tomlinson paper 2012 prl
TRANSCRIPT
-
8/16/2019 Tomlinson Paper 2012 PRL
1/18
1
Temperature and Velocity Dependences in the Tomlinson/Prandtl
Model for Atomic Sliding Friction
Sergio J. Manzi1, Wilfred T. Tysoe
2 and Octavio J. Furlong
1*
1 INFAP/CONICET, Universidad.Nacional de San Luis, Ejercito de los Andes 950, 5700San Luis, Argentina
2 Department of Chemistry and Laboratory for Surface Studies, University of Wisconsin-
Milwaukee, Milwaukee, WI 53211, USA
Manuscript submitted for publication in Physical Review Letters
Keywords: Tomlinson/Prandtl Model, Monte Carlo simulations, periodic sliding potentials
PACS numbers: 68.35.Af, 07.79.Sp, 46.55.+d
* Author to whom correspondence should be addressed
Telephone: +54 266 4436151
E-mail: [email protected]
-
8/16/2019 Tomlinson Paper 2012 PRL
2/18
2
Abstract
The Tomlinson-Prandtl model for nanoscale sliding friction is analyzed for various
sliding regimes defined by the values of the corrugation factor 2/, where E 0 is the height of the sinusoidal sliding potential, k L is the effective stiffness of the
contact, and a is the surface periodicity. For γ < 1, the friction force tends to zero,
defining a so-called superlubricious regime. The most commonly observed behavior, in
which the friction force increases monotonically with increasing sliding velocity up to a
maximum value F *, given by 1/2, is found for γ > 4.603. However,
completely different behavior is found when 1 < γ < 4.603 due to a breakdown of the
usual assumption that the height of the barrier during sliding varies as . Thevelocity and temperature dependences of the lateral force in this regime are investigated
using Monte Carlo simulations. The friction force still tends to F * at T = 0 K, but in
contrast to the behavior found when for γ > 4.603, the friction force is found to increase
with increasing temperature, reach a maximum value, and then decrease monotonically as
the temperature rises further. Unusual velocity dependences are also found in which the
friction force tends asymptotically to zero as the velocity approaches zero, but does not
increase monotonically to a maximum value as the velocity increases; instead, a
maximum friction force is found at intermediate velocities.
-
8/16/2019 Tomlinson Paper 2012 PRL
3/18
3
Applications involving sliding interfaces at the micro- and nano- scale, such as
micro/nano-electro-mechanical systems (MEMs and NEMs), have become of great
importance for technological advances, requiring a better understanding of the frictional
phenomena down to the atomic scale [1]. The introduction of the atomic force
microscope (AFM) [2], has allowed friction phenomena under single asperity conditions
to be studied, [3,4]. The Tomlinson/Prandtl model [5,6] is extensively used to rationalize
AFM measurements of sliding friction [7]. This atomic-scale model assumes that a
harmonically coupled tip slides across a corrugated surface potential. The model is
generally analyzed using a simple sinusoidal potential to mimic the surface corrugation,
with a superimposed, moving parabolic potential for the sliding contact and can be
described as
, 2 cos 2
2
, (1)
where k L is the effective stiffness of the contact, a the surface periodicity, E 0 the potential
corrugation, x the tip position, and v the scanning velocity, so that vt becomes the time-
dependent position of the tip support, X .
The model describes the transition of the tip from one stable position to the next
when it is capable of overcoming the energy barrier ∆ E between the two stable points,
thereby giving rise to energy dissipation, resulting in a stick-slip type of motion. At T = 0
K, spontaneous sliding occurs when the combined sinusoidal+parabolic potential results
in the formation of an inflection point (∆ E = 0) that allows the tip to move spontaneously
to the next minimum in the potential, at a lateral force referred to as critical or maximum
-
8/16/2019 Tomlinson Paper 2012 PRL
4/18
4
lateral force F *
(in the present work it is referred to only as the critical force since it will
be shown that it is not necessarily the maximum value), which is given by
2
1 , (2)
where γ (the corrugation factor) is the ratio between the strength of the tip-sample
interaction and the elastic energy of the system and is defined by
2
. (3)
At some finite temperature T , the tip transition is considered to be thermally
activated, enabling the jump to occur when ∆ E > 0. In general, this is assumed to result
in a monotonic decrease in friction force ( F L < F *) with increasing temperature and
decreasing velocity. To facilitate solution of the Tomlinson/Prandtl model, the energy
barrier is often assumed to depend on the lateral force F L as [8,9],
∆ 1 , (4)
where the parameter β [10,11] takes into account the shape of the tip-sample potential.
