Anastasios
Karakassides
*a,
Hirotaka
Inoue
ab,
Peng
Liu
a,
Zhenyu
Xu
a,
Ghulam
Yasin
a,
Hua
Jiang
a and
Esko I.
Kauppinen
*a
aDepartment of Applied Physics, School of Science, Aalto University, Puumiehenkuja 2, 02150 Espoo, Finland. E-mail: anastasios.karakasidis@aalto.fi; esko.kauppinen@aalto.fi
bSumitomo Electric Industries, Ltd, 1-1-3 Shimaya, Konohana-ku, Osaka 554-0024, Japan
First published on 24th April 2025
In response to the growing demand for novel catalyst designs for the selective growth of single-walled carbon nanotubes (SWCNTs), this study explores the use of ferrocenium (the oxidized state of ferrocene) as a new catalyst precursor for the first time. Utilizing the floating-catalyst chemical vapor deposition (FC-CVD) method, SWCNTs were synthesized and characterized through analytical transmission electron microscopy selected area electron diffraction (TEM SAED) and optical techniques (Raman spectroscopy and UV-vis-NIR absorption). The introduction of ferrocenium led to an enhancement in the metallicity of the nanotubes, increasing the proportion of metallic SWCNTs to 43.1%, while also broadening the nanotube mean diameter from 1.84 nm to 2.10 nm. The key factor behind this improvement lies in the positive charge of Fe3+ in ferrocenium, which has been shown to stabilize metallic nanotube formation. These findings highlight the pivotal role of catalyst charge in controlling SWCNT chirality and electronic properties, paving the way for more precise control in nanotube synthesis for applications in nanoelectronics and materials science.
Chirality is a critical factor in determining the properties of SWCNTs because it directly influences whether they behave as semiconductors (s-SWCNTs) or metals (m-SWCNTs), affecting their potential applications.7 About two-thirds of SWCNTs with specific chiralities exhibit semiconducting behavior, making them ideal for use in transistors,8 solar cells,9 and flexible electronics,10 while one-third behaves like metals, suitable for applications such as interconnects in circuits11 and conductive composites.12 In addition to electronic properties, chirality also affects the mechanical strength of SWCNTs13 and their growth rate during synthesis.14 Given the significant impact chirality has on both the functional properties and production efficiency of SWCNTs, controlling it during synthesis is crucial. Achieving selective control of chirality would enable the tailored design of SWCNTs for specific applications, unlocking their full potential in fields such as electronics, energy storage, and materials engineering.
Lately, one promising route for selective chirality during synthesis that has gained attention is the engineering of novel catalysts that possess a specific structure. Traditionally, these catalysts have comprised combinations of two low-melting temperature transition metals (such as Fe, Ni, or Co), but recent developments have explored the utilization of one high-melting temperature metal (or non-metal) such as Re,15 Rh,16 W,17,18 WC,19 and Mo2C20 alongside a low-melting one or even alone. Depending on the metals selected, the pursuit of bimetallic catalyst combinations involves modifying the morphology and composition of the catalyst and its final melting temperature. This change in melting point may therefore have an impact on the cap formation and carbon incorporation at the open ends of SWCNTs as a result of the different physical states of the metals of the catalyst. These differences in the early formation of carbon rings are closely related to the surface structure of the catalyst, demonstrating a “template effect” marked by a predominance of hollow site occupancy.21
More specifically, Feng Zhang et al.22 managed to synthesize SWCNTs with a narrow-diameter distribution of about 1.35 nm by utilizing a bi-metallic CoRu catalyst supported on an SiO2 substrate. Their results revealed an enrichment of armchair nanotubes by almost 12%, based mostly on Raman spectroscopy characterization. Xin Li et al.15 synthesized m-SWCNTs with an abundance of 80% using a CoRe4 catalyst possessing a hexagonal close-packed structure. According to the authors, the result of their synthesis was achieved because of the lower binding energy between their catalyst and the m-SWCNTs, which promoted their growth rate. Similarly, W6Co7 alloy nanoparticles have shown promising results on selective chiral growth of SWCNTs, where Feng Yang et al.23 reported an abundance of 94.4% (12,6) nanotubes based on Raman and optical spectroscopy methods. According to these authors, by utilizing functional theory simulations, their alloy nanoparticles possessed the perfect geometrical match for (12,6) nanotubes while exhibiting extremely low matches for all other chiralities. These advancements in catalyst design and synthesis methods have paved the way for further exploration of various strategies aimed at achieving selective growth of SWCNTs, such as the introduction of etchants during the synthesis process, commonly referred to as selective etching.
During the last few years, various efforts have been made to achieve the desired selective growth of SWCNTs, including introduction of etchants during the synthesis24–30 (known as selective etching). Selective etching involves the introduction of an oxidizing environment during the synthesis process (methanol, water vapor, oxygen, combinations of H2O and CO2, etc.), in which it is possible to etch smaller diameter reactive m-SWCNTs.31 Moreover, it has been shown that H2O vapor in specific amounts could significantly affect the nucleation of s-SWCNTs, while it should be noted that many gaseous by-products of the decomposition of various alcohols (ethanol and IPA) could include H2O and/or CO2 that contribute to this etching as well.32 Apart from the efforts to selectively etch m-SWCNTs, significant attention has also been given to etch s-SWCNTs. For instance, Yongping Liao et al.33 have grown SWCNT films with a metal/semiconductor ratio of 46.3% by introducing various amounts of CO2 into floating-catalyst CVD (FC-CVD) using CO as a carbon source. They suggested that CO2 could affect the CO disproportionation and nucleation modes of SWCNTs, resulting in various diameter ranges of the SWCNTs. Moreover, Qiangmin Yu et al.34 explored the enrichment of m-SWCNTs using NO2 to selectively etch s-SWCNTs, achieving purity values higher than 90%. Besides selective etching, recent studies have also emphasized controlling chirality by focusing on the chemical state of catalyst precursors.
One recent research study by Sook Young Moon et al.35 explored the effect of oxygen-containing functional groups in ferrocene (FeCp2)-derived catalyst precursors. They utilized ferrocenemethanol (–OH functional group), acetylferrocene (CO functional group), and 1,1′-diacetylferrocene (C
O functional group) to investigate their influence on the synthesis of SWCNTs. According to these authors, the pyrolysis and release of Fe ions were delayed by the functional groups of the ferrocene-derived precursors compared to ferrocene alone, leading to variations in SWCNT growth. The variations in SWCNT growth caused by the delayed release of Fe ions from ferrocene-derived precursors highlight the importance of accurately assessing the chirality of the resulting nanotubes. Given the sensitivity of CNT growth based on catalyst precursor chemistry, more efforts need to be made to investigate the effect of this complex chemistry on the structure of synthesized nanotubes (chirality, diameter, crystallinity etc.).
In this context, we considered ferrocenium, the oxidized form of ferrocene, an iron precursor frequently used in FC-CVD, to be a worthy material for investigation. It has been widely employed in various fields due to its unique redox properties, stability, and ability to act as an electron-transfer mediator.36 It has found applications in areas such as electrochemical sensors37 and materials science,38 particularly as a redox catalyst in polymerization reactions.39,40 These applications leverage ferrocenium's ability to facilitate electron transfer and its compatibility with a range of chemical environments. Despite its versatile use in other domains, ferrocenium has never been explored as a catalyst for single-walled carbon nanotube (SWCNT) synthesis. This gap is particularly intriguing given the critical need for selective growth of metallic SWCNTs (m-SWCNTs) with high yield and precision.
In particular, the evaluation of chirality is quite challenging and more attention needs to be given to the reliability of reported results that have not coupled all of the available techniques (e.g. TEM SAED, Raman spectroscopy, and UV-vis-NIR spectroscopy). It is crucial to pay careful attention to the accuracy of these chirality evaluation techniques, as they directly impact the interpretation of growth outcomes. Raman spectroscopy, one of the most widely used methods for evaluating chirality, has been employed in many studies as mentioned before, which reported high yields of m-SWCNTs.19,41 However, some reports suggest that Raman spectroscopy may have low reliability for quantitative evaluation of chirality42 and diameter.43
Motivated by these different results and addressing the pressing challenges in catalyst design, we investigated ferrocenium for the first time as a novel catalyst to enrich m-SWCNT growth. By utilizing the FC-CVD method, SWCNTs were grown in high yields and the nature of the nanotubes was assessed through analytical TEM SAED studies supported by optical characterization techniques (Raman spectroscopy and UV-vis-NIR light absorption).
The mechanism of ferrocene oxidation can be found in the literature in acetonitrile instead of ethanol and is described by the following:
CuII(NO3)2 + FeCp2 → CuI(NO3)2 + FeCp2+ | (1) |
According to the literature,47 the nitrate ion in this reaction mildly slows down the oxidation process, which is in good agreement with our results in that from the moment the oxidation begins it progresses slowly over time for more than 24 h, as previously mentioned. To monitor the oxidation reaction of ferrocene, both UV-vis-NIR spectroscopy and pH measurements over time were employed. Initially, the pH values of the prepared solutions were measured and then the same solutions were analysed using a light absorption spectrometer. From Fig. 1e, it can be seen that the pH was not stable but rather changes over time, revealing the progress of the oxidation reaction. Moreover, the pH value of a normal ferrocene–ethanol–thiophene solution was measured as well at 6.6, just for comparison purposes. The initial solution of yellowish colour exhibited a pH value of about 4.4 and after 24 h this value gradually decreased to 2.8, where saturation was observed. This was a clear indication that the oxidation reaction occurs slowly over 24 h in this solution. However, more analytical experiments were needed to justify the pH results.
For this reason, UV-vis-NIR spectroscopy was employed. The solution with Cu/Fe of 0.3 atomic ratio was analysed using the spectrometer during oxidation at 3 different stages: (i) the initial solution just after mixing all precursors (yellowish phase); (ii) after 30 min of rest (light green phase); and (iii) after 24 h of rest (dark green phase). Fig. 2a shows the light absorption results of all the samples. As can be seen from the spectra, the band at around 435 nm can be assigned to the characteristic ferrocene signal in ethanol solution, while the bands at 620 nm and 760 nm are assigned to ferrocenium48 and copper(II) nitrate, respectively.49 All the remaining bands can be attributed to the ethanol solution. During the oxidation of ferrocene, the intensity and area of the ferrocene band decreased (Fig. 2b), while that of the ferrocenium band increased, illustrating the progression of the oxidation process over time. By measuring and comparing the areas of these two characteristic bands (Fig. S1†), it is possible to determine the oxidation ratio (%) of the solution over time as follows:
![]() | (2) |
![]() | ||
Fig. 2 (a) UV-vis-NIR absorption spectra of all samples, (b) UV-vis-NIR spectra of all samples during the oxidation process, and (c) oxidation ratio (%) of all samples over time. |
Using eqn (2), the Cu/Fe 0.3 sample exhibited a maximum oxidation value of approximately 35% during 48 hours of testing. Initially, data points were collected every 10 minutes and after the first hour, data were recorded every hour (Fig. 2c). In contrast, the Cu/Fe 0.5 sample reached an oxidation value of nearly 47% over the same time frame, which was expected due to the higher copper content in this second sample, contributing to the increased levels of oxidized ferrocene over the same period.
The first and second interband transitions of semiconducting nanotubes (S11 and S22) and the first interband transition of metallic nanotubes (M11) in the optical absorption spectra are the characteristic features that differentiate SWCNTs from other nanotube assemblies.50–52 The wavelength range of these specific bands was used to estimate the diameters of SWCNTs, where S11 (∼2100–2500 nm), S22 (∼1200–1400 nm) and M11 (∼800 nm) peaks are clearly visible in the absorption spectra, as shown in Fig. 3a. A redshift of the peak position of the bands can be observed with the addition of copper(II) nitrate trihydrate to the solution of precursors, indicating an increased mean diameter of SWCNTs. To estimate this increased mean diameter, the data were analysed utilizing a MATLAB code developed by Tian et al.44 The SWCNTs grown from ferrocene only exhibited a mean diameter of about 1.84 nm, while Cu/Fe 0.3 and Cu/Fe 0.5 showed mean diameters of 2.10 nm and 2.09 nm, respectively (Fig. 3b–d).
To further investigate the light absorption spectra, TEM was employed (Fig. 4a–c). Samples were directly collected on TEM grids placed on the membrane in the collection filter area. To ensure that the sample's density did not impede electron transmission, a short collection time of 6–7 seconds was chosen. Subsequently, the samples were analysed using a 200 kV acceleration voltage in the microscope to obtain high-resolution micrographs. From the analysis of the micrographs, it can be seen that the Fe sample exhibited the smallest mean bundle diameter of 11.20 nm, while the Cu/Fe 0.3 and Cu/Fe 0.5 samples exhibited mean diameters of 18.40 nm and 19.80 nm, respectively (Fig. 4d–f). There is an obvious increase (almost double in the case of Cu/Fe 0.5) in the mean bundle diameter with the introduction of copper(II) nitrate trihydrate into the synthesis process.
This bundle increment did not translate into much worse resistance values of the thin films, as can be seen in Fig. S3a.† It was expected that the samples with larger bundle diameter should exhibit decreased resistance values. It is well proven that reducing the concentration of CNTs during the FC-CVD process can lead to the formation of thinner SWCNT bundles, which is one of the most effective strategies for enhancing the conductivity of SWCNT films.53,54 Apart from this, larger bundles tend to absorb more light because they have a higher density of nanotubes, providing more material to interact with the light, which leads to worse resistance values in the thin films from the moment the transmittance of the films is used for normalization of the resistance values. It should also be noted here that the specific yield of the samples (Fig. S3b†) did not dramatically change, showing only a slight increase for the Cu/Fe 0.5 sample, when benchmarking it against the other two samples (Fe and Cu/Fe 0.3).
Apart from the observation of SWCNTs, TEM was also used to observe the catalyst nanoparticles. Remarkably, the analysis revealed an interesting result: the distribution of catalyst nanoparticles remained unchanged despite the addition of copper(II) nitrate trihydrate to the synthesis process (Fig. 5a–c). It can be seen that the mean nanoparticle diameter was about 7.7–8.0 nm for all samples, demonstrating that there is no impact from the addition of copper(II) nitrate trihydrate to the synthesis process. This can be explained by the fact that the chosen heating line temperature of 250 °C is significantly higher than the initial thermal decomposition temperature of the copper precursor, which is around 170–180 °C,55 and therefore, it is likely that the copper precursor decomposes before reaching the reactor. Instead, the copper oxide formed during decomposition is expected to deposit on the walls of the heating line or at the entrance of the reactor, where temperatures are even higher. Consequently, these oxide nanoparticles do not reach the reactor or participate in the SWCNT synthesis process.
To verify this hypothesis, TEM EDS (Fig. 6) along with SEM EDS mapping (Fig. S4†) and XPS analysis (Fig. S5†) were performed to check for the presence of any traces of copper in the catalyst particles (for all measurements, the Cu/Fe 0.5 sample was selected because it is the one with the highest copper concentration in the solution of precursors). Both SEM EDS mapping and XPS analysis did not show any traces of Cu in the fabricated samples. Fig. 6a shows a representative EDS spectrum of the catalyst nanoparticles and the corresponding TEM micrograph is shown in Fig. 6b. The EDS scan covered the entire image area shown in Fig. 6b, maximizing the inclusion of nanoparticles in a single scan to ensure a comprehensive analysis of their composition. The EDS spectra of the catalyst nanoparticles are shown in Fig. S6.† From these results, it is evident that no copper is present in the composition of the catalyst nanoparticles, confirming the initial assumption that the copper remains in the heating lines of the reactor in the form of copper oxide.
![]() | ||
Fig. 6 EDS analysis of the Cu/Fe 0.5 catalyst nanoparticles. (a) Representative EDS spectrum of the scanned particles and (b) TEM micrograph of the particles in the EDS spectrum. |
To determine the chirality distribution of the SWCNTs, SAED was employed to analyse the fabricated samples.56,57Fig. 7 illustrates the three chirality distribution maps along with the corresponding diameter and chiral angle distributions derived from the electron diffraction patterns of approximately 140 individual SWCNTs for each sample. It should be noted that well-isolated and separated SWCNTs from nanotube bundles were selected for the analysis. Representative SAED patterns are shown in the ESI (Fig. S7†). Notably, no MWCNTs were observed, either through high-resolution imaging or electron diffraction analysis, and the results are in very good agreement with the light absorption results discussed earlier. The reference Fe sample (Fig. 7a–c) exhibited a near-armchair chirality distribution with a mean chiral angle of 20.80° (Fig. 7c) and a mean diameter of 1.84 nm (Fig. 7b). The percentage of metallic nanotubes was at 35%. On the other hand, the samples with the ferrocenium catalyst exhibited increased percentages of metallic nanotubes, reaching 43.1% for the Cu/Fe 0.3 sample (Fig. 7d–f) and 41.4% for the Cu/Fe 0.5 sample (Fig. 7g–i).
![]() | ||
Fig. 7 Chirality maps and the corresponding distributions of diameters and chiral angles for all samples. (a–c) Fe sample, (d–f) Cu/Fe 0.3 sample, and (g–i) Cu/Fe 0.5 sample. |
The mean diameters of these samples also increased, reaching 2.10 nm and 1.97 nm, respectively, highlighting the influence of ferrocenium on SWCNT growth. The significant differences in chirality distributions are evident in the chiral angle histograms (Fig. 7f and i), which showed a noticeable decrease in chiral angles from the Fe sample to 18.30° and 18.80°, respectively, indicating a shift in the distribution from a near-armchair region (as seen in the Fe sample) to more scattered distributions with many nanotubes approaching the near-zigzag region. Last but not least, this increase in the mean diameter is in good agreement once more with the UV-vis-NIR results discussed previously.
Previous studies53 have shown that using ethanol as a carbon source in catalyst-supported CVD methods can lead to the synthesis of near-armchair type SWCNTs with 75–77% being semiconducting, based on TEM SAED. Certain intrinsic properties of ethanol can lead to the preferential growth of near-armchair or semiconducting SWCNTs, where the formation of –OH groups and –H radicals during the initial stage of ethanol decomposition on Fe nanoparticles is thermodynamically and kinetically favourable. These decomposition products, including –OH groups and –H radicals, can selectively etch small-diameter metallic SWCNTs, which have higher curvature energy and lower ionization energy, thereby enriching the fractions of near-armchair or semiconducting SWCNTs. However, our results using a ferrocenium catalyst under the same conditions as ferrocene showed an increase in the amount of metallic SWCNTs, indicating that, in addition to the carbon source and reactor conditions, the choice of catalyst plays a critical role in determining selective chiralities.
This charge difference significantly alters the ability of the catalyst to facilitate helicity transitions in growing SWCNTs.58 When the catalyst is in the Fe3+ state, its higher positive charge leads to a stronger electrostatic potential, which increases the re-nucleation barrier at the nanotube–catalyst interface. As a result, the nanotube structure remains largely fixed, making helicity transitions highly improbable. In contrast, Fe2+ has a lower positive charge and can acquire more negative charge during growth, leading to a reduction in the re-nucleation barrier. This lower barrier allows the occurrence of structural rearrangements at the nanotube–catalyst interface, making helicity transitions more likely. Specifically, the reduced re-nucleation barrier favours a transition from the metallic to semiconducting state in SWCNTs (Fig. 8a). This effect aligns with the induced charge mechanism, where a more negatively charged catalyst raises the Fermi level, shifting the electronic density of states (DOS) and making semiconducting SWCNTs more stable. As a result, the Fe2+ catalyst, with its ability to lower the re-nucleation barriers more effectively, plays a crucial role in facilitating the helicity change, whereas Fe3+ is less effective in this regard.
![]() | ||
Fig. 8 (a) Proposed mechanism for the chirality changes of SWCNTs using ferrocene and ferrocenium catalysts and (b) aerosol electrometer measurements of all samples. |
To justify the proposed mechanism, we investigated the average net charge of the produced nanotubes and catalyst particles using an aerosol electrometer 3068B instrument. The results of the net charge measurements are shown in Fig. 8b. The data in Fig. 8b show that the Cu/Fe 0.3 and Cu/Fe 0.5 samples possessed more positive net charge than the ferrocene samples (−71 fA and −64 fA vs. −103 fA), which is in good agreement with the proposed mechanism above. These experimental data strengthen the conclusion that Fe2+ is more effective in inducing helicity changes, where Fe3+ could retain the initial chirality of grown SWCNTs. It should be noted here that previous FC-CVD works investigated the effect of net charge on SWCNT synthesis and found that more negative net charge was associated with the existence of more s-SWCNTs in the fabricated films, which is in good agreement with our results as well.59 Based on the diameter analysis from TEM (Fig. 7b, e and h) and the observed increase in SWCNT diameter from 1.84 nm to 2.10 nm, it can be assumed that the Cu/Fe samples initially contained a higher proportion of smaller-diameter metallic nanotubes, which were likely etched away by exposure to H2 or by-products of ethanol decomposition, such as CO2 or H2O.32 This etching process selectively removed the smaller metallic nanotubes, resulting in a shift in the overall diameter distribution towards larger diameters and, consequently, the statistical analysis showed different diameter distributions, reflecting this etching effect.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5nr00297d |
This journal is © The Royal Society of Chemistry 2025 |