16 October 2002 Chemical Physics Letters 364 (2002) 568 572 www.elsevier.com/locate/cplett The large-scale production of carbon nanotubes in a nano-agglomerate fluidized-bed reactor Yao Wang *, Fei Wei, Guohua Luo, Hao Yu, Guangsheng Gu Department of Chemical Engineering, Tsinghua University, Beijing 100084, PeopleÕs Republic of China Received 10 September 2001; in final form 27 March 2002 Abstract Carbon nanotubes (CNTs) produced by catalytic chemical vapor deposition (CCVD) can be formed into loose agglomerates that can be fluidized during the growth process. This provides a way to prepare high-quality CNTs on a large scale at low cost in a nano-agglomerate fluidized-bed reactor (NAFBR). With the present fluidized-bed reactor design and catalyst preparation, 50 kg/day of carbon materials was synthesized, and a high yield of 70 80% CNTs was obtained. Fluidization characteristics distinctive to CNT growth in a fluidized-bed reactor are discussed. Ó 2002 Elsevier Science B.V. All rights reserved. 1. Introduction Carbon nanotubes (CNTs) are very promising materials in a wide range of potential applications, e.g., as hydrogen storage media, microelectronic devices, catalyst supports, selective absorption agents, reinforcement materials and so on. Since their discovery in 1991, research in the field of CNTs has undergone an explosive growth. Currently, several techniques, such as electric arc-discharge, laser evaporation and catalytic chemical vapor deposition (CCVD) through the decomposition of hydrocarbons have been successfully developed to synthesize CNTs. The first two methods can produce high-quality nanotubes in yields * Corresponding author. E-mail address: wangyao@flotu.org (Y. Wang). suitable for limited research use [1,2], but they are not adaptable to industrial production. By comparison, the CCVD method requires a lower reaction temperature with the potential for a low cost of production. It is the best possibility for large-scale production, and the CCVD method has been successfully used to produce aligned carbon nanotubes [3] and single-walled carbon nanotubes (SWNTs) [4]. However, only limited carbon nanotubes can be synthesized per day in the reports in the literature till now. The lack of methods for large-scale preparation restricts fundamental research and application development of this unique material. The main theme of our study is to produce nanotubes on a large scale and at low cost. Fluidization is the operation by which solid particles are transformed into a fluid-like state through suspension in a gas or liquid. This method 0009-2614/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved. PII: S0009-2614(02)01384-2
Y. Wang et al. / Chemical Physics Letters 364 (2002) 568 572 569 of contacting has been used in numerous industrial processes for its unusual characteristics. Normal gas solid fluidization is extremely difficult for ultrafines (including nanoparticles) because interparticle forces are greater than the hydrodynamic drag. This is because ultra fine particles adhere to each other tightly in all directions by Van der Waals forces or other forces among ultrafines, which seriously limits the fluidization of ultrafine particles. Fortunately, when particle sizes decrease to the nano-scale, things can be different. For nanoparticles that are not zero-dimensional or that can coalesce into fractal subsets, the interparticle forces vary significantly with the packing structure and can be exploited for the fluidization of ultrafines. Studies by Fitzgerald and Brooks [5] showed near particulate fluidization with a tendrillar carbonaceous material. An aerogel Cu=Al 2 O 3 powder has shown promise as another candidate for improving fluidization [6]. More recently, a synthesized silicon aerosol has been fluidized smoothly via self-agglomeration by Wang et al. [7]. In this work, large amounts of CNTs were successfully produced in a nano-agglomerate fluidized-bed reactor (NAFBR). pressure and ambient temperature. The catalyst particles are pushed apart from one another by the upflow of the gas at a sufficient velocity to cause mobility. Reaction occurs within the catalyst particles which are the sites of growing CNTs. Both the catalyst and the CNTs were smoothly fluidized in the reactor via proper selfagglomeration. The reaction temperature was maintained at 500 700 C, and typical synthesis times for our runs are between 30 and 60 min. The exact amount of carbon deposit formed during the reaction is determined by weighing the catalyst before and after reaction. The yield of deposited carbon is obtained from the total weight of the catalyst after the reaction by subtracting the initial weight of the catalyst before reaction. After reaction, the morphology and microstructure of the CNTs are observed in their as-prepared state using field emission scanning electron microscopy (SEM) and transmission electron microscopy (TEM). 2. Experimental The schematic diagram of the apparatus used in the experiment is shown in Fig. 1. The main body is a fluidized-bed reactor made of quartz glass with an ID of 250 mm and a height of 1 m. There is a sintered porous plate used as the gas distributor at the bottom of the reactor. The gas distributor also is the floor, which supports the weight of the solids above it before they are suspended in a fluid flow. In our experiments, Fe=Al 2 O 3 powder was used as catalyst. This is fed into the reactor before reaction. The gas mixture containing a carbon source reactant enters into the bottom vessel of the reactor, and then passes through the gas distributor, the fluidized-bed units, and finally flows out into the atmosphere. Ethylene and propylene were used as gaseous reactants separately. The flow of reactant feed was 5 10 m 3 =h diluted with 0 0:5 m 3 =h nitrogen and 0 0:5 m 3 =h hydrogen at atmospheric Fig. 1. Schematic representation of the experimental setup.
570 Y. Wang et al. / Chemical Physics Letters 364 (2002) 568 572 3. Results Weight yield (100 2000%) of deposited carbon were obtained according to different operating conditions. It is worthwhile to mention that the TEM and SEM images shown here are of as-grown materials, and no purification was performed before the imaging. With respect to quality, the only target is the diameter of the carbon nanotubes. The TEM photograph shown in Fig. 2 indicates that the carbon nanotubes are 10 nm in outer diameter, about 3 5 nm in inner diameter and more than several micrometers in length. The SEM photograph in Fig. 3 shows entangled carbon filaments and their abundance is very high in all the observed samples. From several SEM images, the yield of these filaments can be estimated approximately to be of the order 70 80%, depending on the catalyst used and the operating Fig. 3. SEM photograph of the tangled carbon nanotubes. conditions. For a 30 min reaction, we obtained a maximum of 5 kg of carbon deposit in the unit and with 10 reactions per day, the daily production is about 50 kg of products containing essentially multi-walled carbon nanotubes (MWNTs). The nanotubes made with this method are usually tangled and in loose powder form. This is exactly one of the important reasons that they can be fluidized smoothly. Another reason is that they have a peculiar agglomerate structure. As shown in Fig. 4, the nanotubes cohere into agglomerates from several microns to several tens of microns, and then further coalesce with each other in a fractal structure. The CNT agglomerates are freeflowing and possess low fluidized and settled bulk densities below 200 kg=m 3. This permits good Fig. 2. TEM photograph of the multi-walled carbon nanotubes. Fig. 4. SEM photograph of the CNT agglomerates.
Y. Wang et al. / Chemical Physics Letters 364 (2002) 568 572 571 Fig. 5. Fluidization regimes of the CNT agglomerates. fluidization at the operating conditions under which CNTs are produced in the fluidized-bed reactor. Fig. 5 shows the typical bed expansion and pressure drop variation with the superficial gas velocity tested under a condition without reaction. These indicate that when the CNTs, produced in a NAFBR, are exposed to an increasing up-flow of gas, the pressure drop and bed expansion will increase with gas flow. A point is reached when the pressure drop is enough to balance the weight of the bed. At this point, the bed is fluidized. The minimal fluidization velocity of CNT agglomerates is about 0.006 m/s. When fluidized, the mass behaves like a fluid that tends to establish a level and flows in response to pressure gradients. With increasing gas velocity, several flow patterns or regimes have been identified, such as particulate fluidization, bubbling fluidization and turbulent fluidization. If the gas velocity is further increased, fast fluidization and pneumatic conveying regimes can be realized. 4. Discussion Agglomerate fluidization is complicated in practice. The hydrodynamic parameters are different from those for non-agglomeration beds, and the agglomerate structure and product type have a major effect on the fluidization parameters. The success of the fluidized-bed process derives from a match of the capabilities of a fluidized-bed reactor with the characteristics of CNT growth by the CCVD method. First of all, a special challenge posed by the gasphase fluidization of CNTs is to keep them fluidized during the growing process within the reactor. The ease with which agglomerates fluidized and the range of operating conditions that sustain fluidization vary significantly among gassolid systems. For the growth of CNTs in a NAFBR, the two-phase flow behavior is mainly decided by the properties of the fluidized agglomerates, and this has a significant effect on the reaction, and vice versa. From the viewpoint of the agglomerate fluidization of nanomaterials, the agglomerate structure is of fundamental importance. If a carbon nanotube exists as a single dispersed tube, it is easily entrained by the fluidizing gas. If nanotubes bundle up too tightly with each other in a compact form, fluidization also cannot be achieved. The agglomerate structure can be controlled by the careful selection of catalyst support and operating conditions. In this case, the challenge is not only to make CNTs with unique, desirable properties, but also to design catalysts that provide the desired operability characteristics in fluidized beds. By the proper preparation of the catalyst, careful design of the