Encapsulated Nanowires Formed by Nanotube-Assisted Oriented

NANO LETTERS

Encapsulated Nanowires Formed by Nanotube-Assisted Oriented Attachment

2004 Vol. 4, No. 11 2299-2302

I. Alexandrou,*,† D. K. H. Ang,‡ N. D. Mathur,§ S. Haq,| and G. A. J. Amaratunga‡ Department of Electrical Engineering & Electronics, UniVersity of LiVerpool, LiVerpool L69 3GJ, UK, Department of Engineering, UniVersity of Cambridge, Cambridge CB2 1PZ, UK, Department of Materials Science and Metallurgy, UniVersity of Cambridge, Cambridge CB2 3QZ, UK, and BAE Systems, AdVanced Technology Centre, PO BOX 5, Filton, FPC 267, Bristol BS34 7QW, UK Received September 3, 2004; Revised Manuscript Received September 22, 2004

ABSTRACT Carbon tubules are used to assemble Co3C or Co particles from an agglomerate into a straight line. The walls of the carbon tubules are grown by the slow catalytic decomposition of a fluorocarbon at 375 °C over the Co3C or Co particles. The nanoparticles then merge via oriented attachment to form nanowires. Depending on the nanoparticle material, nanowires of up to 1 µm long can be formed. Using atomic resolution electron microscopy images, the formed nanowires are seen to be of the same material as the initial nanoparticles.

Recently many methods have emerged for the formation of nanoparticles with control over their size. In their own merit such nanoparticles have significant technological importance.1 Even more impressive is however the manipulation of such nanoparticles to form new structures2,3 and shapes,4 leading to the prediction that new artificial optical and electrical materials can occur.5 The most popular/frequently used method for ‘fusing’ nanoparticles together is “oriented attachment” where two adjacent nanoparticles merge after they have fully aligned,2,6,7 or with a lattice mismatch to form complex structures.8,9 When building nanoscale devices, nanowires are very desired building blocks since one can use them to connect device components or their 1-D structure to extract novel properties.10 Nanowires can be formed by oriented attachment of nanoparticles produced in solution,4,11-15 by filling the voids of preconditioned templates16-21 or by a catalytic process.22,23 A common feature of all methods based on oriented attachment is that the nanoparticles initially seem to be assembled into a line via the help of surfactants or ligands encapsulating the nanoparticles in the solution, with oriented attachment taking over to ensure the particles’ fusion with crystallographic alignment. In the second and third mechanisms, the nanowires are formed because lateral growth is restricted by a rigid template or favorable kinetics, respectively. Even though practical applications in photon* Corresponding author, e-mail: [email protected]. † University of Liverpool. ‡ Department of Engineering, University of Cambridge. § Department of Materials Science and Metallurgy, University of Cambridge. | BAE Systems. 10.1021/nl048554d CCC: $27.50 Published on Web 10/15/2004

© 2004 American Chemical Society

ics11 and electrical circuits24 have been shown, it seems likely that nanowires will have their most profound effect on magnetic applications, e.g., spintronics.17,20,25 The reason is that nanowires of a magnetic material can, due to their shape, be permanently magnetized even if their diameter is close to or below the superparamagnetic limit.26,27 This makes ferromagnetic nanowires distinctively advantageous over any other ferromagnetic nanostructures, even though alternatives for use of ferromagnetic nanoparticles have been proposed.28 Oriented attachment seems to be the best candidate method for forming ferromagnetic nanowires because it allows control over their size and can be used to form extremely thin nanowires. However, in the initial steps required to align the nanoparticles into lines, an organic layer surrounding them is often used. The presence of an organic coating might lead to a reaction with the ferromagnetic core since such materials are known catalysts for the decomposition of organic molecules. In addition, bare ferromagnetic cores exposed to air are susceptible to oxidization, which would greatly affect their magnetic properties. In this paper we report the formation of Co3C and Co nanowires on an amorphous carbon thin film through the self-assembly of nanoparticles which need not be prearranged into lines. The nanowires are seen to form inside carbon nanotubes growing slowly from the catalytic decomposition of a fluorocarbon vapor over the Co3C and Co nanoparticles at 375 °C. Depending on the material of the initial nanoparticles, nanowires of up to 1 µm long can be formed. Using atomic resolution electron microscopy images, the formed nanowires are seen to be of the same material as the initial nanoparticles,

suggesting that it is possible to form nanowires of various materials via simply varying the initial nanoparticles. The single-crystal nature of the produced nanowires suggests that once inside the nanotube the nanoparticles merge via oriented attachment. Our work started during annealing experiments of Co3C35a particles at moderate temperatures (375 °C) for 72 h in vacuum or inert gas environment (initial pressure 8 × 10-6 mbar) to study carbon segregation out of the particle. Routinely these particles were dispersed on a holey carbon transmission electron microscopy (TEM) grid and were examined using high-resolution electron microscopy (HREM) prior to any thermal treatment. In one of our experiments, a fluorocarbon polymer (Kel-F: polychlorotrifluoroethylene, PCTFE) capsule loaded with carbon encapsulated Ni particles was also placed in the vacuum furnace as part of an independent study. However, at the process temperature the PTCFE capsule decomposes and at the end of the experiment (annealing at 375 °C for 72 h) we noted the remains of the decomposed capsule covering the whole chamber. No matter the obvious “contamination” of our Co3C sample, we decided to continue our HREM studies.35b Figure 1a shows carbon tubules encapsulating a metallic core, seen mainly growing radially out of areas where before the process we could only observe Co3C particle clusters. Figure 1b shows an HREM image of the single-crystal core of two nanowires. A continuous crystal wire is seen inside carbon tubules with small inside diameters (between 4.4 and 11.2 nm), while wider tubules are made of nanoparticles assembled via perfect and imperfect oriented attachment, as is discussed in more detail later. Images such as Figure 1b were used for phase identification of the filled core, always consistent with a projection of Co3C. The evident formation of metallic nanowire structures of the same material as the starting particles is an exciting prospect since this could allow control over the nanowire material by simply controlling the phase of the initial particles. To explore the prospect of phase customization and to verify the process itself, we applied the same process using Co-rich nanoparticles produced in the following way: a TEM grid containing Co3C particles was annealed in a preevacuated chamber at exactly the same conditions used for the growth of the nanowires (375 °C, 72 h) under a flow of N2 gas in the absence of fluorocarbon vapor. Phase identification revealed that carbon had almost entirely segregated out of the Co3C lattice, creating effectively carbon encapsulated Co-rich nanoparticles. The HREM images of the metal core of these particles were consistent with the projection of either R-Co or δ-Co. Independent magnetization measurements on bulk powder samples of this material verified the weak magnetization of the initial Co3C material and a subsequent 190× increase in magnetization when the Co3C particles were annealed in N2 gas flow at 375 °C for 72 h; consistent with carbon segregation out of the weakly magnetic Co3C phase. After this process, the particles were still arranged in agglomerates and no nanotubes or nanowires could be detected during TEM examination. This actually proved that the carbon encapsulated Co3C nanowires shown in Figure 1 were formed 2300

Figure 1. Co-annealing of Co3C nanoparticles and fluorocarbon: (a) Low magnification TEM image showing carbon tubules filled to a great extend with a metallic phase, (b) HREM lattice fringe image of two nanowires with apparently single-crystal metallic cores, encapsulated by 4-6 disordered graphitic layers. Phase identification showed that the metallic phase of the core is Co3C. The diameter of metallic single-crystal wires such as those shown here is between 4.4 and 11.2 nmnm as measured from several HREM images.

through the interaction between the Co3C particles and the fluorocarbon. A TEM grid with dispersed Co-rich nanoparticles was annealed in the presence of fluorocarbon vapor at conditions same as for the Co3C particles described above. Figure 2a shows a low magnification TEM image of the resulting material. In this case impressively long nanowires (up to 1 µm) could be seen in abundance. Similar to previous observation, the nanowires that are thicker than 11 nm appear (see Figure 2b) to form by merging nanoparticles without a continuous crystal lattice being formed throughout the volume of the core. Phase identification using HREM images such as the one shown in Figure 2b revealed that the core of the nanowires consists of R-Co encapsulated by an amorphous carbon (a-C) layer. The formation of carbon tubules with walls made of amorphous-like carbon is a common feature at low growth temperatures.29 The appearance of the Nano Lett., Vol. 4, No. 11, 2004

Figure 2. Co-annealing of Co-rich nanoparticles with fluorocarbon: (a) Low magnification TEM image of the resulting nanowires growing up to 1µm long, (b) HREM image of a nanowire like those in (a) showing convincingly that the metallic core is made of assembled nanoparticles. The single crystal Co wires are typically around 10 nm wide while the imperfect core can have a diameter of up to 32 nm. The core is always encapsulated by an amorphous carbon layer.

crystalline walls when the fluorocarbon decomposition took place over Co3C particles suggests a more efficient catalysis by Co3C, in agreement with previously observed Ni3C catalytic particles at the tips of nanotubes with highly crystalline walls.30 However, the improved catalysis seems to come at the expense of the growth rate since nanowires formed using Co-rich nanoparticles are much longer. Images such as Figure 2b are the strongest proof so far that the formed nanowires result from the self-assembly of the respective nanoparticles, while continuous crystalline cores such as those shown in Figure 1b are mainly seen for wires with a core diameter of about 11 nm or less (4.411.2 nm). The similarity between the core structure of our thin nanowires and that previously reported for other materials formed via perfect oriented attachment,11,12 and between our thick nanowires and material formed by imperfect oriented attachment,8,9 suggests that in our case the nanoparticles also merge via oriented attachment. On the other hand, there is little information of how the carbon tubules grow, encapsulating the nanoparticles and at the same time collecting them from their aggregates to assemble them into straight lines. The decomposed fluorocarbon certainly plays an important role because, in the absence of its vapor, nanowires are not formed. The catalytic action of the nanoparticles in the decomposition of the Nano Lett., Vol. 4, No. 11, 2004

fluorocarbon vapor is evident by the growth of the carbon walls around the metallic cores (nanowires). The carbon walls formed are structurally similar to the ones grown by chemical vapor deposition (CVD) over a catalyst. However, in our method we are not aware of the existence of strong carbon etchants such as ammonia and hydrogen, used during the catalytic growth of nanotubes by CVD. Therefore, there is no mechanism that would keep the surface of the nanoparticles clean for the reaction to continue and grow the carbon walls via the tip growth mechanism proposed for CVD nanotubes. Therefore, we anticipate that once the side of any given nanoparticle exposed to the fluorocarbon vapor is covered with carbon, additional carbon for building the carbon walls is provided by the decomposition of fluorocarbon over neighboring nanoparticles. It is the eventual joining of the carbon walls from one nanoparticle to the next, we propose, that gives rise to a base growth mechanism for the carbon tubules. A tip growth mechanism for the carbon walls and progressive encapsulation of nanoparticles they encounter in their path to finally form the nanowires seems unlikely because (a) in order to obtain straight nanowires (as is the case here) the initial nanoparticles should have been arranged in lines, and most importantly (b) the nanowires in many cases grow over holes on the grid where there are no nanoparticles to encapsulate. The joining of the carbon walls actually drives the nanoparticles into assembly inside the carbon tubule. Since carbon walls have formed mainly on the side of the particle exposed to the fluorocarbon, there is still clean surface for oriented attachment to take place. Merging of the nanoparticles via oriented attachment and the growth of the carbon tubule are two processes that are treated separately in our discussion, but it is very likely that they happen concurrently. If the joining carbon comes from individual nanoparticles (involving only two particles at a time), the result is a thin carbon tubule encapsulating a single metallic core since individual nanoparticles are more likely to be able to rotate and position themselves for perfectly oriented attachment. On the other hand, carbon might merge across the surface of more than one nanoparticle, which are exposed to the fluorocarbon vapor simultaneously. With the carbon across these particles now constituting the ‘front’ of the tubule, with additional carbon/particles joining from underneath, the result is a tube with a metallic core that has grown via imperfectly oriented attachment since there are more than one particles along its diameter making difficult their perfect alignment with respect to all their neighbors. Since the size distribution of our initial particles ranges from 4 to 15 nm with a maximum around 10 nm, it is expected that single core nanowires will have diameters in the same range while those with imperfect cores will have a larger diameter (typically 9-30 nm), in agreement with our observations. In this article we have demonstrated that nanowires encapsulated in carbon can be formed via the decomposition of a fluorocarbon over Co3C and Co nanoparticles at relatively low temperatures. This is the first time that selfassembly of nanoparticles out of their agglomerates is 2301

reported without redispersion or the use of external fields to guide the motion of the nanoparticles. Combination of this method with advances in the arrangement of nanoparticles from solution onto specially prepared templates to form custom shapes11,31,32 is expected to provide methods for producing arrays of nanometer-wide nanotubes and most importantly to turn linear assembles of nanoparticles into magnetic nanowires that are safely protected inside carbon walls. The encapsulation of the ferromagnetic material inside the carbon tubules provides protection of the metallic core against oxidization, which might prove crucial in the lifetime of the formed devices. The most attractive aspects of the invented technology are its simplicity and scalability. Acknowledgment. The authors and especially I.A. would like to thank Bill Eccleston in Liverpool University for proofreading the manuscript and suggesting improvements before submission. In this work, I.A. and K.H.A. performed the self assembly experiments; I.A. performed the electron microscopy work, the TEM image processing and phase identification and wrote the manuscript; N.D.M and G.A.J.A. were involved in discussions of the results and the potential use of the material as well as helped in manuscript compilation; S.H. was responsible for funding a large part of the research and enthusiastically supported the work. References (1) Alivisatos, A. P. J. Phys. Chem. 1996, 100, 13226-13239. (2) Banfield, J. F.; Welch, S. A.; Zhang, H.; Ebert, T. T.; Penn, L. R. Science 2000, 289, 751-754. (3) Zhang, H.; Banfield, J. F. Nano Lett. 2004, 4, 713-718. (4) Ghezelbash, A.; Sigman, M. B., Jr.; Korgel, B. A. Nano Lett. 2004, 4, 537-542. (5) Alivisatos, A. P. Science 2000, 289, 736. (6) Huang, F.; Zhang, H.; Banfield, J. F. Nano Lett. 2003, 3, 373-378. (7) Huang, F.; Zhang, H.; Banfield, J. F. J. Phys. Chem. B 2003, 107, 10470-10475. (8) Penn, R. L.; Banfield, J. F. Science 1998, 281, 969-371. (9) Shen, P.; Fahn, Y. Y.; Su, A. C. Nano Lett. 2001, 1, 299-303. (10) Xia, Y.; Yang, P.; Sun, Y.; Wu, Y.; Mayers, B.; Gates, B.; Yin, Y.; Kim, F.; Yan, H. AdV. Mater. 2003, 15, 353-389. (11) Tang, Z.; Kotov, N. A.; Giersig, M. Science 2002, 297, 237-240. (12) Pacholski, C.; Kornowski, A.; Weller, H. Angew. Chem., Int Ed. 2002, 41, 1188-1191. (13) Liu, B.; Yu, S.-H.; Li, L.; Zhang, F.; Zhang, Q.; Yoshimura, M.; Shen, P. J. Phys. Chem. B 2004, 108, 2788-2792. (14) Deng, Y.; Nan, C.-W.; Guo, L. Chem. Phys. Lett. 2004, 383, 572576. (15) Weller, H. Philos. Trans. R. Soc. London A 2003, 361, 229-240. (16) Whitney, T. M.; Jiang, J. S.; Searson, P. C.; Chien, C. L. Science 1993, 261, 1316-1319.

2302

(17) Wernsdorfer, W.; Doudin, B.; Mailly, D.; Hasselbach, K.; Benoit, A.; Meier, J.; Ansermet, J.-P.; Barbara, B. Phys. ReV. Lett. 1996, 77, 1873-1876. (18) Ge, S.; Li, C.; Ma, X.; Li, W.; Xi, L.; Li, C. X. J. Appl. Phys. 2001, 90, 509-511. (19) Lee, K. H.; Lee, H. Y.; Jeung, W. Y.; Lee, W. Y. J. Appl. Phys. 2002, 91, 8513-8515. (20) Vila, L.; Piraux, L.; George, J. M.; Faini, G. Appl. Phys. Lett. 2002, 80, 3805-3807. (21) Nakamura, A.; Mathunaga, K.; Tohma, J.; Yamamoto, T.; Ikuhara, Y. Nature Materials 2003, 2, 453-456. (22) Martensson, T.; Carlberg, P.; Borgstro, M.; Montelius, L.; Seifert, W.; Samuelson, L. Nano Lett. 2004, 4, 699-702. (23) Gudiksen, M. S.; Lauhon, L. J.; Wang, J.; Smith, D. C.; Lieber, C. M. Nature 2002, 415, 617-620. (24) Harnack, O.; Pacholski, C.; Weller, H.; Yasuda, A.; Wessels, J. M. Nano Lett. 2003, 3, 1097-1101. (25) Atkinson, D.; Allwood, D. A.; Xiong, G.; Cooke, M. D.; Faulkner, C. C.; Cowburn, R. P. Nature Materials 2003, 2, 85-87. (26) Gambardella, P.; Dallmeyer, A.; Maiti, K.; Malagoli, M. C.; Eberhardt, W.; Kern, K.; Carbone, C. Nature 2002, 416, 301-304. (27) Snoeck, E.; Dunin-Borkowski, R. E.; Dumestre, F.; Renaud, P.; Amiens, C.; Chaudret, B.; Zurcher, P. Appl. Phys. Lett. 2003, 82, 88-90. (28) Skumryev, V.; Stoyanov, S.; Zhang, Y.; Hadjipanayis, G.; Givord, D.; Nogue´s, J. Nature 2003, 423, 850-853. (29) Hofmann, S.; Kleinsorge, B.; Ducati, C.; Ferrari, A. C.; Robertson, J. Diamond Relat. Mater. 2004, 13, 1171-1176. (30) Ducati, C.; Alexandrou, I.; Chhowalla, M.; Robertson, J.; Amaratunga, G. A. J. J. Appl. Phys. 2004, 95, 6387-6391. (31) Warner, M. G.; Hutchison, J. E. Nature Materials 2003, 2, 272277. (32) Hutchinson, T. O.; Liu, Y.-P.; Kiely, C.; Kiely, C. J.; Brust, M. AdVanced Materials 2001, 13, 1800-1803. (33) Alexandrou, I.; Baxendale, M.; Rupesinghe, N. L.; Amaratunga, G. A. J.; Kiely, C. J. J. Vac. Sci. Technol. B 2000, 18, 2698-2703. (34) Hong, Y.-K.; Kim, H.; Lee, G.; Kim, W.; Park, J.-I.; Cheon, J.; Koo, J.-Y. Appl. Phys. Lett. 2002, 80, 844-846. (35) Fletcher, D. A.; McMeeking, R. F.; Parkin, D. J. J. Chem. Info. Comput. Sciences 1996, 36, 746-749. (a) The Co3C particles were formed by cathodic arc discharge between a graphitic rod anode and a cathode made of compressed Co and graphite powders (Co:graphite ) 1:1, weight) in a high gas pressure region. A more detailed description of the method can be found elsewhere.33 Contrary to the regular arrangement of particles in 2-D patterns reported by other researchers,34 in our case, the Co3C particles dispersed on a TEM grid from evaporation of their solution in toluene were seen to form agglomerates without any evident arrangement. (b) Our HREM analysis was carried out using a JEOL 4000EX II microscope with a resolution limit of 1.4 Å, using the HREM images for phase identification: the experimental HREM images were scanned directly from the negative and their digital diffraction patterns (DDPs) were obtained as the fast Fourier transform of the image. The DDPs were used to measure the lattice spacings and angles of the imaged phases with high accuracy and the obtained values were compared to well documented crystallographic data of known materials obtained from the EPSRC’s Chemical Database Service at Daresbury.

NL048554D

Nano Lett., Vol. 4, No. 11, 2004