The inflection point in the pressure dependence of viscosity under high pressure: A

comprehensive study of the temperature and pressure dependence of the viscosity of

propylene carbonate
,
R. Casalini and S. Bair

Citation: The Journal of Chemical Physics 128, 084511 (2008); doi: 10.1063/1.2834203
View online: http://dx.doi.org/10.1063/1.2834203
View Table of Contents: http://aip.scitation.org/toc/jcp/128/8
Published by the American Institute of Physics
 THE JOURNAL OF CHEMICAL PHYSICS 128, 084511 2008

The inflection point in the pressure dependence of viscosity under high

pressure: A comprehensive study of the temperature and pressure
dependence of the viscosity of propylene carbonate
R. Casalini1,2,a and S. Bair3
1
Chemistry and Biochemistry Department, George Mason University, Fairfax, Virginia 22030, USA
2
Chemistry Division, Naval Research Laboratory, Washington, DC 20375-5342, USA
3
Center for High Pressure Rheology, Georgia Institute of Technology, Atlanta, Georgia 30332-0405, USA
Received 29 October 2007; accepted 18 December 2007; published online 28 February 2008

The pressure dependence of the prototypical glass-former propylene carbonate has been investigated
over a broad range of temperature and pressure that were inaccessible in previous investigations
using dielectric spectroscopy. We find that the viscosity measurements validate the scaling relation,
T , V = ITV, with a scaling parameter close to that found from dielectric relaxation
measurements. In the pressure dependence of the viscosity, we observe an inflection point in the
log versus P response, similar to that found previously for other materials. However, this
inflection has never been observed in dielectric relaxation measurements. Using the scaling property
above, it is possible to determine the behavior of the dielectric relaxation time in this otherwise
inaccessible experimental range and compare it with the viscosity measurements. We find that the
behaviors of and are very similar, and a very good agreement between the function P
calculated for these two quantities is found. Starting from the validity of the scaling properties, we
show that the inflection point in the pressure dependence of the viscosity can be attributed to the
convolution of the pressure dependences of the compressibility T and the apparent activation
energy at constant volume EV. 2008 American Institute of Physics. DOI: 10.1063/1.2834203

INTRODUCTION T. At even lower temperatures TB T Tg, the behavior

When a liquid is cooled below its melting temperature, if
crystallization is avoided through a rapid cooling rate, its
viscosity progressively increases roughly by 14 orders of
magnitude to the glassy state where it has the rigidity of a
solid but retains the disordered structure of a liquid.
The physics behind the formation of a glass is still de-
bated in the literature and, yet, there is not an accepted the-
oretical explanation for the striking increase of the viscosity
on approaching the glass transition. By a careful analysis of
the temperature dependence of the viscosity over a broad
dynamic range, the behavior seems to be divided into three
main temperature regimes separated by the temperatures TA
and TB, both higher than the glass transition temperature
Tg.14 At the highest temperatures T TA, the liquid is con-
sidered as having an ideal behavior characterized by a con-
stant activation energy i.e., Arrhenius. At lower tempera-
tures TA T TB, the activation energy increases with
decreasing temperature, and the temperature behavior in this
regime is found to be well described using the VogelFulcher
VF equation5,6 or equivalently the WilliamsLandelFerry
equation,7
can be described by a second VF equation. The change of
dynamics between the two VF behaviors at TB has been re-
cently the subject of much attention in the literature. TB is
often found to occur at a temperature TB 1.2Tg, which is in
close agreement with the divergence that was predicted by
the mode coupling theory8 and which has not been observed
experimentally. Several phenomena have been observed in
the vicinity of TB, such as the breakdown of the
DebyeStokes2,9,10 and StokesEinsteinDebye11 although
this is not always the case Refs. 12 and 13, the beginning
of a marked broadening of the structural relaxation function
for some materials H-bonded materials tend to be an
exception,14,15 and the splitting of cooperative and non-
cooperative dynamics relaxations.16,17 The change of dynam-
ics at the temperatures TA and TB can be seen also by the
analysis of relaxation time data.
More recently, the crossover was observed in dielectric
relaxation and viscosity data under high pressure. Interest-
ingly, the relaxation time B or viscosity B at which the
crossover occurs is found to be largely independent of
pressure.1821 This is also the case for isochoric curves,20

