Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

ARTICLE IN PRESS

Physica A 373 (2007) 88–96


www.elsevier.com/locate/physa

Consequences of the Brenner modification to the Navier–Stokes


equations for dynamic light scattering
André Bardow, Hans Christian Öttinger
Institute of Polymers, Department of Materials, ETH Zurich, 8093 Zurich, Switzerland
Received 20 January 2006
Available online 22 June 2006

Abstract

A modification of the classical Navier–Stokes equations has recently been proposed by Brenner [Is the tracer velocity of
a fluid continuum equal to its mass velocity? Phys. Rev. E 70 (2004) Art. No. 061201; Kinematics of volume transport,
Physica A 349 (2005) 11–59; Navier–Stokes revisited, Physica A 349 (2005) 60–132] and then formalized by Öttinger
[Beyond Equilibrium Thermodynamics, Wiley, Hoboken, 2005]. In the modified theory, a contribution for mass diffusion
is included in the continuity equation. The argument was based on experimental support from thermophoresis which
however depends on the correct formulation of boundary conditions. The controversy therefore remained. Since such an
additional mass diffusion transport mode should contribute to dynamic light scattering spectra, the consequences of the
modified theory for light scattering spectra are discussed in this work. For liquids, the new theory is consistent with
measured scattering data since the modification to the spectrum is usually negligible. The effect could, however, be
observable in gases.
r 2006 Elsevier B.V. All rights reserved.

Keywords: Hydrodynamics; Navier–Stokes; Light scattering; Diffusion; Bulk viscosity

1. Introduction

The continuum description of hydrodynamics has been developed more than 150 years ago resulting in the
famous Navier–Stokes equations. They have been successfully applied to a wealth of practical applications
and therefore seem to be validated today.
However, Brenner [1–3] recently questioned the universality of the Navier–Stokes equations. In particular, a
diffusive mass flux was introduced that already occurs in single-component fluids. The modified theory
will be called Brenner–Navier–Stokes equations in the following. The equations were put on a sound
thermodynamical basis by Öttinger [4].
The underlying physical picture is based on the fact that the Navier–Stokes equations were derived without
the knowledge about the existence of individual molecules and atoms. The continuity equation therefore only
accounts for the fact that the same mass is contained in volume element over time evolution. However, as we

Corresponding author. Tel.: +41 44 633 68 13; fax: +41 44 632 10 76.
E-mail address: andre.bardow@mat.ethz.ch (A. Bardow).

0378-4371/$ - see front matter r 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.physa.2006.05.047
ARTICLE IN PRESS
A. Bardow, H. Christian Öttinger / Physica A 373 (2007) 88–96 89

know today, the volume element will most likely not contain the same molecules anymore due to the thermal
motion. If two molecules of same mass but of different volume would exchange their position the mass average
velocity would be zero. According to the classical continuity equation the density should thus not change its
value which is in contrast to the molecular picture just sketched.
Experimental support for the new theory has been drawn from the study of thermophoresis [1,3,5,6]. Here,
the motion of a large particle surrounded by small molecules is driven by a thermal gradient. Previously, the
effect could only be described by introducing a slip boundary condition on the surface of the particle. The
Brenner–Navier–Stokes equations allow a description of thermophoresis retaining the usual non-slip
boundary conditions.
Still, the example is not decisive since both theories are able to describe thermophoresis by a change of
boundary conditions. It would be desirable to probe the fluid bulk behavior directly. This would be possible
using dynamic light scattering (DLS) [7]. The transport modes of a fluid are observable in the spectrum of light
scattered from a small fluid volume element. Thus, if there was an additional diffusive contribution to fluid
dynamics, the DLS spectra should change. It has therefore to be shown that the DLS spectra from the new
theory are still consistent with experimental data. Furthermore, there might be a regime where the theories
might be discriminated based on DLS measurements.
In order to clarify these questions in the controversy between the classical and new theory the light
scattering spectra are derived based on the Brenner–Navier–Stokes equations. These equations are therefore
presented next. Then, the DLS spectra are computed. The implications of the modified theory are discussed
for gases and liquids. Finally, conclusions will be given.

2. Brenner–Navier–Stokes equations

For single-component fluids considered in this work, the time evolution equations for the hydrodynamical
variables, the densities of mass r, momentum M and internal energy , are usually written as
qr q
¼   ðvrÞ, (1)
qt qr

qM q
¼   ðvM þ pÞ, (2)
qt qr
 T
q q q
¼   ðv þ j q Þ  p : v , (3)
qt qr qr
where v ¼ M=r is the velocity field, p the momentum flux or pressure tensor and j q the conductive flow of
internal energy. The fluxes will be described here by Newton’s and Fourier’s law respectively, i.e.,
"  T #
q q 2q q
p ¼ p1  Zs vþ v   v1  Zv  v1, (4)
qr qr 3 qr qr

qT
j q ¼ lq . (5)
qr
Here, the thermal conductivity is denoted by lq , the viscosity by Zs and the dilatational viscosity by Zv .
A fresh look onto the flux terms in the balance equations (1)–(3) might already raise the question why the
flux contribution for momentum and energy is split into a convective and diffusive contribution while the
latter is missing in the continuity equation. Further, one can note that only the (self-)diffusion coefficient is
missing from the list of occurring transport coefficients.
These more intuitive observations have been taken up by Brenner [1–3] and led to a modified hydro-
dynamic theory. The approach has then been formalized by Öttinger (Section 2.2.5 in [4]). Based on the
recent GENERIC formalism of beyond-equilibrium thermodynamics [8,9] an additional diffusive
contribution to the irreversible dynamics was introduced into the theory of hydrodynamics. This led to the
ARTICLE IN PRESS
90 A. Bardow, H. Christian Öttinger / Physica A 373 (2007) 88–96

Brenner–Navier–Stokes equations
qr q
¼   ðvr þ j m Þ, (6)
qt qr
qM q
¼   ðvM þ vj m þ pÞ, (7)
qt qr
 T
q q q
¼   ðv þ a0 j m þ j q Þ  p : v , (8)
qt qr qr
where the diffusive mass flux j m is defined by
   
q m q 1 D0 q p q 1
j m ¼ D0  a0 ¼ þ ð  ra0 Þ . (9)
qr T qr T r qr T qr T
The diffusion coefficient is D0 , the chemical potential m and the temperature T. The coupling coefficient
between mass and heat flux is denoted by a0 . Simple expressions are obtained for ra0 ¼  (mass diffusion
driven by p=T) and for ra0 ¼  þ p (mass diffusion driven by p) [4]. The first expression results from kinetic
theory for an ideal gas [10]. In order to characterize the choice for the coupling coefficient a0 , we introduce a
factor k defined as
 þ p  ra0
k¼ . (10)
r
The choices for a0 discussed above lead to k ¼ p=r and k ¼ 0, respectively.

3. Derivation of light scattering spectra

Since the Brenner–Navier–Stokes equations are not widely known yet some intermediate steps of the
derivation of the light scattering spectrum will be presented in order to highlight the differences to the classical
approach. The derivation of the light spectrum is based on the linearized Brenner–Navier–Stokes equations
qr1 q q
¼ r0  v1   j m , (11)
qt qr qr 1
  
qv1 q q2 Zq q
r0 ¼  p1 þ Zs 2 v 1 þ Zv þ s  v1 , (12)
qt qr qr 3 qr qr
2
qs1 q q q
T0 ¼ k  jm
1 þl T 1, (13)
qt qr qr2
where
 
D0 q kr0 q
jm ¼ p  T1 , (14)
1
r0 T 0 qr 1 T 0 qr
where the entropy s has been introduced. The subscript 1 denotes the deviation of a quantity from its
equilibrium value denoted by 0. Note that the linear momentum balance (12) is identical to the one obtained
from the classical Navier–Stokes equations (cf. Eq. (10.4.13b) in [7]) since all changes occur in the nonlinear
convection term dropping out in the linearization.
It should further be noted that Brenner [1–3] also considered the possibility to employ the volume average
velocity vm ¼ v þ j m =r in Newton’s law (4). Here, the original formulation is retained since the mass average
velocity corresponds to the set of thermodynamic variables while the volume average velocity is an additional
concept. For the purpose of this paper, the two choices give only rise to terms of higher order that can usually
be neglected in the spectrum. The analysis given here therefore applies to both formulations.
The correlation function of the density fluctuations describe the light scattering spectrum. There exist at
least two popular routes to derive the spectrum. One route is based on fluctuating hydrodynamics where
ARTICLE IN PRESS
A. Bardow, H. Christian Öttinger / Physica A 373 (2007) 88–96 91

stochastic contributions are added to the balance equations [11]. The spectrum is then obtained by invoking a
fluctuation-dissipation theorem. The other approach, usually associated with Mountain [12], is based on a
Fourier–Laplace transform and studies the decay of initial perturbations. The strength of these initial
perturbations is obtained from thermodynamic fluctuation theory [7]. Here, this latter route will be followed
for convenience.
The five hydrodynamic equations (11)–(13) contain seven fluctuating quantities (r1 , v1 , p1 , s1 , T 1 ). From a
thermodynamic viewpoint, the extensive variables truly fluctuate while the fluctuations of the intensive
variables are induced by their role as respective Lagrange multipliers in equilibrium. Local equilibrium
thermodynamic equations of state are therefore used to express pressure and entropy fluctuations p1 ; s1 in
terms of density r1 and temperature T 1 [7]:
 
r cV ðg  1Þ
p1 ¼ c2T ½r1 þ aT r0 T 1 ; s1 ¼  0 r1  T 1 . (15)
T0 aT r0
Here, cT is the isothermal speed of sound, aT the thermal expansion coefficient, cV =P the specific heat at
constant volume/pressure and g ¼ cP =cV the specific heat ratio. Thermodynamic fluctuations theory can be
used to show that the remaining fluctuations are statistically independent [7].
Introducing the quantity c1 ¼ q=qr  v1 the linearized Brenner–Navier–Stokes equations can then be
rewritten
 2 
qr1 q a q2
¼ r0 c1 þ DM r þ T1 , (16)
qt qr2 1 T 0 qr2

qc1 c 2 q2 q2 q2
¼  T 2 r1  aT c2T 2 T 1 þ DV 2 c1 , (17)
qt r0 qr qr qr
 2 
qT 1 ðg  1Þ qr1 q2 k0 q a q2
¼ þ gDT 2 T 1  DM r þ T 1 , (18)
qt aT r0 qt qr r0 cV qr2 1 T 0 qr2
where the mass diffusion coefficient DM , the effective thermal diffusion ratio a, the longitudinal kinematic
viscosity DV and the thermal diffusivity DT have been introduced:
D0 c2T k0 r0
DM ¼ ; a ¼ aT r0 T 0  , (19)
r0 T 0 c2T
 
4
DV ¼ Zv þ Zs =r0 ; DT ¼ lq =r0 cP . (20)
3
The new terms compared to the Navier–Stokes-based derivation have been underlined. The additional
contribution to the continuity equation (16) has now the familiar (Fickian) form expected for a diffusion flux.
The occurring diffusion coefficient DM will therefore be assumed to be equal to the isothermal self-diffusion
coefficient.
The computation of the light-scattering spectrum from these equations is now an involved one, but
otherwise a standard calculation. Details are presented in Appendix A. The final result for the spectrum is
 
1 2 1 2Lq2
Srr ðq; oÞ ¼ V r kB TwT 1
p g o2 þ ½Lq2 2
 2

1 Gq Gq2
þ þ , ð21Þ
g ðo  oðqÞÞ2 þ ½Gq2 2 ðo þ oðqÞÞ2 þ ½Gq2 2
where L and G are given by
k20
L ¼ DT þ DM , (22)
c2T cV T 0 g
ARTICLE IN PRESS
92 A. Bardow, H. Christian Öttinger / Physica A 373 (2007) 88–96

1
G ¼ ½ðg  1ÞDT þ DV þ DM g0 , (23)
2
with
 
0 aT g1 k0 ðg  1Þ
g ¼ g  k0 þ  . (24)
cV c2T aT T 0 c2T cV T 0 g
A small non-Lorentzian term usually not observable in practice [7] is neglected (cf. Eq. (A.14)).

4. Discussion

For the discussion, the spectrum is analyzed as function of the frequency o for a fixed size of the scattering
vector q.
The spectrum (21) to be expected from light scattering according to the Brenner–Navier–Stokes equations
consists of a sum of three Lorentzian peak shapes. The first term in Eq. (21) is the unshifted Rayleigh or central
line induced by diffusive modes. In the classical theory based on the Navier–Stokes equations, the width of the
Rayleigh line is determined by the thermal diffusivity. The two further terms correspond to the Brillouin
doublet, Lorentzian lines symmetrically shifted by oðqÞ ¼ cS q of width q2 G. The Doppler shift
characterizes the propagating sound modes. In classical theory, the width is given by the acoustic attenuation
coefficient.
Thus, the spectrum expected based on Brenner’s theory has the same triplet form as the one predicted by the
Navier–Stokes equations. Shifts, as well as prefactors are identical for all peaks in both theories. Note that this
implies that the Landau–Placzek ratio, i.e., the ratio of the integrated intensities of Rayleigh and Brillouin
peaks, a result from equilibrium thermodynamics, is unaffected (cf. p. 245f in [7]). However, the half-width at
half-maximum of the Lorentzian lines, Doc ¼ Lq2 and DoB ¼ Gq2 , is expected to contain an additive
contribution due to the additional mass diffusion mode (see Eqs. (22) and (23)). The new theory would thus
only be consistent with the DLS measurements available in the literature if the additional contributions are
negligible. However, if there are regimes where the new factor is significant dynamic light scattering could also
offer an experimental test for the hydrodynamic theories. In Table 1, the size of the modification due to the
mass diffusion effect is quantified for nitrogen and methane in liquid and gaseous state.
The effect on the Rayleigh component of the spectrum depends on the constitutive equation for the
coupling coefficient. For the coupling coefficient given by ra0 ¼ , as expected from kinetic gas theory, the
half-width is modified whereas the choice ra0 ¼  þ p yields a half-width identical to the one predicted from
the classical Navier–Stokes equations (e.g. Eq. (10.4.39) in [7]). With the former choice that is favored here, the
effect is still negligible in liquids. In gases however, a modification of the half-width by 15% can be expected.
In Fig. 1 (left), the central Rayleigh line predicted from Brenner’s theory is shown for gaseous nitrogen in
comparison to the result expected based on the Navier–Stokes equations. The additional mass diffusion mode
leads to a significant broadening and flattening of the Lorentzian line.

Table 1
Contribution to half-width at half-maximum for Rayleigh and Brillouin lines due to mass diffusion for liquid (l) and gaseous (g) nitrogen
and methane (in %)

N2 ðlÞa CH4 ðlÞb N2 ðgÞc CH4 ðgÞc

Rayleigh line ra0 ¼ 0.001 0.003 16.1 15.7


ra0 ¼þp – – – –
Brillouin line ra0 ¼ 1.5 2.5 35.0 31.3
ra0 ¼þp 1.6 2.6 43.4 38.8
a
Data at T ¼ 84:3 K; r ¼ 787 kg=m3 from Refs. [13,14].
b
Data at T ¼ 145:6 K; r ¼ 378 kg=m3 from Refs. [15,16].
c
Data at T ¼ 293 K from Refs. [17,18]. Thermal expansion coefficient estimated by aT ¼ 103 K1 for liquids [19] and ideal gas relation
aT ¼ 1=T for gases. Scattering vector is q ¼ 105 and 103 cm1 for liquids and gases, respectively.
ARTICLE IN PRESS
A. Bardow, H. Christian Öttinger / Physica A 373 (2007) 88–96 93

2 2

1.5 1.5
S(q,ω)

S(q,ω)
1 1

0.5 0.5

0 0
-2 -1 0 1 2 40 41 42 43 44
ω [MHz] ω [MHz]

Fig. 1. Rayleigh (left) and Brillouin line (right) of light scattering spectrum for gaseous nitrogen predicted from Navier–Stokes equations
(dashed) and from Brenner’s theory with k ¼ p=r (full) and k ¼ 0 (dash-dotted) for q ¼ 103 cm1 .

For the Brillouin doublet, an additive contribution to the half-width at half-maximum is also expected from
the diffusion of mass in the Brenner–Navier–Stokes theory. Introducing dimensionless numbers, the
coefficient G characterizing the half-width can be written as
 
n 4 Zv ðg  1Þ g0
G¼ þ þ þ , (25)
2 3 Zs Pr Sc

where the kinetic viscosity n ¼ Zs =r, the Prandtl number Pr ¼ n=DT and the Schmidt number Sc ¼ n=DM have
been introduced. The difference between the classical Navier–Stokes and the Brenner–Navier–Stokes theory is
visible in dynamic light scattering if the last term in the brackets of Eq. (25) is significant compared to the
others. Since the specific heat ratio g typically covers only a small range of values this implies that the Schmidt
number Sc of the fluid should be small. The quantitative results in Table 1 show that in dense liquids the mass
diffusion effect is negligible for the Brillouin lines. The contribution to the half-width is on the order of a few
percent only. In summary, the Brenner–Navier–Stokes theory is consistent with DLS measurements in liquids.
In other words, light scattering experiments in liquids will therefore probably not be sufficient to discriminate
the two hydrodynamic theories.
The situation changes for the Brillouin lines in gases. Here, roughly 40% of the half-width should
be related to the mass diffusion mode. Fig. 1 (right) compares the Brillouin line predicted by Brenner’s
theory and by the Navier–Stokes equations for gaseous nitrogen. The large impact of the additional
mode on height and width of the peak can be seen. Thus, scattering experiments in gases might provide a
viable route to test the hydrodynamic theories. The modifications of the Rayleigh and the Brillouin lines
should contain sufficient independent information to resolve both the mass diffusion effect and the cross
coupling.
The fluid must however be amenable to a description by the linearized hydrodynamic equations which
might be critical for gases (see e.g. discussion in Ref. [20]). Here, small angle light scattering experiments
were therefore assumed with q ¼ 103 cm1 . Light scattering equipment to achieve wave numbers of this
magnitude have been described e.g. in Ref. [21]. If required, smaller wave numbers could be obtained e.g. by
shadowgraphy [22].
As only a sum of transport coefficients is observable by either L (22) or G (23), a practical realization would
further require to separate the mass diffusion from the other contributions. In particular, experimental data on
the bulk viscosity are scarce and of limited precision (errors: 10–50%) [23]. Furthermore, measurements of the
bulk viscosity are usually based on the assumption of the validity of the Navier–Stokes equations [24]. The
analysis of the Rayleigh line contribution seems therefore more straightforward. In any case, the analysis
requires independent measurements of the other transport coefficients.
ARTICLE IN PRESS
94 A. Bardow, H. Christian Öttinger / Physica A 373 (2007) 88–96

5. Conclusions

The recent hydrodynamic theory of Brenner introduces a new mass diffusion contribution to the continuity
equation. Transport modes occur in the light scattering spectrum. It could be shown that the new contribution
would modify the half-width at half-height of the Brillouin doublet and, depending on the cross-coupling
coefficient a0 , that of the Rayleigh line.
In liquids, the modification of the DLS spectra is expected to be negligible. The Brenner–Navier–Stokes
theory is therefore consistent with the available light scattering measurements in liquids.
In gases, the discrimination between the classical Navier–Stokes equations and the new theory based on
DLS spectra seems possible. Since both, Raleigh and Brillouin lines, are affected dynamic light scattering
spectra are expected to contain sufficient information for identification of the mass diffusion effect and the
cross-coupling with thermal diffusion. However, a careful choice of the experimental system is required to
ensure the hydrodynamic regime and to be able to separate the effect from the other transport processes. Due
to the latter difficulty it might be more promising to study the effect in molecular simulations where this
critical separation can be controlled.

Acknowledgments

The authors thank Howard Brenner, Martin Kröger and Henning Struchtrup for helpful discussions.
A.B. gratefully acknowledges financial support by the Deutsche Forschungsgemeinschaft (DFG) though
Grants BA 2884/1-1–2.

Appendix A. Computation of the spectrum

In order to compute the light scattering spectrum from the linearized Brenner–Navier–Stokes equations
(16)–(18) the procedure presented in Chapter 10.4. in [7] is followed. Here, the Fourier–Laplace transforms
~ sÞ defined by
xðq;
Z 1 Z Z 1
~ sÞ ¼
xðq; dtest d3 reiqr xðr; tÞ ¼ dtest xðq; tÞ (A.1)
0 0

are introduced for the fluctuations r1 ðr; tÞ; c1 ðr; tÞ; T 1 ðr; tÞ. The correlation functions hr1 ðq; 0Þr~ 1 ðq; sÞi;
hr1 ðq; 0ÞT~ 1 ðq; sÞi, etc., describe the light scattering spectrum. The dominant contribution to the spectrum
typically stems from the density–density correlation function since the dielectric constant is usually a strong
function of density but only a weak function of temperature [7]. Therefore, the density–density correlations are
usually sufficient for the analysis.
The Fourier–Laplace transform of Eqs. (16)–(18) can be expressed in matrix form as
~ sÞ  /ðq;
Aðq; ~ sÞ ¼ BðqÞ  /ðqÞ, (A.2)
T
where / ¼ ½r1 ; c1 ; T 1  and the matrices have been defined by
2 3
ðs þ DM q2 Þ r0 ðDM a=T 0 Þq2
6 7
6 o2 ðqÞ=gr0 ðs þ DV q2 Þ aT o2 ðqÞ=g 7
~ 6
Aðq; sÞ ¼ 6   7
   7,
4 ðg  1Þs k0 DM 2 k0 DM a 2 5
  q 0 s þ gDT  q Þ
aT r0 r 0 cV r 0 cV T 0
2 3
1 0 0
6 7
6 0 1 07
BðqÞ ¼ 66 ðg  1Þ
7.
7 ðA:3Þ
4 5
 0 1
aT r0

In these equations, oðqÞ has been defined by oðqÞ ¼ cS q where cS is the adiabatic sound speed.
ARTICLE IN PRESS
A. Bardow, H. Christian Öttinger / Physica A 373 (2007) 88–96 95

The solution to the matrix equation (A.2) can be computed from


~ sÞ ¼ A
/ðq; ~ 1 ðq; sÞ  BðqÞ  /ðqÞ. (A.4)
In order to obtain the density–density correlations hr1 ðq; 0Þr~ 1 ðq; sÞi
the first line in Eq. (A.4) is multiplied by
r1 ðq; 0Þ and then ensemble averaged.1 The resulting correlation function is then
hr1 ðq; 0Þr~ 1 ðq; sÞi BðsÞ
¼ , (A.5)
hr1 ðqÞr~ 1 ðqÞi ~
AðsÞ
where

AðsÞ ~ sÞÞ,
~ ¼ detðAðq; (A.6)
    
2 DM a k0 g1 g1 2
BðsÞ ¼ ðs þ DV q Þ s þ gDT   q2 þ o ðqÞ. (A.7)
T 0 r 0 c V aT r 0 g
The spectrum is now proportional to S rr computed from
 
1 Bðs ¼ ioÞ
Srr ðq; oÞ ¼ p SðqÞRe . (A.8)
~ ¼ ioÞ
Aðs
Here, SðqÞ ¼ hr1 ðqÞr~ 1 ðqÞi is the structure factor that can be obtained from thermodynamic fluctuation theory.
Substitution of the expressions for BðsÞ and AðsÞ ~ gives the spectrum. The density–density correlation function
is found here by Laplace inverting Eq. (A.5)
3  
hr1 ðq; 0Þr~ 1 ðq; tÞi X Bðsi Þðs  si Þ
¼ lim exp si jtj, (A.9)
hr1 ðqÞr~ 1 ðqÞi s!si ~
AðsÞ
i¼1

~ ¼ 0. Since the expression is


where s1 ¼ sþ ; s2 ¼ s ; s3 ¼ s0 are the three roots of the dispersion equation AðsÞ
very complex, an approximation valid for most fluids is usually performed [7]. Since DV q2 , DT q2 , DM q2 5oðqÞ
for most fluids the roots of the determinant may be obtained by expanding the solutions in these small
parameters. This condition has been ensured for all cases discussed in the main body of the text. Here, the
approximation derived below is practically indistinguishable from the full solution. Up to first order in these
small parameters the roots of the dispersion equation are then given by

s0 ¼ Lq2 , (A.10)

s ¼ ioðqÞ  Gq2 , (A.11)


where the coefficients L and G have been defined in Eqs. (22) and (23). Inserting these roots into Eq. (A.9) and
expanding to first order in the small parameters finally yields ‘‘after some tedious algebra’’ [7]
 
hr1 ðq; 0Þr~ 1 ðq; tÞi 1
¼ 1  expðq2 LjtjÞ
hr1 ðqÞr~ 1 ðqÞi g
1
þ expðq2 GjtjÞ cosðoðqÞjtjÞ þ bðqÞ expðq2 GjtjÞ sinðoðqÞjtjÞ, ðA:12Þ
g
where bðqÞ is defined as
 
3G  DV  DM ð1 þ g þ aðg  1Þ=aT r0 T 0  k0 aT =cV Þ
bðqÞ ¼ q . (A.13)
gcS

1
Standard thermodynamic fluctuation theory suggests that the density-velocity fluctuations are uncorrelated. In Brenner’s modified
hydrodynamic equations, only one of the two different velocities v and vm can be decorrelated from density fluctuations. From the
fluctuation-dissipation theorem, one would expect v to be the decorrelated velocity variable.
ARTICLE IN PRESS
96 A. Bardow, H. Christian Öttinger / Physica A 373 (2007) 88–96

If the spectrum is rather to be analyzed in the frequency domain the spectral density could be obtained by
directly expanding Eq. (A.8) or by now Fourier transforming Eq. (A.12)
 
1 2 1 2Lq2
Srr ðq; oÞ ¼ V r kB TwT 1
p g o2 þ ½Lq2 2
 2

1 Gq Gq2
þ þ
g ðo  oðqÞÞ2 þ ½Gq2 2 ðo þ oðqÞÞ2 þ ½Gq2 2
 
ðo þ oðqÞÞ ðo  oðqÞÞ
þ bðqÞ  . ðA:14Þ
ðo þ oðqÞÞ2 þ ½Gq2 2 ðo  oðqÞÞ2 þ ½Gq2 2
A comparison with the results obtained from the classical Navier–Stokes [7,12] shows that the spectra only
differ by the modified definition of L, G and bðqÞ where new DM -terms appear. The last term in Eq. (A.14) is
usually difficult to observe [7] and will therefore be neglected in the discussion in the main body of the paper.

References

[1] H. Brenner, Is the tracer velocity of a fluid continuum equal to its mass velocity? Phys. Rev. E 70 (2004) Art. No. 061201.
[2] H. Brenner, Kinematics of volume transport, Physica A 349 (2005) 11–59.
[3] H. Brenner, Navier–Stokes revisited, Physica A 349 (2005) 60–132.
[4] H.C. Öttinger, Beyond Equilibrium Thermodynamics, Wiley, Hoboken, 2005.
[5] H. Brenner, J.R. Bielenberg, A continuum approach to phoretic motions: Thermophoresis, Physica A 355 (2005) 251–273.
[6] J.R. Bielenberg, H. Brenner, A hydrodynamic/Brownian motion model of thermal diffusion in liquids, Physica A 356 (2005) 279–293.
[7] B.J. Berne, I. Pecora, Dynamic Light Scattering, Wiley, New York, 1976.
[8] M. Grmela, H.C. Öttinger, Dynamics and thermodynamics of complex fluids. I. Development of a general formalism, Phys. Rev. E 56
(1997) 6620–6632.
[9] H.C. Öttinger, M. Grmela, Dynamics and thermodynamics of complex fluids. II. Illustrations of a general formalism, Phys. Rev. E 56
(1997) 6633–6655.
[10] H.C. Öttinger, Hydrodynamics from Boltzmann’s kinetic equation after proper coarse-graining, unpublished, 2006.
[11] L.D. Landau, E.M. Lifshitz, Fluid Mechanics, Pergamon Press, London, 1959.
[12] R.D. Mountain, Spectral distribution of scattered light in a simple fluid, Rev. Mod. Phys. 38 (1966) 205–214.
[13] J.R. Singer, J.H. Lunsford, Ultrasonic attenuation and volume viscosity in liquid nitrogen, J. Chem. Phys. 47 (1967) 811–814.
[14] K. Krynicki, E.J. Rahkamaa, J.G. Powles, Properties of liquid-nitrogen .1. self-diffusion coefficient, Mol. Phys. 28 (3) (1974) 853–855.
[15] J.R. Singer, Excess ultrasonic attenuation and volume viscosity in liquid methane, J. Chem. Phys. 51 (1969) 4729–4733.
[16] K.R. Harris, N.J. Trappeniers, The density dependence of the self-diffusion coefficient of liquid methane, Physica A 104 (1980)
262–280.
[17] E.B. Winn, The temperature dependence of the self-diffusion coefficients of argon, neon, nitrogen, oxygen, carbon dioxide, and
methane, Phys. Rev. 80 (1950) 1024–1027.
[18] G.J. Prangsma, A.H. Alberga, J.J. Beenakker, Ultrasonic determination of volume viscosity of N2, CO, CH4 and CD4 between 77 and
300 K, Physica 64 (2) (1973) 278–288.
[19] M. Taravillo, V.G. Baonza, M. Caceres, J. Nunez, Thermodynamic regularities in compressed liquids: I. the thermal expansion
coefficient, J. Phys. Condens. Matter 15 (2003) 2979–2989.
[20] L. Letamendia, J.P. Chabrat, G. Nouchi, J. Rouch, C. Vaucamps, S.H. Chen, Light-scattering-studies of moderately dense gas-
mixtures—hydrodynamic regime, Phys. Rev. A 24 (1981) 1574–1590.
[21] W.B. Li, P.N. Segrè, R.W. Gammon, J.V. Sengers, Small-angle Rayleigh-scattering from nonequilibrium fluctuations in liquids and
liquid-mixtures, Physica A 204 (1994) 399–436.
[22] J. Oh, J.M.O. de Zarate, J.V. Sengers, G. Ahlers, Dynamics of fluctuations in a fluid below the onset of Rayleigh–Benard convection,
Phys. Rev. E 69 (4) (2004) Art. No. 021106.
[23] R.E. Graves, B.M. Argrow, Bulk viscosity: past to present, J. Thermophys. Heat Treatment 13 (1999) 337–342.
[24] J.F. Xu, X.B. Ren, W.P. Gong, R. Dai, D.H. Liu, Measurement of the bulk viscosity of liquid by Brillouin scattering, Appl. Opt. 42
(2003) 6704–6709.

You might also like