1: \documentclass[prl,twocolumn,showpacs]{revtex4}

2: %\documentclass[aps,showpacs]{revtex4}

3: %\documentclass[prb]{revtex4}% Physical Review B

4: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}

5: 

6: \usepackage{graphicx}

7: \usepackage{amsmath}

8: \usepackage{amssymb}

9: 

10: 

11: \begin{document}

12: \def\a{\alpha}

13: \def\b{\beta}

14: \def\e{\varepsilon}

15: \def\d{\delta}

16: \def\l{\lambda}

17: \def\m{\mu}

18: \def\t{\tau}

19: \def\n{\nu}

20: \def\o{\omega}

21: \def\r{\rho}

22: \def\S{\Sigma}

23: \def\G{\Gamma}

24: \def\D{\Delta}

25: \def\O{\Omega}

26: % Macro varie

27: 

28: \def\ra{\rightarrow}

29: \def\ua{\uparrow}

30: \def\da{\downarrow}

31: \def\pd{\partial}

32: \def\bk{{\bf k}}

33: \def\bp{{\bf p}}

34: \def\bn{{\bf n}}

35: 

36: \def\be{\begin{equation}}\def\ee{\end{equation}}

37: \def\bea{\begin{eqnarray}}\def\eea{\end{eqnarray}}

38: \def\nn{\nonumber}

39: \def\lb{\label}

40: \def\pref#1{(\ref{#1})}

41: 

42: 

43: 

44: %\preprint{HEP/123-qed}

45: 

46: 

47: \title{Boundary Friction on Molecular Lubricants: Rolling Mode?}

48: 

49: 

50: \author{V.M. Loktev$^1$ and Yu.G.~Pogorelov$^2$}

51: \affiliation{$^1$Bogolyubov Institute for Theoretical Physics, 

52: National Academy of Sciences of Ukraine, Metrologichna Str. 14-b, 

53: Kiev, 03143 Ukraine, $^2$CFP/Departamento de F\'{i}sica, Universidade 

54: do Porto, 4169-007 Porto, Portugal}

55: %\email{ypogorel@fc.up.pt}

56: %\homepage{http://www.Second.institution.edu/~Charlie.Author}

57: 

58: \date{\today }

59: 

60: 

61: \begin{abstract}

62: A theoretical model is proposed for low temperature friction between two

63: smooth rigid solid surfaces separated by lubricant molecules, admitting

64: their deformations and rotations. Appearance of different modes of energy

65: dissipation (by ''rocking'' or ''rolling'' of lubricants) at slow relative

66: displacement of the surfaces is shown to be accompanied by the stick-and-slip 

67: features and reveals a non-monotonic (mean) friction force {\it vs} external load

68: \end{abstract}

69: 

70: 

71: \pacs{46.55.+d, 81.40.Pq}

72: 

73: \maketitle

74: 

75: 

76: \section{\label{sec:1} Introduction}

77: 

78: In the modern tribology a still increasing interest is put to the studies of

79: wearless friction on atomically smooth surfaces \cite{bhu},\cite{mclel} as a

80: possibility to provide an information about the basic processes of energy

81: losses on microscopic level, important for the purposes of optimization in

82: many technological applications. This is also connected with the search for

83: best coating and lubricant materials. The principal physical picture,

84: usually considered in relation with boundary friction on few molecular

85: layers of lubricant liquid, is the sequence of ''freezing-melting''

86: processes on the lubricant, giving rise to discontinious (stick-and-slip)

87: displacement of sliding surfaces \cite{isr}. Recently a new theoretical

88: approach was proposed for microscopic sliding processes at extremely low

89: velocities of motion and upon a monolayer of lubricant atoms \cite{pog}, as

90: can be the case for friction force microscopy (FFM). Based on the adiabatic

91: formation of metastable states (similar to dislocations in usual deformed

92: crystals or defects in the Frenkel-Kontorova model \cite{fk},\cite{sok}) and

93: their following relaxation, this treatment shows how the (average) microscopic 

94: friction coefficient depends upon the material parameters of the contacting solids

95: and lubricants and also how stick-and-slip jumps with atomic periodicity can

96: develop in the microscopic friction force. It should be noted that a similar

97: microscopic mechanism of dislocation-assisted sliding was recently proposed

98: for contacting asperities in dry friction \cite{bhu2}.

99: 

100: Ultimately, with the discovery of almost spheroidal molecule of fullerene C$%

101: _{60}$ \cite{kro} (and/or cylindrical carbon nanotubes \cite{Ii}) a hope had

102: arosen to use such closed molecular structures, as ''free rotating''

103: lubricants, for considerable reduction of the friction coefficient. For

104: instance, a reduction of sliding friction coefficient was already discussed,

105: due to involvement of spinning motion of surfaces in contact \cite{far}. 

106: However, the FFM experiments with use of C$_{60}$ monolayers deposited

107: over atomically smooth solid surfaces brought some contradictory results 

108: \cite{bhu1}, \cite{lut},\cite{sch}. To get their better understanding, a

109: further theoretical insight is desirable on the elementary processes of

110: boundary friction.

111: 

112: The present communication is aimed on extention of the above mentioned

113: adiabatic approach to the processes of boundary friction which include the

114: internal degrees of freedom of the lubricant molecules. Within a simplest

115: model, we make an attempt to show that, due to the discrete atomic structure

116: of such a molecule, there are possible qualitatively different modes of slow

117: motion, either dissipative or non-dissipative, depending on the applied

118: external load on the contact. For dissipative modes, there are energy losses

119: resulting from stick-and-slip discontinuities, but these losses turn out

120: much lower then for similar processes at sliding solid surfaces upon atomic

121: lubricants. Besides, the mean value of friction force is found to be a

122: non-monotonic function of the external load.

123: 

124: \section{\label{sec:2} The Model}

125: 

126: Let us consider a two-dimensional model for the boundary friction system

127: which includes two semi-infinite atomic arrays, the ''solids'', with

128: identical triangular lattice structure and a spatial separation $d$ between

129: their srufaces, and clusters of four atoms, the ''molecular lubricants'',

130: confined between the solids (Fig. {\ref{Fig.1}). The distances between nearest

131: neighbour atoms in both solids and in clusters are supposed invariable 

132: \cite{note} (that means, corresponding to ''absolutely rigid'' bonds), and 

133: the bond length for the molecule equals to the lattice parameter $a$ for the 

134: solid surfaces.

135: 

136: \begin{figure}

137: \centering{

138: \includegraphics[width=6.cm, angle=0]{rol1.eps}}

139: \caption{Schematic of the friction system: two solid arrays (grey circles)

140: with identical lattice structures are separated by a four-atomic lubricant

141: molecule (white circles). The molecule center rests at the origin, and

142: its rhombic configuration is determined by the angles $\theta$ and $\phi$,

143: while the symmetric displacements of solids are described by the parameter $p$

144: (see the text).}

145: \lb{Fig.1}

146: \end{figure}

147: 

148: The model includes certain ''soft links'', which determine the energy

149: variation {\it vs} relative displacement of the solids. In the first turn,

150: this is weak Lennard-Jones interaction between an atom of solid and a

151: lubricant molecule at distance $r$:

152: 

153: \begin{equation}

154: V_{LJ}(r)=\varepsilon _{0}\left[ \left( \frac{r_{0}}{r}\right) ^{6}-2\left( 

155: \frac{r_{0}}{r}\right) ^{12}\right] ,  \label{1}

156: \end{equation}

157: characterized by the adhesion energy $\varepsilon _{0}$ and equilibrium

158: distance $r_{0}$. Besides, a lubricant molecule can experience rhombic

159: deformation which changes its energy as $k\sin ^{2}\phi $, where $k$ and $%

160: \phi $ are the elastic constant and rhombic angle, respectively. We neglect

161: the interaction between different lubricants, that is omit their possible

162: collective modes. Thus the molecular lubricants are supposed to contribute

163: independently into the total friction force. Then we choose one particular

164: molecule and set its center of mass as the origin of reference frame, so its

165: configuration is determined by the ''internal'' rhombic angle, $\phi $, and

166: the ''external'' orientation angle, $\theta $ (say, with respect to the

167: normal to interface). In such a frame, the two solids are supposed to be

168: displaced symmetrically with respect to the lubricant molecule and their

169: configuration is fully defined by the displacement parameter $p$ chosen,

170: e.g., as the smallest positive longitudinal coordinate of the atoms of upper

171: solid.

172: 

173: All this permits to write the full energy (per one lubricant) in the simple

174: form: 

175: \begin{equation}

176: E=\sum_{n,m}V_{L-J}(|{\bf r}_{n}-{\bf R}_{m}|)+k\sin ^{2}\phi ,  \label{2}

177: \end{equation}

178: where ${\bf r}_{n}$ and ${\bf R}_{m}$ are respectively positions of atoms of

179: solids and atoms in the lubricant molecule. Obviously, this extremely

180: simplified model does not pretend to give a quantitative description and

181: explanation of friction experiments in real systems with complex and

182: typically incommensurate solid and molecular structures. It serves mainly to

183: illustrate some new qualitative possibilities for weakly dissipative

184: processes, associated with the internal (rolling or rotational) degrees of

185: freedom of molecular (nanotubes including) lubricants, in contrast to the

186: dissipation by only translational motions of the ''point-like'' (atomic)

187: lubricants.

188: 

189: \section{\label{sec:3} The Adiabatic Dynamics}

190: 

191: The adiabatic treatment of the system, corresponding to the expression (\ref

192: {2}), follows the lines suggested in Ref. \cite{pog}. The equilibrium

193: distance for Lennard-Jonnes interaction (\ref{1}) is taken equal to the

194: interatomic distance in the solids: $r_{0}=a$. At given separation $d$

195: between the surfaces, we calculate numerically the total energy profile $%

196: E(p,\theta ,\phi )$ as a function of the displacement parameter $p$, and

197: also of the angles $\theta $ and $\phi $. Next this function is optimized

198: with respect to the deformation angle $\phi $ to result in the profiles 

199: $E(p,\theta )$, such as displayed in Figs. \ref{Fig.2}-\ref{Fig.5} (for 

200: different values of $p$, through the whole displacement period from $0$ 

201: to $a$, and at different separations $d$).

202: 

203: The primary optimization in $\phi $ refers to the stronger elastic

204: deformation constant (we took $k=0.5\varepsilon _{0}$, while the amplitude

205: of relevant energy oscillations in $\theta $ is $\sim 0.1\varepsilon _{0}$),

206: and hence to a faster relaxation in $\phi $ than in $\theta $.\cite{note1} 

207: Then the system behavior at very slow uniform variation of the parameter $p$ 

208: with time (that is, the slow dynamics) is obtained from the analysis of the 

209: profile $E(p,\theta )$. Below we analyze how this profile changes with growing

210: external load, which is here simulated by a gradual decrease of the

211: separation distance $d$.\newline

212: {\it i}) At greatest separations (evidently corresponding to the lowest

213: loads), it is seen from Fig. \ref{Fig.2} that the energy profile has a single $\pi /2$

214: periodicity in $\theta $. This means that, for any given $p$, there is a

215: single equilibrium state for the molecule, characterized by its orientation $%

216: \theta \left( p\right) $ (within to a $C_{4}$ rotation) and energy $E\left(

217: \theta \left( p\right) \right) $ , such that $\left( \partial E/\partial

218: \theta \right) _{\theta \left( p\right) }=0$, $\left( \partial

219: ^{2}E/\partial \theta ^{2}\right) _{\theta \left( p\right) }>0$. In this

220: case, the phase trajectory $E(\theta )$ (shown by the sequence of bold dots 

221: linked by arrows, for growing displacement) is closed and continious. Thus, the 

222: system energy changes in a fully reversible way and, though some forces are 

223: exerted in the process, their mean value over the cycle and so the mean friction 

224: force are exactly zero. This reversible variation of the angle $\theta $ around 

225: its median position (such that the long diagonal of the rhomb points vertically, 

226: inset to Fig. \ref{Fig.2}), corresponds to a ''rocking mode'' of the molecular motion.

227: 

228: \begin{figure}

229: \centering{

230: \includegraphics[width=6.cm, angle=0]{rol2.eps}}

231: \caption{Series of profiles of potential energy {\it vs} orientation (at optimal

232: deformation, see the text) and the related equilibrium states (bold circles) of 

233: a lubricant molecule at growing displacement of the solids (the numbers 

234: indicate $p/a$ values) and at highest separation between them. The phase 

235: trajectory (arrows) shows that the molecule performs continuous "rocking" motions and 

236: after the full cycle (at $p/a=1$) returns to its initial state with no energy 

237: dissipation. Inset: the rhombic configuration at $p=0$.}

238: \lb{Fig.2}

239: \end{figure}

240: 

241: \noindent{\it ii}) When the two solid surfaces are getting closer (at increasing

242: load), the first qualitative change that appears in the system behavior is 

243: the doubling of its energy minima. Now there are two splitted minima at the

244: initial configuration $p=0$ (Fig. \ref{Fig.3}), which correspond to a two-fold

245: degenerate equilibrium state of the deformed lubricant molecule: the long

246: diagonal of the rhomb can deviate by a finite angle $<\pi /2$ to both sides

247: from the vertical (inset to Fig. \ref{Fig.3}). Since the molecule is considered a

248: classic object, it initially occupies only one of the minima (the left one

249: is chosen in Fig. \ref{Fig.3}).

250: 

251: \begin{figure}

252: \centering{

253: \includegraphics[width=6.cm, angle=0]{rol3.eps}}

254: \caption{The same as in Fig. \ref{Fig.2} but at closer separation between the 

255: solids (higher external load). The lubricant molecule after "rocking" returns 

256: to its initial state, but in this course it experiences a thermally activated 

257: jump (the dotted arrow over the potential barrier for $p/a=0.4$) between the

258: metastable (0.4) and stable (0.4$^\prime$) minima, and the corresponding energy 

259: difference is irreversibly lost. Inset: the configuration at $p=0$ corresponds to 

260: one of the splitted minima}

261: \lb{Fig.3}

262: \end{figure}

263: 

264: However the degeneracy gets lifted for $p>0$, so that one of the splitted

265: minima turns stable ({\it s}) and another a metastable ({\it m}) equilibrium

266: state. As it is seen from the consecutive curves in Fig. \ref{Fig.3}, the energy 

267: barrier $h$ between the $m$-state and the nearest $s$-state (here that to the 

268: left from $m$) decreases by many times with growing $p$, as a certain function 

269: $h\left( p\right) $. Since the adiabatic lifetime for the $m$-state is: $\tau

270: _{m}=\tau _{a}\exp (\beta h)$ (where $\tau _{a}\sim 10^{-12}$ s is the

271: atomic oscillation time and $\beta $ is the inverse temperature), it

272: decreases in this course by many orders of magnitude. Eventually, this

273: lifetime gets comparable to the characteristic time $\tau _{0}$ of slow

274: displacement by an atomic period, at a very sharply defined instant when 

275: $p=p_{0}$ so that $h\left( p_{0}\right) =h_{0}=\beta ^{-1}\ln \left( \tau

276: _{0}/\tau _{a}\right) $. Hence it is almost exactly at this instant that a

277: thermally activated jump from $m$- to $s$-state is realized. After the jump,

278: the energy difference between $m$- and $s$-states (to the moment of

279: transition) is irreversibly lost, through the creation of quasiparticles

280: (phonons, for insulating solids) which are finally thermalized in the bulk.

281: For typical displacement rates in FFM $\sim 10^{2}$ ${\rm \AA }$/s, one has 

282: $\tau _{0}\sim 10^{-2}$ s, so that the barrier to the transition moment is

283: still as high as $h_{0}\sim 23\beta ^{-1}$.

284: \begin{figure}

285: \centering{

286: \includegraphics[width=6.cm, angle=0]{rol4.eps}}

287: \caption{The same as in Figs. \ref{Fig.2} and \ref{Fig.3} but at still higher load. 

288: The thermally activated jump changes its direction and the lubricant molecule rotates 

289: by the angle $\pi/2$ after every period in $p$ ("rolling mode"). The energy loss 

290: and hence the friction force is somewhat larger than that in Fig. \ref{Fig.3}.}

291: \lb{Fig.4}

292: \end{figure}

293: For the situation presented in Fig. \ref{Fig.4}, the transition corresponds to 

294: $p_{0}=0.4a$ (the potential profile shown by the solid line). It is also seen

295: that, with further growing $p$, up to $a$, the system returns to its initial

296: state; thus the phase trajectory is still closed (''rocking mode'' again)

297: though discontinious. This discontiniuity produces stick-and-slip features

298: in the microscopic behavior of the force {\it vs} displacement and it is the

299: only source of the irreversible losses in our model system, in a full

300: similarity to the model of sliding solids upon atomic lubricants.\newline

301: There are however some distinctions between the two systems. Firstly, the

302: stick-and-slip profile for molecular lubricants is more complicated than the

303: simple triangular sawtooth for atomic lubricants. But especially important

304: is the fact that the irreversible forces for molecular lubricants are 

305: {\it smaller} than elastic, reversible forces (the jump heights in Figs. 

306: \ref{Fig.3}-\ref{Fig.5} are noticeably lower then the amplitudes of smooth 

307: oscillations), while the irreversible forces for atomic lubricants are orders 

308: of magnitude {\it higher} than the reversible ones. Since the reversible forces 

309: in both cases are characterized by the same energy scale $\varepsilon_{0}$, this 

310: indicates a possibility to essentualy reduce the dissipation by molecular (rotating)

311: lubricants {\it vs} that by atomic lubricants.

312: \begin{figure}

313: \centering{

314: \includegraphics[width=6.cm, angle=0]{rol5.eps}}

315: \caption{The same as in Figs. \ref{Fig.2}-\ref{Fig.4} at futher growing load. 

316: The molecule returns from "rolling" to "rocking" regime with an intermediate jump, 

317: but at much smaller energy loss than in Figs. \ref{Fig.3} and \ref{Fig.4}.}

318: \lb{Fig.5}

319: \end{figure}

320: Note, at least, that if the molecule occurs initially (at $p=0$) in the

321: right minimum, it simply spends a period, $0<p<a$, in the $s$-state, but

322: then passes to the $m$-state during the next period and eventually comes to the

323: same closed discontinuous rocking.\newline

324: {\it iii}) Now let us bring the surfaces yet a little bit closer, as by a

325: very small change of the distance $d$ between the solids (Figs. \ref{Fig.3} and 

326: \ref{Fig.4}). Then the system evolution acquires yet a new quality: now at the 

327: transition moment the nearest $s$-state is to the right from the given $m$-state. 

328: This results in that after the transition, at $p>p_{0}$, the molecule does not

329: return to its initial state, but will be eventually rotated by the angle 

330: $\pi /2$ (the dashed circles and arrow in the inset to Fig. \ref{Fig.4}), and 

331: this rotation will be repeated each next period. Thus, in this case we have

332: both unclosed (corresponding to the ''rolling mode'') and discontinious

333: regime of lubricant motion.\newline

334: {\it iv}) With still increasing loads, the situation of a single minimum for 

335: $E(0,\theta )$ will be restored again (though displaced by the angle $\pi /4$), 

336: corresponding to a ''square'' molecule (inset to Fig. \ref{Fig.5}). But now, unlike

337: the case of low loads, this minimum gets splitted with increasing

338: displacement $p$ (here at $p\approx 0.3$). The molecule returns from

339: ''rolling'' to ''rocking'' regime (''rock-n-roll dance''), but with

340: considerably reduced energy dissipation: the energy loss at a jump for $d=2.1a$ 

341: (Fig. \ref{Fig.5}) is about 2.5 times smaller than for $d=2.28a$ (Fig. \ref{Fig.4}),

342: i.e. it {\it \ decreases} with load increasing. We do not present here detailed 

343: results of numerical simulations for even higher loads, resuming only that they 

344: reveal a number of subsequent dissipative regimes, either with growing and falling 

345: friction forces.

346: \section{\label{sec:4} Conclusions}

347: The above simple analysis demonstrates that boundary friction with

348: participation of molecular (spheroidal or cylindrical) lubricants can

349: possess quite unusual properties, such as existence of various regimes of

350: molecular motion, either non-dissipative and dissipative, with abrupt

351: transitions from one regime to another at continious variation of external

352: load. In the sequence of regimes (''rocking'' and ''rolling''), the obtained

353: friction force {\it vs} load displays non-monotonic and hence non-linear

354: behavior. This model (of course, with due improvements to be inserted) may

355: provide a mechanism for explaining the data of real experiments with the

356: fullerenes $C_{60}$ as lubricants and give indications for an optimal regime

357: of their practical applications.

358: 

359: \section*{Acknowledgments}

360: 

361: V.M.L. acknowledges partial support of this investigation by CRDF grant 

362: SCOPES 7UKPJ062150.00/1.

363: 

364: \begin{thebibliography}{99}

365: \bibitem{bhu}  B.~Bhushan, J.N.~Israelashvili, and U.~Landman. Nature {\bf %

366: 374}, 607 (1995).

367: 

368: \bibitem{mclel}  G.M.~McClelland, and J.N.~Glosli. In: {\it Fundamentals of

369: Friction: Macroscopic and Microscopic Processes} /Eds. I.L.~Singer and 

370: H.M.~Pollock (Kluwer, Dordrecht, 1992). P. 405.

371: 

372: \bibitem{isr}  J.N.~Israelachvili. {\it Intermolecular and Surface Forces},

373: Academic Press, London (1991).

374: 

375: \bibitem{pog}  Yu.G.~Pogorelov. Preprint cond-mat/0309117 (2003).

376: 

377: \bibitem{fk}  Y.I.~Frenkel and T.~Kontorova, Zh.~ Eksp.~Teor.~Fiz., {\bf 8},

378: 1340 (1938).

379: 

380: \bibitem{sok}  J.B.~ Sokoloff, Phys.~Rev. B{\bf 42}, 760 (1990).

381: 

382: \bibitem{bhu2}  B. Bhushan and M. Nosonovsky. Acta Mater. {\bf 51}, 4331

383: (2003).

384: 

385: \bibitem{kro}  H.~Kroto, J.R.~Heath, S.C.~O'Brian, R.F.~Curl, and R.E.~Smalley. 

386: Nature {\bf 318}, 162 (1985).

387: 

388: \bibitem{Ii}  S.~Iijima. Nature {\bf 354}, 56 (1991).

389: 

390: \bibitem{far}  Z.~Farkas, G.~Bartels, T.~Unger, and D.E.~Wolf. Phys.~Rev.~Lett. 

391: {\bf 90}, 248302 (2003).

392: 

393: \bibitem{bhu1}  B.~Bhushan, B.K.~Gupta, G.W.~van~Geef, C.~Capp, and J.V.~Coe. 

394: Appl.~Phys.~Lett. {\bf 62}, 3253 (1993).

395: 

396: \bibitem{lut}  R.~Luthi, E.~Meyer, H.~Haefke, L.~Howald, W.~Gutmannsbauer,

397: M.~Guggisberg, M.~Bammerlin, and H.-J.~Guntherodt. Surface~Sci. {\bf 338},

398: 247 (1995).

399: 

400: \bibitem{sch}  U.D.~Schwarz, W.~Allers, G.~Gensterblum, and R.~Wiesendanger.

401: Phys.~Rev.~B {\bf 52}, 14976 (1995).

402: 

403: \bibitem{note} Here mean distances are meant, as the only relevant for adiabatic

404: equilibrium states, while small thermal fluctuations become important for

405: activation of relatively fast transitions between these equilibrium states.

406: 

407: \bibitem{note1}Similar hierarchy of relaxation rates between two degrees of 

408: freedom was also reported for point-like lubricants, corresponding to their 

409: normal (transversal) and longitudinal displacements \cite{pog}.

410: 

411: \end{thebibliography}

412: 

413: \end{document}

414: