E,xtracting Resistances of Carbon 1Nanostrucures in Vias

Wen Wu, Shoba Khshnai Ke Li, Xuhui Siun Raymond Wu, Toshishige Yamada, aid Cary Y. Yang

Cente for Nanostrucwtres, Santa CIara niversitySanta Clara, CA 95053, United StaTes

ABSTRACT

This paper describes a curreit-sensing technique toextract the resistances of carbon nanostructures in viaintercnnects. Test structures designed and fabricated forvia applications contain carbo nanofiber (CNF)-metalcomposites emrTbedded in silicon dioxide (SiOf2.Electrical characterization of sinigle CNFs is perfrmedusing an atomic force microscope (AFM. This techniqueyields a metal-CNlF contact resistance of 6.4 kG and alowest CNF resistivity of L.89e-4 Q-cm.

INTRODUCTION

Carbo nanotubes (CNTs) and CNFs exhibit high currnt-carrying capacily (1& A/ckm) [1-3] and robust thermaland mechanical properties [4,5]. Becauise of theirsuperior proprties, carbon nanostructures have becomeattractive alternative materials fbr nex-generation sub-30-nm on-chip interconnects. Electrical characterization ofDC and AC behavior is essential for any new material tobe incorporated in chip technologies. Unlike macro andmicro-scale components nanostructures require specialtechniques for characterizaion because it is difficu lt todirectly probe them using conventional measur ementtechniques. Moreove, additional fabrication steps andextensive sample preparation may be nieeded bdore thenanostructures are ready for measurement. Imaging withmicroscopy such as scanning electron microscope (SEM)and scanning transmission electron microscopy (STEM)only provid1estrctLIral and compositlonal information Ofthe nanoscale components [6].For I-V characterization of nanostructures, it is necessarto apply a low current or voltage to the device under test(DUT) in orde to lower the power an1d minimize Jouleheating [7]. The low-curTret (r lowvoltage)specification requires a probing system for rapid, high-sensitivity and low-noise measurements. To obtainresistance of nanostructures, typical two-pointom-wafermeasurements cannot differenitiate between intrinsic DUTresistances and contact resistances between the DUT andpads/electrodes. Four-point test structures minimize thevoltage drop across contacts, and have been widelyadopted in horizontal one-dimension nanostructures [8,9].However, it is challenging to fbricate equivalent four-point structures for vertical nanostructures such as viasdue to complex 3-D integration of electrodes with the via.

Therefore, finding an altemative way to extract contactresistance in a via is critically needed.Electrical characteristics of CNT vias have been reported[10,11], but the magtuicdes of the contact resistanceswere unknown. In this woik, we propose a resistanceextraction methodology for carbon nanostructures thatseparates the interfcial resistances from the CNF bulkresistance. Using an AFM on a CNF array, currentdistribution and surface topographies of the sample areobtained simultaneously. This allows the users to probeindividual CNFs precisely and measure their electricalcharacteristics.

CNF ARRAY TEST STRUCTURE

Fig. 1. (a) Cross-section Of CNIF arrys encaps-ulted in Si2, and (b)the SEM irmge of the sanpie surface, where one CNF prot-rding out oftheo ide is hi ihiehht d by a dashed circle.

19 , ,.

for electrical characterization. First we deposit a 20 nmnickel (Ni) catalyst layer on a silicon (1 OQ) wafie which is

coated with a 30 nm titanium (Ti) electrode layer. BoffiNi and Ti layers are deposited by ion beam sputtering. Nicatalyst film was annealed in ammonia (NH) to form Ni

particles, prior to CNF growth by plasma-enhanced

chemical vapor deposition (pECVD). We used acetylene

(C2KH) as the carbon feedstock and NI-I as the carrier gas.The iniherent Vertical alignment f CNFs makesintegration into planar silicon processing technologiesfeasible. The CNF arrays are 6mbedded in

tetraethylorthosilicate (PES) oxide to achieve structuralrigidity and electrical isolation. We polish the sample toexpose the CNFs by about 30-50 nm on the surface fbrAFM probing. The SEM image of the samplIesurfae is

shown in Fig. 1 (b). To fbom an electrical path from AFMsubstrte chuick to the CNF base electrode (i), patterned

metal pads (Mo/Ti) are deposited on part of the sampletop surfce. Then we connect the metal pads to the AFM

978-142444259-1/09/$25.00 @2009 IEEE 27

chuck wit Cu tape. The parasitic resistances induced byMo/Ti pads and te paralleled CNFs beneat tem are

aroud 50-100 which is negligible compared wvith large

resistances (kQ) of individual CNFs.

AFt CURRENT-SENsJNGThCEJINMIQUE

The measurement setup for te AFM is shown in Fig. 2.

Platinum (Pt) coating on the surace ofte AFM probe tip

(tip radius < 60 nm) forms a nanoscale electricalmeasurement probe. With the probe tip virtuallygrounded, a selectable DC bas voltage is applied to the

sample, and an amplifier im the controller senses the

curr:ent flow though te sample. Whei the pobe scans

the saple surface in contact mode, te surface

topography and current1tdistribution are obtained

simultaneously. Fig. 3 shows the 3m x 3gm AFM

images of one CNT sample. By applying a constantvoltage between the probe tip and the sample base

electode, the current through sevefal sinzle CNFs

protruding out ofte oxide is obtained.

300-

..%200-CE 100-_ 0-a)0 -100 -

U_C u

-200--300-

At %d

-3 -2 -1 0 1Voltage (mV)

2 3

Fig 4. I-Vcharacte i soigl CNF

PARAMETER ExTRACTION

The resistance (Rrota) iferred from the measured I-Vcurves consists of the CNF bulk resistance (RUNE in serieswit the total contact resisnce (R) The followingequation holds forRmtd.

Aimplifier Extra Gain & FilterTobe I

DC Bias

AIM

DI 3100

Fig 2S AM cuirent-sensing measurement setp.

Fig. 3. (a) Surface topographic image and QJ) the coresponding currentdistributica of cae CNTE sample. The white dots in (5) denote the1ocations of CNFs. The scan size is 3ktmx 3km.

DC MEAsuRzEMENTs

Based on te AFM images obtained (Fig. 3), we reducete scan size, and systematically locate single CNFs withthe probe tip on the top surace of the CNF. We conductI-V measurements by sweeping the DC bias voltagesbetween -3 to +3 mV. Fig. 4 shows a linear I-V behaviorobtained for one particular CNF of diameter DUE = 120nm.

Rot/ - C -RE (1)

For a CNF, when the length (LE is larger tan the meanfree path (%NE), RUNE can be calculated from geometricalparameters as follows.

RUNE- J E (h >2 ) (2)

Where p is te resistivity of te GNT, and DUNE is thediamter of a CNF. For each sample, LUN is very nearlyconstant.

R represents the sum of CNF-metal and probe tip-CNFcontact resistances. The contatt resistance between the Ptprobe tip anid te CNF is dependent on te pressureapplied oi the tip, Which can be minimized empirically.In maacro-scale, contact resistanice is expected to beinversely proportional to the contact area. For carbonnanostructures, tis contact area or diameter dependenceis more complicated [12-14]. Previous Stdy [15] shbowedtat tunneling is the do6minant transport mechanismbeWteen a CNT tip and an electrode, Which is similar tothe present CNF-metal interfce. If the tunneling bamferthickness w has a certain correlation wiDtUDN such thatw is thicker fo larger DUNE, then RU can be independentof D This is because tunneling resistance isproportional to 4/(itDU2) exp(2osw, where o is thedecay constant of an electron wave fuction in the barrier[15]. If ow ln(DU r), then the tunneling resistance canbe alhost constant. When the interace asperity issignificant, as expected for the as-grown contact between

28

I...............................................................N

CNT and electrode, the efctive w for tunneling tends tobe larger for larger DCAT making the constant Rcassumption a reasonable starting poim for our daaanalysis. Thus, RTo,a in (1) becomes a linear function of1/!lDcj,r.f. Plotting RTr0, versus )21 1W / andextrapolating to zero yields Rc.

'3n nfl

16.0k-

0 12.0k-C) -

u

cc 8.0k-~4n 40 4.Ok-

nfl50 100 150

lI(DCN (gm2)200 250

Fig. 5. Fitted curve of measured resistances versus 1/(Ddo2. Rc is they-intercept.

I-V measurements were carried out on twenty-two 4.5gm-long CNiFs of varying diameters. Thesemeasuremens are plotted as a function of 1/(wf in Fig.5. The extacted Rc is 6.4 ko, with an error bar of about4000. It is fbund that the lowest resistivity amongmeasured CNIFs is 1 .89e-4 0-cm. Unlike the low-resistance behavior exhibited by metal contacts, R, valuesobtained here are several times larger than the bulkresistances of carbon nanostructures with similardimensions. The contact resistance therefbre dominatethe electrical characteristics of carbon nanostructures invia interconnects.For extraction of nanostructure bulk and contactresistances from measured total resistance, the methodhere can be extended to extract resistances from measuredresistances versus varying lengths. In this case, twoCNTs with similar diameters will hve similar contactresistances if process variations are ignored. Thelimitations of this extraction method are that we need toprepare several samples with different lengths. Besides,process variations mut be minimized because we assumethat the difference in measured resistances of the twoCNTs only results from the length difference. Study ofthis extraction tecbniqe usig measured resistance versuslength is ongoing.

CONCLUSIONS

In this work, electrical characterization of carbonnanostructures m via interconnects has been conducted

using an AFM. The current-sensiqg techrique enables usto probe individual UNFs and obtain I-V curves for singlenanofibers. The CNT bulk resistivity and contactresistance are then extracted fom the total resistancemeasurements. This method is applicable to other one-dimensional nanostructures. Our results show tat theContct resistances at the metal-carbon interfice dominatetie electrical characteristics of carbon nanostructures invia interconnects.

ACKNOWLEDGEMENT

This work is supported by the United States Army Spaceand Missile Defense Command (SMDC) and carriesDistribtifon Statement A, approved for public release,distribution limited.

REFERENCES

[1] Z. Yao, C. L. Kane, and C. Dekker, "High-fieldelectrical transport in. single-wall carbon nanotubes",Phys. Rev. Left, vol. 84, March 2000, pp. 2941-2944.

[2] 1B. Q. Wei, R. Vajtai, and P. M. Ajaan, "Reliabilityand current carrying capacity of carbon nanotubes",AppL Ph Left vol. 79, August 2001, pp. 1172-1174.

[3] A. Javey, P. Qi Q. Wang, and H. Dai, "Ten- to 50-nm-long quasi-ballistic carbon nanotube devicesobtained without complex lithography", Proc. NatEAcad SI. UISA vol. 101, September 2004, pp.13408-13410.

[4] Q. Ngo, T. Yamadca, M. Suzuki, Y. Ominami, A. M.Cassell, J. Li, M. Meyyappan, and C. Y. Yang,"Structural and electrical characterization of carbonnanofibers for ihterconnect via applications", IEEETrans. Pano., vol. 6, November 2007, pp. 688-695.

[5] M. Suzuki, Y. Ominami, Q. Ngo, C.Y. Yang, A.M.Cassell, and J. Li, "Current-carying capacity ofcarbon nanofiber interconnects". J ApL Phys, vol.101, June 2007, pp. 143 07-11431 1.

[6] Nanotechnology Measurement Handbook: A Guideto Electrical Measurements for anoscienceApplications, 1 t Edition, Keithley Instruments, Inc.

[7] M. A. Kuroda and Jean-Pierre Leburton, "jouleheatina induced negative differential resistance mnfeestanding metallic carbon nanotubes", AppL Ph.Lett.- vol. 89, September 2006, 103102.

[8] A. Bachtold,M.Heny, C. Terrier, C. Strunk,andC.Schonenberger, J. P. Salvetat, J. M. Bonard, and L.Forro, "Contacting carbon nanotubes selectivelywith low-ohmic contacts for four-probe electricmeasuremets", AppL Phys. Left., vol. 73, July 1998,pp. 274-276.

[9] A. S. Walton, C. S. Allen, K. Critchley, M. L.

29

* MeasurementsLi Linear filling

0 0~~lp

f.

Y-.F7F-

Gorzny, J. E. McKendry, R. M. D. Brydson, B. J.Hickey, and S. D. Evans, "Four-probe electricaltransport measurements on individual metallicnanowires", Nanotechnology, vol. 18, February2007, 065204.

[10] A. Kawabata, S. Sato, T. Nozue, T. Hyakushima, M.Norimatsu, M. Mishima, T. Murakami, D. Kondo, K.Asano, M. Ohfuti, H. Kawarada, T. Sakai, M. Nihei,and Y. Awano, "Robustness of CNT viainterconnect fabricated by low temperature processover a high-density current", in proceedings ofIEEE2008 International Interconnect TechnologyConference, pp. 237-239.

[11] J. C. Coiffic, M. Fayolle, H. Le Poche, S. Maitrejean,and S. Olivier, "Realization of via interconnectsbased on carbon nanotubes", in Proceedngs ofIEEE2008 Int. Interconnect Tech. Conf, pp. 153-155.

[12] Q. Ngo, S. Krishnan, A. M. Cassell, Y. Ominami, J.

Li, M. Meyyappan, and C. Y. Yang, "Electricalcharacterization of carbon nanofibers for on-chipinterconnect applications", in proceedings ofIEEE-Nano 2005, pp. 431-434.

[13] Y. Massoud and A. Nieuwoudt, "Accurate resistancemodeling for carbon nanotube bundles in VLSIinterconnect", in proceedings of IEEE-Nano 2006,pp.288-291.

[14] A. Nieuwoudt and Y. Massoud, "Evaluating theimpact of resistance in carbon nanotube bundles forVLSI interconnect using diameter-dependentmodeling techniques", IEEE Trans. ElectronDevices, vol. 53, October 2006, pp. 2460-2466.

[15] T. Yamada, "Modeling of electronic transport insaning tunneling microscope tip-carbon nanotubesystems", AppL Phys. Lett., vol. 78, March 2001, pp.1739-1741.

30