Raman Spectroscopy and Imaging of Carbon Allotropes

1.1 Nanostructures

In the recent years, the synthesis and characterization of nanomaterials have been one of the most efficacious ways to produce new materials with improved or completely new properties [1]. Their physical dimensions can be used to classify the nanomaterials in subgroups. One-dimensional (1D) nanostructures are systems in which one of the spatial dimensions has less than 100 nm, such as carbon nanotubes, metallic nanowires, or zeolites having 1D cavities (Figure 1). Lamellar materials are classified as two-dimensional (2D) nanostructures, because there are formed by platelets piled up in one crystallographic direction, as the graphite and clays. For materials having nanocavities or structures those follow in all directions, they are named as three-dimensional nanostructures (3D; as some zeolites). When the material is symmetric in all directions, it is considered as zero-dimensional (0D) nanostructures, as found in quantum dots, fullerenes, or cyclodextrins (Figure 1).

Nanostructures are systems in which at least one of the spatial dimensions is smaller than 100 nm [1]. The synthesis of controlled dimensional nanostructures and the characterization of the intrinsic and potentially peculiar properties of these nanostructures are central themes in nanoscience. The study of different nanostructures has great potential to test and understand fundamental concepts about the role of particle dimensionality on their physicochemical properties. Among the various materials studied in the literature, undoubtedly, carbon-derived materials, especially fullerenes and nanotubes, and more recently graphenes, are of particular note.

There are two central questions for the study of chemistry and physics of these nanostructures: (i) how controlled dimensionality nanostructures can be fabricated, and (ii) what are the intrinsic and potentially peculiar properties of these nanostructures. ID structures have great potential for testing and understanding fundamental concepts about the role of dimensionality and size over the properties. For example, 1D systems must have singularities in their electronic density of states. There are also several applications, as 1D systems are smaller structures that can be used for efficient electrical transport. Their selected properties can be explored, for example, in nanoelectronics. Among the carbon nanostructures, we highlight carbon nanotubes and graphenes [2, 3, 4].

1.2 Carbon allotropes

The discovery of fullerenes in 1985 opened a new field in chemistry [5]. Since this discovery, research with carbon structures has grown rapidly. In 1991, Sumio Iijima was the first researcher to observe some unusual carbon structures under a transmission electron microscope. Iijima called these structures as carbon nanotubes (CNTs) [2], because they consisted of many cylindrical coiled carbon layers. The layers have carbon atoms attached by six-membered rings and are stacked in the c-crystallographic direction, such as in graphite. CNTs can be open-ended or closed-ended with closed ends having five-membered carbon rings as in fullerenes.

The CNTs can be multiwalled carbon nanotubes (MWCNTs), double-walled carbon nanotubes (DWCNTs), or single-walled carbon nanotubes (SWCNTs). In fact, depending on the way, the graphite layers are coiled, i.e., the different combinations of the vectors (defined by the integers n, m) that define each tube geometrically, they can be classified as “armchair,” “zigzag,” and “chiral” (Figure 2). As consequence of this peculiar geometry, the electronic conductivity and optical properties, such as luminescence and light scattering, are dependent of the tube geometry [6, 7, 8].

CNTs attract great attention because they are model systems for nanoscience and have great potential for applications such as composite materials, batteries, sensors, and nanoscale electronics. Interest in CNTs is in their unique structure and properties, their small size (from 0.8 to 2 nm in diameter, see Figure 1), their ability to be metallic or semiconductor depending on their geometric structure (Figure 2), their exceptional ballistic transport properties and thermal conductivity, optical polarizability, and high structural perfection [6, 7].

SWCNTs have a relatively simple structure, allowing detailed calculations of their electronic structure [7, 9, 10, 11, 12]. The unique optical properties of CNTs are due to the confinement of electronic states in one direction, resulting in the so-called van Hove singularities [10, 13]. The presence of a large number of electron states with very close values ​​in energy leads to an intensification of corresponding photophysical processes, and it is possible through electron absorption, photoluminescence, resonance Raman, and photoelectron spectroscopy techniques to obtain detailed information about the electronic structure and vibration analysis of nanotubes [7, 14, 15].

Theoretical predictions about the one-dimensional electronic structure of nanotubes have been experimentally verified. The most conclusive evidence comes from Scanning Tunneling Microscopy/Spectroscopy (STM/STS) studies, which showed atomic resolution images and the corresponding electronic structures for metallic and semiconductor nanotubes, and verified the dependence of electronic properties on diameter and helicity [16]. Therefore, the electronic structure of nanotubes depends only on their symmetry, being quite peculiar to solid-state physics. Specifically, the electronic structure may be metallic or semiconductor, depending on diameter and chirality, although there is no difference in the chemical bonds between carbon atoms in different nanotubes.

Despite the unique properties of CNTs, and these are already produced in macroscopic quantities, allowing the study of their physicochemical properties, they still have low solubility in most solvents, as a consequence of their high aggregation, limiting the possibility of chemical manipulation and technological application. Thus, different approaches have been employed to separate or disaggregate CNTs, such as chemical modifications of nanotubes or through interaction with polymers.

Graphite (3D), which is the most well-known allotropic form of carbon, has layers of several carbon atoms bonded with sp² hybridization (Figure 3), one carbon atom joining three other atoms forming a planar array of hexagons. The layers remain connected via the interaction of van der Waals forces. These two-dimensional sheets have the thickness of a carbon atom, which allows them to have different properties that differ from graphite, such as high electrical, thermal conductivity, and mechanical stiffness [3, 4, 18, 19, 20, 21, 22]. These layers have been called graphene (2D), which were discovered by scientists André Geim and Konstantin Novoselov at the end of 2004 at the University of Manchester.

Graphene is the allotropic form of carbon most recently studied, due to its wide application in the scientific environment. This material is obtained via graphite oxidation and exfoliation processes. Depending on the synthesis parameters (oxidant type, reactant proportions, temperature, etc.), graphene sheets or their oxide (GO) or graphene oxide will have distinct properties. Brodie performed the first documented synthesis in 1859 [23]. Since then, several other scientists have prepared graphene oxide (GO) synthesis experiments from graphite (G) in order to reduce the environmental hazards that this synthesis presents in the use of strong oxidants and concentrated acids. Modifications were later made by Staudenmaier (1898), who focused on improving the reaction introduced H₂SO₄ to the mixture and some aliquots of solid KClO₃ throughout the reaction. With these modifications, the author achieved the synthesis of a more oxidized graphic material and a simplification in the reaction [24]. However, there is release of ClO₂ gas, with risks of explosions during the process. In 1958, Hummers proposed some more modifications to this synthesis with the intention of making it safer and more profitable, replacing the oxidizing agent used in the Brodie method with MnO₄ and an additive, NaNO₃. All the methods mentioned so far make use of strong and toxic reagents for the production of GO, however the use of thermal expansion of graphite, or sonic spacing, those are more green routes, it takes around 6–12 h of preparation.

1.3 Raman spectroscopy and imaging

Through the years, the infrared and Raman spectroscopies have been the techniques par excellencefor the investigation of the vibrational structure of conjugated materials, such as dyes [25], metallic complexes [26, 27, 28], conducting polymers [29, 30], nanocomposites [31, 32, 33], and carbon allotropes [34, 35, 36, 37, 38, 39]. The laser is the common source in Raman spectroscopy, the radiation interacts with the sample through the scattering process. The incident light (laser source) has much more energy than the vibrational levels, however, by the scattering process; the information about the vibrational modes can be accessed. This behavior is very different from the IR vibrational spectroscopy, where the radiation energies have the same magnitude of vibrational levels, and by absorption process, the vibrational modes can be studied (Figure 4).

The intensities of the Raman bands are proportional to the fourth power of the scattered frequency λs4when the laser energy is very far from a permitted molecular electronic transition [40, 41]. This behavior changes dramatically when the laser energy is close to an electronic transition, this condition is known as Resonance Raman. Hence, the intensities of the vibrational modes associated with the excited electronic state in resonance are amplified by 10^5–7 times. For multichromophoric systems, like conducting polymers, it is possible to screen the different chromophores, just by selecting the appropriated laser line energy.