This assumption along with a transition-state theory analysis, yields the commonly used
relationship between friction force, temperature and sliding velocity [9],
1
ln 12 ln 1
, (5)
where v0 is a characteristic velocity given by 2/3√ , and f 0 is thetransition attempt frequency. This equation has been shown to successfully describe the
velocity and temperature dependence in sliding friction on mica and HOPG [9,12]. We
demonstrate that this equation is applicable only over a certain range of conditions due to
the assumptions involved in deriving Eq. (4) to describe ∆ E as a function of F L. A more
-
8/16/2019 Tomlinson Paper 2012 PRL
5/18
5
detailed analysis reveals regimes where both an increase of the lateral force with T , and a
decrease with scanning velocity v, can also be described by the classical Tomlinson
model for a simple sinusoidal sliding potential.
Different regimes can be identified within the framework of the Tomlinson model
based on the values of the corrugation factor γ (Eq. (3)). It is well known that for γ ≤ 1 the
model describes a superlubricious or ultra-low friction regime [13], where friction tends
to zero since the instabilities, that under stick-slip conditions give rise to the dominant
energy dissipation mechanism, disappear. Another important range of γ values are those
at which the tip, under fully relaxed conditions ( x = 0 and X = 0), has no neighboring
minimum in the potential to jump to, so that a finite minimum force ( F min) must be
applied before a transition can occur regardless of the values of T and v. This is
illustrated in Figure 1, where the parameter values E 0 (0.317 eV), a (0.29 nm), and k L
(0.6, 2.59 and 6 N/m) have been chosen to be consistent with previous experimental
conditions [14]. This plots the positions at which the tip is at a minimum of the potential
(Eq. (1)), given by ∂V ⁄ ∂ x = 0, so at X = 0, x is given by
sin 2 . (6)
In Figure 1, the right hand side of Eq. (6) is represented by the solid black (sinusoidal)
curve, while the solid straight lines represent k L x for different values of k L. When k L <
2.59 N/m ( , k L = 0.6 N/m), there are multiple solutions to Eq. (6), indicating that
forward and backward transitions are allowed, even when the tip is fully relaxed ( x = 0, X
= 0, therefore F L = 0). However, when k L = 2.59 N/m ( ), only one possible solution
remains for each direction at x = 0, such that transitions to neighboring stable points for
k L > 2.59 N/m are not allowed since there are no stable positions to move to from x = 0
-
8/16/2019 Tomlinson Paper 2012 PRL
6/18
6
when F L = 0 ( , k L = 6.0 N/m). The characteristic limiting value of γ, when there is
only one possible solution, can be calculated by taking the first derivative of Eq. 6, since
it is the only solution with the slopes of the two terms having the same values with
respect to the position x. This leads to
2 tan
2 4.4934… (7)
Introducing the resulting value of x (in this case 0.2074 nm) from Eq. (6), gives the value
of k L = 2.59 N/m; substituting into Eq. (3), yields characteristic value of γ = 4.603. Thus
for values of γ < 4.603, a finite lateral force has to be applied for neighboring stable
points to exist. This is illustrated in Figure 1 for the case of k L = 6.0 N/m (γ = 1.987),
where the support has to be displaced a distance X min = 0.114 nm ( F Lmin = 0.433 nN) to
induce a neighboring stable position in the displacement direction (dashed straight line).
Also shown in Figure 1, for k L = 6.0 N/m, is the maximum support displacement ( X max =
0.176 nm) before the energy barrier decreases to zero (∆ E = 0), at which point the tip
jumps regardless of the values of T and v, when a critical lateral force ( F
*
= 0.475 nN) is
obtained.
Thus, three distinct friction regimes can be defined depending on the value of γ.
For γ ≤ 1, the friction force becomes zero, producing so-called superlubricity. For 1 < γ
< 4.603, the normally assumed variation in energy barrier with lateral force (Eq. (4)) does
not apply, giving rise to unusual temperature and velocity effects, while for γ ≥ 4.603, Eq.
