificial graphite. In the second step, the artificial graphite was oxidized and exfoliated with a modified Hummers method. Subsequently, aqueous solutions of the resultant GOs were heated to reflux with oligomeric poly(ethylene glycol) diamine and then reduced with hydrazine.

There is another original methodology to produce GQDs that consists in opening fullerenes. Loh et al. synthesized a series of atomically-defined GQDs by metal-catalyzed cage-opening of fullerenes.¹⁰⁶  The fragmentation of the embedded molecules at elevated temperatures produced carbon clusters that underwent diffusion and aggregation to form graphene quantum dots onto a Ruthenium surface. The equilibrium shape of the graphene could be tailored by optimizing the annealing temperature and the density of the carbon clusters. The main problem is that this methodology is not easily scalable due to the raw material employed.

Considering the future challenges of GQD synthesis methods, the main handicap is the synthesis of high-quality GQDs with well-controlled size, shape, and surface functionalization. In this sense, GQDs prepared by the top-down approaches are usually difficult to control in terms of size and shape, so the bottom-up methods seem to be the best option.

Finally, the synthesis of graphene quantum dots and other similar structures (carbon dots) is a cutting-edge research field in which new, improved, and imaginative methodologies are continuously emerging. Inquisitive readers are invited to read these recent reviews.^29–31 

There is a vast body of literature describing the methodologies to fabricate CNTs. Among them, it is relevant to mention contributions from Endo, Iijima and Dresselhaus,¹⁰⁷  Guldi and Martin,¹⁰⁸  Kar,¹⁰⁹  Morris and Iniewski,¹¹⁰  Harris,¹¹¹  Mishra,¹¹²  and Zhang.¹¹³ 

Briefly, CNTs can be fabricated by a number of techniques,^114,115  including chemical vapor deposition (CVD), laser-ablation, or carbon arc-discharge.¹¹⁶  However, due to the simpler requirements (lower temperature) and versatility (better control of the length, diameter, alignment, purity, density, and orientation), the former method is largely preferred. The basic mechanism is similar to the one described in Section 3.1.2 (Bottom-up Methods) with the main difference being that metallic nanoparticles are used as catalysts for the breakdown and nucleation of the feedstock and the tube, respectively. For this purpose, feedstock gases such as methane or acetylene,¹¹⁷  temperatures <800 °C,^118,119  and metallic nanoparticles (usually Au, Pd, Pt, Cu, Ni, Fe, Co, and/or Mo)^120,121  with targeted dimensions¹²²  are typically employed. In this regard, a recent report from Wang et al.¹²³  suggests that a potential charge-transfer mechanism between the catalyst (Fe) and the formed CNTs could have a significant influence on the growth and resulting chirality. Other groups have also provided evidence that oxygen-containing groups may be deleterious for the synthesis of vertically-aligned CNTs¹²⁴  and that oxygen-deficient catalysts could direct the growth towards SWCNTs.¹²⁵ 

An additional advantage of CVD is the versatility of the instrumental setup, which has allowed the implementation of several modifications and etching procedures^126,127  including water-assisted,^128–131  oxygen-assisted,¹²⁴  hydrogen-free,¹³²  hot-filament,^133,134  or microwave plasma CVD.¹³⁵  It is also important to mention that although most reports point to a mechanism that allows growing the CNT from the base,¹²⁵  several authors have also considered the penetration of carbon atoms (and clusters) into the metallic nanoparticle, a phenomenon that can be controlled by both the melting point of the material and the solubility of carbon. The former mechanism set the basis for the development of a carbon-coated Co nanoparticle¹³⁶  that enabled not only vertical growth of SWCNTs but also a narrow band-gap distribution.

Albeit being one of the most popular synthetic approaches, CVD still suffers from rather low yields (around 20% feedstock to CNT ratios) and the product is typically a mixture of different structures¹³⁷  that require further manipulation with strong acids (HCl), bases (KOH) and/or mild oxidants (air ¹³⁸  or H₂O₂¹³⁹ ). In general, these steps are implemented to remove other carbonaceous materials (amorphous carbon, fullerenes, nanocrystalline graphite) as well as metallic residues used as catalysts. A step forward in the development of clean-up procedures was recently presented by Gomez et al., who utilized a microwave oven and a chlorination step to remove unwanted material.¹⁴⁰  It is also important to mention that a careful selection of the experimental conditions is required because these purification steps may lead to significant differences in the electrochemical properties of the resulting material.¹⁴¹  Other alternatives include current injection/acid washing,¹⁴²  sonication¹⁴³  and processing using microwaves.¹⁴⁴ 

After synthesis of the carbon nanomaterials, they need to be fully characterized to assure the product quality (purity, defects, chemical species on the surface…) and to elucidate their structure. Every technique provides different morphological, physical, and chemical information so several techniques need to be employed for a complete characterization.

