Wear 259 (2005) 738–743

Short communication

Friction properties of surface-fluorinated carbon nanotubes

R.L. Vander Wala, K. Miyoshib, K.W. Streetb, A.J. Tomaseka, H. Pengc,Y. Liu c, J.L. Margrave�, V.N. Khabasheskuc,∗

a The National Center for Microgravity Research at NASA-Glenn Research Center, Cleveland, OH, USAb The NASA-Glenn Research Center, Cleveland, OH, USA

c Department of Chemistry and Center for Nanoscale Science and Technology, Rice University,6100 Main street, Houston, TX 77005, USA

Received 17 October 2004; received in revised form 8 February 2005; accepted 14 February 2005Available online 10 May 2005

Abstract

Surface modification of the tubular or sphere-shaped carbon nanoparticles through chemical treatment, e.g., fluorination, is expected tosignificantly affect their friction properties. In this study, a direct fluorination of the graphene-built tubular (single-walled carbon nanotubes)s ef .002–0.07,a ovided anu licationso©

K

1

tTlposmiwtss

nderout.rmalng-

oSerrialsntlycture

rbonenedionsheticles,l

ori-the

0d

tructures has been carried out to obtain a series of fluorinated nanotubes (fluoronanotubes) with variable CnF (n= 2–20) stoichiometries. Thriction coefficients for fluoronanotubes, as well as pristine and chemically cut nanotubes, were found to reach values as low as 0ccording to evaluation tests run in contact with sapphire in air of about 40% relative humidity on a ball-on-disk tribometer which prnidirectional sliding friction motion. These preliminary results demonstrate ultra-low friction properties and show a promise in appf surface modified nanocarbons as a solid lubricant.2005 Elsevier B.V. All rights reserved.

eywords:Friction; Carbon nanotubes; Fluorination; Fluoronanotubes

. Introduction

The performance of solid lubricants largely depends uponhe particle size and their shape at the nanoscale level[1–6].he modeling and experimental studies suggest that the cage-

ike nanosize particles having either spherical or tubular mor-hology will provide great application advantages in the fieldf tribology [4]. Their seamless structure helps to inhibit theticking and burnishing of the nanoparticles by the rubbingetal or other surfaces. These tubular particles slide and roll

n part during sliding contact, resulting in a low friction andear. The primary structures may also crack open during con-

act between the tribocouples leading to small graphitic lowurface energy layers, which behave similar to graphite. Thepherical nanoparticles can also serve as effective spacers,

∗ Corresponding author. Tel.: +1 713 348 3486; fax: +1 713 348 5155.E-mail address:khval@rice.edu (V.N. Khabashesku).

� Deceased.

prohibiting contact and wear of the metal surfaces uheavy loads where fluid lubricants are normally squeezedOther advantages include the superior oxidation and thestability of the tubular and spherical nanoparticles, proloing their wear life. For example, the hollow spheres of M2and WS2 exhibit ultra-low friction and wear (by an ordof magnitude) in comparison with the macro-scale matebecause the curved S–M–S (M = Mo, W) planes significareduce the oxidation and preserve the layered stru[5,6].

The discovery of synthetic methods to produce cananotubes, structurally built of graphene cylinders, opthe research opportunities for a variety of applicat[7–10], including lubricants. Surface modification of tcarbon nanotubes as well as other carbon nanoparnano-onions[11] and nanodiamonds[12], through chemicatreatment[13], e.g., fluorination[14,15], is expected taffect their friction and degradation behavior in lubcant application. This expectation is supported by

043-1648/$ – see front matter © 2005 Elsevier B.V. All rights reserved.

oi:10.1016/j.wear.2005.02.082

R.L. Vander Wal et al. / Wear 259 (2005) 738–743 739

experimental studies of graphite fluoride (CF)x powder filmsthat demonstrate longer wear lives, low friction coefficientsand superior load-carrying capacities in comparison withgraphite both in humid air and at elevated temperatures dueto weakened interlamellar van der Waals forces[16,17].The intent of this work is to examine the effect of surfacemodification of SWNTs on their tribological properties.In the present work, we report the evaluation data onfriction coefficients and wear life of neat lubricating filmsprepared from fluorinated single-walled carbon nanotubes(fluoronanotubes) with variable CnF (n= 2–20) stoichiome-tries. The samples of pristine and oxidative acid treatednanotubes, and the nanotubes which were short cut by thefluorination-pyrolysis method[18] were also tested forcomparison.

2. Experimental

2.1. Materials

Single-walled carbon nanotubes (SWNTs), preparedby the HiPCO process, were purified prior to use by wetair oxidation and subsequent hydrochloric acid treatmentfollowed by washing and vacuum drying to remove alln ingt ice’sC asf hefl ep derc ork[ dtc hsw3 si ) ofp

2.2. Friction property testing

The evaluation tests were run on a ball-on-disk tribometer(Fig. 1). A 6 mm sapphire ball was loaded (1.4 N) and putin contact with a rotating (120 rpm) quartz disk of 12.7 mmdiameter to provide the average Hertzian contact pressure ofabout 0.3 GPa. The force of the ball on the disk and tangen-tial force were used to determine the friction coefficient. Allexperiments were performed at a track diameter of 6 mm anda sliding velocity of 38 mm/s. The wear life of the lubricantwas estimated by the number of rotating passes at which thefriction coefficient increased to 0.15.

The films were prepared by deposition of nanotube sam-ple suspensions either in toluene (SWNTs, cut-SWNTs,and C20F) or isopropanol (C2F, C5F, and F-cut-SWNTs)on the quartz disk followed by evaporation of solvent un-der dry nitrogen flow. Films of four different relative thick-nesses (0.63, 1.26, 1.89, and 2.52�g/mm2) were fabricatedand tested. The most unambiguous measure of thickness isweight per unit area, which we use to define the coatingload. Bonding between a coating and the substrate is dueto van der Waals’ forces. The nanotubes are mechanicallytangled with each other. All tests were run in air environ-ment of about 40% relative humidity at room temperature(∼23◦C).

3

3

d tedn NTs.Sg acid( tesm intos image

as co

ontubular forms of carbon and iron catalyst accordo a documented detailed procedure developed at RNST [19]. The iron content in so purified SWNTs w

ound to be lower than 1 wt.% according to TGA. Tuoronanotubes of the C2F and C5F stoichiometry werrepared by direct fluorination of purified SWNTs unontrolled conditions well established in the earlier w14,18]. Hydrazine treatment of the C2F sample was useo reduce the fluoronanotubes to approximately a C20Fomposition[13,14]. Cut-SWNTs of 100–300 nm lengtere prepared by treatment of SWNTs in 2 M HNO3 by0 min sonication[13]. Pyrolysis of C5F fluoronanotube

n argon produced shorter nanotubes (F-cut-SWNTsredominantly 20–80 nm length distribution[18].

Fig. 1. The ball-on-disk tribometer (the disk surface w

. Results and discussion

.1. TEM studies

High-resolution TEM images (Fig. 2) show significantlyifferent surface morphologies for all chemically treaanotube samples in comparison with the pristine SWmooth sidewall surfaces of thick SWNT bundles (Fig. 2a)enerate slight protrusions as a result of oxidativeHNO3) treatment under mild conditions which creainor defects on the sidewalls of nanotubes, cuts them

horter lengths and opens their ends, as shown by the

ated by a thin film of a neat nanotube sample prior to testing).

740 R.L. Vander Wal et al. / Wear 259 (2005) 738–743

Fig. 2. TEM images of SWNT samples studied: (a) Pristine SWNTs, (b) cut-SWNTs, (c) C2F, (d) C5F, (e) C20F, and (f) F-cut-SWNTs. The SWNTs formbundles of different diameter due to the random nature of the way they are collected and mounted. These bundles are held together by strong wall-to-wall vander Waals’ forces.

of cut-SWNT sample inFig. 2b. Fluorination causes moredramatic changes in SWNT surface appearance by increasingboth the size and number of protrusions on the sidewallsof nanotubes in C2F and C5F samples (Fig. 2c and d) thatcan be related to a great number of sp3 carbon-carbon bondsformed via the sidewall functionalization[13,14]. Significantreduction in bundle size is also noted for these fluorona-notubes.

Treatment of the C2F sample with hydrazine helps to re-move fluorine without causing any significant destruction tothe sidewalls of nanotubes in the resulting C20F sample. TheTEM image (Fig. 2e) of this sample shows a largely restoredsmooth surface morphology, as in pristine SWNTs. In con-trast with this sample, some surface areas of F-cut-SWNT(Fig. 2f), are partially damaged and appear as split open tubes.

This can be correlated with the persistent showing of the side-wall “defect” mode in the Raman spectra of nanotubes cut byfluorination[18].

3.2. Friction coefficients

The friction coefficients determined for all tested neatnanotube samples at four relative disk coating thicknessesare summarized inFig. 3. These data show that almost allmeasured friction coefficients lie well below the range ofvalues for graphite (0.13–0.3)[20], the most commonlyused solid lubricant in air environment, with the exceptionof only a very few values falling into a low vicinity of thisrange. This clearly indicates that all tested SWNT samplesexhibit very good friction properties. However, for most

R.L. Vander Wal et al. / Wear 259 (2005) 738–743 741

Fig. 3. Friction coefficients of SWNT samples determined at different rela-tive coating thickness in contact with sapphire in air.

samples tested the relationship between friction coefficientsand coating thicknesses was found to be inconsistent.

Thus, in the case of pristine SWNTs a steady increasein coating thickness results in consistent lowering of thecoefficient of friction, while similar increases in coatingthickness for cut-SWNTs sample cause negligible effect, al-ways yielding a friction coefficients as low as 0.07–0.09. Asteady decrease of friction coefficient is observed for fluo-ronanotubes C2F and C20F samples with increase in coatinglevel from 0.63 to 1.89�g/mm2. At the highest coating level(2.52�g/mm2) for C2F and C20F and at 1.89�g/mm2 forF-cut-SWNTs, the friction coefficient again increases. TheC5F samples demonstrate a very low friction (0.02–0.05) at0.63, 1.26 and 2.52�g/mm2 thicknesses, but show an ele-vated value (0.2) at 1.89�g/mm2. The observed discrepancyis likely related to the difficulty in applying a uniform coatingof the nano-materials onto the disk in some experiments.

Nevertheless, the measured lowest friction coefficients forSWNT samples in air (Fig. 4) were found to either fall into therange known for the best lubricants, such as DLC (0.05–0.15)and PTFE (0.03–0.1), or even outperform the latter. For in-stance, F-cut-SWNTs and C5F samples show very low valuesof 0.013 and 0.016, respectively, pristine SWNTs—0.011,

F ntactw

while a C20F sample yields an ultra-low friction coefficientof 0.002. It is likely that the lowest friction shown by thesefour SWNT samples in a humid air environment can be re-lated both to the smooth (or relatively smooth) sidewall sur-face morphology and to the drastically increased surface areain the case of very short F-cut-SWNTs. It also seems thathaving a small amount of residual fluorine on the sidewallof the nanotube assists in obtaining an ultra-low friction, asobserved for C20F (0.002). Finally, asFigs. 3 and 4show,all materials tested exhibit friction coefficients substantiallylower than traditional carbon materials and generally betterthan diamond like carbon films. The SEM observations of thewear tracks indicated that although most nanotubes were gen-erally plowed out and piled at both sides of the wear track, asmall amount of tubular particles remained in the wear track.

The measured relative order of friction coefficients can beunderstood on the basis of material composition. By refer-ence to bulk materials, fluorination can improve the lubricityof carbon materials. Yet fluorination can introduce “defects”within the SWNT sidewalls. These defects may result in pre-mature degradation of the material (tubular structure) andcorresponding loss of lubrication. Upon degradation, reac-tive edge sites will be created leading to binding between thelubricant and adjacent surfaces. Furthermore, sp3 hybridiza-tion of sidewall carbons will reduce the flexibility, modulusand tensile strength of the SWNT. Hence a lower degree offl ablet rial.N risesf nsid-e thinb ossi-b rfacer reventc ithinb eass logi-c

3

ance( ple( -t lowerw

ese ts:l ond sa anb cingd Ts aticf rada-t l

ig. 4. Lowest friction coefficients determined for SWNT samples in coith sapphire in air.

uorination may provide a more resilient material, bettero resist buckling than a more highly fluorinated mateotably this stands in contrast to bulk materials and a

rom the nanosize scale of these lubricants. A second coration is that fluorination will act to separate tubes wiundles. While some separation may be desirable, it is ple that individual SWNTs may not be best because suoughness exceeds their diameter. Hence they cannot pontact between surface asperities. Moreover, tubes wundles permit the possibility of intertube sliding wheringle SWNTs necessarily slide between adjacent triboal surfaces.

.3. Wear lives

Under the conditions studied, the highest endurwear) life was demonstrated by the pristine SWNT samFig. 5). This was followed by the C20F and F-cut-SWNT maerials. The other three nanotube samples show muchear lives as lubricants.These preliminary data show that the wear lifetim

xhibit a rough inverse correlation with friction coefficienower friction coefficient correlates with longer lubricatiurability (lifetime). Interestingly, the C20F material showlifetime a bit lower than the pristine SWNTs. This c

e understood on the basis of the fluorination introduefects, namely sp3 hybridization sites within the SWNidewalls. These sites disrupt the integrity of the aromramework and create sites where buckling or other degion can occur. The modest amount of F in the C20F materia

742 R.L. Vander Wal et al. / Wear 259 (2005) 738–743

Fig. 5. Estimated endurance (wear) life for SWNT lubricating coatings incontact with sapphire in air.

results in its performance having somewhat less durationthan the SWNT material. Notably, it has a longer wear lifethan any of the other fluorinated materials, all of whichpossess a higher F:C stoichiometry. Increasing F contentnecessarily introduces larger amounts of wall disruption,hastening degradation and shortening lubricant lifetime.The nascent nano-materials were characterized by Ramanspectroscopy, which was also employed to examine thewear tracks in an attempt to detect changes in the chemicalnature of the SWNT materials resulting form the tribologicaldegradation. Unfortunately due to the low sensitivity of thetechnique, no chemical signature of the SWNT materials wasdetected.

While the F-cut-SWNT materials nominally have the samecomposition as the C5F materials, their smaller length mayaid lubrication. As they are produced by pyrolysis, the tubeends are expected to be closed by either bonding to adja-cent carbon or the carbon atoms are terminated byF or Hatoms. Notably, these groups are more thermodynamicallystable than oxygen containing functional groups. For the cutSWNTs, the cutting action has been shown to introduce sig-nificant amounts of oxygen containing functional groups atthe tube ends, terminating the carbon valences. Apart fromtheir reactivity, these groups can readily oxidize the SWNTends during friction testing, thereby degrading the lubricant.Moreover, upon oxidation, radical sites will be created thatw ntingf

4

ofS reat-m ir lu-b hef tinea es asl tiono asedo and

C20F samples show the best lubrication performance in airamong all nanotube samples investigated thus far.

We interpret the trend in SWNT materials as lubricantsbeing partly related to the degree to which they can fill inpits and scratches, and between surface asperities, etc. be-tween sliding surfaces. If this were the only mechanism,then all tubes would behave similarly; hence there is addi-tional effect from the chemical nature of the surface modifi-cations. A modest amount of fluorination can serve to breakup SWNT bundles, without significant tube wall degrada-tion, thereby creating a better lubricant. While cut SWNTsand F-cut-SWNTs should similarly have tube segments com-mensurate with surface roughness values, their productionalters the graphitic quality of the SWNTs (on the tube ends),leading to a poorer lubricating ability. Magnifying this trendis the higher fluorination stoichiometry of the C2F and C5Fmaterials. Though producing a nominally better lubricant byreference to bulk materials, their synthesis leads to substan-tial SWNT degradation (by introduction of sp3 hybridization,etc.). Consequently, these highly fluorinated SWNTs can beless resilient towards deformation and more apt to rupture,producing dangling bonds and resulting stiction.

More extensive studies of surface-modified SWNTs andother nanocarbons, such as fluorinated nano-onions, poly-fullerenes[21] and fluoronanodiamond[15], both in neatform and as additives to liquid lubricants are now in progress.

A

lica-t ber0 r C-0 entN eatlya Pro-g tiveE lennR lly ac-k neD tingi

R

ls,

tionfolk,

gy,

om-truc-

R.ri-

ill cause stiction with adjacent surfaces, thereby accouor the relatively high friction coefficient.

. Conclusion

In the present work, tribological studies of a seriesWNT samples in air revealed that the type of chemical tent of nanotube surface has a significant effect on therication properties, friction coefficients and wear life. T

riction coefficients for fluoronanotubes, as well as prisnd chemically cut nanotubes, were found to reach valu

ow as 0.002–0.07, thus showing a promise for applicaf surface modified nanocarbons as solid lubricants. Bn combined friction and wear life data, pristine SWNTs

cknowledgements

The support of Rice research on chemistry and appions of nanocarbon materials by Texas ATP (grant num03604-0026-2001), Welch Foundation (grant numbe109), and in part by NASA-URETI (cooperative agreemCC-1-02038) and Carbon Nanotechnologies, Inc. is grppreciated. Other funding through the Vehicle Systemram’s Low Emission Alternative Power Project, Alternanergy Foundation Technologies Subproject at NASA Gesearch Center is acknowledged. The authors gratefunowledge David R. Hull for the HRTEM imaging, Duaixon for the SEM images and Richard Mondry for assis

n collection of tribological data.

eferences

[1] W.A. Glaeser, Characterization of Tribological MateriaButterworth-Heinemann, Boston, 1993, p. 1.

