Various bonding configurations of transition-metal atoms on carbon ...

APPLIED PHYSICS LETTERS

VOLUME 76, NUMBER 26

26 JUNE 2000

Various bonding configurations of transition-metal atoms on carbon nanotubes: Their effect on contact resistance Antonis N. Andriotisa) Institute of Electronic Structure and Laser, Foundation for Research and Technology-Hellas, P.O. Box 1527, 71110 Heraklio, Crete, Greece

Madhu Menonb) Department of Physics and Astronomy and Center for Computational Science, University of Kentucky, Lexington, Kentucky 40506-0045

George E. Froudakis Department of Chemistry, University of Crete, P.O. Box 1470, Heraklio, Crete, Greece 71409

共Received 24 January 2000; accepted for publication 2 May 2000兲 Our investigations reveal that the bonding of the transition-metal atoms on a single-wall carbon nanotube 共SWCN兲 depends on the detailed contact conditions. On the basis of our results, we suggest that the early 3-d elements 共Sc, Ti, and V兲 can be expected to be good candidates for making metal–SWCN contacts of low resistance, while contacts employing the late 3-d elements 共Fe, Co, and Ni兲 and Cu are expected to exhibit large contact resistance. © 2000 American Institute of Physics. 关S0003-6951共00兲03326-X兴

Recently reported measurements1–3 of the single-wall carbon nanotube 共SWCN兲 conductivity have shown unexpectedly high values for the contact resistance R c 共of the order of 1 M⍀兲. At the same time, R c values as low as one resistance quantum (h/2e 2 ⬇12.9 k⍀) have also been reported for multiwalled nanotubes 共MWNT兲 immersed into liquid mercury or gallium,4 contrary to the expected value of half a resistance quantum. Observation of such diverse results for R c and the technologically important issue of the contact resistance in the development of potential nanoscale electronic devices using SWCNs has led to various considerations of the R c problem. Thus, Tersoff5 pointed out that in case where the one-dimensional translational symmetry is conserved 共i.e., for SWCNs laying on top of a planar surface or embedded in a metal jellium兲, wave-vector conservation at the metal–SWCN contact may play an important role in R c . Xue and Datta,6 on the other hand, addressed the question of how the Fermi level E F alignment at the metal–SWCN interface affects the R c . In particular, they pointed out that the E F position depends on the detailed contact geometry and may or may not be located at the valence-band edge. They argued that charge-transfer effects can shift E F to the valence-band edge of the semiconducting SWCN while, when the metal–SWCN bond is strong, metallic screening by metal-induced gap states may lineup E F at the midgap. In the former case high-R c , and in the latter case low-R c , values can be expected. Finally, the charge-transfer effects were employed by Choi et al.7 as a partial justification for the absence of substantial ␲ character of the ␲-band complex in the metal–SWCN region. In particular, Choi et al. suggested that charge-transfer effects may give the ␲-band complex a downwards shift, which brings this band lower than E F , thus a兲

Electronic mail: [email protected] Electronic mail: [email protected]

b兲

leaving the ␲* band as the only transmission channel. In this way, Choi et al. explained the low R c (⬇h/2e 2 ) observed experimentally.4 All these works, however, ignore the important details of the metal–SWCN contact geometry, which can have a profound effect on the R c value. It is the purpose of the present Letter to remedy this situation by demonstrating the effects of the contact conditions on the R c values. Our conclusion is based on the results of our recent investigations of the interaction of the transition-metal atoms 共TMAs兲 with the SWCNs, the C60 molecule, and graphite. In these works,8–10 we have demonstrated that the detailed contact condition between a TMA and a (n,n)-SWCN 共or C60 or graphite兲 plays a crucial role in the development of their bonding character. In particular, it was shown that the Ni-共5, 5兲 SWCN bond exhibits a covalent character for Ni on a bridge site 共i.e., for a Ni located over the center of a C–C bond兲, while the Ni-tube bond becomes ionic for a Ni at the atop cite 共i.e., over a C atom兲. Among these two stable contact geometries for the Ni-共5, 5兲 SWCN system, the atop site was found more stable and suffering less charge transfer than at the bridge site. Similar results were obtained for the Ni on a 共10, 10兲 SWCN with the exception of greater charge transfer than that on the 共5, 5兲 tube. Furthermore, we have also shown that the electron density of states ␳ (E F ) at the Fermi energy is much smaller when the Ni atom is adsorbed at the atop site of a SWCN as compared with the ␳ (E F ) for Ni adsorption at a bridge site.11 We have used this observation to explain the anomalous temperature dependence of the resistance R(T) of the SWCNs. Analogous findings for the Cu-tube contact were reported by Kong, Ham, and Ihm,12 who attributed the observed high-R c values to the opening of a pseudogap at E F as a result of the metal–SWCN interaction. Our results for Ni on a 共5, 5兲 and a 共10, 10兲 SWCN do not indicate any gap development at E F and, therefore, do not support the conclusions of Kong, Ham, and Ihm. Our findings allow us to elucidate the relationship be-