Graphene and CNTs show a strong absorption band around 230 nm, which corresponds to a π–π* transition of aromatic CC bonds.^145,146  Interestingly, graphene is less transparent in the UV–visible region than GO, which is attributed to a lower proportion of sp² carbons in the structure. This different behaviour was used for monitoring reduction reactions¹⁴⁷  or for estimating the oxidization grade of GO.¹⁴⁸  Moreover, the transmittance of graphene decreases as the number of layers increases. Sun et al. estimated the number of graphene layers by UV–visible absorbance at 550 nm, taking advantage of this feature.¹⁴⁹ 

In the case of CNTs, UV–vis spectrophotometry was employed for monitoring the dispersion protocols of SWCNTs¹⁵⁰  and MWCNTs¹⁵¹  due to an increase in absorbance intensity when the dispersion is improved. Additionally, the concentration of CNTs in solutions can be measured by UV–vis absorbance.^150,152 

Mainly, IR spectra allow one to characterize the functional moieties present in graphene and CNT samples.^153,154  In particular, it is employed to determine the presence of oxygen-containing functional groups such as carboxylic acid or carboxylate because CO stretching vibrations are characterized by strong absorptions at 1719 cm⁻¹ and 1620 cm⁻¹, respectively.¹⁵⁵  IR spectroscopy is not usually used in carbon-nanomaterials’ characterization because the intensity signals are weak and there are more suitable techniques for this purpose such as Raman spectroscopy and X-ray photoelectron spectroscopy.

Raman spectroscopy is one of the techniques most commonly employed for the characterization of carbon nanomaterials. Among the reasons, it is a non-destructive technique that allows one to probe purity, structure, and the degree of functionalization. Carbon allotropes possess their identity at D, G, and 2D peaks around 1350, 1580 and 2700 cm⁻¹, respectively, by Raman spectroscopy. The G-band corresponds to the tangential stretching (E_2g) mode of graphite (G-graphite), while the D-band (D-diamond or disorder) arises from the out-of-plane vibrational modes and it is indicative of sp³ carbon presence. This causes a disorder in the sp²-hybridized carbon atoms, characteristic for lattice distortions in the curved graphene layer and/or tube ends. The 2D band is at almost double the frequency of the D band and originates from a second order Raman scattering process.^156,157 

For graphene, the higher the number of graphene sheets, the higher the G band intensity. Furthermore, this increase causes a broadening of the 2D band to high wavenumber direction.¹⁵⁷  Ferrari et al. took advantage of this effect to develop a methodology for calculating the number of layers in graphene samples (see Figure 1.10).¹⁵⁸  Another interesting parameter is the ratio of peak intensities I_D/I_G, because it can be used to evaluate the level of disorder in graphene.^159–161  Two different stages have been described when defects are introduced: the first stage shows the transition from pristine graphite to nanocrystalline graphite (low defect density) and the second stage considers the transition from nanocrystalline graphite to mainly sp² amorphous carbon (high defect density).^160,162  In the first stage, the I_D/I_G ratio follows the Tuinstra–Koenig equation:¹⁶³ 

In the second stage, an ever-increasing number of defects generate carbon clusters, which become smaller and the rings fewer and more distorted, until they begin to open up. The G peak is due to the relative motion of the sp² carbon atoms, while the D peak is linked to the breathing modes of rings. So, it can be assumed that I_G is approximately constant as a function of disorder. Thus, with the loss of the sp² rings I_D will now decrease with respect to I_G, and the Tuinstra–Koenig equation will not be valid. For small L_a, the D mode strength is proportional to the probability of finding a six-fold ring in the cluster, i.e. to the cluster area. Thus, in amorphous carbons the development of a D peak indicates ordering, exactly the opposite to the case of graphite.^160,162  The second stage follows the next equation:

In the case of CNTs, the I_D/I_G ratio is also used to measure their quality but is simpler to interpret.^165,166  The higher the I_D/I_G ratio, the higher the number of defects. Besides the D and G bands, there is another interesting band due to the radial breathing modes (RBM). This peak is found at low wavenumber, from 160 to 350 cm⁻¹.¹⁶⁷  These modes are created by the symmetrical expansion and contraction of the tubes around the tubes axis. RBM can be used to study the nanotube diameter (d_t) through its frequency (ω_RBM), to probe the electronic structure through its intensity (I_RBM) and to perform an (n, m) assignment of a single isolated SWCNT from analysis of both d_t and I_RBM. Furthermore, the detection of the RBM band is an unequivocal signal of SWCNT presence. However, the RBM signal is hardly detectable for the MWCNT. This is due to the fact that the RBM band from large diameter tubes (MWCNTs possess a higher diameter than SWCNTs) is usually too weak and the ensemble average of the inner tube diameter broadens the signal. One of the most attractive features of RBM is that its frequency is inversely proportional to the diameter of the tube, allowing the calculation of the SWCNT diameter by the following equation:¹⁶⁸ 

XPS has been mainly used for the characterization of functional group moieties, especially oxygen-containing groups, present in CNTs and graphene.^58,169  XPS can distinguish the groups of bonds and the oxidation state of carbons (C–C, C–O, CO). In fact, XPS of C 1s is employed to study the reduction of GO to rGO.^58,170 

TGA is used for studying the thermal stability of CNTs but its application to graphene is scarce. The oxidation temperature (T_o) of the sample depends on several parameters such as the nanotube diameter (smaller nanotubes are believed to oxidize at a lower temperature due to higher curvature strain), defects and derivatization moiety in the nanotube walls,^171,172  metal particles present in the nanotube (they catalyze carbon oxidation) and the presence of amorphous carbon.¹⁷³  The parameter T_o offers information about the overall quality of nanotube sample.¹⁷⁴  Higher oxidation temperature is always associated with purer, less defective samples. The typical T_o for high purity SWCNT is ≈600 °C,¹⁷⁵  for MWCNT≈700 °C,¹⁷⁶  and for fullerene (C₆₀)≈420 °C.¹⁷⁷  Furthermore, it is possible to detect the presence of amorphous carbon because the corresponding T_o is around 350 °C (see Figure 1.11).^178,179  By TGA analysis, the percentage of CNTs, metal particles, and amorphous carbon occurring in the CNT samples can be estimated.^175,178,180  In addition, it has been used to demonstrate the success of functionalization protocols.¹⁷¹ 

For graphene, TGA has been applied for evaluating the reduction of GO to rGO. GO is thermally unstable and loses mass at 200 °C due to pyrolysis of the labile oxygen-containing functional groups, yielding CO, CO₂, and steam.⁵⁸ 

This technique is suitable for the characterization and identification of polycrystalline phases and can measure interlayer distance. For these reasons, XRD has been exhaustively employed for monitoring the graphite oxidation process and the consequent exfoliation of graphite oxide to GO. Pristine graphite presents a basal reflection (002) peak at 2θ=26° (interlayer distance 0.34 nm) in the XRD pattern and graphite oxide at 2θ≈12° (interlayer distance 0.7 nm). The intercalation of oxygen species between the graphite layers generates an interlayer expansion in graphite oxide. As oxidation proceeds, the intensity of the (002) diffraction line gradually weakened and finally disappeared. At the same time, the intensity of the diffraction peak at 12° increased with oxidation. When the graphite oxide is fully exfoliated, the peak at 12° disappeared.^60,181–183 

With respect to CNTs, as-synthesized SWCNTs tend to form crystalline bundles, which show a peak at 2θ≈6° in the XRD pattern.^42,184  By this technique, the number of SWCNTs contained in each bundle can be calculated.¹⁷⁵  MWCNTs show a reflection (002) peak at 2θ=26° as graphite. This signal has been used to monitor the MWCNTs growing.^185,186 

High-resolution scanning electron microscopy (HRSEM) is a characterization tool used to examine the topography, morphology, and orientation of nanomaterials. High-resolution transmission electron microscopy (HRTEM) offers higher resolution and allows more accurate morphological (inner structure) and topographical evaluation. Additionally, energy dispersive X-ray analysis (EDX) is usually coupled to electron microscopy for determining the sample composition (elemental analysis).

HRSEM was used for visualizing the orientation/alignment of CNTs¹⁸⁷  and the patterned surfaces due to the different growing of CNTs.¹⁸⁸  By HRTEM, SWCNTs and MWCNTs can be distinguished and their diameter can be measured.^175,189  In addition, the presence of carbonaceous particle and catalytic nanoparticles can be detected and identified by HRTEM-EDX.^174,190  More interesting is the ability of HRTEM to assign the chiral indices of SWCNTs.¹⁹¹  Furthermore, it allowed the identification of fullerenes (C₆₀) inside SWCNTs (peapod structures),¹⁹²  and even the visualization of conformational changes of small hydrocarbon molecules confined in carbon nanotubes.¹⁹³ 

With respect to graphene, single-layer graphene can be observed as semi-transparent sheets by TEM analysis, although atomic resolution imaging can be also achieved.¹⁹⁴  The number of graphene layers can be observed in the foldings at the rim of the membrane, where the sheet is locally parallel to the electron beam.^{53,75,194,195}  Moreover, single-layer and two-layer graphene can be differentiated by their corresponding nano-area electron diffraction patterns (see Figure 1.12). These patterns are obtained by changing the incidence angles between the electron beam and the graphene sheet.^53,194,195  Graphene roughness is also monitored by electron diffraction patterns.¹⁹⁵ 