[2] A.R. Lansdown, in: B.S. Edmunds (Ed.), Lubrication and LubricaSelection. A Practical Guide, Mechanical Eng. Publ. Ltd., SufUK, 1996.

[3] J.D. Summers-Smith, An Introductory Guide to Industrial TriboloMechanical Eng. Publ. Ltd., London, UK, 1994.

[4] R. Tenne, M. Homyonfer, Y. Feldman, Nanoparticles of layered cpounds with hollow cage structures (inorganic fullerene-like stures), Chem. Mater. 10 (1998) 3225–3238.

[5] L. Rapoport, Yu. Bilik, Y. Feldman, M. Hoyofer, S.R. Cohen,Tenne, Hollow nanoparticles of WS2 as potential solid-state lubcants, Nature (London) 387 (1997) 791–793.

R.L. Vander Wal et al. / Wear 259 (2005) 738–743 743

[6] M. Chowalla, G.A.J. Amaratunga, Thin films of fullerene-like MoS2

nanoparticles with ultra-low friction and wear, Nature (London) 407(2000) 164–167.

[7] M.S. Dresselhaus, G. Dresselhaus, P.C. Eklund, R. Saito, in: J. Shi-nar, Z.V. Vardeny, Z.H. Kafafi (Eds.), Optical and Electronic Prop-erties of Fullerenes and Fullerene-Based Materials, Marcel Dekker,Inc., New York, 2000, p. 217.

[8] P.J.F. Harris, Carbon Nanotubes and Related Structures, CambridgeUniversity Press, Cambridge, 1999.

[9] T.W. Ebbesen (Ed.), Carbon Nanotubes, CRC Press, Boca Raton,1997.

[10] M.S. Dresselhaus, G. Dresselhaus, P.C. Eklund (Eds.), Science ofFullerenes and Carbon Nanotubes, Academic Press, New York,1996.

[11] K.W. Street, M. Marchetti, R.L. Vander Wal, A.J. Tomasek, Evalu-ation of the tribological behavior of nano-onions in Krytox 143AB,Tribology Lett. 16 (2004) 143–149.

[12] V.Yu. Dolmatov, Detonation synthesis of ultradispersed diamonds:properties and applications, Russ. Chem. Rev. 70 (2001) 607–626.

[13] V.N. Khabashesku, J.L. Margrave, Chemistry of carbon nanotubes,in: H.S. Nalwa (Ed.), Encyclopedia of Nanoscience Nanotech-nology, vol. 1, American Scientific Publishers, 2004, pp. 849–861.

[14] V.N. Khabashesku, W.E. Billups, J.L. Margrave, Fluorination ofsingle-wall carbon nanotubes and subsequent derivatization reactions,Acc. Chem. Res. 35 (2002) 1087–1095.

[15] Y. Liu, Z. Gu, J.L. Margrave, V.N. Khabashesku, Functionalizationof nanoscale diamond powder: fluoro-, alkyl-, amino-, and aminoacid-nanodiamond derivatives, Chem. Mater. 16 (2004) 3924–3930.

[16] H. Gisser, M. Petronio, A. Shapiro, Graphite fluoride as a solidlubricant, Lubr. Eng. 28 (1972) 161.

[17] P. Kamarchik, J.L. Margrave, Poly (carbon monofluoride), a solidlayered fluorocarbon, Acc. Chem. Res. 11 (1978) 8.

[18] Z. Gu, H. Peng, R.H. Hauge, R.E. Smalley, J.L. Margrave, Cut-ting single-wall carbon nanotubes through fluorination, Nano Lett. 2(2002) 1009–1013.

[19] I.W. Chiang, B.E. Brinson, A.Y. Huang, P.A. Willis, M.J.Bronikowski, J.L. Margrave, R.E. Smalley, R.H. Hauge, Purifica-tion and characterization of single-wall carbon nanotubes (SWNTs)obtained from the gas-phase decomposition of CO (HiPco process),J. Phys. Chem. B 105 (2001) 8297–8301.

[20] H.O. Pierson, Handbook of Carbon, Graphite, Diamond andFullerenes. Properties, Processing and Applications, Noyes. Publ.,1993.

[21] Z. Gu, L. Zhang, J.L. Margrave, V.A. Davydov, A.V. Rakhmanina, V.Agafonov, V.N. Khabashesku, Fluorination of pressure-polymerizedC60 phases, Carbon, in press.