6.6
Configuration 5: Plasma Jet in Continuous
Reactor Tube
The fifth configuration is designed to support the existence of a plasma jet in a wider range of conditions, specifically with hydrogen present in the growth furnace and precursors passing through the plasma. It provides somewhat of a compromise between the plasma jet in configuration 4 and the more traditional setups such as configuration 1. Shown in Figure 6.9, the plasma reactor tube producing the jet is housed in a larger tube directly passing to the furnace. The hope is that with the jet closer to the microwave source, it will be less prone to contraction from hydrogen or catalyst precursor addition and moreover, heat will not be lost through conduction to the mating flange as with configuration 4. The diameter of the alumina tube housing the plasma has also been further reduced, from 6 mm to 3 mm in an effort to further increase the gas velocity and the plasma’s propensity to form a jet. It is now also possible to inject the carbon source as a sheath gas similar to configuration 3 since the thin tube does not heat to the same degree as the axial torch in configuration 3, and therefore methane breakdown and soot deposition are considerably reduced.
Fe / S / Ar
Nanoparticle
nucleation Catalytic nanotube growth on Fe / S particles Furnace Axial Torch Plasma Counterflow gas injection
Continued growth and CNT agglomeration C / H2 Furnace Tube Reactor Tube
Fig. 6.9 Configuration 5 featuring 3 mm reactor tube and argon plasma jet situated inside larger furnace tube. Carbon and hydrogen injected as sheath gases.
Using this configuration, it is possible to maintain a small (<1 cm) argon plasma jet when hydrogen is present in the furnace tube and while precursors are passed through the plasma. Conversely, the total particle concentration emitted from the plasma is low in this configuration (< 1 × 106cm−3) and also appears to be somewhat unsteady. This is exemplified in Figure 6.10a which depicts the total concentration output from
the plasma over time using a CPC. At the 50 s mark, ferrocene was introduced into the system which should produce several orders of magnitude more particles, yet in this case the concentration never exceeds 7 × 105cm−3. Moreover, this concentration is only transient and the value soon decreases back towards that of the background. The conclusion is then that penetration of the ferrocene and its constituent atoms is not effective in this configuration.
To eliminate the chance of contamination and in this case early deposition of the iron caused by the carbon within the ferrocene, the iron precursor was replaced by a hot wire particle generator [211]. This type of particle generator simply features a resistively-heated wire, in this case of iron, with a gas flow across it to remove the nascent particles. The peak size and number concentration (and therefore the mass concentration) of the iron particles can be controlled with the temperature of the wire, and this is normally accomplished by varying the current through the wire. Since iron is readily oxidized, hydrogen must be included in the gas flow to ensure the wire does not oxidize and break. As discussed previously, hydrogen exhibits a much higher thermal conductivity than other carrier gases (argon in this case) so the convective cooling on the wire is improved. As a result, the temperature of the wire, and thus its outgoing particle distribution is also dependent on the hydrogen fraction in the carrier gas. This phenomenon is shown in Figure 6.10b which presents size distributions using approximately 9% and 17% H2in Ar. Note that this increase in hydrogen fraction
results in the mean diameter decreasing to 56% of its previous value (42.9 nm to 24.1 nm). The quantity of iron sent to the plasma is therefore controllable over a large range of mass throughputs.
Unfortunately, the requirements of the hot wire generator do not agree well with the requirements of the plasma system. As discussed for configuration 4, the inclusion of even very modest amounts of hydrogen (<4%) are sufficient to cause significant heating of the plasma tube and possibly even its catastrophic failure. Attempts were made to operate the generator using 2% H2. However, this was not sufficient to prevent
oxidation and failure of the wire within minutes. As a result, it does not appear that any overlap of hydrogen fraction exists which is suitable for the hot wire generator and this plasma configuration simultaneously. Moreover, even during the time the generator was able to function, particle generation from the plasma appeared similar to that from ferrocene. Once again, very few particles are produced by the plasma so the conclusion can be made that it is not solely the ferrocene’s carbon that is responsible for the questionable catalyst production from configurations using thin plasma tubes. Instead, precursor material is likely lost to the tube walls. Three mechanisms could be responsible for this wall loss: diffusion, charging, and thermophoresis. Each of these phenomena are examined here.
6.6 Configuration 5: Plasma Jet in Continuous Reactor Tube 117
Diffusion via Brownian motion is ubiquitous for aerosol samples and losses to tube walls occurs continuously. Classical diffusion theory [212] dictates that the penetration efficiency through a given length of tube is dependent only on the volumetric flow rate. Penetration is not dependent on tube diameter so there is no reason the thin tubes would have higher diffusion losses than any other reactor tube. Instead, ions or particles may be lost from space charging effects. This phenomenon occurs when many charges are present in a given volume, and can also be common when non- conducting tubes are used. Both of these criteria are met for the reactor tube since alumina is an insulator and the plasma has a high concentration of charges. Given the thin cross section of the reactor tube, the electrostatic forces may be large and could be responsible for particle deposition on the tube walls and a significant decrease in number concentration. Finally, themophoresis may also cause the high particles losses, and this is an effect which is certainly stronger given the small tube cross section. With a large temperature gradient such as that existing between the hot plasma and relatively cool tube wall, particles experience more energetic atomic collisions from the hot side and a net force is created towards the cold side. In this case, this force will result in deposition on the tube wall. The thermophoretic force is a function of the temperature gradient and no doubt an extremely large gradient exists in the reactor tube. It could be as high as several thousand Kelvin per millimetre. For comparison, dedicated thermophoretic sampling devices employ temperature gradients up to 100 to 200 K mm−1. Therefore it seems highly likely that thermophoresis is causing the majority of the iron mass to be deposited on the walls. Further, if the particles manage to evaporate before they contact the tube wall, the iron vapour may still be lost to the wall since the wall temperature is well below the iron’s melting point and the iron would remain here upon contact.
Given these findings, it appears that the use of a thin reactor tube to house the plasma and create a jet for efficient linkage of the catalyst nucleation and CNT growth stages exhibits limitations that outweigh its ability to produce a favourable temperature profile. Precursor penetration of the plasma is limited both for ferrocene and thiophene, and the material that is produced does not possess the correct morphology and number concentration for optimal high-throughput CNT production. It is clear that temperature gradients within the plasma region must not be too severe, and it is likely helpful if hydrogen is present during not only the CNT growth phase, but also the catalyst formation phase (either to suppress soot formation upon catalyst decomposition or prevent catalyst oxidation). A configuration is required which reduces the plasma-to- wall temperature gradients and can accommodate higher hydrogen fractions.
dg= 42.9 nm dg = 24.1 nm a) b) Ferrocene Addition N= 3.01×107cm-3 dg= 33.5 nm c) 300 nm d) 5 μm e) f) (c o u n ts W -1m 2 s -1)
Fig. 6.10 (a) Total particle concentration over time for output of configuration 5 including discrete addition of ferrocene, and (b) size distributions of particles produced from hot wire generator at two hydrogen fractions. Data fit with two lognormal distributions each; σg1 = 1.68, σg2 = 1.49 for low H2 and σg1 = 1.73, σg2 = 1.45
for high H2. (c) catalyst particles from ferrocene and thiophene using swirl torch
(configuration 6) fit with single lognormal distribution with σg= 1.37. All SMPS
distributions are averages of two scans each and error bars represent one standard deviation. Characterization of material from configuration 6 including (d) Raman spectrum (average of two scans) and SEM with (e) high magnification and (f) low magnification.