0003-6951/2000/76(26)/3890/3/$17.00 3890 © 2000 American Institute of Physics Downloaded 01 Feb 2005 to 128.163.209.112. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp

Appl. Phys. Lett., Vol. 76, No. 26, 26 June 2000

Andriotis, Menon, and Froudakis

3891

TABLE I. Contact resistances for representative metal–tube contacts.

Metal

Bonding configuration

Type of nanotube

Exp/theory

Contact resistance

Reference

Aua Aua,c Au Pt Pt Ti Cu V, 共Sc, Ti兲 Ni 共Fe, Co兲

Not specified Not specified Not specified Not specified Not specified Not specified Ionicg Covalenth Covalenth

MWNTb MWNT SWNTd Metallic–SWNT Metallic–SWNT Metallic–SWNT Metallic–SWNT Metallic–SWNT Metallic–SWNT

Exp Exp Exp Exp Exp Exp Theory Theory Theory

⬎1 M⍀ 共4–30兲 k⍀ 1 M⍀ 1 M⍀e 共1–4兲 M⍀ ⬇150 k⍀f ⬇1 M⍀ 共10–20兲 k⍀ 共0.2–0.4兲 M⍀

3 3 2 13 14 15 12 Present work Present work

a

Two-terminal resistance. Multiwall carbon nanotube 共MWNT兲. c After exposing the metal–tube contact to an electron beam. d Single-wall carbon nanotube 共SWNT兲. e Low-temperature value; R c ⬇300 k⍀ at room temperature. f Room-temperature value obtained at zero-backgate voltage. g With a simultaneous opening of a gap at Fermi energy. h For the energetically more favorable bonding configuration. b

tween the bonding character and the value of ␳ (E F ). Furthermore, utilizing our results for the relationship determined between ␳ (E F ) and the bonding character, we can point out the following: 共i兲 For a TMA–tube contact of covalent type, the conduction is metallic and the contact resistance may be of the order of a quantum resistance. 共ii兲 For a TMA–tube contact of ionic type, the conduction is of the tunneling type and may be responsible for the high-R c values observed. 共iii兲 With the appropriate choice of the transition metal 共TM兲 and the growing conditions of the TMA on the SWCN, one may be in a position to prepare metal–tube contacts with the desired value of R c . In fact, our calculated values of ␳ (E F ) for Ni at the atop and bridge sites are consistent with the corresponding bonding character exhibited at these sites. Thus, as expected, at the atop site 共ionic bonding兲 the Ni d resonance appears shifted lower than E F and, therefore, the Ni contribution to ␳ (E F ) becomes negligible. On the other hand, for Ni at the bridge site 共covalent bonding兲 the Ni d resonance appears closer to E F contributing thus, substantially, to ␳ (E F ). Repeating our calculations by replacing the Ni atom by a V atom, we find that V can be adsorbed at atop, bridge, and hole sites with the hole site being more stable than the atop site and the atop site more stable than the bridge site on the 共5, 5兲 SWCN. At these sites, the corresponding ␳ (E F ) values of the V–tube system do not differ appreciably. Furthermore, it is found that contrary to the Ni, in the V case the larger ␳ (E F ) values are associated with the more stable hole and atop sites. Finally, it is observed that the V–tube bonds exhibit covalent character at all these sites 共although the bond strengths differ from site to site兲. According to our proposed explanation,11 the downward turn of the tube resistance, R(T), obtained as the temperature T increases, is attributed to the Ni 共and Co兲 atom adsorption at the walls of the tube, and the changes in ␳ (E F ) induced by these adsorption sites as the Ni atoms change their position by changing the temperature. Our results for V, on the other hand, do not justify similar R(T) dependence since ␳ (E F ) does not change appreciably when the V atom changes its adsorption site. However, our results indicate V to be a very

good candidate for developing Ohmic tube contacts with low R c . This is because the bonding at the most stable site 共hole兲 for the V atom on the 共5, 5兲 tube is strongly covalent. This is in contrast to the Ni case where the stable atop site exhibits ionic bonding. Also, our calculated charge transfer between V and the SWCN can be regarded as a reasonable justification of the suggestion by Choi et al.7 It is expected that the properties exhibited by V on the SWCNs will be shared by the early 3-d elements 共Sc, Ti, and V兲, if it is recalled that these elements exhibit qualitatively the same bonding character on graphite and C60 共see, for example, Ref. 9, and references therein兲. For the same reason, the late 3-d elements 共Fe, Co, and Ni兲 can be expected to follow the properties described for Ni. Also, on the one hand, the late 3-d elements when adsorbed on a SWCN can be expected to exhibit the R(T) anomaly and large R c values. This behavior should be expected for Cu, too, according to the results of Kong.12 On the other hand, low-R c values and absence of the R(T) anomaly should be expected when the early 3-d elements are bonded to the surface of a SWCN. Table I2,3,13–15 contains experimental and qualitative numerical estimates for resistances for some elements, including Au, Pt, and Ti. The present work is supported through grants by NSF 共OSR 98-62485, OSR 99-07463, and MRSEC Program under Award No. DMR-9809686兲, DEPSCoR 共OSR 99-63231 and OSR 99-63232兲, the Semiconductor Research Corporation 共SRC兲, and the University of Kentucky Center for Computational Sciences.

S. J. Tans, R. M. Verschueren, and C. Dekker, Nature 共London兲 393, 49 共1998兲. 2 R. Martel, T. Schmidt, H. R. Shea, and Ph. Avouris, Appl. Phys. Lett. 73, 2447 共1998兲. 3 A. Bachtold, M. Henny, C. Terrier, C. Srtunk, C. Schonenberger, J.-P. Salvetat, J.-M. Bonard, and L. Forro, Appl. Phys. Lett. 73, 274 共1998兲. 4 S. Frank, P. Poncharal, Z. L. Wang, and W. A. de Heer, Science 280, 1744 共1998兲. 5 J. Tersoff, Appl. Phys. Lett. 74, 2122 共1999兲. 6 Y. Xue and S. Datta, Phys. Rev. Lett. 83, 4844 共1999兲. 7 H. J. Choi, J. Ihm, Y.-G. Yoon, and S. G. Louie, Phys. Rev. B 60, R14009 共1999兲. Downloaded 01 Feb 2005 to 128.163.209.112. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp 1

3892 8

Appl. Phys. Lett., Vol. 76, No. 26, 26 June 2000

A. N. Andriotis, M. Menon, and G. E. Froudakis, Chem. Phys. Lett. 320, 425 共2000兲. 9 A. N. Andriotis and M. Menon, Phys. Rev. B 60, 4521 共1999兲. 10 A. N. Andriotis, M. Menon, G. E. Froudakis, and J. E. Lowther, Chem. Phys. Lett. 301, 503 共1999兲. 11 A. N. Andriotis, M. Menon, and G. E. Frudakis, Phys. Rev. B 61, R13393 共2000兲.

Andriotis, Menon, and Froudakis K. Kong, S. Ham, and J. Ihm, Phys. Rev. B 60, 6074 共1999兲. S. J. Tans, M. H. Devoret, H. Dai, A. Thess, R. E. Smalley, L. J. Geerligs, and C. Dekker, Nature 共London兲 386, 474 共1997兲. 14 A. Bezryadin, A. R. M. Verschueren, S. J. Tans, and C. Dekker, Phys. Rev. Lett. 80, 4036 共1998兲. 15 T. W. Tombler, C. Zhou, J. Kong, and H. Dai, Appl. Phys. Lett. 76, 2412 共2000兲. 12 13

Downloaded 01 Feb 2005 to 128.163.209.112. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp