Kinetics of catalyst size dependent carbon nanotube growth by growth ...

4 downloads 465 Views 273KB Size Report
CNT arrays.2,3 These processes usually involve different catalysts, barrier layers ..... This work was supported by Samsung Advanced Insti- tute of Technology ...
APPLIED PHYSICS LETTERS 96, 094101 共2010兲

Kinetics of catalyst size dependent carbon nanotube growth by growth interruption studies S. P. Patole,1,a兲 Hyeongkeun Kim,2 Jaeboong Choi,1,2 Youngjin Kim,1,2 Seunghyun Baik,1,2,3 and J. B. Yoo1,4,a兲 1

SKKU Advanced Institute of Nanotechnology (SAINT), Sungkyunkwan University, Suwon 440-746, Republic of Korea 2 School of Mechanical Engineering, Sungkyunkwan University, Suwon 440-746, Republic of Korea 3 Department of Energy Science, Sungkyunkwan University, Suwon 440-746, Republic of Korea 4 School of Advanced Materials Science and Engineering, Sungkyunkwan University, Suwon 440-746, Republic of Korea

共Received 19 November 2009; accepted 1 February 2010; published online 5 March 2010兲 The dependence of the growth kinetics of carbon nanotubes 共CNTs兲 on the size of the Fe-catalyst in the H2 assisted atmospheric pressure chemical vapor deposition was studied. A growth interruption method was used to determine the in situ growth rate. The formation of a compact scale contaminant layer around the catalyst hinders the diffusion of the reactant species required to grow the CNTs. The high temperature metal oxidation behavior observed using parabolic curve fitting was attributed to the size dependent catalyst activity. The parabolic rate constant shows linear dependence on the catalyst size. Details of the analysis are presented. © 2010 American Institute of Physics. 关doi:10.1063/1.3330848兴 The synthesis of vertically aligned, ultralong carbon nanotubes 共CNTs兲 is of technological importance, because they can be spun into fibers, cables and sheets for lightweight and high-strength material applications.1 Since 1996, various methods have been reported to produce vertically aligned CNT arrays.2,3 These processes usually involve different catalysts, barrier layers, carbon sources, and operation parameters, resulting in products with different morphologies and qualities. However, none of these CNT growth processes can overcome the gradual deceleration and eventual termination of growth. The ability to understand and overcome the underlying deactivation mechanisms is one of the key steps in the application of nanoscale tubes to real macroscopic materials. Until now, many theories have been proposed to account for the observed CNT growth mechanism.4 In addition, specific techniques have been developed for the in situ diagnostics and, thus, control of the CNT growth.5,6 Recently, we reported the in situ growth interruption method, which can be used to examine the kinetics of CNT growth in water assisted chemical vapor deposition 共CVD兲.7 This paper reports the size dependent catalytic activity using growth interruption studies in H2 assisted atmospheric pressure CVD 共AP-CVD兲. The details of the substrate preparation were reported elsewhere.8 In summary, initially, a 15–18 nm thick Albarrier layer was deposited on a Si/ SiO2 wafer by electron beam evaporation. Subsequently, the Si wafer was cut into several pieces to deposit the Fe-catalyst. Fe was deposited at different thicknesses in a similar manner to that of Al. The deposition rate was kept fixed at 0.1 Å/s and a 0.5 to 3 nm thick Fe layer was deposited. The synthesis was carried out in a horizontal quartz tube furnace with an inner diameter of Authors to whom correspondence should be addressed. Tel.: ⫹82-31-2907396. FAX: ⫹82-31-290-7410. Electronic addresses: [email protected] and [email protected].

a兲

0003-6951/2010/96共9兲/094101/3/$30.00

29 mm. He 关150 SCCM 共SCCM denotes cubic centimeter per minute at STP兲兴 was used to purge the tube furnace for 13 min while it was ramped up to 750 ° C. After reaching 750 ° C, H2 共55 SCCM兲 was supplied for 5 min. The gas flow was then switched to C2H4 共30 SCCM兲, He 共150 SCCM兲, and H2 共55 SCCM兲 for the CNT growth. After 4 min of growth, the C2H4 feed stock was turned off for 1 min. During this interval, the Fe-catalyst was treated exclusively with He and H2. This growth interruption cycle was repeated four times to obtain five stacks with four marked regions. After 40 min, the supply of H2 and C2H4 was turned off and the system was cooled to room temperature in an inert He atmosphere. Supporting Fig. S1 illustrates the growth interruption scheme.9 After removing the samples from the reactor, they were analyzed by scanning electron microscopy 共SEM; JSM6700F, JEOL兲 to determine the height and morphology of the CNT forest. There are four indistinct lines on the SEM image of the as-grown CNT array 关Figs. 1共a兲–1共d兲兴. The resulting CNT arrays were divided into five stacks with different lengths through a series of parallel lines. The stack heights 共from the top of the CNT forest兲 for the different Fe thicknesses are summarized in Table S-I of the supporting information.9 The summation of the heights of the consecutive stacks gave the variation in the height of the CNT forest with the growth time, which is plotted in Fig. 1共e兲. In the case of a thickness of Fe of 0.5 nm after the initial 4 min growth time, the CNT height is around 222⫾ 10 ␮m and it reached 1018⫾ 10 ␮m after 20 min. The CNT height profile varied with the Fe thickness. In the initial 4 min, the CNT heights for Fe thicknesses of 1, 2, and 3 nm are around 184, 157, and 87⫾ 10 ␮m, respectively, and reached 844, 716, and 394⫾ 10 ␮m in the next 20 min. It was observed that the efficiency of the CNT growth process 共rate of graphitic wall growth兲 depends on the thickness of the Fe layer.8,10 The thickness of the Fe layer governs the number of walls in the vertically aligned CNT growth, which ultimately affects the

96, 094101-1

© 2010 American Institute of Physics

Downloaded 08 Mar 2010 to 115.145.200.148. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp

094101-2

Patole et al.

Appl. Phys. Lett. 96, 094101 共2010兲

FIG. 1. 共Color online兲 CNT stack formation on 共a兲 0.5, 共b兲 1, 共c兲 2, and 共d兲 3 nm Fe thick samples. The white line indicates the stack boundaries, which are not seen clearly in the image. 共e兲 The evolution of the CNT height with the growth time for the samples with different Fe thicknesses. The dotted lines are guidelines for the eye.

growth rate 共required carbon per unit length of a nanotube兲. The present results are concomitant with those in previous reports.8,10 To minimize the error originating from the misaligned CNTs, the average growth rate 共total stack height/stack growth time兲 in ␮m / min was calculated and plotted as a function of the growth time in Fig. 2. It can be clearly seen that the growth rate varied with the thickness of the catalyst, as well as the growth time. In the first 4 min, the Fe layer with a thickness of 0.5 nm leads to a growth rate of around 56⫾ 3 ␮m / min and changes to 47⫾ 3 ␮m / min at the 20th min. Similarly, the samples with Fe thicknesses of 1, 2, and 3 nm show growth rates of 46, 39, and 22⫾ 3 ␮m / min after the initial 4 min and change to 39, 33, and 18⫾ 3 ␮m / min at the 20th min, respectively. The observed variation in the growth rate with the Fe thickness can be assigned to the size dependent catalyst activity. It should be noted that the amount of feed stock gas used was the same for all of the samples and, therefore, the apparent change in the growth rate might be related to the different rates of consumption of the C species required for the formation of the CNTs. This quantitative conclusion is supported by the observed variation in the number of walls with the Fe thickness.8,10 The number of walls in the CNTs changed from 2 to 17 as the Fe thickness increased from 0.5 to 3 nm.8 The larger sized catalyst consumes more C species and forms a larger number of walls than the smaller sized catalyst, resulting in an overall decrease in the growth rate. It is assumed that the available number of C species near the catalyst per unit time is constant. The change in the CNT growth rate with time cannot

FIG. 2. 共Color online兲 CNT growth rate as a function of the growth time for samples with different Fe thicknesses. The dotted lines are guidelines for the eye.

