- © 2016 E. Schweizerbart’sche Verlagsbuchhandlung Science Publishers
Natural nanostructured carbon materials from large Eurasian deposits were studied by Raman spectroscopy, complemented with high-resolution transmission electron microscopy (HRTEM) and X-ray diffraction (XRD). It is shown that natural carbonaceous materials of relatively high metamorphic grade can be divided into shungite-like and anthraxolite-like materials on the basis of the main Raman spectral parameters. Among them, the tortuosity factor characterizes the degree of curvature of the graphene layers in natural nanostructured carbon and is in good agreement with the results of HRTEM.
Disordered natural carbonaceous materials are one of the most controversial objects of mineralogy. These materials at a high degree of metamorphism have an almost pure carbon composition and a three-dimensional disordered structure with graphite-like domains. The finding of fullerene (Buseck et al., 1992) and onion-like structures (Buseck & Huang, 1985; Kovalevski et al., 1994, 2001; Beyssac et al., 2002a; Kovalevski, 2007) and the demand for new functional materials stimulated interest for such natural nanostructured carbonaceous materials. Disordered carbon materials with high conductivity similar to glassy carbon are especially interesting. Among such natural materials, a large amount of research was devoted to the carbon from shungite rocks (named shungites) from Karelia, Russia (e.g. Buseck et al., 1997). This carbon has a great application potential in various fields of technology and medicine (e.g. Melezhik et al., 2004).
Furthermore, many carbonaceous materials in the geological environment have an interplanar distance in graphite-like domains, a chemical composition, pressure (P) and temperature (T) conditions of formation and some physicochemical properties like the shungites. These materials are combined into a single classification group called anthraxolites (Uspensky et al., 1964; Cherevko, 1999; Melezhik et al., 2004; Filippov et al., 2007, 2012). The grouping is based primarily on physical and chemical properties. Anthraxolites and shungites are not soluble in organic solvents, contain less than 5 % of hydrogen, and are electrically conductive (Uspensky et al., 1964). Additionally, shungites and anthraxolites were combined into one group in other classifications that are based on various atomic ratios, for example (H/C)at, versus (O/C)at in Van Krevelen diagram (Cornelius, 1987). The supramolecular structure of shungites and anthraxolites is also similar, with globular particles which are 30–100 nm in size (Golubev, 2009). The shungite and anthraxolite deposits are formed as a result of mobilisation of primary liquid hydrocarbons. Migrating carbonaceous matter fills joints (it is referred to as vein-type bitumen), and fills rock cavities and sedimentary voids ranging in size from 1 mm to several tens cm (it is referred to as clastic-type bitumen).
Other authors (Meyer & De Witt, 1990; Kovalevski et al., 2001; Kovalevski, 2007) found fundamental structural differences between shungites and anthraxolites. Kovalevski et al. (2001) have divided natural carbonaceous materials into five groups according to X-ray diffraction (XRD) and high-resolution transmission electron microscopy (HRTEM) data. Carbonaceous materials of high degree of metamorphism were also divided into two groups according to the number and length of graphene layers in the stack and to nanodiffraction pattern (shungites and cokes versus anthraxolites and anthracites). The shungites and cokes are considered as non-graphitizing carbon (Kovalevski, 2009) following Franklin (1951), i.e. non-graphitizing at atmospheric pressure and at a temperature not higher than 2800 °C. In general, a structural criterion of differentiation between the graphitizing and non-graphitizing carbon is derived from XRD data (as the intensity (I) ratio of the (002) and (004) basal reflections), and from the HRTEM data. Non-graphitizing carbon has an I004/I002 ratio less than 0.5 (Inagaki, 1972), nanoscale porosity (Oberlin & Terrier, 1975), and ordered stacks of graphene layers with sizes less than 5 nm (Bustin et al., 1995). However, shungites and many anthraxolites are like in these criteria. Raman spectroscopy is used successfully for studying the nanostructured carbonaceous materials (Ferrari & Robertson, 2004), including natural carbonaceous materials (Beyssac et al., 2002a and b, 2003; Jehlička et al., 2003; Urban, and Pokorny 2003; Jehlička & Edwards, 2008). The method allows obtaining information about the prevailing hybridization type of carbon atoms, the proportion of defective aromatic rings, the degree of defectiveness (ordering) and length of graphene layers. For nanostructured (nanocrystalline) graphite-like carbon materials the sp2-hybridization corresponds to ordered parts of the structure, and sp3-hybridization is caused by disordered (amorphous) carbon (Ferrari & Robertson, 2004). This approach allows an evaluation of the structural state of nanostructured carbonaceous materials.
This paper presents the results of Raman spectroscopy, complemented with XRD and HRTEM, for shungite and anthraxolite samples. Earlier results of infrared spectroscopy on anthraxolite samples (Martirosyan, 2012) and XRD and HRTEM on Karelian shungite samples (Kovalevsky et al., 2001; Kovalevski, 2007) show that the studied samples underwent a relatively high degree of metamorphism. All the samples have electrical conductivity. The aim of the present work is a comparison of Karelian shungites with natural carbonaceous materials that are similar in physical and chemical properties, composition and basic structural features, through a detailed characterisation of their Raman spectra.
2. Sampling and analytical methods
The samples were collected from large deposits of carbonaceous material in Karelia, Novaya Zemlya islands, Polar Ural, Kazakhstan, Crimea, and Siberia (Fig. 1). Samples 1 to 5 were selected by Ye.A. Golubev from shungite occurrences in Karelia; samples 8 to 10 were provided by Dr. V.O. Ilchenko, Russian Petroleum Exploration Research Institute, St. Petersburg; samples 6 and 11 to 14 (from the collection of N.K. Cherevko) were provided by Dr. O.V. Martirosyan, Institute of Geology of Komi SC of RAS, Syktyvkar; sample 15 was granted by Dr. A.S. Glebashev, Central Research Institute of Geology of Industrial Minerals, Kazan. Additionally, we used a sample of commercial, synthetic glassy carbon. Geological information for each sample is summarized in Table 1.
The samples were pieces of pure carbon material ranging in size from several millimeters (for the clastic type) to a few centimeters (for the vein type). Raw samples were broken, and pieces of a few millimeters in size were taken from their inner portions. The isolated pieces were prepared as samples for XRD, Raman spectroscopy, SEM/EDS and HRTEM study.
The XRD patterns were obtained with a Shimadzu XRD-6000 diffractometer using CuKα radiation. Powder samples were placed onto a glass plate and scanned from 10 to 60° 2θ, at a speed of 1 °/min; signals were recorded at intervals of 0.05° 2θ. The absolute intensity of the 004 and the 002 diffraction lines was evaluated.
Raman spectroscopy was carried out with a LabRam HR800 instrument (Horiba, Jobin Yvon) at room temperature. The system was equipped with an Olympus BX41 optical microscope and a Si-based CCD detector (1024 × 256 pixels). A 50× objective (working distance ~3 mm, numerical aperture 0.75) was used. Spectra were recorded in the 100–4000 cm−1 range using a spectrometer grating of 600 g/mm, with a confocal hole size of 300 μm and a slit of 100 μm. As exciting radiation external Ar+ laser (514.5nm 1.2 mW), and onboard He–Ne laser (632.8 nm, 2 mW) were used. Each spectrum was the result of three accumulations with a 10 s exposure. After background correction, individual lines were deconvolved using a curve-fitting procedure from the software provided by LabSpec 5.36.
The cleaved surfaces of raw samples were studied with a Tescan Vega LMH scanning electron microscope with W heated cathode. X-ray energy-dispersive spectrometry (EDS) was used to investigate the chemical composition with AZTEC software (Oxford Instruments) processing. The microprobe was operated at an emission current of 100 μA, a specimen current of 100 pA, an accelerating potential of 20 kV, and a spot size of 100 nm in diameter for elemental analysis. Element peaks and backgrounds were measured with counting times of 120 s. The following standards were used: glassy carbon rod (for C), SiO2 (O, Si), albite (Na), wollastonite (Ca), FeS2 (S), NaCl (Cl).
Elemental analysis of H was conducted by gas chromatography with an EA1110 CHNO-S (Carlo-Erba Co, Cornaredo, Italy) instrument.
The nanostructure of anthraxolite samples from Perya (No. 8) and from Yuzhnyi (No. 10) was examined in a FEI CM300UT FEG transmission electron microscope (300 kV field emission gun, 0.65 mm spherical aberration, and 0.17 nm resolution at Scherzer defocus). The samples for HRTEM were broken into small pieces. The thin grains (20–50 nm) were placed onto carbon films on copper grids.
To eliminate the delocalization effect in HRTEM experimental images, which is caused by the microscope contrast transfer function, we obtained aberration-corrected high-resolution micrographs near the Gaussian focus value of the objective lens (Brydson, 2011) in a FEI Titan Themis 200–80 transmission electron microscope operated at 80 kV and equipped by spherical aberration (Cs) corrector and FEI Ceta 16M CMOS camera. Thin foils for HRTEM investigations were prepared from particles of shungite samples from Maxovo (No. 2) and Nigozero (No. 3) by mechanical thinning followed by ion-beam final milling in Gatan PIPS Model 691, the accelerating voltage of the Ar beam being gradually changed from 4 to 0.5 keV.
The results of XRD (Fig. 2) and elemental analysis of C, H and O atoms are summarized in Table 2 and show that the samples have similar structural parameters and chemical composition. All samples plot into the “shungite area” in the Van Krevelen diagram (H/Cat ≤ 0.02; O/Cat ≤ 0.03). The XRD results indicate that the samples are non-graphitizing carbon (I004/I002 < 0.5).
Figure 3 shows the band parameters of the Raman spectra for the samples at a wavelength of 514.5 nm. Similar relationships were obtained for the wavelength of 632.8 nm, but at a wavelength 514.5 nm the second-order bands were more clearly visible. Therefore, the results of the work are based on the analysis of Raman spectra at the wavelength of 514.5 nm.
The Raman spectra of the studied samples are typical for disordered carbon materials. Decomposition of first-order scattering using convolution of Lorentz and Gauss functions revealed that the spectra usually represent a superposition of five overlapping wide lines, labeled G, D1, D2, D3 and D4 following Beyssac et al. (2002a and b, 2003). In general, the first-order Raman spectrum for poorly organized carbonaceous material consists of five peaks (Fig. 2a,c): the fundamental G-band (centered in the area 1580–1600 cm−1); the D1-band (1330–1350 cm−1) – in publications often called the D-band (Ferrari & Robertson, 2004); the D2-band (1610–1620 cm−1) – according to Ferrari the D’-band, which is a shoulder of the G-band; the D3-band (centered at about 1500 cm−1) – sometimes called the A-band (Ungar et al., 2005); and the D4-band (1180–1200 cm−1). In the second-order spectrum shown also in Fig. 2b and d, the overtones and combined tones of the bands are resolved at about 2500 cm−1 (D4 + D1), 2700 cm−1 (2D1 or S1 after Beyssac), 2850 cm−1 (D3 + D1), 2950 cm−1 (D1 + G or S2 after Beyssac), and 3230 cm−1 (2D2).
The G-band is related to the fundamental E2g phonon mode (Ferrari & Robertson, 2004), which is responsible for the vibrations of carbon atoms within the graphitic layer. The position and half-width of the G band are considered as a measure of disorder, caused by a distortion of the hexagonal rings and chains of carbon atoms. A Raman up-shift of the G-band with respect to its position for high-crystallinity graphite (at 1582 cm−1) is associated with the appearance of the D2-band for a small crystallite size (a few nanometers).
The D1-band intensity was originally associated with the degree of graphite disorder. The D1-mode relates to the finite size of the crystallites and disappears in an ideal crystal (Reich & Thomsen, 2004). The problem of its origin has not lost its topical significance. Initially it was believed that it is a breathing mode of the sp2-bonded carbon atoms in the hexagonal aromatic rings (Tuinstra & Koenig, 1970). Ferrari (2007) established that the D1 (D) band is associated with the longitudinal optical phonon (LO phonon) around the point K (the Brillouin zone). It is activated by double-resonance and its scattering depends on the excitation energy by Kohn anomaly at the point K.
The origin of the D4-band is still under discussion. It was supposed that this line is associated with curved graphene layers in carbon materials (Tan et al., 2004), as in transpolyacetylene-(TPA)-like structures, which are formed on the zigzag edges or in an onion-like structures (Ferrari & Robertson, 2001), or in a tape, similar to the glassy carbon structure. Dippel & Heinzenberg (1999) observed the band at 1190 cm−1 in the Raman spectra of flame soot and tentatively attributed it to sp2–sp3 bonds or C–C and C=C stretching vibrations in polyene-like chains.
The broad D3-band in the region of 1400–1500 cm−1 is usually matched by amorphous carbon (Ungar et al., 2005), which is present, for example, in the form of interstitial disordered carbon atoms with sp3-bonding outside or inside the plane of the aromatic rings (Cuesta et al., 1994). The relative integrated intensity of this band (ID3/IG), together with a conventional ratio (ID1/IG), can be used as a parameter to estimate the degree of ordering of carbon materials (Yoshikawa et al., 1988).
Figure 4 shows to difference between shungite and anthraxolite on the basis of the main Raman spectral parameters.
The ID1/IG ratio appears to be the most important parameter of the Raman spectra even if, for integrated intensity, we have not found a difference between shungites and anthraxolites. However, for the ID1/IG ratio in absolute units, there is a fundamental difference between shungites spectra and anthraxolites spectra (Fig. 4c). Shungite samples have an ID1/IG ratio ≫ 1, except for the sample from Shardonskie islands (No. 6), whereas for anthraxolite samples the ID1/IG ratio is less than 1, except for the sample from Kozhim River (No. 12). These contrasting ratios probably reflect a difference in size of the ordered domains. Additionally, shungites have an ID1/IG ratio similar to that of glassy carbon. The half-width of the D1-band and the position of the G-band also indicate essential distinction between shungites and anthraxolites. Indeed, the half-width of D1-band, which is correlated with the degree of disorder of the carbon structure (Sadezky et al., 2005; Sato et al., 2006), clearly distinguishes shungites from anthraxolites (Fig. 4c). The G-band position depends on the crystallite size (Ferrari & Robertson, 2004). Because crystallite size (La) is traditionally determined from Raman spectra by the (ID1/IG)−1 ratio (Tuinstra & Koenig, 1970; Cancado et al., 2006), anthraxolites should have a greater crystallite size than shungites (Fig. 4c).
Comparison of the disordered structural component (with sp3-bonds) also allows distinction between shungites and anthraxolites. Shungites (and glassy carbon) are distinguished from anthraxolites both by the ID3/IG ratio and the ID4/IG ratio (Fig. 4b,d). This indicates that the amorphous phase contents in shungites and glassy carbon is less than in anthraxolites.
The IS1/IG ratio is another important parameter of the Raman spectra. In a number of studies it is proposed to relate this ratio to the coefficient of tortuosity (Rtor) of graphene layers (Tan et al., 2001; Larouche & Stansfield, 2010): (1)
A normalization parameter D(λ) = 1 was set for a wavelength of 514.5 nm (Larouche & Stansfield, 2010). The coefficient of tortuosity for the samples studied here is shown in Table 2; it is generally higher for the shungites than for anthraxolites. A correlation of Rtor with the G-band position is apparent in Fig. 5, revealing the influence of tortuosity (curvature) of graphene layers on the degree of order within the graphene stacks. Additionally, the curvature of graphene layers influences the nanocrystallite size (~Rtor · La) as evaluated from the Raman spectra (Larouche & Stansfield, 2010). Therefore, we tested the Raman spectroscopy results by HRTEM, on samples showing different coefficients of tortuosity.
The HRTEM images of shungites from Maksovo (Fig. 6a) and from Nigozero (Fig. 6b) and anthraxolites from Novaya Zemlya (Fig. 7a,b) show different degrees of ordering of graphite-like nanostructures and different sizes of crystallites. The size of crystallites was estimated in HRTEM images as areas occupied by at least three periodically located fringes, the distance of 0.34 nm between the fringes corresponding to the (002) interplanar spacing of graphene planes.
Figure 6 shows that shungites consist of numerous crystallites with a linear size up to 10 nm. Statistical analysis showed that the size of crystallites in shungite from Maksovo is ~30 % larger than that in Nigozero shungite. The curvature of (002) planes in Maksovo shungite is also slightly lower than in Nigozero shungite. The XRD studies (Kovalevski et al., 2001; Kovalevski, 2007) also showed that samples from Maksovo and Chebolaksha are the most structurally ordered among all shungites.
The HRTEM study demonstrates that the level of ordering in anthraxolites (Fig. 7) is much lower than in shungites (Fig. 6), and that the crystallite size in anthraxolites hardly reaches 3 nm. The anthraxolite samples from Novaya Zemlya islands (No. 8 and No. 10) have a relatively small tortuosity ratio and mostly straight graphite-like nanocrystallites (Fig. 7).
It is interesting to note that such ordering regions in HRTEM micrographs of anthraxolite and shungite samples is inherent in artificial, carbon-based materials such as pyrolytic carbon, in which the interplanar distance between (002) basal planes is also 0.34 nm (Kukin et al., 2004).
The result concerning shungite samples agrees with previously published HRTEM data by Kovalevski et al. (2001) and Kovalevski (2007): the HRTEM images show that the curved carbon layers are more frequent in shungites from Zazhogino, Nigozero and Shunga. The agreements of HRTEM and Raman spectroscopy results (with respect to the coefficient of tortuosity) show that the Rtor coefficient has a physical meaning and is useful for the study of homogeneous samples of natural disordered carbonaceous materials with a broad range of tortuosity values. It is known that the curvature of the carbon layer influences its electronic properties, and from this point of view Rtor is an important structural parameter for the evaluation of physical properties of carbonaceous materials (Wakabayashi et al., 1999; Tomita et al., 2001; Golubev, 2013).
The Raman spectra of carbon materials have been described within a three-stage model of increasing disorder (Ferrari & Robertson, 2004). The authors called this the amorphization trajectory, consisting of three stages: (1) graphite → nanocrystalline graphite (nc-G); (2) nc-G → amorphous carbon (sp2 a-C); (3) a-C → tetrahedral amorphous carbon (ta-C). The transformation stages of carbon materials in the reverse sequence can be considered as the ordering trajectory for the natural process of bitumen carbonization. Following Ferrari & Robertson (2004), we believe that the parameters of the Raman spectra (Fig. 4) characterize the stage of transition from amorphous carbon (a-C) to nanocrystalline graphite (nC-G). The structural transformations from anthraxolites to shungites occur in the direction of a decrease of the sp3-bonded phase, with G-band displacement as a result of the increasing proportion of the D2-band. Note that in the natural carbonaceous materials the G-band position is not only controlled by structural factors, but also by the presence of impurities. Thus, if the G-band position and the ID1/IG ratio fall into one point on the trajectory of amorphization (ordering) of graphite for shungite samples, then the G-band position and the ID1/IG ratio fall on different points of the trajectory for anthraxolite samples.
A sample of shungite from Shardonskie Islands (No. 6) has Raman spectral parameters that are similar to the parameters of anthraxolites (Fig. 4). This can be explained by the fact that this shungite sample and anthraxolites were formed during relatively high-temperature hydrothermal processes in the catagenesis field (Table 1), whereas the other shungites were transformed at still higher temperatures, during greenschist-facies metamorphism and by contact with volcanic intrusions, under higher pressure conditions for some samples.
Natural carbonaceous materials of relatively high degree of metamorphism can be divided into shungite-like and anthraxolite-like materials on the basis of their main Raman spectral features. The G-band position, the half-width of the D1-band, the ID4/IG, ID3/IG and ID1/IG (in absolute units) ratios can serve for distinction. Shungites have a smaller proportion of disordered (sp3-bonds) phase as comparied to anthraxolites. Unlike anthraxolites, the shungites are analogues of glassy carbon in their Raman spectra. Although Raman spectroscopy suggests that the crystallite size in anthraxolites is a little more than in shungites, the results of HRTEM do not support this inference. This discrepancy is probably due to the influence of the tortuosity and curvature of graphene layers in shungites. The tortuosity factor Rtor, derived from Raman spectra, is useful for the study of disordered carbonaceous materials with widespread tortuosity.
In general, the main formation process of the anthraxolite-like materials is a relatively high-temperature hydrothermal process, but still in the catagenesis realm. The shungite-like materials were transformed at higher temperatures of the greenschist facies and in the zone of influence of volcanic intrusions, occasionally under higher pressure conditions.
Financial support for the research was provided by RFBR (Grant No. 15-05-04369). Authors thank S.S. Shevchuk for the SEM/EDS research and B.A. Makeev for XRD patterns. We also thank reviewers of this work for their remarks.
- Received 15 November 2015.
- Modified version received 25 January 2016.
- Accepted 15 February 2016.