Rotation of Cu Nanorods during Growth - American Chemical Society

Mar 19, 2008 - Most Cu nanorods formed during growth rotate nearly monotonically around a fixed axis, and their instantaneous rotation speed follows a...
1 downloads 0 Views 404KB Size
J. Phys. Chem. C 2008, 112, 5459-5462

5459

Rotation of Cu Nanorods during Growth J.-X. Fu,†,‡ J. S. Wu,†,‡,§,| and Y.-P. Zhao*,†,‡ Nanoscale Science and Engineering Center, Department of Physics and Astronomy, The Center for Ultrastructural Research, UniVersity of Georgia, Athens, Georgia 30602 ReceiVed: December 20, 2007; In Final Form: January 23, 2008

The dynamic growth of Cu nanoparticles at high temperatures was observed in situ by a transmission electron microscope. During growth, Cu nanorods are formed and their aspect ratio increases from 1.24 to 4.59 with growth time and temperature (T e 920 °C). With further temperature increasing (at T ) 960 °C), all the particles become a spherical shape. Most Cu nanorods formed during growth rotate nearly monotonically around a fixed axis, and their instantaneous rotation speed follows a Gaussian distribution. The variation of the rotation speed is 3 × 10-3 rad/s at T ) 800 °C and 1.7 × 10-3 rad/s at T ) 920 °C, which infers a more active rotation at lower temperature. This phenomenon is believed to be caused by the difference in Cu adatom concentrations around the Cu nanorods during growth.

Introduction With a greater understanding of nanostructure growth mechanisms, it is easier to obtain nanostructures with desired morphologies and properties and develop desired heterogeneous structures. In most cases, growth of nanostructures is under the nonequilibrium condition and the growth mechanism is determined by a wide range of surface and interfacial interactions. In particular, for metallic or semiconductor nanostructure growth, the initial growth stage can be associated with the motion of the nanoclusters. Metallic clusters can move on the surface due to coalescence or Ostwald ripening.1-4 For heteroepitaxy growth of semiconductor nanoclusters, the nanoclusters can move on the surface due to the strain field.5 Surfacemediated alloying can also drive nanoclusters to move.6-8 The motion of the nanoclusters plays an important role in nanostructured thin film growth, and the motion mechanism can be used to design nanomechanical systems.9,10 In general, the mobility D of nanoclusters and the size N of the cluster follow D ∝ N-3/2.11 In most experimental observations, the size of the nanoclusters was small when compared to the size of the nanorod structures. For a nanoparticle or nanorod with millions of atoms, it is difficult to make the nanoparticle move. Here, we show the in-situ observations of Cu nanorod rotations during growth of the nanorod at varying temperatures. To our knowledge, this is the first report on this kind of phenomenon, and the rotation is believed to be caused by the asymmetric distribution of Cu adatom concentration gradient during Cu nanorod growth. Experimental Method A Cu grid with an ultrathin layer of carbon film (Lacy carbon), purchased from Ted Pella (with reference number of 01822), was heated gradually from room temperature to 960 °C in situ in a FEI Tecnai 20 transmission electron †

Nanoscale Science and Engineering Center. Department of Physics and Astronomy. § The Center for Ultrastructural Research. | Currently at Department of Materials Science and Engineering, Northwestern University, Evanston, IL 60208. ‡