fluidized-bed reactor and specification of the operating range, CNTs with an ideal agglomerate structure, as shown in Fig. 4, can be obtained. Secondly, the typical agglomerates-fluid twophase flow patterns exhibit complexity through self-organization of agglomerates and fluid. The maintenance of a constant average agglomerate diameter is essential for stable operation while the CNTs are developing in the fluidized-bed. Possessing a loose structure, the agglomerates are not very firm and when fluidized, large agglomerates can be broken up by the drag force or interaction among them. On the other hand, small pieces can escape elutration by sticking to other pieces. Although a large amount of nanotubes are formed and grow in the fluidized-bed reactor, the average agglomerate size in a fluidized-bed reactor does not vary significantly with time because larger agglomerates are broken into pieces in the intense flow. Fluidization is a process of choice for the largescale production of CNTs because such reactors
572 Y. Wang et al. / Chemical Physics Letters 364 (2002) 568 572 provide a large effective surface area and plenty of space for the growth of CNTs. Many works indicate that the presence of a substrate is essential for nanotube formation, and the yield in any given method is determined by the surface area that is available in the hot zone of the reactor. In a horizontal fixed-bed reactor, the diffusion of the carbon source gas to the catalyst particles becomes rate limiting as more and more nanotubes are grown on the surface, and only catalyst particles on the surface can be efficiently used. In an NA- FBR, a large amount of nano-metal catalysts are highly dispersed on the support and suspended in a gas flow, so that all the catalysts are equally effective. During the growth of the CNTs, the reaction zone augments the expansion of the fluidized-bed. Thus much space is available for carbon deposition. Another important advantage of an NAFBR is that heat and mass transfer is rapid. Good heat and mass transfer maintains the entire fluidized bed at the same temperature and reactant concentration. The whole vessel of well-mixed solids represents a large flywheel that limits rapid temperature and concentration changes, and responds slowly to abrupt changes in operating conditions. Thus, the operation can be controlled simply and reliably. On one hand, it assures that carbon atoms can diffuse rapidly over and through the metal, driven by a temperature or concentration gradient, to a location where they assemble into an ordered structure. The even conditions in the reactor are responsible for the uniform products of CNTs with good quality. On the other hand, it maintains a rapid process with the use of high-activity catalysts with the rapid movement of the reactant to the active catalyst site. The smooth, liquid-like flow of the agglomerates allows continuous automatically controlled operations with easy handling. Thus it is suitable for continuous large-scale operations. A process of this nature can be easily scaled up to several thousand tons per year, which will open up new opportunities for the industrial application of carbon nanotubes. Although MWNTs has been successfully produced on a large scale in a NA- FBR, further investigations are needed for SWNTs synthesis and to achieve a continuous process. 5. Conclusions Agglomerate fluidization is a competitive method for engineering processes for producing and handling nano-materials. A nano-agglomerate fluidized-bed reactor (NAFBR) can be simply and inexpensively operated at atmospheric pressure and moderate temperatures. It gives 70 80% multi-walled nanotubes with a high production rate of 50 kg/day. Acknowledgements Project 20006009 supported by National Natural Science Foundation of China (NSFC). References [1] Y.S. Park, Y.C. Choi, K.S. Kim, et al., Carbon 39 (2001) 655. [2] C. Journet, W.K. Maser, P. Bernier, A. Loiseau, et al., Nature 388 (1997) 756. [3] R. Andrews, D. Jacques, A.M. Rao, et al., Chem. Phys. Lett. 303 (1999) 467. [4] J.-F. Colomer, C. Stephan, S. Lefrant, et al., Chem. Phys. Lett. 317 (2000) 83. [5] E.F. Brooks, T.J. Fitzgerald, in: Elsinore, K. Denmark, Ostergaard, Ansgarsorensen (Eds.), Fluidizaiton V, Engineering Foundation, New York, 1986, p. 217. [6] J. Chaouki, C. Chavarie, D. Klvana, G. Pajonk, Powder Technol. 43 (1985) 117. [7] Y. Wang, F. Wei, Y. Jin, et al., in: Y. Jin, N.N. Li (Eds.), Proceedings of the Third Joint China/USA Chemical Engineering Conference (CUChE-3), Beijing, 2000, p. 12-006.