Carbon nanotubes have been at the forefront of materials research for the better part of three decades now, with single-walled carbon nanotubes (SWCNTs) taking precedence over their multi-walled variants. Indeed, a major part of this interest has been due to the unique electronic properties possessed by SWCNTs, dictated by the chirality of the individual carbon nanotube [1
], that has led to potential applications including solar power [3
], fuel cells [4
], water filtration [5
], and thin-film transistors [6
]. As-synthesized SWCNTs have also been further analyzed via various methods of separation to develop enriched ensembles of metallic- and semiconducting-SWCNTs [7
], to further facilitate research and a move from the lab to the retail market. Synthesis of SWCNTs has thus always been key in progressing this field of research, with various techniques having been developed for this, including arc discharge [8
], laser ablation [10
], and chemical vapor deposition (CVD) [12
All three of these processes have been successfully used for commercial SWCNT synthesis; however, CVD offers the most control over the final product when it comes to chirality and diameter alike. One specific approach of the CVD process was the high-pressure carbon monoxide (HiPco) gas-phase synthesis of SWCNTs developed at Rice University in Houston, TX in the early 2000s [15
]. This method, resulting in the catalytic production of SWCNTs via the decomposition of Fe(CO)5
in the presence of a continuous flow of carbon monoxide at high temperature and pressure, has been synonymous with SWCNT research due to the university providing a stable, inexpensive supply of consistent, high-quality SWCNTs for researchers and companies around the world.
HiPco SWCNTs continue to be the de-factor standard even in 2019; however, things are quickly turning as a result of the reactor, associated IP, and all remaining stock of HiPco SWCNTs being sold by Rice University to Atom OptoElectronics, Inc. [16
] as the sole-supplier of the remaining materials (at $
899/g compared to the $
150–250 when available directly from Rice University). The expiration of the core patents for the HiPco process combined with a typical 25–30 wt% residual metal catalyst impurity [17
], create the potential for a new source of less expensive, high-quality HiPco SWCNTs available globally.
Erstwhile members of the Smalley group at Rice University, which developed the original HiPco process, helped create a new entity (NoPo Nanotechnologies, India), which aims to update the HiPco process using information developed over the last decade, and produce what they call NoPo HiPCO® SWCNTs, which are herein termed NoPo HiPco SWCNTs for consistency of naming. Since variability in SWCNT sources can make a change in research results (due to catalyst residue content, diameter and chirality distribution, and length), making back-to-back comparisons between batches is critical for researchers to obtain reproducible results, it is in the interest to all researchers to understand how the presently available NoPo HiPco SWCNTs compare to historically available Rice HiPco SWCNTs that have been the subject of a great range of academic studies. Here, we report a detailed comparison of the two HiPco materials to provide a baseline for researchers.
3. Results and Discussion
A photograph of the as-prepared NoPo HiPco SWCNTs is shown in Figure 1
in comparison to the Rice HiPco SWCNTs. The as-received physical appearance is of importance with regard to handling and as may be seen, they are both bulk low-density nanomaterials. However, the Rice HiPco SWCNTs come available in a more porous and powder-nature, with a measured average bulk density of 0.09 g/cm3
which made it fluffier and harder to handle relative to the NoPo HiPco SWCNTs that have an average bulk density of 0.11 g/cm3
and also a longer, Bucky paper-like consistency for most of the product. It is still suggested that the handling precautions for the NoPo HiPco SWCNTs should be considered similar to the Rice SWCNTs [18
Tapping-mode atomic force microscopy (AFM) analysis of ca. 500 nanotubes of each HiPco sample was used to determine the diameter distribution, revealing the predominant presence of SWCNTs in both samples. A typical example of the analysis done for both samples is shown in Figure 2
. The NoPo sample had an average diameter of 0.86 ± 0.1 nm, while Rice University’s sample (batch HPR 194.3) had a slightly larger average diameter, 0.97 ± 0.1 nm.
TEM imaging, with some examples shown in Figure 3
, supports the AFM measurements, with the NoPo HiPco samples having smaller diameter SWCNTs (0.88 ± 0.1 nm across ca. 50 CNTs) relative to those from Rice University (0.96 ± 0.09 nm across ca. 50 CNTs). SEM images collected during EDS analysis (Figure 4
) also showed a thick net of CNTs in both samples, with the NoPo HiPco sample showing thicker bundles and more pristine, individual SWCNTs with residual metal catalyst showing up as large, more discrete chunks instead.
EDS elemental analysis for both samples (Table 1
) showed a typical SWCNT composition with a majority of carbon, as expected, followed by residual iron catalyst and oxygen associated with both the CNT side walls and catalyst alike [20
]. The Rice HiPco sample (HPR 194.3) typically had a higher fraction of residual catalyst as compared to the NoPo HiPco sample. We note that certain “starved-catalyst” batches prepared as research experiments with the Rice HiPco reactor have had much lower iron content (~10 wt%). These batches were never sold as a retail product, however, and the use of HPR 194.3 was deliberately chosen to help represent the average Rice University HiPco product used by a wider group of researchers.
Thermogravimetric analysis (TGA, Figure 5
) corroborated the results from EDS elemental analysis, with a final residue of 30 wt% and 18.4 wt% for the Rice University and NoPo HiPco samples, respectively. Given that the TGA was performed in air, these numbers account for the formation of iron oxides during heating [17
], and the actual iron content in both samples are on the order of 21 wt% and 12.9 wt%, respectively, which is in good agreement with the EDS analysis (Table 1
). The nature of the weight loss plot for the two samples was also different, which also helps describe the carbon content in the samples. Amorphous carbon burns off prior to SWCNTs in air, and the Rice sample appeared to have more amorphous carbon impurity relative to the NoPo sample. The relatively lower oxidation temperature for the CNTs in the NoPo sample can be attributed to the lower average SWCNT diameter here, and the lower residual metal content in the NoPo sample can also be an explanation in the delayed catalyzed thermal decomposition of amorphous and graphitic carbon relative to Rice HiPco [10
Raman spectroscopy of the two as-produced HiPco SWCNT samples revealed a mixed bag of information, but again, it must be noted that the slightly lower average diameter for the NoPo SWCNTs compared to those from Rice University would play a role in resonance. This, combined with bundle interference of a detailed analysis of the radial breathing modes (RBMs) in the absence of a liquid Raman spectroscopy holder, also meant it is prudent to limit discussions on the nature of the carbon content and the diameter distribution, rather than chirality distribution.
At an excitation wavelength of 514 nm, as seen in Figure 6
, smaller-diameter metallic-SWCNTs (for SWCNTs of average diameter 1 nm) tend to resonate [7
] and evidence of a more pronounced activity was observed in the NoPo sample in the RBM range (Figure 6
b). The higher-intensity peaks at a higher Raman shift for this sample also were indicative of a relatively lower average diameter compared to the Rice University sample (Figure 6
a). At 633 nm, larger-diameter metals and smaller-diameter semiconducting-SWCNTs tend to resonate for an ensemble of average 1 nm diameter. This is where the majority of SWCNTs in the two samples would be expected to resonate (Figure 6
c,d). Excitation at 785 nm (Figure 6
e,f) is where larger-diameter semiconductors tend to resonate for a SWCNT ensemble of average diameter 1 nm.
A common figure of merit IG
(degree of order) was employed for the purposes of evaluating the defect density across different diameters [23
]. We have previously reported that in order for meaningful IG
values, a number of measurements need to be made across a sample [17
]. Table 2
gives the IG
value (with associated error) for both samples as a function of the excitation wavelength. The IG
value was found to be lower for the NoPo sample at 514 nm, suggesting greater defects for smaller-diameter metallic-SWCNTs. In contrast, for excitation at 633 nm and 785 nm, the NoPo SWCNTs now having higher IG
ratios (Figure 7
), suggesting lower defects for larger-diameter metals and both smaller- and larger-diameter semiconducting-SWCNTs [7
The availability of a UV-vis-NIR spectrometer and expertise in usage of the same meant that we were able to prepare surfactant-stabilized aqueous dispersions of both, as described in the experimental section. Once again, it must be noted that the slightly lower average diameter for the NoPo HiPco SWCNTs relative to those from Rice University would make assignment of individual (n,m) chiralities a non-trivial matter, but there is still useful information to be gained from Figure 7
, which covers the entire collected spectrum for both dispersions. The procedure used to generate this data has been described previously [7
], and we can see that once the CNT bundles and impurities are removed, the individual SWCNTs in both ensembles generally had a similar chirality distribution in the three electronic transitions we see in this range for SWCNTs in the ~1 nm average diameter: M11
for the first metallic transition that would show peaks corresponding to the metallic-SWCNTs, as well as the first (S11
) and second (S22
) semiconducting transitions for the semiconducting-SWCNTs. There is a small overlap between M11
for HiPco SWCNTs, which further makes it hard to assign individual peaks completely even for an entirely homogenous sample. It can be noted that there was a small shift towards chiralities favoring smaller diameters for both metallic- and semiconducting-SWCNTs for the NoPo HiPco sample relative to Rice University’s HPR 194.3, and this too agreed with all the characterization data obtained so far.