microscope (TEM) using a Gatan double-tilt hot stage system (Gatan Inc., Warrendale, PA). During the experiment, the temperature of the TEM heating stage was subsequently set to T ) 800, 920, and 960 °C. From room temperature to 800 °C, the heating speed was 200 °C/min. An in-situ low-resolution TEM image of the carbon film was taken at a fixed temperature as a function of the annealing time. The image was taken, for a process lasting 20-30 min, every 0.5-1 min under a working voltage of 200 kV until no further changes were observed. Results and Discussions When a carbon-coated Cu grid was heated to T ) 800 °C, we started to observe formation of nanoparticles on the carbon film as shown in Figure 1a. From EDX and electron diffraction analysis, those nanoparticles were Cu particles. As the annealing time and temperature rose, the nanoparticles increased in size and the well-known Ostwald ripening process was observed. Although many particles took a circular shape or irregular polygon shape initially, some of the particles in fact took a rod shape, as shown in Figure 1a. In fact, at T ) 920 °C, most of the particles took the rod shape, as shown in Figure 1b. Particles with circular and irregular polygon shapes were also observed. Figure 2 plots the evolution of the average lengths of the long axis l and short axis a of all the Cu nanoparticles observed at T ) 800 and 920 °C as a function of annealing time. In order to see the trend more clearly, we plot the l-t and a-t relationships at T ) 800 and 920 °C in the same figure, and the time axis is relative. At T ) 800 °C, the average length of the short axis stays at a ≈ 376 nm and for the most part remains unchanged during the entire annealing process while the average long axis length increases monotonically with annealing time, from l ≈ 465 nm at t ) 7 min to l ≈ 670 nm at t ) 17 min, i.e., the average aspect ratio (γ ) l/a) of the particle increases from 1.24 to 1.78. The aspect ratio of the particles continues to grow at T ) 920 °C as shown in Figure 2. In the initial annealing period (t ) 19 and 24 min) when T ) 920 °C, the average length of the short axis remained almost unchanged, a ≈ 320 nm, while the average long axis length increased drastically from l ≈ 760 nm at t ) 19 min to l ≈ 1470 nm at t ) 24 min, i.e., the average aspect ratio γ of the particle increases from 2.38 to

10.1021/jp711970x CCC: $40.75 © 2008 American Chemical Society Published on Web 03/19/2008

5460 J. Phys. Chem. C, Vol. 112, No. 14, 2008

Fu et al.

Figure 1. Representative low-resolution TEM images for T ) (a) 800, (b) 920, and (c) 960 °C.

Figure 2. Average length of the long axis and short axis of the Cu nanorods versus growth time for different temperatures: T ) 800 and 920 °C.

4.59. The observed phenomenon demonstrates that during nanoparticle growth and annealing the short axis of the elongated particles remains approximately constant while the long axis keeps on growing. After t ) 24 min, l decreases slightly while a increases slightly as a function of time, which results in a decrease in the average aspect ratio γ (3.23 at t ) 31 min). This decrease in aspect ratio is due to the limited space for lateral growth of the nanorods, which will be demonstrated later. From the shape of the Cu nanorod and selective area electron diffraction pattern, the sidewalls of the Cu nanorods were associated with the (110) planes, while facets on the tips were associated with the (111) planes, which is similar to the Cu nanorods observed under electron beam irradiation.12 When the temperature further increases to 960 °C, all the particles, including the elongated particles, become spherical-shaped particles as shown in Figure 1c. Those spherical particles are probably the molten particles at such a high temperature. When the rod-shaped nanorods are kept at T ) 800 and 920 °C, we observed an intriguing phenomenon: the elongated particles rotate. This is exemplified in Figure 3, which shows the rotation of such a particle at T ) 800 °C. In the beginning (t ) 0 min), the particle has a short axis length of a ) 276 nm and long axis length of l ) 745 nm; the long axis angle with respect to the horizontal direction is θ ) 36.2°. At t ) 2 min, θ changes to 43.0° while a ) 270 nm and l ) 776 nm. Subsequent measurements for t ) 3, 8, and 9 min give θ ) 43.1°, 46.6°, and 57.5°, respectively. Clearly this demonstrates that the angle of the nanorod increases monotonically with the annealing time. During this period of time the projected shape of the nanorod

