Arc Discharge Production of SWNTs

SWNTs are produced in an arc process utilizing covaporization of graphite and metal in a composite anode [30,31] commonly made by drilling an axial hole in the graphite rod and densely packing it with a mixture of metal and graphite powders. Various pure elements and mixtures have been used to fill the rod, including Fe, Co, Ni, Cr, Mn, Cu, Pd, Pt, Ag, W, Ti, Hf, La, Ce, Pr, Nd, Tb, Dy, Ho, Er, Y, Lu, Gd, Li, B, Si, S, Se, Zn, Sn, Te, Bi, Cd, Ge, Sb, Pb, Al, In, Fe/Co, Fe/Ni, Fe/Co/Ni, Co/Ni, Co/Pt, Co/Cu, Co/ Bi, Co/Pb, Co/Ru, Co/Y, Ni/Y, Ni/La, Ni/Lu, Ni/B, Ni/Mg, Ni/Cu, Ni/Ti [32-46], but at present only Ni/ Y and Co/Ni catalysts are commonly used in SWNT production.

These preferred systems share many common features. Both perform better in helium atmosphere than in argon, at about 0.6 bar pressure. The process is most efficient with a stable arc discharge and a constant anode erosion rate, corresponding to ca. 2 A/mm2 current density and ca. 3-mm gap width. This is better achieved by maintaining a constant arc current and anode feed rate and thus constant arc gap width.

A cylindrical deposit grows at the surface of the cathode. The weight of the deposit constitutes about one half of the weight of the anode consumed in the process. The deposit consists of a hard gray shell and a soft core. The core has poorly developed columnar structure and contains MWNTs, MPPs, and graphitic particles. In the Co/Ni system, commonly operating with ca. 2 at% metals in the anode, the core is diamagnetic and does not contain any metal components. High temperature prevents precipitation of metal vapor on the growth surface, and all metal escapes the gap. However, the outermost layer of the gray shell contains trace amounts of metals, introduced by gas convection onto the exposed side surface. In many other systems, especially with high metal loading of the anode, the deposit may contain metal-filled MWNTs and MPPs or bare spherical metallic particles [47]. In the Ni/Y system the deposit contains large amounts of yttrium [48], whereas metal nanoparticles found in the chamber are enriched in nickel [46,48,49].

The mixed carbon and metal vapor, which has escaped the gap, then condenses into the product, which moves to the reactor surfaces and deposits on them. The product is divided into three distinct structural types, depending on the deposition area. A spongy soft belt called collaret is formed around the cylindrical deposit and represents about 20% of the product weight. Relatively strong clothlike soot on the chamber walls represents another 70%, and the remaining 10% of the product is a weblike structure suspended in the chamber volume between cathode and walls. All three types contain varying amount of SWNTs, fullerenes, amorphous carbon, empty and metal-filled MPPs, naked metal particles, and graphitic nano- and micro-particles. Thermogravimetric analysis (TGA) and near-infrared (NIR) spec-troscopy [50] are used to accurately determine the metal and SWNT content in the arc material, whereas electron microscopy (TEM and SEM) and Raman spectroscopy (RS) generally provide only rough assessments of the concentration of product components. These methods appear to be the most useful for analyzing arc-product composition and structure. The collaret contains more SWNTs and metal particles than other components of the product. This difference is about 20% for Co/Ni and 50% for Ni/Y systems. The collaret is formed from an electrostatically deflected flow of soot that is propagating away from the gap. SWNT growth continues in the material that is deflected and captured on the deposit surface for a longer time, leading to enrichment of the collaret in SWNTs. The SWNT yield averaged among all three structural types of the product is about the same in these systems, roughly ca. 20 ± 5% of the total product weight. Fullerenes C60 and C70 contribute ca. 5 to 10 wt% to the total product weight in the Co/Ni system and ca. 1% in the Ni/Y system. The average mass fraction of the metal catalyst is ca. 20% of the total product weight in the Co/Ni system and 35% in the Ni/Y system. Amorphous carbon constitutes ca. 50 wt% of the total product weight and graphitic particles ca. 5 wt% in both cases.

SWNTs are generally organized in bundles consisting of a few dozen tubes, tightly compounded in a honeycomb lattice with an average separation between tube axes of ca. 1.7 nm. Rare isolated tubes can be observed. The bundles are covered with an amorphous carbon layer ca. 2 to 5 nm thick, which contains embedded fullerenes. The majority of tubes have diameters in the range of 1.2 to 1.5 nm and lengths reaching up to 5 ^m in the Ni/Y system and 20 ^m in the Co/Ni system. The tips of individual tubes are closed, in the rare cases when they can be observed. Some bundles terminate with a spherical metal particle of 10 to 30 nm in diameter, which exceeds the bundle diameter. Such termination is characteristic of Co/Ni [51] (Figure 3.4), Ni [70], Ni/La [51], and Ni/Y [49] arc systems, as well as of a nickel-catalyst-based laser vaporization system [52].

SWNT diameter depends on the temperature of the catalytic site at which growth occurs. This temperature is regulated by many factors, including heating of the reaction zone with an externally controlled heat source. The constant temperature thus set by external heating (environment temperature) provides a minimum temperature for the reaction site. It was established that the mean diameter of SWNTs increases with the environment temperature [53,54]. Other factors affecting reaction-site temperature cause the same kind of dependence, namely, production of thicker SWNTs is favored at higher temperatures [49,55-59]. The environment temperature also affects the yield of SWNTs. Higher SWNT production rate at elevated environment temperatures is generally observed both with arc and laser techniques [53,54,60].

3.4.1 Metal Catalyst Particles

The efficiency of SWNT synthesis in a carbon/metal arc is determined primarily by the choice of metal catalyst for the process. It was found [33,34] that catalysts composed of two metals produce much higher yields of SWNTs than did individual metals. The ratio of metals in the original catalyst mixture also produces a drastic effect on SWNT yield. In the Co/Ni catalyst arc system the highest yield is obtained with Co:Ni = 3:1 ratio in the original mixture (Figure 3.5). For the Ni/Y catalyst the optimum molar ratio of the original mixture was determined to be Ni:Y = 4.2:1 [46]. The predominant diameter of metal particles in the arc SWNT product is 10 to 30 nm, and the metal ratio in these particles differs from that

FIGURE 3.4 TEM image of 3Co/Ni catalytic metal alloy particle (CP) with three SWNT bundles (b1, b2, b3) grown from it (left) and magnified view of the same particle showing a bundle root with about 20 active sites arranged in a triangular pattern on the surface from which SWNTs of the bundle b1 originate (right). The 1.65-nm separation of active sites reveals dense grouping of nanotube roots [51].

FIGURE 3.4 TEM image of 3Co/Ni catalytic metal alloy particle (CP) with three SWNT bundles (b1, b2, b3) grown from it (left) and magnified view of the same particle showing a bundle root with about 20 active sites arranged in a triangular pattern on the surface from which SWNTs of the bundle b1 originate (right). The 1.65-nm separation of active sites reveals dense grouping of nanotube roots [51].

FIGURE 3.5 Dependence of the yields of SWNTs (squares) and fullerenes (circles) on the metal composition of the catalyst with the total metal content held constant at Ni + Co = 2.3 at.% in the 8-mm diameter anode vaporized in the arc-discharge process under optimized conditions [61].

Ni/Co Ratio

FIGURE 3.5 Dependence of the yields of SWNTs (squares) and fullerenes (circles) on the metal composition of the catalyst with the total metal content held constant at Ni + Co = 2.3 at.% in the 8-mm diameter anode vaporized in the arc-discharge process under optimized conditions [61].

in the original metal mixture. The energy dispersive x-ray fluorescence (EDX) analysis has shown that the metal composition of different particles varies within 30% from the 3:1 ratio of cobalt to nickel in the original mixture [61]. The same-size metal catalyst particles in the Ni/Y system are essentially depleted of yttrium [46,49], regardless of the Ni:Y composition of the original catalyst mixture [49]. This implies that there are additional functions for yttrium in the arc synthesis, besides forming catalytic particles of the optimum composition. The presence of yttrium may facilitate an adjustment of the optimal process temperature through the highly exothermic reaction of yttrium carbide formation. This reaction brings about higher catalytic particle temperature, which results in a generally observed 0.1- to 0.2-nm shift to larger average tube diameter, compared to that produced with yttrium-free catalysts under otherwise similar conditions. The 3Co/Ni mixture produces only long SWNT bundles with 1.2- to 1.3-nm diameter nanotubes. The Ni/Y system, in addition to long bundles, under certain conditions can generate [49] short bundles (~ 50 nm in length) composed of thicker SWNTs, with an average nanotube diameter of 1.8 nm, radiating from metal catalyst particles (sea urchin structures). Long bundles of narrow tubes generally grow from particles containing less yttrium (below 15%) than metal particles in sea urchin structures (over 11% yttrium) [49]. These facts may imply that sea urchin structure tubes grow from hotter metal particles and that the presence of yttrium in the metal particle retards the start of tube growth until a higher carbon supersaturation is attained than for a pure nickel particle. As a result, sea urchin tube growth starts simultaneously over the whole metal particle surface. The sea urchin metal particle cools down very quickly to temperatures at which dissolution of carbon becomes too slow to sustain steady tube growth. In addition, the bundle roots covering the whole metal surface prevent the outside carbon feedstock from coming in contact with the metal particle surface. When the carbon inside the particle is exhausted, growth is terminated, resulting in short bundles. The amount of carbon in these bundles is commensurate with that dissolved in the metal particle at the start of growth.

3.4.2 Dynamics of SWNT Growth in the Arc Process

Carbon vapor flowing from a sufficiently narrow arc gap can be idealized as a turbulent jet of cylindrical symmetry [22,62] and described in the framework of a semiempirical theory [63] of heat and mass transfer in a free turbulent jet. These transfer phenomena control the dynamics of carbon vapor mixing with helium gas and the resulting cooling. In the framework of the turbulent jet model, an analytical relationship between the essential parameters of the arc process is valid [22]. These parameters include the volume rate of carbon vapor flow from the gap Vsoot, the carbon vapor temperature To and velocity Uo at the cylindrical boundary of the gap, the helium pressure in the reactor P, the gap width ho and electrode diameter 2ro, and the characteristic time for turbulent mixing and cooling of carbon vapor Tmix. The value of Tmix is connected to other parameters by Equation (3.1):

Tmix = ro1'5/ Uoho0-5 = 2ft ro2-5ho0-5P/VsootRTo. (3.1)

The rate of carbon vapor cooling in a turbulent jet outside the gap is directly proportional to the linear rate of carbon vapor propagation Uo and thus to the volume rate of vapor generation Vsoot and to the inverse value of the chamber pressure P. Temperature To can be either determined experimentally or evaluated under the assumption that the equilibrium carbon/metal vapor pressure in the gap is equal to the ambient gas pressure in the chamber. The volume of carbon/metal vapor produced should allow for vapor composition. It should be noted that the true value of the cooling rate in a real arc system may differ from that calculated with Equation (3.1), because the model considers only turbulent cooling and because of uncertainties in the values of the parameters involved. In this respect, it is noteworthy that laminar diffusion of cold environment helium into the hot zone, which also includes the gap area, can substantially contribute to the rate of reagent cooling [22]. This contribution becomes stronger with wide gaps [22]. However, the laminar diffusion cooling rate has similar pressure and temperature dependence to that of the turbulent case, so in the first approximation this should change only the proportion between the characteristic cooling time and the other process parameters in Equation (3.1), without disturbing the underlying dependence. Assessments show that Tmix value, as determined by Equation (3.1) functional dependence, can be treated as having an accuracy better than a factor of 3 in the analysis of arc synthesis performance in the whole range of arc process conditions.

The parametric study of SWNT yield in the 3Co/Ni arc system was performed in a wide range of helium pressures, arc currents, and FRs, and the data were evaluated in relation to the turbulent jet model [61]. The gap width and rate of vapor generation were measured in each run, and vapor temperature inside the gap was assessed as previously to calculate the Tmix value.

The dependence of the SWNT yield on the Tmix value is presented on Figure 3.6. All SWNT yield data points are grouped around a common curve peaking at ca. 6 ms. This means that Tmix is the sole parameter determining the yield of SWNTs. The largest SWNT yields occurring near Tmix = 6 ms correspond to fairly different sets of helium pressures, arc currents, and gap widths. This implies that no unique set of externally controlled parameters exists that would produce the highest SWNT yield for a given system. The parameter to optimize the process for the highest SWNT yield is Tmix. SWNT production in the arc process is totally controlled by the cooling rate of carbon vapor escaped from the gap, and formation of nanotubes occurs at a distance of a few centimeters from the arc gap. The cooling time Tmix = 6 ms, which

FIGURE 3.6 Dependence of the yields of SWNTs (squares) and fullerenes (circles) on the characteristic time of turbulent mixing of the hot carbon vapor with helium gas in the fan jet flow moving from the interelectrode gap. Data points correspond to a wide variation of externally controlled arc discharge process parameters (helium pressure, arc current, and electrode feed rate) [61].

FIGURE 3.6 Dependence of the yields of SWNTs (squares) and fullerenes (circles) on the characteristic time of turbulent mixing of the hot carbon vapor with helium gas in the fan jet flow moving from the interelectrode gap. Data points correspond to a wide variation of externally controlled arc discharge process parameters (helium pressure, arc current, and electrode feed rate) [61].

is optimal for SWNT production, gives characteristic time of the SWNT growth process. Assuming 20 |im as the average SWNT length, the linear growth rate of SWNTs is ca. 3 mm/s. This value is many orders of magnitude higher than the linear growth rate of MWNTs formed on metal catalyst particles at lower temperatures, which implies that diffusion of carbon through the metal particle is the rate limiting step in catalytic nanotube growth.

A similar dependence of fullerene yield on Tmix was observed for fullerene synthesis with a metal-free graphite anode [22]. The optimum value of Tmix for high fullerene yield was determined to be ca. 0.2 ms, at which point the fullerene yield peaked at ca. 23%. Therefore, the characteristic time of fullerene formation in the arc process is much shorter than that of SWNT formation. The carbon component of the mixed carbon/metal vapor totally condenses into fullerenes and soot particles much before the growth of SWNTs is initiated. Fullerenes and soot particles are the only carbon feedstock available for SWNT growth. No low-molecular carbon species — including such components of hot carbon vapor as C3 and C2 molecules or other C„ clusters in the form of chains, rings etc. — are available in the SWNT growth zone. Soot particles are used preferentially to fullerenes C60 and C70 for the carbon feedstock. This conclusion is supported by the following data. Figure 3.6 shows the yield of fullerenes C60 and C70 in relation to Tmix. The fullerene yield in the total collected soot product is nearly constant under widely varied conditions in the 3Co/Ni system and equal to ca. 7 wt%. In the control runs on fullerene, production performed under identical conditions, but with a metal-free anode the yield of fullerenes was within 1 wt% standard deviation, the same as in the corresponding 3Co/Ni runs. The SWNT production in the 3Co/Ni arc system is independent of the presence of fullerenes. The molar ratio C60/ C70 for fullerenes produced in carbon arc in helium is perfectly constant and equal to 5.0, independent of arc conditions [64]. The same constancy and value of the C60/C70 ratio was determined for SWNT runs presented in Figure 3.6. This provides further evidence that fullerene and SWNT formation processes are independent. Amorphous carbon is ca. 150 kJ/mol less stable than fullerenes [65], thus providing much higher thermodynamic driving force to the process of SWNT growth. Therefore, the observed preference for soot particles over fullerenes for the SWNT precursor in the arc synthesis has a strong thermodynamic basis.

3.4.3 Carbon "Dissolution-Precipitation" Model

The carbon "dissolution-precipitation" (DP) kinetic model developed a long time ago for the growth of carbon filaments and MWNTs on bulk metal nanoparticles [66-69] was subsequently adapted with some modifications for SWNT growth from similarly massive catalytic nanoparticles in the arc discharge and laser ablation processes [49, 51,70-74]. The model includes three consecutive steps: dissolution of carbon coming into contact with the bare metal surface, transfer of dissolved carbon to another location on the particle surface, and precipitation of carbon as nanotubes at this location. Dissolution of amorphous carbon into metals of the iron group is a highly exothermic process. It keeps the particle hot enough to produce SWNTs in the tube formation zone a few centimeters away from the arc, where otherwise the particle temperature would be too low. The round shape of particles found in the product attests to their molten state during the process. Particle melting is facilitated by the incorporation of carbon into metal. The temperature of the carbon-containing, molten catalytic particle in the SWNT synthesis zone is assumed to be near 1300°C, the eutectic temperature of the metal-carbon mixture [49,51,75,76]. Carbon solubility in the particle at SWNT synthesis temperature can be estimated as 2 to 3 wt%, using the binary M-C phase diagrams of cobalt and nickel. Small supersaturated catalytic particles may contain up to four times as much carbon [75,77].

Carbon diffusion through the catalytic metal particle has been shown to be the rate-limiting stage of the overall DP process in the case of vapor-grown carbon fiber (VGCF) and MWNT synthesis by catalytic pyrolysis of hydrocarbons [66-69]. In the framework of the DP model, the rate of SWNT growth can be estimated for arc synthesis as well. The carbon diffusion flux through the catalytic nickel particle at 1300°C can be evaluated as Q = D •grad C ~ 2-10-6 cm2/s-1 • 105 g/cm4 ~ 0.2 g/cm2s. The carbon diffusivity of 2 • 10-6 cm2/s was obtained from the temperature dependence D = 0.1 exp (-140 kJ • mol-1/ RT) for nickel [78]. The carbon concentration gradient 1 • 105 g/cm4 was estimated using the assumption that carbon concentration at the dissolution spot exceeds that at the precipitation spot by ~0.2 g/cm3, which is about the value of carbon solubility, and the distance between these spots is 20 nm. Division of carbon flux value 0.2 g/cm2 s by the gravimetric density of the SWNT bundle (~ 1 g/cm3) yields 0.2 cm/s for the value of the SWNT linear growth rate. This value (0.2 cm/s) practically coincides with the value for SWNT linear growth rate of 3 mm/s independently assessed in Section 3.4.2. This coincidence implies that carbon diffusion through the catalyst nanoparticle is the limiting stage of the overall SWNT production process.

Given the carbon flux of ~0.2 g/cm2 s through the typical size 20-nm particle, and the characteristic SWNT growth time of 6 ms, it is easy to calculate that the fraction of particles active in SWNT production should be only about 1% of the total particles. This is close to what is observed by TEM in the arc synthesis products.

Precipitation of the dissolved carbon as SWNTs is an endothermic process (by ca. 40 kJ/mol), and the particle surface in the precipitation spot is somewhat colder than in the dissolution spot. The temperature gradient in nanometer-size metal particles is too small to cause the directional diffusion of carbon, and the carbon concentration gradient is the sole reason for diffusion in the catalyst particle [68]. Precipitation starts with nucleation of the SWNT caps.

Carbon atoms, brought by diffusion from inside the liquid particle to the surface, start to arrange into a small carbon network containing a few hexagons and pentagons. When this network grows to the diameter of a nanotube and incorporates six pentagons, it has a more or less hemispherical structure with carbon atoms on the circumference of the hemisphere chemically bonded to the surface atoms, which can be either metal or carbon atoms. Other carbon atoms of the hemisphere form a carbon network peeled away from the surface. This scenario is very close to the construction of a SWNT cap by a computer through the molecular dynamics modeling of the segregation of carbon from hot metal-carbon alloy [49]. The SWNT cap diameter is energetically "locked up" at this point, and further addition of carbon atoms to the hemisphere root result only in formation of hexagons in the wall of the growing SWNT, independent of slight changes of synthesis conditions at the root area.

At a higher temperature, a larger number of hexagons are included in the hemisphere structure together with the six pentagons, because the more intense supply of carbon atoms reduces the probability of pentagon formation. Thicker SWNTs therefore emerge at higher temperatures.

Mutual disposition of the six pentagons on the hemisphere determines the tube chirality and is unlikely to easily change after the cap is formed, for the migration of carbon atoms in the isolated hemisphere and Stone-Wales-type rearrangements necessary for this change are essentially suppressed at 1300°C.

Growth of all SWNTs in a bundle starts simultaneously [49] from an active spot on the particle surface. The local temperature of multiple SWNT nucleation sites belonging to an active spot does not vary much, and therefore one-spot bundles consist of nanotubes that are very close in diameter [46,60,79,80] with a strong tendency to contain tubes of uniform chirality [60,79,80]. Several small neighboring spots can coalesce, giving rise to a combined bundle composed of discernible narrow bundles [46,73]. For a substantial portion of their length, bundles originating from different particles can also combine into a thick bundle while drifting in the gas phase or during ultrasonic treatment of the SWNT product in a liquid. Combined bundles of both types can contain SWNTs of substantially different diameters and chiralities [81-88].

The above-proposed molecular model for nanotube growth from a molten metal/carbon particle explains origin of the preferred chirality in tubes. Insertion of carbon atoms into the M-C bonds at the nanotube root cannot proceed without the relative displacement of metal atoms tangentially along the tube circumference, when the tube is chiral. With achiral tubes this displacement can be smaller, or negligible for armchair tubes with a certain type of M-C coordination bonding. For a larger helical pitch of chiral tubes this displacement is larger. The displacement should generate torque that will cause mutual rotation of the tube and the ensemble of metal atoms bound to the tube root, about the tube axis. The rotation will be hindered because of the high viscosity of molten metal and van der Waals interactions among the tubes in the bundle. At a certain bundle lengths the van der Waals retardation of tube rotation prevails over circular metal ensemble drag friction, and mutual rotation of tubes in the bundle will cease. Activation energy of circular metal ensemble motion in the surrounding liquid can exceed or even far surpass that of achiral (armchair) tube growth, thus slowing down or preventing further growth of chiral tubes. As a result, the armchair tubes grow in preference to zigzag and chiral tubes, especially to those chiral tubes that have large helical pitch. This preference has been observed in many studies and is commonly attributed to higher thermodynamic stability of the armchair tubes, particularly the (10,10) tube. The kinetic mechanism for tube chirality selection outlined above finds support in several observations. The core of a bundle is likely composed of only armchair tubes, whereas the outer shell of a bundle consists of chiral tubes [79-88]. This is expected with the kinetic mechanism, as outer tubes in the bundle have more opportunities for rotation due to weakened van der Waals interactions. Also anticipated is more frequent occurrence of nonzero chirality in tubes that are seen lying apart from bundles [79-88], as these tubes most likely have been individually grown or detached from the outer layer of a bundle during sample preparation. In either case, nonzero chirality is presumed by the kinetic selection mechanism. It also predicts that a higher metal particle temperature would favor formation of chiral tubes and larger helical pitch, whereas lower temperature should facilitate armchair tube dominance in the product. Some support for these expectations can be found in the literature; however, more research is required to verify the predictions.

For steady tube growth conditions, the rates of all three consecutive stages of the DP kinetic scheme (dissolution, diffusion, and precipitation) are equal. This rate balance results in production of long tubes and is possible only when particle temperatures are in an appropriately high range and certain other conditions are met. The rate of carbon dissolution should not exceed the maximum achievable carbon diffusion rate in the particle. The maximum achievable rate of carbon precipitation should be higher than the diffusion rate. Failure to observe either of these relationships may, under some circumstances, result in fast termination of tube growth. The particle carbon concentration will increase when either or both of the relationships are violated, and high supersaturation may be reached. This may lead to a sudden precipitation of a graphitic shell, which will eventually prevent further dissolution of external carbon into the particle. The concomitant temperature drop and exhaustion of dissolved carbon lead to fast cessation of the growth process. This temperature drop is likely to occur with acceleration and irreversibly, which is similar to the temperature rise in a thermal explosion; therefore, the process in the particle can be referred to as "cold explosion." This term also presumes the catalytic particle to be a highly nonequilibrium thermal system, with particle temperature substantially exceeding that of the environment.

3.4.4 Scaled-Up SWNT Production Process

The dynamic model of the carbon arc (Section 3.4.1) implies no principal limitations for obtaining high SWNT yields with much larger-diameter graphite/metal anodes than are commonly used in laboratory scale practice. In essence, it predicts that a high SWNT yield would be retained if the value of characteristic cooling time Tmix is maintained at ca. 6 ms while using a larger rod diameter 2ro. According to expression (1) for Tmix, to compensate for the growth of the term ro25, the value of the term ho0-5P/VsootRTo should be decreased appropriately. By adjusting the gap width ho, helium pressure P, the rate of vapor generation Vsoot and arc temperature To through deliberate variation of externally controlled parameters (electrode FR, helium pressure, and arc current), this compensation may become possible. These adjustments were implemented with 25-mm diameter anodes of the same 3Co/Ni composition and resulted in a product that contained on the average ~15 wt% of SWNTs, which is nearly the same as the SWNT yield obtained with commonly used 8-mm diameter rods under optimal conditions. Experiments were done with the large-scale arc discharge apparatus shown in Figure 3.2. Clothlike soot with ~12 wt% SWNT yield can be peeled off the walls of the reactor chamber as hand towel-sized sheets (Figure 3.7). The average SWNT yield in the large collaret on the cathode deposit is ~25 wt%. The soot production rate with 25-mm-diameter rods makes ~100 g/h; thus a 20-fold scaling factor for the process is attained compared with 8-mm diameter rods [26]. The process using 25-mm-diameter rods can be deemed semi-continuous, as loading a new rod takes a small fraction of the operation cycle. Moreover, in this sense the process can be rendered virtually continuous by arranging the automatic change of rods and continuous harvest of the soot, technical improvements that are already implemented at some laboratory scale arc installations. Finally, a further increase in rod diameter to 50 to 75 mm appears quite feasible and expedient.

Was this article helpful?

0 0

Post a comment