(5) provides a good solution to the Tomlinson/Prandtl model. This is emphasized by the
results in Figure 2 which show the calculated values of ∆ E as a function of F L for
different values of γ (solid dotted lines) where, for comparison, the results of Eq. (4) have
been included using the expression for β derived by Furlong et.al. [10] (open dotted
-
8/16/2019 Tomlinson Paper 2012 PRL
7/18
7
lines). Although Eq. (4) does predict a monotonic decrease in ∆ E as a function of F L,
between F L = 0 and F *, the results are quite different in the range 1 < γ < 4.603. First, it is
only when a finite value of F L ( F Lmin) has been achieved that a neighboring stable position
appears and therefore a ∆ E value can be defined. The magnitude of F Lmin increases as the
parameter γ decreases from the particular value, γ = 4.603, deviating from the behavior
predicted by Eq. (4), which assumes a finite value for ∆ E starting from F L = 0. However,
F Lmin starts to show a decrease when γ approaches 1 (e.g. γ = 1.19), as it approaches the
superlubricious regime, γ ≤ 1. The corresponding values of F Lmin are indicated by the
vertical dashed lines intercepting the x-axis. Second, the decrease in ∆ E is not
monotonic, but increases with F L for higher values of F L when 1 < γ < 4.603. Since the
critical force F * is defined as that required to decrease the energy barrier ∆ E to zero, the
plot indicates that, under certain conditions, the lateral force can become larger than F *,
so that F * cannot be considered to be the maximum possible value under all conditions.
In order to emphasize this behavior, Figure 3 expands the plot in Figure 2 to show larger
values of F L. Note that, as expected, the values of F * predicted by Eq. (2) for the
different values of k L show agreement between the calculated results and those resulting
from Eq. (4). As a result, it can be concluded that the use of Eq. (4) and (5) is only valid
when γ > 4.603.
The intervening regime (when 1 < γ < 4.603) gives rise to unusual variations of F L
with T and v, that have not been previously described by the classical Tomlinson model.
In order to study the effects of T and v on the lateral force in this regime, Monte Carlo
simulations based on the one-dimensional classical Tomlinson model were performed.
This strategy has been previously used to explore velocity effects on sliding friction, and
-
8/16/2019 Tomlinson Paper 2012 PRL
8/18
8
it has been demonstrated that it precisely reproduces the solution of the sinusoidal
Tomlinson/Prandtl model [10,15].
Figure 4 shows the simulated results of lateral force ( F L) with velocity (ln(ν)) for
different values of k L (and γ), using f 0 = 50 kHz to be consistent with previous
experimental results [14]. It can be observed that when γ > 4.603, the friction versus
velocity results are in agreement with the predictions of the semiempirical equation (Eq.
(2)), where there is a monotonic increase in friction force with velocity to a maximum
value equal to F *. However, when γ < 4.603, the maximum in lateral force appears at
intermediate velocities, after which friction decreases to the corresponding F
*
value and
similar effects have been observed experimentally [16]. In particular, when k L = 6.0 N/m
(γ = 1.987), a well-defined maximum in lateral friction is observed at ln(ν) ~ -1. In accord
with Figure 2, the value of the upper limit in lateral force ⁄ is indicated by ahorizontal dotted line. As expected, the value of F
* decreases with k L but clearly does not
correspond to the maximum lateral force that can be obtained as a function of v in the
range 1 < γ < 4.603.
Figure 5 displays Monte Carlo simulation results of lateral force versus
temperature at different velocities. All the parameters were kept the same as above,
except for f 0, which has been increased to 500 kHz, still well within the normal range of
f 0 values. As expected, the lateral force at T = 0 corresponds to F *, where based on Eq.
(5), a monotonic decrease to zero with temperature should be observed. However, under
these conditions (γ = 1.987), the simulation shows an initial increase in lateral force from
F * with temperature up to a maximum value, from which it then decreases monotonically
with temperature. Similar behavior has been observed experimentally [17]. Such effects
-
8/16/2019 Tomlinson Paper 2012 PRL
9/18
9
have been previously ascribed to two competing processes acting at the interface: the
thermally activated formation and rupture of an ensemble of atomic contacts, each
occurring with different activation energies [18]. Here, it is shown that by properly
considering the classical one-dimensional Tomlinson model, it is possible to predict such
behavior by a single activation process. In addition, it has been shown that F * is not
necessarily the maximum force value (see Figure 4) and that under certain conditions (1 <
γ < 4.603), the lateral force increases as ∆ E increases from zero (see Figure 3). Therefore,
as the temperature increases from zero, transitions at higher values of ∆ E are favored
resulting in an increase in the lateral force at the low temperatures.
Acknowledgments
We gratefully acknowledge the National Science Foundation under grant number CMMI
0826151 and the CONICET (Argentina) for support of this work.
-
8/16/2019 Tomlinson Paper 2012 PRL
10/18
10
Figure Captions
Figure 1: Graphic representation of Eq. (6). The right hand side of Eq. (6) is represented
by the solid black (sinusoidal) curve, while the solid dotted straight lines represent k L x for
different values of k L, using E 0 = 0.317 eV and a = 0.29 nm. The y-intercept of the dashed
straight line represents the minimum support displacement required to induce the
appearance of neighboring stable position, and the y-intercept of the dotted straight line
indicates the support displacement at which ∆ E becomes cero.
Figure 2: Calculated values of ∆ E as a function of F L for different values of k L where E 0
= 0.317 eV and a = 0.29 mm (solid dotted lines). For comparison, the results of Eq. (4)
have been included using the expression for β derived by Furlong et.al. [10] (with the
corresponding open dotted dashed lines). The vertical dashed lines represent the
minimum lateral force required to induce the appearance of a stable neighboring position
to which to jump.
Figure 3: Expanded plot of Figure 3 for larger values of F L. (k L = 0.6 N/m, solid black
line; k L = 2.59 N/m, ; k L = 4.0, ; k L = 6.0 N/m, ; k L = 8.0 N/m, ; and k L
= 10.0 N/m, ), where the results for k L = 8.0 N/m have been added for clarification
purposes. The value of F * obtained from Eq. (2) is indicated for each curve.
Figure 4: Simulated results of lateral force ( F L) with velocity (ln(ν)) for different values
of k L (γ), using f 0 = 50 kHz, E 0 = 0.317 eV, a = 0.29 nm and T = 298 K.
-
8/16/2019 Tomlinson Paper 2012 PRL
11/18
11
Figure 5: Simulated results of lateral force ( F L) with temperature, T , for different values
of velocity (ln(ν)), using f 0 = 500 kHz, E 0 = 0.317 eV, a = 0.29 nm and k L = 6.0 N/m.
Indicated in the y-axis is corresponding critical force F *.
-
8/16/2019 Tomlinson Paper 2012 PRL
12/18
12
Figures
Manzi et al ,Figure 1
-
8/16/2019 Tomlinson Paper 2012 PRL
13/18
13
Manzi et al , Figure 2
-
8/16/2019 Tomlinson Paper 2012 PRL
14/18
14
Manzi et al , Figure 3
-
8/16/2019 Tomlinson Paper 2012 PRL
15/18
15
Manzi et al , Figure 4
-
8/16/2019 Tomlinson Paper 2012 PRL
16/18
16
Manzi et al , Figure 5
-
8/16/2019 Tomlinson Paper 2012 PRL
17/18
17
References
[1] I. Szlufarska, M. Chandross and R. W. Carpick, J. Phys. D: Appl. Phys. 41, 123001(2008)
[2] C. M. Mate, G. M. McClelland, R. Erlandsson and S. Chiang, Phys. Rev. Lett. 59,1942 (1987).
[3] R.W. Carpick and M. Salmeron, Chem. Rev. 97, 1163–1194 (1997).
[4] E. Gnecco, R. Bennewitz, T. Gyalog and E. Meyer, J. Phys.: Condens. Matter13, R619–R642 (2001).
[5] G. A. Tomlinson, Philos. Mag. 7, 905 (1929).
[6] L. Prandtl, Z. Angew. Math. Mech. 8, 85 (1928).
[7] Y. Dong, A. Vadakkepatt and A. Martini, Tribol. Lett. 44, 367 (2011).
[8] Y. Sang, M. Dubé, and M. Grant, Phys. Rev. Lett. 87, 174301 (2001).
[9] E. Riedo, E. Gnecco, R. Bennewitz, E. Meyer and H. Brune, Phys. Rev. Lett., 91,084502 (2003).
[10] O. J. Furlong, S. J. Manzi, V. D. Pereyra, V. Bustos and W. T. Tysoe, Phys. Rev. B
80, 153408 (2009).
[11] Y. Dong, D. Perez, H. Gao, A. Martini, J. Phys.: Condens. Matter 24, 265001(2012).
[12] L. Jansen, H. Hölscher, H. Fuchs and A. Schirmeisen, Phys. Rev. Lett. 104, 256101
(2010).
[13] A. Socoliuc, R. Bennewitz, E. Gnecco and E. Meyer, Phys. Rev. Lett. 92, 134301(2004).
[14] Q. Li, Y. Dong, D. Perez, A. Martini and R. W. Carpick, Phys. Rev. Lett. 106,
126101 (2011).
-
8/16/2019 Tomlinson Paper 2012 PRL
18/18
18
[15] O. J. Furlong, S. J. Manzi, V. D. Pereyra, V. Bustos, and W. T. Tysoe, Tribol. Lett.
39, 177 (2010).
[16] S. Sills, T. Gray and R. M. Overney, J. Chem. Phys. 123, 134902 (2005).
[17] X. Zhao, S. R. Phillpot, W. G. Sawyer, S. B. Sinnott and S. S. Perry, Phys. Rev. Lett.
102, 186102 (2009).
[18] I. Barel, M. Urbakh, L. Jansen and A. Schirmeisen, Phys. Rev. Lett. 104, 066104
(2010).