In recent years, the use of Raman spectrometers coupled with different microscopes, from optical to force atomic type, has increased. In Figure 5, the configuration of an optical microscope coupled to a laser source used for Raman measurements is schematically illustrated. The laser lines reach to the sample on the microscope stage by optical elements. The Raman scattered radiation is collected in a scattering angle of 180° by the same microscope objective and captured by an opening of the spectrometer using a beam splitter. It is necessary that the instrument has a high lighting efficiency, and the collection of scattered radiation must be precisely done, owing to the very small Raman cross section (typically a factor of 10⁻⁶ to 10⁻¹² of the incident radiation) and the small volume of the sample. Raman microscopy can be considered a nondestructive technique; however, sometimes the laser power can destroy the sample or change its structure during the measurement. The use of microscopy opens the possibility to search different areas of the sample. In a conventional microscope, it is possible to investigate a very small part of the sample (1 μm approximately or smaller). The high lateral resolution and depth of field (the order of a few micrometers) are very useful for the study of multilayered polymeric thin films or other complex materials [42, 43, 44].

2.1 Resonance Raman of modified single- and double-walled carbon nanotubes

The resonance Raman spectroscopy technique is sensitive to the electronic, structural, and vibrational properties of CNTs. Our group is using resonance Raman spectroscopy to characterize the interactions between nanotubes and different kinds of molecules [25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39], such as the conducting polyaniline (PANI), molecular magnets such as (NBu₄)₂[Cu(opba)], [MnCu(opba)]_n chains, where opba = ortho-phenylenebis(oxamate), dyes such as phenosafranine (PS) and Nile Blue (NB), and CNTs doped with bromine or iodine.

For all samples investigated, the resonance Raman spectra are dominated by SWCNT or DWCNT bands at different laser excitation energies (E_laser) [37, 38, 39]. The Raman spectra obtained with laser lines from 790 to 514.5 nm (E_laser from 1.57 to 2.41 eV) are dominated by the characteristic bands from the SWCNTs or DWCNTs. This behavior is associated with the strong resonance Raman with the van Hove singularities of the single-walled nanotubes [6, 7]. The spectra can be divided into four groups of bands: (1) sharp bands from 120 to 350 cm⁻¹ are assigned to the radial breathing modes (RBMs); (2) strong band in the frequency range from 1500 to 1650 cm⁻¹ is attributed to stretching modes of carbon atoms (tangential G band); (3) the mode at ca. 1300–1350 cm⁻¹ is forbidden for symmetry reasons and is related to the disorder-induced D-band feature; and (4) finally, the second-order G’band (or 2D band), which is the highly dispersive harmonic of the D-band frequency, and it is observed here from ca. 2600 to 2700 cm⁻¹.

As a consequence of the intense CNT bands, the analysis remains in the comparison of standard CNTs before and after any chemical change. For instance, Figure 6 exhibits the Raman spectra, in the RBM region, obtained for the composite between the SWNTs and [MnCu(opba)]_n chains. At higher E_laser (2.33 eV) or 532.0 nm, it was observed many changes in the RBM bands, shifts from 254 to 257 cm⁻¹ (Δ = +3 cm⁻¹), from 266 to 270 cm⁻¹ (Δ = +4 cm⁻¹), and from 276 to 281 cm⁻¹ (Δ = +5 cm⁻¹) were seen for tubes assigned to E11Mfamily with minor diameter than observed for the other laser lines. In Figure 6, it is also observed that RBM bands assigned to E33Sfamily also have changed their intensities and frequencies from 155 to 160 cm⁻¹ (Δ = +5 cm⁻¹) and from 143 to 149 cm⁻¹ (Δ = +6 cm⁻¹) in the presence of the heterobimetallic complex.

At E_laser = 2.33 eV also, semiconducting tubes interact with the heterobimetallic polymer, contrarily to that observed for lower E_laser energies. Hence, the RBM data suggest that the heterobimetallic polymer interacts mainly with metallic tubes independently of the diameter of the tube and excitation energy, however, for semiconducting tubes, solely for tubes with diameter higher than ca. 1.47 nm (Figure 5). In certain circumstances, it is possible to use UV laser line, and some bands from the metallic complex emerge in the spectra. For instance, the resonance Raman spectra, at higher E_laser (3.82 eV), of SWCNTs-(NBu₄)₂[Cu(opba)] samples show bands at 1430 and 1474 cm⁻¹. These bands can be assigned to the vibrational modes related to the benzene-like ring in the molecular structure of the [Cu(opba)]²⁻ anion [34].

The CNT families are assigned by using the Kataura plot (Figure 7a). The Kataura plot is the result of calculated energy separations between the van Hove singularities [E_ii(d_t)] for SWCNTs obtained from tight-binding calculations [15, 45]. For instance, according to the Kataura plot, at 514.5 nm (E_laser = 2.41 eV) or 532.0 nm (E_laser = 2.33 eV), the DWCNTs in resonance have inner tube metallic and outer tube semiconducting. The Br₂ doping was reported for SWCNTs [47, 48] and for DWCNTs [49, 50]. A very large Raman bands upshifted at 514.5 nm (2.41 eV) laser line in both the RBM and G⁺ band for brominated SWCNTs are observed. Indeed, the doping induces a Fermi level depression of 1.2–1.4 eV in the DWCNTs after bromination, mainly in the outer tubes. The Br–Br molecular vibration is reduced to ca. 233 cm⁻¹ in comparison with molecular bromine (323 cm⁻¹) [37, 38, 39], as a consequence to the charge transfer with the carbon tubes.

The use of a large variety of laser lines [37, 38, 39] permits to monitor differences in the charge transfer behavior to the inner and outer tubes from the adsorbed bromine or iodine. Figure 7b shows that the Br₂ molecules are interacting with the outer tube surface of the DWCNTs and that the adsorbed bromine molecules act as an electron acceptor. In addition, it is observed that metallic tubes are extremely sensitive to doping process. In fact, the presence of Br₂ molecules changes the Raman spectra of the metallic tubes even when they are in the inner configuration surrounded by semiconducting outer tubes of DWCNTs.

The differences between metallic and semiconducting tubes can also be analyzed in the other regions of the Raman spectra of CNTs. For instance, the presence of the lower frequency component of the G-band spectra is associated with the metallic tubes. This is the result of the strong electron-phonon coupling observed in metallic nanotubes, and it gives rise to Kohn anomalies in the phonon dispersions [51, 52]. Hence, the influence of the metallic tubes can easily be distinguished from semiconducting G⁻feature, which shows a Breit-Wigner-Fano (BWF) line shape [53, 54]. The changes in relative intensities, dispersion, and linewidth of the G’ (2D) band can also be used as a probe to extract information about interactions between the tubes and the graphene layers.

2.2 Raman imaging of commercial sample of graphene

More recently, our group is dedicated to prepare modified graphene samples with a myriad of molecules. However, due to the limitations of chemical procedures for the preparation of chemically modified graphene, the resulting samples are very inhomogeneous. Graphene consists in a two-dimensional single-layer sheet of sp² hybridized carbon atoms having excellent electron transport properties and very high carrier mobility and thermal conductivity. However, the interesting properties exhibited by graphene are only observed for graphene films that contain only one or a few graphene layers [3, 4]. It is optically difficult to observe low number of layers, but single-, double-, and multilayer graphenes can be differentiated by their Raman fingerprints. The graphene Raman spectrum is dominated by G band (ca. 1582 cm⁻¹) and 2D band (ca. 2685 cm⁻¹). Ferrari and coworkers [55] used Raman to characterize samples consisting of varying number of graphene layers and found that the 2D band width and shape modify with increasing the number of layers. Therefore, the spectral shape of the 2D band is representative of the number of graphene layers and can be used to determine that number. By using Raman imaging has been possible to analyze a large area of the graphene samples in order to search irregularities. For example, Figure 8 shows the Raman imaging obtained from a commercial sample of graphene (Sigma-Aldrich). The D- and G-band intensities were used as probe to see the differences in a large area of the sample.

The D and G images can be combined to form an image from the D/G ratio of intensities (red-yellow image). The Raman spectra of two spots of the sample exemplify the point; yellow regions have more intense D bands than red regions.