© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

phys. stat. sol. (b) 245, No. 10, 2311–2314 (2008) / DOI 10.1002/pssb.200879582 p s sbasic solid state physics

b

statu

s

soli

di

www.pss-b.comph

ysi

ca

Dielectrophoretically assembled carbon nanotube-metal hybrid structures with reduced contact resistance

Nitesh Ranjan and Michael Mertig*

Max Bergmann Center of Biomaterials and Institute of Materials Science, Dresden University of Technology, 01062 Dresden,

Germany

Received 29 April 2008, revised 17 June 2008, accepted 18 June 2008

Published online 26 August 2008

PACS 72.80.Tm, 73.40.Cg, 73.63.Fg, 81.07.De, 81.16.Dn

* Corresponding author: e-mail michael.mertig@nano.tu-dresden.de, Phone: +49 351 463 39404, Fax: +49 351 463 39401

© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction Due to their unique structural, me-chanical and electrical properties [1, 2], carbon nanotubes (CNTs) are considered to be basic building blocks for the bottom-up fabrication of nanoelectronic devices which possibly may replace silicon-based components from cur-rent microelectronics in the future. The build-up of CNT-based devices requires CNT integration into main circuit components which has two basic routes. In one of them single CNTs are dispersed over a substrate, and then elec-trodes are deposited over them by standard lithography techniques [3, 4]. This method provides low CNT metal-contact resistance, but has the disadvantage of no site specificity, i.e., CNT interconnects are not made at desired locations of the substrate. Therefore, lot of work has been done on site-specific deposition of CNT between pre-fabricated electrodes that can, e.g., be accomplished by ap-plying dielectrophoresis (DEP) [5, 6]. Deposition of CNTs over electrodes, however, is usually associated with the occurrence of high contact resistance which considerably limits the electrical performance of CNT-based devices like multi-tube field-effect transistors (FETs) [6-9]. Re-cently, improvement of the switching ratio of CNT FETs

has been achieved by applying a specific nano-welding technique [10]. Here we report on the improvement of the contact resistance of CNT forming electrical conducting bridges between micro-fabricated gold contact pads by ap-plying a post-dielectrophoretic deposition of metal from an aqueous metal salt solution. The process of metal deposi-tion is self-regulating and leads to a considerable reduction of the contact resistance.

2 Experimental

2.1 Dielectrophoresis Here DEP was used both for the deposition of CNTs between the electrodes and also for the deposition of metallic sheath over CNT-CNT and CNT-electrode interfaces. DEP deals with forces generated due to the polariza-tion of particles when kept in an inhomogeneous electric field [11]. In the case of spherical particles, the induced di-poles experience a dielectrophoretic net force given by

22 Re | ( ) | ( )DEP p m rmsF V K Eπ ε ω= ∇ , (1)

where Vp is the volume of the particle, εm is the permittiv-ity of the suspending fluid medium, and Erms and ω are the

We report the bottom-up assembly of single-walled carbon

nanotube interconnects between micro-fabricated gold elec-

trodes by dielectrophoresis followed by a post-dielectro-

phoretic deposition of metal from an aqueous Pd salt solution.

The dielectrophoretic metal deposition along interconnects

takes preferentially place at sites where high contact resis-

tances occur. These create sufficiently high voltage drops,

and thus, large field gradients which lead to a local deposition

of metal. Since in turn the metal deposition leads to a reduc-

tion of the local resistances, the reported post-treatment proc-

ess is self-regulating, that is, the deposition of metal stops

when the contact resistances become sufficiently reduced.

Applying this process to single interconnects of carbon nano-

tubes a reduction of the overall resistance below 1 MΩ has

been achieved.

2312 N. Ranjan and M. Mertig: Dielectrophoretically assembled carbon nanotube-metal hybrid structures

© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-b.com

ph

ysic

ap s sstat

us

solid

i b

root-mean-squared value and the angular frequency of the applied electrical field, respectively. Re [K(ω)] is the real part of the Clausius-Mossotti factor which describes the frequency dependence of the force. Here we work at a rela-tively low frequency (300 kHz) where both metallic and semiconducting CNTs experience positive DEP, i.e., they move into the direction of increasing electric field that is towards the electrodes where they are deposited. Once a short CNT has been deposited at the rim of one electrode, due to its nanometer-sized diameter it creates a local field of higher intensity and gradient at its tip than the electrical field at the rim does. As a consequence, the CNT deposi-tion gets self-regulated [6]. Now, the following nanotube gets deposited at the tip of the already deposited CNT, consequently an electrical bridge of in-series assembled CNTs grows towards the opposite electrode. Recently, we have demonstrated that one-dimensional metallic nanowires can be directly grown from a metal salt solution when both high electrical field magnitude and gradient is accomplished to achieve a sufficiently high enough dielectrophoretic force to manipulate and deposit hydrated palladium ions by DEP [12]. Here, we apply this method as post-treatment to the already deposited CNT bridges to achieve a reduction of the contact resistances. 2.2 Sample preparation and characterisation Single-walled CNTs produced by laser ablation method were used for the experiment. The as-produced CNT bun-dles were dispersed in water by sonication in presence of RNA as described in Reference [13]. The preparation of the palladium salt stock solution in 10 mM HEPES buffer (pH 6.2) is described elsewhere [12]. The planar gold electrodes with a gap distance of 2 μm were manufactured by standard optical lithography using either thermally oxidized n-type silicon wafers (100) or glass slides as the substrates. The electrodes have a thick-ness of 25 nm. A 5 nm thick Cr grafting layer was used to accomplish adhesion of the Au film to the substrate.

Figure 1 Schematic viewgraph of the DEP experiment. AC elec-

tric field is applied to the gold electrodes by a function generator

(FG). The current flowing through the circuit is measured by an

oscilloscope (Osc). For the dielectrophoretic deposition, the solu-

tion containing CNT or metal salt is put on top of the electrodes.

The experimental setup is schematically depicted in Fig. 1. In a typical experiment, first 10 μl of the CNT solu-

tion was placed on the electrodes and an AC voltage of 300 kHz and 4 V peak-to-peak amplitude was applied using a function generator AGF320 (Tektronix). The formation of one CNT bridge between the electrodes was monitored by measuring the current flowing in the circuit with an oscil-loscope TDS3014B (Tektronix) added in series. The single connection was detected by observing the beginning of the flow of an offset current in the circuit, and then the deposi-tion was stopped. After assembling a CNT bridge, excess solution was blotted with filter paper. Thereafter, the sam-ple was rinsed and air dried. The morphology of the CNT interconnect was then characterized by atomic force mi-croscopy (AFM) using a NanoScope IIIa (Digital Instru-ments) operated in tapping mode. The electrical characteri-zation of the sample was performed by measuring the cur-rent-voltage (I-V) characteristics using a DC Keithley 2602 digital source meter unit. Then 10 μl of a 40 times diluted Pd salt stock solution was placed on the electrodes and DEP was repeated at 30 kHz and peak-to-peak amplitude of 1 V. The metal deposi-tion process could clearly be monitored by an increase of the current in the circuit during the deposition. After blot-ting, rinsing and drying the sample as described above, the formed CNT-metal hybrid structure was again structurally and electrically characterized. 3 Results and discussion Figure 2 compares the structures of one and the same sample after the DEP for-mation of one CNT interconnect and after the DEP post-treatment with Pd salt solution. In the AFM image shown in Fig. 2a, a single CNT bridge connects the two electrodes. As depicted in the inset with enhanced z-resolution the CNT bridge is deposited on top of the gold electrode over-lapping at a distance of a few hundred nanometers. We suppose that the single CNT bridge consists rather of a se-ries of short CNTs than of one long nanotube because the mean length of the CNTs after ultrasonic dispersion is a few hundred nanometers well below the distance between adjacent electrodes. This assumption is supported by the fact that on both electrodes short CNTs were initially de-posited by DEP which did not grow to a complete CNT bridge because dielectrophoretic wire growth is a physical self-regulating process. The fastest growing bridge experi-ences the highest electrical field, and thus, inhibits the fur-ther growth of neighbouring bridges. In addition, the AFM image shows that the height of the single CNT interconnect is not constant. At several sites along the CNT bridge indi-cated by arrows the height is locally enhanced indicating locations where adjoining nanotubes are deposited with one end lying on top of the other nanotube’s end. After the DEP post-treatment with Pd salt solution (Fig. 2b), metal is deposited at the Au contacts, partially along the CNT bridge in the form of discontinuous sheaths and on randomly distributed locations of the substrate. Since the metal deposition is not observed when no electrical AC voltage is applied to the sample, we can conclude that the metal deposition is caused by the occurrence of dielectro-

phys. stat. sol. (b) 245, No. 10 (2008) 2313

www.pss-b.com © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Original

Paper

phoretic forces by which hydrated palladium ions are at-tracted into steep local gradients of the electrical field and get reduced when reaching conducting area of the sample surface. We assume that these conditions can be satisfied at different sites of the sample: (i) In agreement with the recently reported mechanism for the initiation of metal nanowire growth [12], metal will be deposited to small in-homogeneities at the gold electrodes as these cause high electric field gradients. (ii) Metal is also expected to be de-posited by DEP at regions of the CNT bridge having high resistance which is - as sketched in Fig. 3 - at the CNT gold-contact interface and at locations where CNTs adjoin each other because in a close circuit loop a region having high resistance R is associated with high potential drops (ΔV = IR). Hence, a high electric field is generated around this region where hydrated metal ions can undergo positive DEP, thus being dragged, reduced and deposited over the CNT bridge. As the force is extremely short range (F ~ 1/r5), we can expect the deposition to be locally restricted to the high resistance regions on the bridge. In this case, the deposition process is self-regulated because it contin-ues only until the local resistance drop decreases due to the metal deposition to a critical value where the field, and hence, the net force is no longer strong enough to deposit more Pd. (iii) Since the applied external AC field also in-duces electrical dipoles, and therefore, the formation of electrical field gradients in nanotubes which are not part of the CNT interconnection, metal can also be deposited there as long as the field magnitude and gradient exceed the critical value for Pd deposition. This third mechanism ex-plains why CNTs which did not form complete intercon-nects or are non-specifically adsorbed between the two gold electrodes might be covered with Pd, too. In Fig. 4 the I-V characteristics of the structure de-picted in Fig. 2 is given for the two states, before and after the DEP post-treatment. The I-V characteristics of the as deposited CNT bridge is slightly asymmetric at positive and negative bias voltages. This could be attributed to the fact that the electronic transport barrier created at CNT metal contact may be different at the two electrodes since we deposit single-walled CNTs with a wide variety of elec-tronic properties ranging from metallic to semiconducting. The resistance of the bare CNT interconnect varies be-tween 2 and 40 MΩ reaching the highest value around zero bias. After the DEP post-treatment the bridge resistance is decreased in the whole range of bias voltage. It ranges from 0.8 to 14 MΩ. The I-V characteristics is non-linear around zero bias where the resistance has its maximum value. For a quantitative analysis, we consider the resis-tance change at the bias voltage of –3 V. At these condi-tions we observe a decrease of the overall resistance of about 76% after the deposition of metallic sheaths onto the CNT bridge. We have performed more than 10 experi-ments which all gave comparable results.

Figure 2 AFM images of a CNT interconnection before (a) and

after (b) the dielectrophoretic post-treatment with a Pd salt solu-

tion. Sites along the CNT bridge indicated by arrows mark loca-

tions where adjoining nanotubes are deposited with one end lying

on top of the end of the other nanotube. The insets show the

regions where the CNTs are deposited on the surfaces of the

electrodes in larger magnification. The scale bars of the insets are

260 nm.

Figure 3 Sketch of the structure of CNTs deposited between the

electrodes before (a) and after (b) the post-treatment. In (a) two

CNTs are deposited in series. One end of each of the nanotubes is

deposited on one gold contact. The other ends are disjoined in the

middle forming a single interconnect. Large contact resistances

are expected to occur at the interfaces between the CNTs and the

Au contacts as well as at the CNT-CNT junction. At these three

locations metal deposition is accomplished in the post-treatment

process. The resultant structure after the sealing is shown in (b).

2314 N. Ranjan and M. Mertig: Dielectrophoretically assembled carbon nanotube-metal hybrid structures

© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-b.com

ph

ysic

ap s sstat

us

solid

i b

Figure 4 I-V characteristics for the structure shown in Fig. 2 be-

fore (black line) and after the post-treatment (red line).

Transmission of electrons through a molecular channel depends on the number of channels Nch through it which in turn depends on the electronic structure of the channel. In CNT experiments the transmission T is usually much smaller than the channel conductance (T < T(Nch)) as there exists reflection of electrons when injected from the con-tact into the nanotube. Thus, improper contacts between the metal electrodes and the CNTs lead to a poor charge transfer, and thus, to a contact resistance which dominates the overall resistance of the investigated CNT bridges. In these experiments Pd was chosen as the deposition mate-rial because of two different reasons: one is simply related to the fact that hydrated Pd ions are well suited for the di-electrophoretic growth of metallic nanowires as we could demonstrate recently [12]. The more important, second reason is that the Schottky barrier between semiconducting nanotubes and Pd is known to be lower than that with Au or Pt contact pads [14]. Therefore, Pd has superiority over other metals as contact-electrode material due to the more effective metal-nanotube hybridisation. Theoretical calculations have shown that coating a CNT over the electrode with metallic palladium decrease the reflection of injected electrons, and hence, make the contact more transparent due to a better charge transfer [15]. The calculations show that if the contact length L is larger than a critical value which is typically in the region of a few hundred nanometers, the reflection of electron at-tains a minimum saturating value. This condition is ful-filled in most of our experiments (cf. Fig. 2). Therefore, we can expect to reach this lower limit of attainable resistance by the described DEP post-treatment method. However as suggested in Fig. 3, we also have to take into account the contact resistance at adjoining CNTs. Brought together by a physical force during dielectrophoretic deposition, the CNT joints have to be considered as non-perfect interfaces between adjacent CNTs which also might lead to consider-able contact resistances adding to the overall resistance of the assembled CNT bridge. After the DEP post-treatment the CNTs at the electrodes are covered from all sides with Pd. Similarly, the CNT joints are completely covered with

metal which also leads to a reduction of the overall bridge resistance. However, since the overlaps between the ad-joining nanotubes and the deposited Pd sheaths is much shorter than one hundred nanometer as it can be seen in Fig. 2, one might expect that these locations determine the overall bridge resistance now. 4 Conclusions We have shown that the contact resis-tance of CNT bridges deposited over adjacent electrodes by DEP can be reduced by roughly a factor of four by ap-plying a DEP post-treatment with Pd salt solution. During the post-treatment metallic Pd sheaths are formed at loca-tions where large potential drops existed before. As a result, the CNTs on the electrode surface become coated with a Pd film and CNT-CNT joints are sealed, both leading to a reduction of the overall resistance of the assembled CNT interconnect. This process may not only be applied to CNT bridges, but will have a lot of implication for the future nanoelectronics as it can serve as an effective way to re-duce the contact resistance for tuning the device character-istics.

Acknowledgements We thank O. Jost for the synthesis of

the single-walled carbon nanotubes used and S. Taeger for taking

part in the early stage of the experiment. This work is supported

by the DFG within the framework of the European ERA NA-

NOSCIENCE NET (Project S5; ME 1256/11-1).

References

[1] S. Iijima, Nature 354, 56 (1991).

[2] R. Saito, G. Dresselhaus, and M. S. Dresselhaus, Physical

Properties of Carbon Nanotubes (Imperial College Press,

London, 2005).

[3] T. W. Ebbesen, H. J. Lezec, H. Hiura, J. W. Bennett, H. F.

Ghaemi, and T. Thio, Nature 382, 54 (1996).

[4] R. V. Seidel, A. P. Graham, J. Kretz, B. Rajasekharan, G. S.

Duesberg, M. Liebau, E. Unger, F. Kreupl, and W. Hoen-

lein, Nano Lett. 5, 147 (2005).

[5] R. Krupke, F. Hennrich, H. B. Weber, M. M. Kappes, and

H. v. Löhneysen, Nano Lett. 3, 1019 (2003).

[6] S. Taeger and M. Mertig, Int. J. Mater. Res. 98, 742 (2007).

[7] E. Artukovic, M. Kaempgen, D. S. Hecht, S. Roth, and

G. Grüner, Nano Lett. 5, 757 (2005).

[8] Q. Cao, S.-H. Hur, Z.-T. Zhu, Y. Sun, C. Wang, M. A. Meitl,

M. Shim, and J. A. Rogers, Adv. Mater. 18, 304 (2006).

[9] A. Star, E. Tu, J. Niemann, J.-C. P. Gabriel, C. S. Joiner,

and C. Valcke, Proc. Natl. Acad. Sci. USA 103, 921 (2006).

[10] C. Chen, D. Xu, E. S.-W. Kong, and Y. Zhang, IEEE Elec-

tron Device Lett. 27, 852 (2006).

[11] H. A. Pohl, Dielectrophorosis (Cambridge University Press,

1978).

[12] N. Ranjan, H. Vinzelberg, and M. Mertig, Small 2, 1490

(2006).

[13] S. Taeger, O. Jost, W. Pompe, and M. Mertig, AIP Conf.

Proc. 723, 185 (2005).

[14] B. Shan and K. Cho, Phys Rev. B 70, 233405 (2004).

[15] N. Nemec, D. Tomanek, and G. Cuniberti, Phys. Rev. Lett.

96, 076802 (2006).