changed slightly, starting with the rectangular projection and ending with the polygon-shaped projection. In addition, the position of the left bottom tip of the nanorod does not significantly change with annealing time, as indicated in Figure 3: the red circle in each figure shows a reference point on the Cu grid and has a distance of ∼50 nm to the nearest Cu nanorod surface. This distance remains relatively constant during the annealing process. A similar phenomenon has also been observed for T ) 920 °C, as shown in Figure 4. The behaviors of two particles indicated by red arrows in Figure 4 are quite different. For the large aspect ratio nanorod on the left, the axis angle of the long axis with respect to the horizon decreases monotonically with the annealing time while the length of the nanorod keeps on growing and the short axis length is almost fixed. The long axis angle decreases monotonically from 38° at t ) 0 min to 3.6° at t ) 2.5 min. Then the long axis angle had almost no change, while the length of the nanorod keeps on growing for further annealing. In addition, during the growth process, the left end of the nanorod almost stays at the same location while the upper right end of the nanorod keeps on extending, which implies that one end of the nanorod is a pinning end and the other is a growth tip. The rotation axis goes through the pinning end and is perpendicular to carbon surface. At t ) 4 min (Figure 4h), the long nanorod starts to touch a big circular-shaped particle on the right and stop to grow longer. In the meantime, the circular particle on the righthand side, as indicated in Figure 4a, also kept on growing and even merged together with other small spherical-shaped particles on its left due to Ostward ripening (Figure 4c and d). However, it is very hard to observe the rotation of the spherical particle due to its topological symmetry. To further demonstrate this effect, Figure 5 plots the major axis angles versus time for three selected elongated particles at T ) 800 and 920 °C, respectively. One can observe that the angles change as a function of the annealing time. In general, the angle of the long axis changes almost monotonically with the annealing time, which means that the angular motion has certain directionality. The monotonic change of the long axis angle is associated with the growth of the long axis of the nanorod. When the long axis stops growing (t > 26 min in Figure 2), the nanorod rotation slows down significantly (t > 6 min in Figure 5b). However, the statistics on the angular speed for different temperatures show no preferred motion. Figure 6 plots the instantaneous angular speed distribution from 10 particles at each temperature obtained at different time intervals. The distribution is symmetric around ω ) 0°/min and can be closely fitted by a Gaussian distribution. For T ) 800 °C the angular speed distribution has a standard deviation of 10.3°/ min (or 3 × 10-3 rad/s), while for T ) 920 °C the standard deviation is 5.8°/min (or 1.7 × 10-3 rad/s), which is about one-

Rotation of Cu Nanorods during Growth

J. Phys. Chem. C, Vol. 112, No. 14, 2008 5461

Figure 3. Rotation of a Cu nanorod formed at T ) 800 °C. Frames a, b, c, d, and e correspond to observation times t ) 0, 2, 3, 8, and 9 min, respectively.

Figure 4. Rotation of a Cu nanorod formed at T ) 920 °C. Frames a, b, c, d, e, f, g, and h correspond to observation times t ) 0, 1, 1.5, 2, 2.5, 3, 3.5, and 4 min, respectively.

half of that at T ) 800 °C. This result demonstrates that at T ) 800 °C the nanorod rotates more actively than that at T ) 920 °C. From the above observation one can conclude that the nanorod rotation is closely related to the nanorod growth and that it has the following characteristics: (1) the growth of a nanorod starts from a pinning site, and the rod extends away from the pinning site during growth (no further growth at the pinning site); (2) the surface normal at the pinning site serves as the rotation axis for nanorod rotation; (3) the nanorod rotates during growth of the long axis of the nanorod. An illustration in Figure 7 summarizes the growth and rotation. The thermal fluctuation-induced rotation mechanism can be excluded since nanorod rotation is closely associated with the growth process and the angular change has obvious directionality. The surface rearrangement during growth can also be excluded since the length of the short axis remains unchanged and the number of Cu atoms in the nanorods was fairly large, on the order of ∼106-107. One possible mechanism to be considered results from the Cu adatoms diffusing from the surrounding environment and impinging the Cu nanorods, as illustrated in Figure 7. Figure 7 shows that Cu nanorod growth starts from a pinning site and extends to an active grown tip. This active tip could be extended into the vacuum or along the

carbon surface due to the local temperature difference. Diffusion of the Cu adatoms from the surrounding environment of a nanorod could induce diffusion flux fluctuation depending on the makeup of the surrounding environment of the nanorods. For example, there are excessive small Cu islands on one side of the Cu nanorod (Figure 7). These islands are not stable, and the adatoms can easily diffuse out to generate a higher Cu adatom flux on one side of the nanorod due to Ostwald ripening. There may be a large Cu cluster on one side of the nanorod which will deplete the Cu adatoms. Thus, a net impinging Cu adatom flux could induce a net torque and make the nanorods rotate.13 This difference of adatoms diffused toward the two long sides of the nanorod could be the main contribution of the driving force to rotate the nanorod. When the adatom impinges to the surface, it translates its kinetic energy during diffusion into the rotary energy of the nanorod. Statistically the number of adatoms attached to the two long sides should be randomly and uniformly distributed, and one can treat the diffusion of an adatom onto the side of the nanorod as a total inelastic collision, i.e., the drift momentum of the adatom contributes to the angular momentum of the nanorods. The net contribution is due to the number of fluctuated particles per unit time from both sides of the nanorod, denoted as ∆N, and, according to the conservation of energy, one has, 1/2∆NmaV2 ) Iω2 + Wf, where ma is the

5462 J. Phys. Chem. C, Vol. 112, No. 14, 2008

Fu et al.

Figure 7. Phenomenological model for Cu nanorod growth and rotation. The nanorod could be protruded from the carbon surface toward the vacuum.

one could obtain, ∆N ≈ Fll2ω/maV2. Since the temperature difference T ) 800 and 920 °C is fairly small, both maV2 and Fl at these two temperatures are comparable, while l ≈ 600 nm at T ) 800 °C and l ≈ 1200 nm at T ) 920 °C. If one uses the angular speed fluctuation as ω, then ∆N920/∆N800 ≈ 2, i.e., the average net flux at T ) 920 °C is two times that at T ) 800 °C. However, this is just a very preliminary explanation for this phenomenon. Conclusions

Figure 5. Angular motion of selected Cu nanorod particles for T ) (a) 800 and (b) 920 °C.

In-situ growth and rotation of Cu nanorods has been observed by transmission electron microscopy. Nanorod growth starts from a pinning site and extends out away from the pinning site. During the growth phase, the Cu nanorod also rotates almost monotonically about the surface normal on the pinning site. At 800 °C, rotation is more active than that at 920 °C. Rotation is believed to be caused by the Cu adatom concentration difference around the Cu nanorod during growth. This kind of rotation behavior could be used to synthesize nanomotors or other nanomechanical parts by engineering the local environment close to Cu or other metal nanorods. Acknowledgment. This work was supported by the National Science Foundation under contract no. CMMI-072670. The authors would like to thank Chris Hoffmann for proof reading the manuscript. References and Notes

Figure 6. Angular speed distribution obtained from 10 Cu nanorod particles at T ) 800 and 920 °C.

mass of a single adatom, I is the momentum of inertia of the nanorod, ω is the angular speed, and Wf is the energy dissipated by friction when the nanorod is moving on the carbon surface. Since the nanorod is uniform along the long axis, Wf ) 1/2Fll2ω, where Fl is the friction force per unit length. The dissipation is to overcome the physisorption or chemisorption of the Cu nanorod on the carbon surface, and in general, Wf . Iω2. Thus,

(1) Jose-Yacaman, M.; Gutierrez-Wing, C.; Miki, M.; Yang, D. Q.; Piyakis, K. N.; Sacher, E. J. Phys. Chem. B 2005, 109, 9703. (2) Trushin, O. S.; Salo, P.; Alatalo, M.; Ala-Nissila, T. Surf. Sci. 2001, 482, 365. (3) Morgenstern, K.; Braun, K. F.; Rieder, K. H. Phys. ReV. Lett. 2004, 93, 056102. (4) Karim, A.; Al-Rawi, A. N.; Kara, A.; Rahman, T. S.; Trushin, O.; Ala-Nissila, T. Phys. ReV. B 2006, 73, 165411. (5) Liu, C. H.; Wu, W. W.; Chen, L. J. Appl. Phys. Lett. 2006, 88, 023117. (6) Schmid, A. K.; Bartelt, N. C.; Hwang, R. Q. Science 2000, 290, 1561. (7) Denker, U.; Rastelli, A.; Stoffel, M.; Tersoff, J.; Katsaros, G.; Costantini, G.; Kern, K.; Jin-Phillipp, N. Y.; Jesson, D. E.; Schmidt, O. G. Phys. ReV. Lett. 2005, 94, 216103. (8) Tu, Y. H.; Tersoff, J. Phys. ReV. Lett. 2007, 98, 096103. (9) Regan, B. C.; Aloni, S.; Jensen, K.; Ritchie, R. O.; Zettl, A. Nano Lett. 2005, 5, 1730. (10) Regan, B. C.; Aloni, S.; Jensen, K.; Zettl, A. Appl. Phys. Lett. 2005, 86, 123119. (11) Pal, S.; Fichthorn, K. A. Phys. ReV. B 1999, 60, 7804. (12) Wang, P. I.; Zhao, Y. P.; Wang, G. C.; Lu, T. M. Nanotechnology 2004, 15, 218. (13) Golestanian, R.; Liverpool, T. B.; Ajdari, A. Phys. ReV. Lett. 2005, 94, 220801.