indicating that the nature of this crossover is not related to a
D TT 0 characteristic temperature or volume.
T = 0 exp , 1
T T0 Here, we investigate the viscosity of a well known glass
former, propylene carbonate14,22 PC under high pressure in
where DT is a constant, T0 is the Vogel temperature at which the regime of low viscosity, where for isobaric processes
should diverge, and 0 is the viscosity in the limit of high there is observed to be a crossover between Arrhenius and
VF behavior. We investigate whether such a crossover can be
a
Electronic mail: casalini@nrl.navy.mil. observed under high pressure in the same manner that the

0021-9606/2008/1288/084511/7/$23.00 128, 084511-1 2008 American Institute of Physics

084511-2 R. Casalini and S. Bair
FIG. 1. Color online Viscosity of PC as a function of the hydrostatic
pressure. Measurements were made at three fixed temperatures.
crossover at higher viscosity, at TBP, was observed under
high pressure. Henceforth, we refer in the paper to these two
crossovers as crossover A and crossover B, since the tem-
peratures of the crossover change with pressure.
EXPERIMENTAL
Propylene carbonate PC was purchased from Aldrich
and used as received. Viscosity measurements were per-
formed in a pressurized falling body viscometer which has
been thoroughly described elsewhere.23 Briefly, a guided cy-
lindrical sinker made of magnetic material falls within a non-
magnetic tube contained within a nonmagnetic pressure ves-
sel. The position of the sinker is detected by a linear variable
differential transformer to measure the terminal velocity. The
viscosity of the sample is determined, with a correction for
buoyancy, from the terminal velocity using calibrations with
liquids of known viscosity.
RESULTS
In Fig. 1, the isothermal viscosity for PC versus pressure
is reported for three temperatures. In all three curves at lower
pressures, an inflection point is present in the viscosity be-
havior. This inflection point has been first reported for vis-
cosity measurements under high pressure in 1925 for a num-
ber of materials by Bridgman The curve of log viscosity
against pressure is at first concave toward the pressure axis,
but above a few thousand kilograms the curve becomes
nearly straight for most liquids, and for a number even re-
verses curvature slightly.24 Then successively confirmed for
other materials such as silicone oils,25 dibutyl phthalate,26,27
glycerol,28 isopropanol, and methanol.27,29,30 The presence of
the inflection has been attributed to the nonlinearity of the
pressure dependence of volume28 in the low pressure re-
gime. This inflection can be made more evident by calculat-
ing the activation volume V = RT ln / PT, which is
reported in Fig. 2 for the three isotherms. The inflection point
in the pressure dependence of corresponds to a minimum
J. Chem. Phys. 128, 084511 2008
FIG. 2. Color online Activation volume of the viscosity calculated from
the data in Fig. 1.
in the activation volume V with respect to pressure. This
type of behavior has not been observed in other types of
measurements, such as dielectric relaxation measurements,
in which V is observed to either increase or remain constant
with increasing pressure.31 The reason for this is probably
the fact that viscosity measurements are currently the only
measurements that can explore this specific dynamic range at
low viscosity under high pressure. This range of viscosity
corresponds to relaxation times of the order of nanoseconds
while because of experimental limitations, dielectric relax-
ation measurements under high pressure are limited to
100 ns.
We observe that, in the low pressure limit, V at differ-
ent temperatures seems to collapse into a single curve from
which a deviation occurs at increasing pressure with increas-
ing temperature. However, in previous cases, it has been ob-
served that V decreases with T until a minimum is reached
and then increases,32 therefore, the constancy of V in this
range of T may be due to the fact that the temperature range
of these measurements is close to a minimum. However, fur-
ther measurements are necessary to assess this point. Below
the inflection point pressure, V increases with decreasing
temperature, as generally observed for liquids and polymer
melts in the higher viscosity larger regime.31 In some
cases, the value of V in the limit of zero pressure has been
found to be of the same order as the molecular or repeating
unit size when compared close to Tg.31,3336 However, in this
case, we observe that the molar volume of PC is
V M 25 C = 85.86 cm3 mol1 which is more than five times
larger than the value of V in the limit of zero pressure.
Using the equation of state EOS previously reported,22
we calculated the volume corresponding to the high pressure
measurements. The viscosity of PC versus the specific vol-
ume is reported in Fig. 3. In this figure, we reported also the
data at atmospheric pressure previously published by Barthel
et al.37 and by Bondeau and Hick,38 the three sets of data
agree well with each other. Since the data in Fig. 3 do not
collapse into a single master curve, evidently, V is not the
unique variable that can describe the dependence of the vis-
084511-3 Temperature and pressure dependence of PC viscosity J. Chem. Phys. 128, 084511 2008

FIG. 3. Color online Viscosity of PC as a function of the specific volume.
The data at atmospheric pressure are from Refs. 37 and 38 and the specific
volume is calculated using the equation of state parameters reported in Ref.
22. Note that the data from Ref. 38 extend to much higher viscosity
1012 mPa s but in this figure only the viscosity data close to that of the
high pressure measurements are reported.
FIG. 4. Color online Viscosity of PC vs the function of temperature and
specific volume, 1000/ TV. The parameter = 4.3 was found empirically to
give the best superposition. The data at atmospheric pressure are from Refs.
37 and 38 and the specific volume is calculated using the equation of state
parameters reported in Ref. 22. Note that the data from Ref. 38 extends to
much higher viscosity 1012 mPa s but in this figure only the viscosity data
close to that of the high pressure measurements are reported.

cosity; however, the curvature of the isothermal measure-

ments is a clear indication that the viscosity is affected in
part by the change in V. Otherwise, if the viscosity was
dominated by thermally activated processes, the isotherms
would be horizontal lines. This behavior is in accord with
what has already been found from dielectric relaxation time
measurements under high pressure closer to the glass
transition.39
Increasing the pressure slows down the dynamics as
does decreasing the temperature, with the important differ-
ence that by increasing pressure at constant temperature the
dependence of the dynamics reflects only the sensitivity of
dynamics to changes of volume. So by doing measurements
under variable pressure, and knowing the equation of state,
the effect of temperature and volume on the dynamics can be
quantified.3944 Probably the more common way to do this
deconvolution is the one first used by Williams45 calculating
the ratio EV / E P of the constant volume apparent activation
energy EVT , V = R ln / T1V and the constant pres-
sure apparent activation energy E PT , V = R ln / T1 P.
The ratio EV / E P is always smaller than 1 and larger than 0,
representing the two limit cases EV = 0 V controlled dynam-
ics and EV = E P T controlled dynamics.
Moreover, it has been found that for more than 40 glass
formers with the exceptions of some hydrogen bonded
materials46 there is a material constant such that the T and
V dependence of viscosity and relaxation time follows the
scaling equation47,48
T,V = ITV,
where I is an unknown function although some forms of I
have been verified to be valid in the dynamic range below
the crossover A Ref. 49. The parameter has been found
to be smaller than 8.5 Refs. 31 and 46, and for the case of
molecular glass formers excluding hydrogen bonded 3.55
8.5.50 We have shown previously that following Eq.
2, and EV / E P satisfy the relationship20,47
E PT,V
= 1 + T PT,V, 3
EVT,V
where P = lnV / T P is the isobaric expansion coeffi-
cient. Since P is slowly varying with T, according to Eq.
3, under isobaric conditions, with increasing temperature,
becomes more sensitive to change of V than to change of T
i.e., EV / E P decreases. Although this may seem inconsistent
with the decrease of V with increasing T, it results from the
decrease of E P with increasing T.
In Fig. 4, we verify that for the viscosity data of PC, Eq.
2 is validated. In this case, the data scale into a single
master curve for = 4.3. This value is close to that found
previously for dielectric relaxation data for which = 3.7
Ref. 22 and = 4.2.51
An extensive characterization of the temperature depen-
dence to determine TA and TB of the dielectric relaxation of
PC at atmospheric pressure was presented by Stickel et al.
using the derivative function T = log / 1000/ T1/2.14
The function, T1 / T, has the property of transforming a
VF behavior into a straight line and an Arrhenius behavior
into a constant. More recently, using T together with the
pressure analog function P = log / P1/2 it has been
found for PC that TB increases with pressure; however, the
2
relaxation time of crossover logB = 7 1 is independent
of T and P.20,22
In Fig. 5, we report a direct comparison of the and
data at atmospheric pressure using the Stickel function T.
There is quite good agreement of the behaviors of and .
Using the function T, there is a clear evidence of the cross-
084511-4 R. Casalini and S. Bair J. Chem. Phys. 128, 084511 2008

known EOS. The modified Stickel function P for both

and data are reported in Fig. 6. In the figure the arrows
indicate the crossover B in this case, between two behaviors
both described by the empirical equation of Johari and Whal-
ley JW Ref. 52 that is the pressure analog of the VF
equation

T, P = T,0exp DPP
P0 P
, 4

where D P is a constant and P0 is a pressure at which a di-

FIG. 5. Color online Stickel function for the viscosity and relaxation time
data at atmospheric pressure. The data are from Refs. 14, 37, and 38.
over at TB between two VF behaviors and the crossover at TA
between the high temperature VF behavior and the Arrhenius
behavior.
Direct comparison between and data at high pressure
vergence of is expected. Calculating P for the JW equa-
tion, a linear function of P with a negative linear coefficient
is obtained. Therefore, two JW equations, one above and one
below the crossover B occurring at the same viscosity as the
crossover at atmospheric pressure, are needed to describe
the pressure dependence of as shown by the linear fits
reported in Fig. 6. Moreover, at low pressure, a maximum in
the P is observed, in correspondence with the inflection
point in the viscosity behavior. To give an empirical descrip-
tion of this inflection point, an equation has been proposed
previously by Bair and Kottke29 which adds, at low pressure,
a less-than-exponential behavior with a divergence at a nega-
tive pressure, P, superposed onto the greater-than-
exponential JW behavior Eq. 4

is not possible since there are no relaxation time data at short
times available. However, as shown before20 using Eq. 2, it
is possible to determine the behavior of at high pressure
using the data at atmospheric pressure which are in a more
extended dynamic range, making the comparison between
and possible.
In the inset of Fig. 6, the data at high pressure are
shown at the temperature corresponding to the measure-
ments reported in Fig. 1. These data were determined from
the data at atmospheric pressure using Eq. 2 with the
M N
P = 0 exp + 5
P P P P
where 0, M, N, P, and P are constants all larger than
zero and P is a constant less than zero. Using Eq. 5 it is
possible to give an accurate description of the behavior of
at low pressures around the inflection point. This relation has
been useful in tribological calculations of elastohydrody-
namic friction.29 Calculating the function P for the addi-
tional term at low pressure yields a linear function of P like
the JW equation but with a positive linear coefficient, so that
P of Eq. 5 has a maximum at low pressures. However, no
physical meaning was given to this additional singularity at
P.
An understanding of the behavior of the viscosity at low
pressure can be found by starting from Eqs. 2 and 3 to-
gether with the equation from Naoki et al.43

EV
EP
=1
P
T
V
T
P
6

Substituting Eq. 3 into Eq. 6 yields

=
1 EP P
PTT EV T

V
T
P
7

and using the thermodynamic equality P / TV = P / T,

where T= lnV / PT, is the isothermal compressibil-
FIG. 6. Color online Modified Stickel function P vs pressure for relax- ity, gives

ation time open symbols and viscosity data solid symbols. The relaxation
time data were determined from the atmospheric pressure data using Eq. 1 1 T EP
together with the known EOS. The solid lines are the best linear fit, and the = . 8
arrows indicate the crossover B. In the inset are reported the dielectric re- TT P EV
laxation time data for which P is calculated in the main figure same
symbols as in the main figure. Finally, using the property of the derivatives,
084511-5 Temperature and pressure dependence of PC viscosity J. Chem. Phys. 128, 084511 2008

FIG. 7. Color online Apparent activation energy at constant volume EV

and compressibility T vs pressure.
FIG. 8. Color online Test of the DebyeStokes relation. Main figure: loga-

rithm of the viscosity divided by the temperature vs the logarithm of the
T T dielectric relaxation time. The lines are linear fit, the linear coefficients are
= , reported in the figure. Top inset: logarithm of the viscosity divided by the
P P T P temperature and the dielectric relaxation time vs the inverse temperature.
The lines are the same fit reported in the main figure.
and the definitions of E P and V given above, it follows that
1 V curve. To better show the superposition of the high pressure
= . 9 data, we report the data in more detail in Fig. 9, since the
T EV
high pressure data are in a more limited dynamic range.
Therefore, since is constant, the pressure dependence of From the analysis of the data, we find that the linear coeffi-
the activation volume is given by the product of EV and T. cient is very close to 1 only in the range between the cross-
In Fig. 7 we show T and EV versus pressure, where EV was over B and the glass transition, while above the crossover B
calculated using the scaling property Eq. 2 as B, B the coefficient is 1.25 0.01. To show this
EV = RT1 log / log V. As shown in Fig. 7, EV mono- we also reported in the top inset of Fig. 8 the quantity
tonically increases with pressure, while T monotonically de- logT11 versus the inverse temperature. The product
creases with pressure and the convolution of these two op- T11 increases with decreasing temperature, similarly to
posite behaviors gives rise to the minimum in the behavior of what was observed for other glass formers.11 At high tem-
V shown in Fig. 2 and also to the maximum in the function perature, logT11 8.2, while at low temperature close
P in Fig. 6. to Tg, logT11 9.3. According to Eq. 10 it is possible
Thus, the inflection point in the pressure behavior of the
viscosity is mainly due to the higher compressibility of the
liquids at low pressure; however, this is more evident at high
temperatures T TA where EV tends to be less pressure
dependent. This effect is also probably the reason why, dif-
ferently from the crossover B, the crossover A is not evident
in isothermal measurements of relaxation time.
The behavior of the viscosity and the dielectric relax-
ation time at different conditions of temperature and pressure
according to the StokesEinsteinDebye SED
relationship11 should satisfy
k BT
= 4rH
3
, 10

where kB is the Boltzmann constant and rH is the hydrody-
namic radius of the diffusing molecule assumed to have
spherical shape. To verify Eq. 10, we report in Fig. 8 the
logarithm of the ratio / T versus the logarithm of , which
according to SED, assuming rH is the same for different dy-
FIG. 9. Color online Check of the DebyeStokes relation. Detail of Fig. 8
namic conditions, should have a linear coefficient equal to 1.
showing the high pressure data. Logarithm of the viscosity divided by the
We find that all the data at different conditions of tem- temperature vs the logarithm of the dielectric relaxation time. The line is a
perature and pressure superimpose onto a single master linear fit and the coefficient is reported in the figure.
084511-6 R. Casalini and S. Bair J. Chem. Phys. 128, 084511 2008

to estimate the hydrodynamic radius rH. We find that in the efficient between logT1 and log is 1.25 0.01 instead
limit of high temperature T TA rH = 0.19 nm, from which of unity as predicted by Eq. 10, and it is independent of
assuming a spherical form for the molecule we calculated pressure. Moreover, the estimated hydrodynamic radius rH is
VH = 17.8 cm3 mol1. We observe that the value of VH is re- of the same order of magnitude of the van der Waals radius
markably close to the value of the activation volume V. rVW as found for other glass formers,11 and that in the region
Taking the data in the literature of salol and orthoterphenyl above the crossover A, the hydrodynamic volume assuming
OTP,11,53 we find V1.5Tg = 25 1 cm3 mol1 and VH a spherical form of the molecule is quite close to the acti-
= 27 cm3 mol1 for salol and V1.6Tg = 33 2 cm3 mol1 vation volume for PC. This was found to be valid also for
and VH = 31 cm3 mol1 for OTP; therefore, the fact, that for OTP and salol.
PC, V and VH have similar values may not be fortuitous,
however further investigations are necessary to understand ACKNOWLEDGMENTS
the generality of this observation. A comparison found in the
literature is that between VH and the van der Waals volume The work at Naval Research Laboratory was supported
VVW calculated as discussed in Refs. 54 and 55. For PC we by the Office of Naval Research.
calculated VVW = 44.7 cm3 mol1, therefore, rVW 1.4rH,
1
which is similar to what has been obtained before for other D. J. Plazek and J. H. Magill, J. Chem. Phys. 45, 3038 1966.
2
K. L. Ngai, J. H. Magill, and D. J. Plazek, J. Chem. Phys. 112, 1887
glass formers with rH always smaller than rVW.11 Interest- 2000.
ingly, while from the data in Fig. 8 the SED relation seems to 3
F. Stickel, E. W. Fischer, and R. Richert, J. Chem. Phys. 102, 6251
give a better description of the low temperature data than the 1995.
4
high temperature data, if we calculate the value of rH in the W. T. Laughlin and D. R. Uhlmann, J. Phys. Chem. 76, 2317 1972.
5
H. Vogel, Phys. Z. 22, 645 1921.
limit of low temperature we find a value VHTg 6
G. S. Fulcher, J. Am. Ceram. Soc. 8, 339 1923.
1.4 cm3 mol1 that is extremely small in comparison with 7
J. D. Ferry, Viscoelastic Properties of Polymers Wiley, New York, 1980.
8
the actual molar volume. Therefore, the quantitative applica- W. Gtze and L. Sjgren, Rep. Prog. Phys. 55, 241 1992.
9
bility of Eq. 10 in the vicinity of Tg seems unlikely. M. D. Ediger, C. A. Angell, and S. R. Nagel, J. Phys. Chem. 100, 13200
1996.
10
K. L. Ngai, J. Phys. Chem. B 103, 10684 1999.
11
I. Chang and H. Sillescu, J. Phys. Chem. B 101, 8794 1997.
CONCLUSIONS 12
J. R. Rajian, W. Huang, R. Richert, and E. L. Quitevis, J. Chem. Phys.
126, 014510 2006.
The pressure dependence of the prototypical glass- 13
R. Casalini, P. G. Santangelo, and C. M. Roland, J. Chem. Phys. 117,
former propylene carbonate has been investigated over a 4585 2002.
14
broad range of temperature and pressure that was inacces- F. Stickel, E. W. Fischer, and R. Richert, J. Chem. Phys. 104, 2043
1996.
sible in previous investigations using dielectric spectroscopy. 15
K. L. Ngai, J. Phys. Chem. B 103, 5895 1999.
We find that the viscosity measurements validate the scaling 16
C. Hansen, F. Stickel, T. Berger, R. Richert, and E. W. Fischer, J. Chem.
equation Eq. 2 Fig. 2 with a scaling parameter close to 17
Phys. 107, 1086 1997.
G. P. Johari and M. Goldstein, J. Chem. Phys. 53, 2372 1970; 55, 4245
that found from the dielectric relaxation measurements.
1972.
In the pressure dependence of the viscosity, we observe 18
R. Casalini, M. Paluch, and C. M. Roland, J. Chem. Phys. 118, 5701
an inflection point similar to that found previously for other 2003.
19
materials. However, this inflection has never been observed R. Casalini and C. M. Roland, Phys. Rev. Lett. 92, 245702 2004.
20
R. Casalini and C. M. Roland, Phys. Rev. B 71, 014210 2005.
in dielectric measurements. Using the scaling property Eq. 21
S. Bair, C. M. Roland, and R. Casalini, Proc. Inst. Mech. Eng., Part J: J.
2 it is possible to determine the behavior of the dielectric Eng. Tribol. 221, 40 2007.
22
relaxation time in this otherwise inaccessible experimental S. Pawlus, R. Casalini, C. M. Roland, M. Paluch, S. J. Rzoska, and J.
range and compare it with the viscosity measurements. We Ziolo, Phys. Rev. E 70, 061501 2004.
23
S. Bair, High-Pressure Rheology for Quantitative Elastohydrodynamics
find that the behaviors of and are very similar, and a very Elsevier Science, Amsterdam, 2007, pp. 8086.
good agreement between the function P calculated for these 24
P. W. Bridgman, Proc. Natl. Acad. Sci. U.S.A. 11, 603 1925.
two quantities is found Fig. 6. 25
P. W. Bridgman, Proc. Am. Acad. Arts Sci. 77, 115 1949.
26
Starting from the validity of the scaling properties Eq. J. B. Irving and A. J. Barlow, J. Phys. E 4, 232 1971.
27
R. L. Cook, C. A. Herbst, and H. E. King, Jr, J. Phys. Chem. 97, 2355
2, we showed that the inflection point in the pressure de- 1993.
pendence of the viscosity i.e., the maximum in P can be 28
C. A. Herbst, R. L. Cook, and H. E. King, Jr., Nature London 361, 518
attributed to the convolution of the pressure dependences of 1993.
29
the compressibility T and of the apparent activation energy S. Bair and P. Kottke, Tribol. Trans. 46, 289 2003.
30
R. L. Cook, H. E. King, Jr., C. A. Herbst, and D. R. Herschbach, J. Chem.
at constant volume EV. Since, following Eq. 2, the activa- Phys. 100, 5178 1994.
tion volume is proportional to the product of T and EV Eq. 31
C. M. Roland, S. Hensel-Bielowka, M. Paluch, and R. Casalini, Rep.
9, the inflection point is due to the crossover from the low 32
Prog. Phys. 68, 1405 2005.
S. Bair, J. Tribol. 123, 50 2001.
pressure regime where T depends more strongly on pressure 33
G. Floudas and T. Reisinger, J. Chem. Phys. 111, 5201 1999.
than EV to the high pressure regime where T is less pressure 34
R. Casalini, M. Paluch, and C. M. Roland, Phys. Rev. E 67, 031505
dependent than EV, as shown in Fig. 7. 2003.
35
Finally, we investigated whether the SED equation Eq. R. Casalini and C. M. Roland, J. Chem. Phys. 119, 4052 2003.
36
10 is valid for PC. We find that the behavior of T1 versus R. Casalini, C. M. Roland, and S. Capaccioli, J. Chem. Phys. 126,
184903 2007.
at high pressures superimposes with the behavior at atmo- 37
J. Barthel, M. Utz, K. Gro, and H. J. Gores, J. Solution Chem. 24, 1109
spheric pressure. In the investigated dynamic range, the co- 1995.
084511-7 Temperature and pressure dependence of PC viscosity J. Chem. Phys. 128, 084511 2008

