Synthesis and Properties of Single-Walled Carbon Nanotubes Filled with Metal Halogenides and Metallocenes

2.1 Synthesis of metal halogenide-filled SWCNTs

Metal halogenides were encapsulated inside SWCNTs by several methods. Among them are the gas phase and liquid phase approaches. The latter includes the solution and melt techniques.

The gas phase approach implies the encapsulation of a substance in a gas state inside SWCNTs. In the filling process, SWCNTs and substances are heated up to the temperature that is higher than the boiling or sublimation point of the substance. During dwelling at the synthesis temperature, the vapor of the substance condenses and diffuses inside SWCNTs. Then, the system is cooled, which leads to the crystallization of the substance. The advantages of this method are that it is rather simple, and it allows obtaining high filling ratios of nanotubes. However, it has several restrictions. Firstly, the maximal boiling or sublimation point of substances cannot exceed 1000–1200°C, because at higher temperatures the destruction of nanotubes occurs. Secondly, there should be no decomposition of the substance during the evaporation or sublimation process. Thirdly, the substance should have a high vapor pressure at synthesis temperature [13]. These restrictions limit the list of substances that can be encapsulated inside SWCNTs by this approach.

The list of introduced substances can be widened by the use of the liquid phase approach. The parameters of the filling process can be broadly varied. This approach is the most popular filling technique of nanotubes.

The solution method implies the filling of SWCNTs with a chosen substance dissolved in a solvent. During the filling process, several parameters are important. For the successful filling, the solubility of the substance in the solvent should be high, whereas the surface tension coefficient of the solution and its viscosity should be low. The choice of appropriate solvents allows encapsulating different substances inside SWCNTs. However, the solution method has several disadvantages. The first of them is the contamination of filled SWCNTs with solvent molecules. This can be critical for further characterizations of filled SWCNTs. The second is low filling ratios of SWCNTs, which usually do not exceed 30%. The third is inhomogeneous morphology of introduced substances inside SWCNTs [13].

These disadvantages are absent in the melt method. This method implies the filling of SWCNTs with molten salt. In a typical experiment, opened SWCNTs are mixed with an excessive amount of salt in quartz ampoules. The ampoules are evacuated, sealed, and heated up to a temperature above the melting point of the salt. The liquid salt is pulled inside SWCNTs by capillary forces. After dwelling at high temperature for some time, the SWCNTs are saturated with filling ratios up to 90%. The final cooling rate is a way to control the crystallinity of the solidified salt inside SWCNTs. Very low cooling rates are required for obtaining one-dimensional nanocrystals. Several parameters are important for the successful filling. Firstly, the melting temperature of the salt should not exceed 1000–1200°C, because at high temperatures, the nanotubes start to degrade. Secondly, the surface tension coefficient of the melt of the salt and its viscosity should be low enough [13].

The melt method is the most popular method of the filling of SWCNTs with metal halogenides. It was applied for the filling of nanotubes in all papers dedicated to the investigation of the electronic properties of metal halogenide-filled SWCNTs.

2.2 Electronic properties of metal halogenide-filled SWCNTs

In the literature, there are the reports on the investigation of the electronic properties of SWCNTs filled with MnCl₂, MnBr₂ [45, 46], FeCl₂, FeBr₂, FeI₂ [47], CoBr₂ [48], NiCl₂ [49], NiBr₂ [49, 50], CuCl [51, 52], CuBr [51], CuI [51, 53, 54], ZnCl₂ [55, 56], ZnBr₂, ZnI₂ [55], RbI, RbAg₄I₅ [57], AgCl [58, 59, 60, 61], AgBr, AgI [58], CdCl₂ [56, 62, 63], CdBr₂, CdI₂ [62], SnF₂ [64], TbCl₃ [56, 65, 66], TbBr₃, TbI₃ [66], PrCl₃ [65, 67], ErCl₃ [68], TmCl₃ [65, 69], and HgCl₂ [70]. The characterization of the electronic properties of the filled SWCNTs was performed by three main techniques: OAS, Raman spectroscopy, and XPS. These three methods are complementary, and they give a comprehensive picture of the modification of the electronic properties of SWCNTs upon their filling. OAS gives information about structure-dependent optical transitions of nanotubes. Raman spectroscopy allows studying the vibronic properties of SWCNTs. XPS investigates the Fermi level shift and bonding environment in filled SWCNTs.

Figure 1 shows the typical OAS spectrum of metal halogenide-filled SWCNTs by an example of the ZnI₂-filled nanotubes in comparison with the spectrum of the pristine SWCNTs [55]. The spectrum of the pristine SWCNTs includes the characteristic peaks corresponding to optical transitions between the first (E^S₁₁) and second (E^S₂₂) van Hove singularities (vHs) in the valence and conduction band of semiconducting SWCNTs and the first vHs (E^M₁₁) of metallic nanotubes. In the spectrum of the filled SWCNTs, there is the suppression of the E^S₁₁ peak, which corresponds to the canceling of the optical transitions between the first vHs of semiconducting SWCNTs. This is a result of the shift of the Fermi level of SWCNTs below the first vHs in the valence band or above the first vHs in the conduction band of semiconducting SWCNTs. Thus, the OAS data testify to the presence of the charge transfer in the filled SWCNTs; however, the direction of the charge transfer cannot be determined. The same modifications were observed in the OAS spectra of SWCNTs filled with FeCl₂, FeBr₂, FeI₂ [47], CoBr₂ [48], ZnCl₂, ZnBr₂ [55], AgCl, AgBr, AgI [58], CdCl₂, CdBr₂, CdI₂ [62], CuCl, CuBr, CuI [51], CuCl [52], and TbCl₃ [56].

Raman spectroscopy allows obtaining further information about the charge transfer in the filled SWCNTs. A Raman spectrum of SWCNTs includes two main characteristic bands: radial breathing mode (RBM), which corresponds to radial vibrations of carbon atoms, and G-band, which belongs to longitudinal and tangential vibrations of carbon atoms [71]. Figure 2 demonstrates the RBM and G-bands of Raman spectra of CdCl₂-filled SWCNTs in comparison with the spectra of the pristine nanotubes acquired at seven different laser wavelengths between 458 and 785 nm [63]. It is visible that all spectra of the filled SWCNTs are significantly modified as compared to the spectra of the pristine SWCNTs. The shift and change of the relative intensity of the peaks of the RBM band, the shift of the peaks of the G-band, and the change in the profile of the G-band are observed. These modifications are common for metal halogenide-filled SWCNTs. They were reported for SWCNTs filled with MnCl₂, MnBr₂ [45, 46], FeCl₂, FeBr₂, FeI₂ [47], CoBr₂ [48], NiCl₂, NiBr₂ [49], CuCl [51, 52], CuBr [51], CuI [51, 53, 54], ZnCl₂ [55, 56], ZnBr₂, ZnI₂ [55], CdBr₂, CdI₂ [62], SnF₂ [64], RbAg₄I₅ [57], TbCl₃ [56, 65, 66], TbBr₃, TbI₃ [66], TmCl₃ [65, 69], PrCl₃ [65, 67], AgCl [58, 59, 60, 61], AgBr, AgI [58], and HgCl₂ [70].

The fitting of the RBM and G-bands of Raman spectra with individual components allows investigating in detail the observed modifications. Figure 3 shows the fitting results of the spectrum of the TbBr₃-filled SWCNTs in comparison with the spectrum of pristine nanotubes acquired at laser wavelength of 633 nm [66]. The RBM-band of the pristine SWCNTs is fitted with two components at 156 and 172 cm⁻¹, which correspond to the nanotubes with diameters of 1.5 and 1.4 nm, respectively [72]. The G-band of the pristine SWCNTs is fitted with three components. The peak at 1540 cm⁻¹ (G⁻) belongs to longitudinal phonon in metallic SWCNTs, and the peaks at 1567 and 1591 cm⁻¹ (G⁺_TO and G⁺_LO) are assigned to tangential and longitudinal phonons in semiconducting SWCNTs, respectively [73]. The RBM band of the filled SWCNTs is fitted with two components, as the spectrum of the pristine SWCNTs. However, the peak positions are shifted to 164 and 175 cm⁻¹, and also the ratio of the relative intensities of the peaks is changed from 0.32:0.68 to 0.53:0.47. These modifications are due to the changes in resonance conditions of the filled SWCNTs, which are caused by the charge transfer in the filled nanotubes. The G-band of the filled SWCNTs is fitted with three components, as the spectrum of the pristine nanotubes. However, their positions are upshifted to 1558, 1576, and 1602 cm⁻¹. This can be attributed to the p-doping of SWCNTs by the encapsulated compound. Additionally, there is the change of the profile of the G-band of the filled SWCNTs from the Breit-Wigner-Fano shape, which is typical for metallic SWCNTs [71, 74] to the Lorentzian shape, which is a fingerprint of semiconducting SWCNTs [71, 74, 75]. This is probably a result of the filling-induced transition of metallic nanotubes into semiconducting state due to the opening of a band gap in their band structure.

Although similar modifications of Raman spectra of metal halogenide-filled SWCNTs testify to p-doping of nanotubes by the encapsulated salts, the doping level varies for different compounds. In Ref. [66], the doping level of SWCNTs by introduced TbCl₃, TbBr₃, and TbI₃ was compared. Authors analyzed the modifications of RBM and G-bands of Raman spectra of the filled SWCNTs acquired at laser wavelength of 633 nm. Figure 4 shows the results of the analysis. In Figure 4(a), the relative intensities of the RBM peaks of the pristine and filled SWCNTs are presented. In the case of the pristine SWCNTs, the second peak of the RBM band has the largest intensity and the ratio of relative intensities of two RBM peaks amounts to 0.32:0.68. In the case of the TbCl₃- and TbBr₃-filled SWCNTs, in contrast, the first peak has the largest intensity and the ratio is changed to 0.55:0.45 and 0.53:0.47, respectively. In the case of the TbI₃-filled SWCNTs, the ratio of the pristine SWCNTs (0.27:0.73) is recovered. Thus, the largest differences as compared to the pristine SWCNTs are observed for the TbCl₃-filled SWCNTs, and the smallest differences are observed for the TbI₃-filled nanotubes. In Figure 4(b), the shift of the G-band peaks and the relative area intensity of the G⁻-peak of the pristine and filled SWCNTs are presented. It is visible that the largest changes as compared to the pristine SWCNTs are again observed for the TbCl₃-filled SWCNTs and the smallest changes are observed for TbI₃-filled nanotubes. For example, the shift of the G⁻-peak decreases from 22 to 19 to 16 cm⁻¹ in the line TbCl₃-TbBr₃-TbI₃. The relative intensity of the G⁻-peak decreases from 0.70 for the pristine SWCNTs to 0.01 for TbCl₃, 0.11 for TbBr₃ and 0.11 for TbI₃. On the basis of these data, the authors of Ref. [66] concluded that TbCl₃ causes the largest doping of SWCNTs, whereas TbI₃ results in the smallest doping.

Authors also investigated the influence of metal cation of metal halogenides on the changes of the electronic properties of SWCNTs. The above-described analysis of RBM and G-bands of Raman spectra was conducted for SWCNTs filled with FeBr₂, CoBr₂ and NiBr₂ [50] and TmCl₃, TbCl_3, and PrCl₃ [65]. It was revealed that the doping level of SWCNTs increases in the lines with NiBr₂-CoBr₂-FeBr₂ and PrCl₃-TbCl₃-TmCl₃.

Thus, Raman spectroscopy allowed revealing p-doping of SWCNTs and elucidating the differences in the doping efficiency of different metal halogenides on nanotubes. However, similar to OAS, Raman spectroscopy does not give quantitative information on the doping level of SWCNTs.

XPS spectroscopy allows quantifying the doping level of nanotubes. Figure 5(a) presents the typical C 1s XPS spectrum of the metal halogenide-filled SWCNTs by an example of MnCl₂-filled nanotubes in comparison with the spectrum of the pristine SWCNTs [46]. The spectrum of the pristine SWCNTs is a single peak positioned at a binding energy of 284.38 eV. The spectrum of the filled nanotubes can be fitted with three components. The component I resembles the position and full width at half maximum of the pristine SWCNTs and it is assigned to the unfilled SWCNTs. The component II is downshifted by 0.43 eV relative to component I. This is attributed to an increase in the work function of the filled SWCNTs, which is caused by the charge transfer-induced downshift of the Fermi level of nanotubes. The origin of the component III is possibly caused by local interactions of carbon atoms of SWCNTs with atoms of the incorporated salt. Similar modifications were observed in the C 1s XPS spectra of SWCNTs filled with MnBr₂ [45, 46], FeCl₂, FeBr₂, FeI₂ [47], CoBr₂ [48], NiCl₂, NiBr₂ [49], ZnCl₂ [55, 56], ZnBr₂, ZnI₂ [55], CdCl₂ [56, 62], CdBr₂, CdI₂ [62], AgCl, AgBr, AgI [58], CuCl, CuBr, CuI [51], RbAg₄I₅ [57], TbCl₃ [56], TmCl₃ [69], and PrCl₃ [67]. The measured shifts of the component II relative to the component I amounted to 0.3–0.4 eV, and they were attributed to p-doping of SWCNTs by the encapsulated compounds.

Figure 5(b) compares the shifts of the component II relative to component I for SWCNTs filled with MnCl₂, FeCl₂, CoCl₂, NiCl_2, and ZnCl₂ [46, 47, 49, 55]. It is visible that the shift decreases in the line with MnCl₂-FeCl₂-CoCl₂-NiCl₂-ZnCl₂. This may testify that the doping level of SWCNTs decreases in this line. This conclusion is in agreement with the above-discussed data of Raman spectroscopy, which showed that among bromides of Fe, Co, and Ni, the doping efficiency decreased in the line with FeBr₂-CoBr₂-NiBr₂.

Thus, a comprehensive characterization of the electronic properties of SWCNTs filled with halogenides of 3d-, 4d-, and 4f-metals by OAS, Raman spectroscopy, and XPS showed that they lead to p-doping of SWCNTs accompanied by the downshift of their Fermi level. The differences in the doping level for different metal halogenides depending on metal cation and halogen anion were revealed.

3.1 Synthesis of metallocene-filled SWCNTs

Metallocene molecules are not stable at high temperatures. The powders of substances decompose and could not be melted. At the same time, they sublime in vacuum at low temperatures. Taking into consideration this fact, researchers introduced metallocenes inside the nanotubes by the gas phase method. In a typical experiment, the SWCNTs were mixed with the powder of substances, sealed in an ampoule under vacuum, and heated at low temperatures (50–200°C) during several days. This method allowed filling SWCNTs with ferrocene [44, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86], cobaltocene [87, 88], nickelocene [89, 90, 91, 92], and cerocene [93, 94]. It should be noted that ferrocene was also incorporated inside SWCNTs by the liquid phase method using its solution in acetone [95].

3.2 Temperature-dependent inner tube growth inside metallocene-filled SWCNTs

The temperature-dependent inner tube growth inside furnace- or laser-annealed ferrocene- [76, 77, 79], cobaltocene- [87], and nickelocene-filled SWCNTs [89, 90] was investigated by Raman spectroscopy. Authors traced modifications of Raman spectra of the filled SWCNTs that occurred at increasing annealing temperature or laser power at fixed annealing time. Figure 6 presents the example of the investigation of the inner tube growth inside nickelocene-filled SWCNTs [89]. Figure 6(a) shows the RBM-bands of Raman spectra of the pristine, filled SWCNTs, and the samples annealed at temperatures between 400 and 1200°C for 2 h acquired at laser wavelength of 633 nm. The RBM-band of the pristine SWCNTs includes the peak at frequencies ranging from 125 to 160 cm⁻¹. This peak is shifted by 4 cm⁻¹ after filling with nickelocene, which is usually observed for molecule-filled SWCNTs. Annealing at 400°C and higher temperatures results in an appearance of new peaks centered at 212, 216, and 253 cm⁻¹. These peaks belong to inner nanotubes with chiralities of (12,3), (13,1) and (11,1) and diameters of 1.081, 1.064, and 0.909 nm, respectively. The gradual increase in annealing temperature leads to an increase of the intensity of the peaks. Figure 6(b) demonstrates the dependence of the relative intensity of the peak of the (12,3) and (13,1) inner tubes on annealing temperature. It is visible that the intensity increases at temperatures ranging from 400 to 700°C. It saturates at 700°C and stays almost unchanged at higher annealing temperatures. Thus, inner tubes grow in the temperature range between 400 and 700°C.

Authors of Ref. [90] compared the growth temperatures of eight different inner tubes with chiralities of (7,5), (8,4), (7,6), (10,3), (12,3), (11,5), (14,2), (12,6) inside nickelocene-filled SWCNTs. They evaluated the growth temperature as the temperature at which the intensity of the RBM peak of the inner tube reaches the half of its maximum. Figure 7(a) and (b) shows the dependences of the growth temperature of the inner tubes on their diameter and chiral angle. It is visible that the growth temperature increases with increasing tube diameter, but it does not depend on their chiral angle. The same trend was observed for inner tubes grown inside ferrocene- [77] and cobaltocene-filled SWCNTs [87].

3.3 Growth kinetics of inner tubes inside metallocene-filled SWCNTs

A detailed investigation of the growth kinetics of inner tubes inside in situ annealed nickelocene- [91], cobaltocene- [88], and ferrocene-filled SWCNTs [78] was conducted by Raman spectroscopy. Authors traced modifications of Raman spectra of the filled SWCNTs at increasing annealing time at a fixed temperature. Figure 8 presents the example of the investigation of the growth kinetics of inner tubes inside nickelocene-filled SWCNTs [91]. Figure 8a shows the RBM-bands of Raman spectra of the pristine, filled SWCNTs, and the samples annealed at 540°C for the time periods between 2 and 4094 min acquired at laser wavelength of 568 nm. The spectrum of the pristine SWCNTs includes two peaks positioned at frequencies ranging from 125 to 185 cm⁻¹. They are shifted by 10 cm⁻¹ for the filled SWCNTs. The spectra of the annealed samples include new peaks of inner tubes at frequencies between 205 and 295 cm⁻¹. The intensity of the peaks increases with increasing annealing time. Figure 8(b) demonstrates the dependence of the normalized area intensity of the RBM peaks on annealing time (growth curves) for nine different inner tubes with chiralities of (8,8), (12,3), (13,1), (9,6), (10,4), (11,2), (11,1), (9,3), and (9,2). It is visible that after an increase in the first minutes of annealing the intensity saturates and stays uncharged at further annealing. The growth curves differ for different inner tubes. The time period required for the saturation of the intensity gradually decreases with decreasing tube diameter.

The observed growth curves of inner tubes do not follow a self-exhausting growth model that was reported for the SWCNT growth in the chemical vapor deposition method. Authors of Ref. [91] modeled the growth of inner tubes by a new mathematical model including two growth rates α and β. The dependence of the amount of carbon in the form of grown inner tubes on time is expressed by the formula:

where A₀ is the initial amount of carbon that can be transformed to inner tubes (at t = 0), α is the rate that determines the fast growth of inner tubes at the beginning, β is the rate that determines the slow growth over longer annealing hours, and χ describes which parts of carbon processed with rates α and β.

The fitting of the experimental growth curves with this model (Figure 8(b)) allowed calculating two rates α and β of the growth of inner tubes. Figure 9 summarizes the calculated rates of the growth of inner tubes with chiralities of (8,8), (12,3), (13,1), (9,6), (10,4), (11,2), (11,1), (9,3), and (9,2) inside nickelocene-filled SWCNTs at different annealing temperatures [91]. Two trends are observed in these plots. Firstly, the growth rates α and β increase with increasing annealing temperature from 480 to 600°C, which is caused by the fact that the inner tube growth is a thermally activated process. Secondly, they increase with decreasing the inner tube diameter from 1.1 to 0.8 nm. This can be explained by the increased catalytic activity of smaller-diameter nanoparticles [91]. It should be noted that the growth rates α and β do not depend on the chiral angle of inner tubes. The same trends were observed for the inner tube growth inside cobaltocene-filled SWCNTs [88].

Using logarithmic plots of the growth rates, two activation energies E_α and E_β of the growth of inner tubes inside nickelocene- [91] and cobaltocene-filled SWCNTs [88] were calculated. The values of E_α and E_β for the inner tubes with chiralities of (8,8), (12,3), (13,1), (9,6), (10,4), (11,2), (11,1), (9,3), and (9,2) amounted to 2.02–2.57 and 1.23–1.84 eV in the case of nickelocene, and 1.72–2.71 and 0.46–1.59 eV in the case of cobaltocene, respectively. Two activation energies were attributed to the energy barriers for solid-state diffusion of carbon through carbidic and purely metallic catalytic nanoparticles. Figure 10(a) and (b) presents the dependences of the activation energies of the inner tube growth inside nickelocene-filled SWCNTs on the tube diameter and chiral angle [91]. It is visible that the activation energy E_α gradually decreases with decreasing the tube diameter, which was explained by the size effect, whereas E_β does not show a clear dependence. Both activation energies do not seem to depend on chiral angle of inner tubes.

The identical growth mechanism of inner tubes inside nickelocene- and cobaltocene-filled SWCNTs allowed authors of Ref. [88] to compare the rates and activation energies of the growth on Ni and Co catalysts. The activation energies E_α of the inner tube growth on two catalysts were in line with each other, whereas E_β values were larger for Ni catalyst. This was in agreement with the slightly different activation energies reported for solid-state carbon diffusion through face-centered cubic nickel and cobalt with hexagonal close packed lattice. Major differences were observed for the growth rates of inner tubes on Ni and Co catalysts at a given temperature. The temperature at which inner tubes started to grow differed by 60°C. It amounted to 480°C for Ni and 540°C for Co catalyst. As a result, at a given temperature, the growth rates of inner tubes on Ni catalyst were significantly larger than those on Co catalyst. This was explained by different thermal stabilities of nickel and cobalt carbides, and different diffusion coefficients of carbon in the two metals.

3.4 Electronic properties of metallocene-filled SWCNTs

The electronic properties of SWCNTs filled with ferrocene [82], nickelocene [89, 90, 92], and cerocene [93, 94] were studied by XPS and UPS. It was shown that the encapsulated metallocenes cause n-doping of SWCNTs. Authors of Refs. [82, 89, 90, 92, 93] investigated the modification of the electronic properties of the filled SWCNTs upon annealing. Figure 11(a) shows the UPS spectra of the pristine, nickelocene-filled SWCNTs, and the samples annealed at temperatures between 250 and 1200°C for 2 h [89]. The spectrum of the pristine SWCNTs includes π- and σ-peaks positioned at binding energies of 3.18 and 8.0 eV, respectively. The spectrum of nickelocene-filled SWCNTs demonstrates the shift of the π-peak by 0.07 eV toward higher binding energies. The annealing of the filled SWCNTs at 250°C leads to a further upshift of the π-peak by 0.18 eV. At increasing annealing temperature, the π-peak gradually shifts toward lower binding energies and reaches the position of the pristine SWCNTs at 600°C. At further increase in annealing temperature, the π-peak downshifts and reaches the maximal shift of 0.18 eV at 1200°C. The change in the position of the π-peak testifies about the change in the doping level of SWCNTs upon annealing. Authors of Ref. [89] suggested that this change is caused by three processes: (i) the chemical modification of the filler of SWCNTs, (ii) the inner tube growth, and (iii) the evaporation of the filler. The annealing of the nickelocene-filled SWCNTs leads to the formation of nickel carbide that causes the largest n-doping level of SWCNTs. As it was discussed above, at 400°C, the inner tube growth starts and nickel carbide decomposes to pure nickel. This leads to the formation of nickel-filled DWCNTs. In DWCNTs, there is p-doping of outer tube by inner tube, which leads to decreasing the n-type doping. At high temperatures, nickel evaporates, and it leaves empty DWCNTs. It explains the largest p-doping of nanotubes at 1200°C. Figure 11(b) presents the schematics showing the charge transfer in the filled SWCNTs and annealed samples. Figure 11(c) demonstrates the shift of the Fermi level, calculated number of transferred electrons per carbon atom of SWCNTs (N(e⁻/C)) and charge transfer density per nanotube length (CT(e⁻/Å)) plotted versus annealing temperature as well as the schematics of n- and p-doping in the filled and annealed samples.

Similar thermally induced modifications of the electronic properties were reported for metallicity-mixed SWCNTs filled with ferrocene [82] and cerocene [93] as well as for metallicity-sorted semiconducting SWCNTs filled with nickelocene [92].