AFM provides accurate 3D topographic data. It has demonstrated to be crucial for the unequivocal identification of monolayer graphene.¹⁹⁶  A pristine monolayer graphene has a van de Waals thickness of ≈0.34 nm corresponding to the inter-plane spacing of graphite.⁴⁷  AFM has proven to be inaccurate with a wide range of measured values for single layer graphene thickness reported (between 0.4 and 1.7 nm). This discrepancy has been attributed to tip-surface interactions, image feedback settings, surface chemistry, and the presence of a layer of absorbed water.^196,197  For example, Valles et al. reported different values of monolayer graphene thickness depending on the substrate used (mica or Si/SiO₂).⁶⁶  They obtained a value of ≈0.4 nm when mica is used and ≈1 nm for Si/SiO₂. Furthermore, the defects and oxygen species present in GO increase its thickness up to ≈1.2 nm^198,199  and, after a reduction step (rGO), this value decreases to ≈0.8 nm (see Figure 1.13).^59,198–200 

AFM also gives information about the presence of macromolecules on the graphene surface,²⁰⁰  and about the mechanical, frictional, electrical, magnetic, and elastic properties of graphene sheets.¹⁴ 

With respect to CNTs, AFM enables one to distinguish between SWCNTs and DWCNTs.²⁰¹  Moreover, metallic particles and biomolecules attached to CNTs have been detected by AFM,^202–205  as in the case of graphene.

STM is a powerful technique that provides atomic resolution images of graphene and CNTs allowing the observation of the hexagonal structure of these materials. By this technique, it is possible to identify defects in graphene surfaces (see Figure 1.14).²⁰⁶  In addition, a more precise measurement of graphene thickness has been achieved by STM.^66,199  In the case of CNTs, it allows the measurement of their chiral angles and diameter. In this sense, Dekker's group confirmed the correlation between the electronic properties of SWCNTs and their (m, n) indices.²⁰⁷  They also examined the bandgaps of metallic and semiconducting SWCNTs by this technique.²⁰⁸  Furthermore, STM has been used for the characterization of functionalized CNTs.^209–211 

In this section, we would like to sketch some of the possibilities of carbon-based nanomaterials in the field of chemical analysis according to their amazing characteristics. Among them, there are some that are not so interesting, such as high strength and high elasticity. Moreover, we must take into account that some of these characteristics depend on the synthesis protocol used.

Due to the large surface area, high adsorption capacity, chemical inertness, and thermal stability, carbon-based nanomaterials have been employed in extraction protocols (solid phase extraction) and chromatography. One key point is the full characterization of the carbon nanomaterial surface because the presence of oxygen functionalities can change the predominant hydrophobic interactions (π–π) by hydrophilic ones. In this sense, CNTs purified by strong acids and graphene obtained from GO (rGO) are rich in these functionalities. This means that it is possible to tune the hydrophobic/hydrophilic character by selecting the adequate purification and/or synthesis method. This fact allows one to design stationary phases with different behaviour for solid phase extraction or chromatography.

Moreover, carbon-based nanomaterials have been exhaustively used in developing electrochemical (bio)sensors. Their conductivity (“ballistic transport”) makes them ideal for electrochemical detection, obtaining higher sensitivity and, sometimes, selectivity (decrease of redox overpotentials). Pristine CNTs and graphene exhibit the highest conductivity, but most applications require the presence of defects or functional groups for developing biosensors. The biorecognition element should be anchored on the carbon nanomaterial surface and this is made mainly through the oxygen functionalities. Again, CNTs purified with strong acids and rGO are better candidates than pristine CNTs and graphene.

In addition, we must mention the younger members of the “graphene family”: graphene quantum dots and graphene nanoribbons. The former could play an important role in bioimaging because of their biocompability,²¹²  and the latter could allow one to further the development of new field emission transistor (FET) sensors due to their tuneable band gap.²¹³ 

Currently, it could be thought than graphene is winning the race against CNTs, mainly, because of the lower cost of graphene production but, in our opinion, there is at least one major obstacle impairing the widespread use of graphene. rGO or pristine graphene tends to stack by π–π interaction so that the graphite structure is recovered. This means that it is still necessary to employ some kind of barrier to avoid re-stacking; most often chemical intercalators, surfactants, or embedding in polymers. These components could hinder the properties by which graphene was chosen. Moreover, the success and amount of information related to CNTs cannot be ignored. They have demonstrated to be excellent electrical connectors between redox enzymes and the transductor elements (bulk electrode) facilitating the electron transfer due to their 1D configuration. Furthermore, label-free FET sensors have been successfully developed using CNTs as sensing material. All things considered, it is absolutely clear that carbon-based nanomaterials have revolutionized the field of analytical chemistry, but their consolidation in this field will depend on the production of well-characterized carbon nanomaterials with reproducible properties, which in turn will allow one to develop robust analytical methods.