38
A. Bondeau and J. Hick, J. Phys. Paris 46, 1717 1985. 2006.
39 47
C. M. Roland, M. Paluch, T. Pakula, and R. Casalini, Philos. Mag. 84, R. Casalini and C. M. Roland, Phys. Rev. E 69, 062501 2004.
1573 2004. 48
R. Casalini and C. M. Roland, Colloid Polym. Sci. 283, 107 2004.
40
G. Floudas, K. Mpoukouvalas, and P. Papadopoulos, J. Chem. Phys. 124, 49
R. Casalini, U. Mohanty, and C. M. Roland, J. Chem. Phys. 125, 014505
074905 2006. 2006.
41
G. Williams, Trans. Faraday Soc. 60, 1556 1964. 50
R. Casalini, K. J. Mc Grath, and C. M. Roland, J. Non-Cryst. Solids 352,
42
M. L. Ferrer, C. Lawrence, B. G. Demirjian, D. Kivelson, C. Alba-
4905 2006.
Simionesco, and G. Tarjus, J. Chem. Phys. 109, 8010 1998. 51
43
M. Naoki, H. Endou, and K. Matsumoto, J. Phys. Chem. 91, 4169 A. Reiser, G. Kasper, and S. Hunklinger, Phys. Rev. B 72, 094204
1987. 2005.
52
44
M. Paluch, R. Casalini, and C. M. Roland, Phys. Rev. B 66, 092202 G. P. Johari and E. Whalley, Faraday Symp. Chem. Soc. 6, 23 1972.
53
2002. K. U. Schug, H. King, and R. Bhmer, J. Chem. Phys. 109, 1472 1998.
54
45
G. Williams, Trans. Faraday Soc. 60, 1548 1964. J. T. Edward, J. Chem. Educ. 47, 261 1970.
55
46
C. M. Roland, S. Bair, and R. Casalini, J. Chem. Phys. 125, 124508 A. Bondi, J. Phys. Chem. 68, 441 1964.