be related to the change in the number of walls in the CNTs, and indeed we do not observe any change in the number of walls at the bottom and top of the CNT forest. Therefore, this type of peculiar trend can be understood with the further analysis of the data. The catalytic activity can be monitored using the stack height.7 It is assumed that the apparent CNT stack height is representative of the catalyst efficiency 共see supporting information S2兲.9 The high temperature CNT growth environment caused the formation of an oxide and carbonaceous layer around the catalyst, which acts as a diffusion barrier layer to the reactant species. The formation of an oxide and carbonaceous layer around the catalyst depends upon the partial pressure of the feed stock and, under balanced reaction conditions, both phenomena occur simultaneously, resulting in the highest CNT growth rate.3 Without catalytic deactivation, the growth rate would be expected to be constant, in the case where there is no resistance to diffusion, and proportional to t1/2 共where t is the growth time兲 in the strong diffusionlimited regime.11 At high temperatures, the oxidation or carbonization of many metals was found to follow a parabolic time dependence,12 x2 = k pt + C, where x represents the thickness of the oxide or carbonaceous film, k p represents the parabolic rate constant, t denotes the time, and C denotes the constant. High temperature parabolic oxidation or carbonization indicates that the rate depends on the thermal diffusion. Such a process might include uniform diffusion of one or both reactants through a growing compact scale. In contrast with those of other research groups,13 we report that the diffusion of the reactant species occurs across the barrier layer formed around the catalyst and not across the CNT forest.7 In the present case, however, in which we use H2 as the reducing gas, the carbonization or oxidation of Fe might have caused the formation of the diffusion barrier layer around the catalyst. The C species has to overcome this barrier layer, in order to form the CNTs. An increase in the thickness of the barrier layer causes a decrease in the CNT growth rate, because the CNT growth rate is inversely proportional to the barrier layer thickness. If the barrier thickness follows parabolic rate behavior, then the inverse of the CNT growth rate has to follow parabolic behavior. If there is a greater amount of barrier scale, the diffusion of C species will be reduced and, consequently, the growth rate will be decreased. Concomitant with this analogy, the inverse of the CNT growth rate was plotted as a function of square root of the growth

Downloaded 08 Mar 2010 to 115.145.200.148. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp

094101-3

Appl. Phys. Lett. 96, 094101 共2010兲

Patole et al.

FIG. 3. 共Color online兲 The inverse of the CNT growth rate as a function of square root of the growth time. The dotted lines represent the best linear fit to the data points.

time to obtain a clearer insight into the growth kinetics 共Fig. 3兲. A straight line was fitted to the data points as a guide. The fitting parameters are shown in Table S-II of the supporting information.9 The parabolic rate constant determines the formation of the diffusion layer around the catalyst. The parabolic rate constant depends on the Fe thickness. The rate of diffusion of the reactant species across the hindering film around the catalyst determines the growth kinetics. The diffusion mechanism leads to a parabolic relationship, i.e., the inverse of the CNT growth rate is proportional to the square root of the growth time. Figure 3 shows the parabolic relationship, which supports the existence of diffusion controlled kinetics. The slope of the graph decreases with the Fe thickness. The plot of the parabolic rate constant as a function of the Fe thickness is shown in Fig. 4. The diameter of the Fe clusters determined from the high resolution transmission electron microscopy 共HRTEM兲 observations from previous report 共Ref. 8兲 is also plotted along the X⬘-axis in the same graph. The values of K p for Fe thicknesses of 0.5, 1, and 2 nm are around 1.7, 2.6, and 3.9⫻ 10−6 ␮m2 / min, respectively. The value is around 1.4⫻ 10−5 ␮m2 / min for an Fe thickness of 3 nm. The increase in K p with increasing catalyst size is due to the difference in the catalyst activity. The formation of a diffusion barrier layer around the catalyst depends on the catalyst size. The larger the size of the catalyst,

FIG. 4. 共Color online兲 Parabolic rate constant as a function of Fe thickness. In the same graph, X⬘ represents the Fe cluster diameter derived from the HRTEM analysis from Ref. 8. The shaded oval region indicates the longer sustainability of the catalyst for the CNT growth, and also shows linear tendency with the Fe thickness. The inset shows the 1.3 cm high CNT forest grown on the 1 nm Fe thick film after a growth time of 13 h.

the greater is the tendency to form a diffusion barrier layer and the lower is the CNT growth rate. More analysis is required to understand the exact phenomenon. It was also observed that the parabolic rate behavior is dominant in the initial growth process and follows paralinear rate behavior in the long growth regime given by the following:14 x = 共k p / kl兲ln共k p / k p − kl共x − klt兲兲 where Kl denotes the paralinear rate constant. In such type of reactions the formation of a diffusion barrier layer 共carbonization or oxidation兲 and the removal of the diffusion barrier layer by H2 共reduction兲 took place simultaneously and led to the paralinear behavior, and further research into this phenomenon is underway. It is interesting to note that such types of reaction lead to the formation of infinitely long CNT arrays. Indeed, we could not grow the CNTs beyond the diameter of the quartz tube for the shaded region of catalyst thickness shown in Fig. 4. A typical CNT forest with a height of 1.3 cm grown in the case of the sample with an Fe thickness of 1 nm for 13 h duration is also shown in the inset of Fig. 4. In summary, growth interruption studies were carried out to examine the dependence of the growth kinetics of CNTs on the size of the catalyst in H2 assisted AP-CVD. The CNT growth shows catalyst size dependent activity. The formation of a compact scale of carbonaceous or oxide layer around the catalyst hinders the diffusion of the reactant species required to form the CNTs. A larger catalyst size shows a greater tendency to form a diffusion barrier layer, which ultimately deteriorates its ability to grow the CNTs. The balance between the formation and removal of the diffusion barrier layer leads to paralinear behavior and is thought to be the basis for infinite CNT growth. This work was supported by Samsung Advanced Institute of Technology and Basic Science Research Program through the National Research Foundation of Korea 共NRF兲 funded by the Ministry of Education, Science and Technology 共Grant No. 2009-0083540兲. The authors also appreciate the project and equipment support from Gyeonggi Province through the GRRC program in Sungkyunkwan University. 1

M. Zhang, K. R. Atkinson, and R. H. Baughman, Science 306, 1358 共2004兲. 2 R. Xiang, E. Einarsson, J. Okawa, Y. Miyauchi, and S. Maruyama, J. Phys. Chem. C 113, 7511 共2009兲. 3 S. P. Patole, P. S. Alegaonkar, H. C. Lee, and J. B. Yoo, Carbon 46, 1987 共2008兲. 4 O. A. Louchev, T. Laude, Y. Sato, and H. Kanda, J. Chem. Phys. 118, 7622 共2003兲. 5 T. Iwasaki, J. Robertson, and H. Kawarada, Nano Lett. 8, 886 共2008兲. 6 E. Einarrson, Y. Murakami, M. Kadowaki, and S. Maruyama, Carbon 46, 923 共2008兲. 7 S. P. Patole, J. H. Park, T. Y. Lee, J. H. Lee, A. S. Patole, and J. B. Yoo, Appl. Phys. Lett. 93, 114101 共2008兲. 8 S. P. Patole, P. S. Alegaonkar, H. C. Shin, and J. B. Yoo, J. Phys. D: Appl. Phys. 41, 155311 共2008兲. 9 See supplementary material at http://dx.doi.org/10.1063/1.3330848 for the growth interruption scheme, stack heights, CNT morphology, and fitting parameters. 10 B. Zhao, D. N. Futaba, S. Yasuda, M. Akoshima, T. Yamada, and K. Hata, ACS Nano 3, 108 共2009兲. 11 R. Xiang, Z. Yang, Q. Zhang, G. Luo, W. Qian, F. Wei, M. Kadowaki, E. Einarsson, and S. Maruyama, J. Phys. Chem. C 112, 4892 共2008兲. 12 P. Kofstad, High Temperature Corrosion 共Elsevier, London, 1988兲. 13 L. Zhu, D. W. Hess, and C. P. Wong, J. Phys. Chem. B 110, 5445 共2006兲. 14 E. W. Haycock, J. Electrochem. Soc. 106, 771 共1959兲.

Downloaded 08 Mar 2010 to 115.145.200.148. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp