1
Microstructure of fluid systems
Gerhard Nägele
Research Centre Jülich, IFF, Soft Matter Division University of Duesseldorf
Modified version of my lecture at IFF Spring School „Soft Matter“ , Jülich, March 4, 2008 Adapted to Theoretical Soft Matter Course, U of Düsseledorf, summer term 2011
Outline
1. Introduction
- Model systems
2. Pair Distribution Function - Basic properties
- Potential of mean force
- Relation to scattering & thermodynamics 3. Ornstein-Zernike Integral Equations
- Direct correlations - Critical opalescence
- Various closure relations 4. Literature
3
1. Introduction
- Model systems: atomar and colloidal
Model systems: atomar
Lennard - Jones 6 - 12 potential of simple liquids: Ar, Xe, …
r
12 6
r r
u(r) = ε 4 σ − σ
electronic repulsion van der Waals attraction
-ε
σ
2
1/6r [nm]
u(r)
σ
5
Colloidal Dispersions
Definition:
1 nm < ∅ < 5 µm Brownian erratic motion
Examples:
proteins Beispiele: viruses (fd)
Beispiele: inorganic particles:
(plexiglass, ...)
solvent: H2O, ...
Industrial products: paint / ink dairy products
Theory: particle interactions
⇒
microstructure⇒
thermodynamics Colloids, proteins and most bacteria share: - inertia-free, laminar hydrodynamics - strong Brownian motion
granular media
atoms
1 nm 10 nm 100 nm 1 µ m
radiusbacteria, protozoa Colloidal dispersions (including proteins & viruses)
molecules
(0.1 – 800 µm)
human cell: ∼10 µm
Colloidal length scales
7
Length scales in micro - biology
P. Nelson, Biological Physics: Energy, Information, Life W.H. Freeman & Company, N.Y. (2008)
Model systems: colloidal
Charge-stabilized particles:
-20 0 20
1 2 3
40
4
r/ σ
u(r) ≈ u (r)
el for |Z | >> 1
B
2
el 2
= B
exp[ a] exp[ r]
k T L 1 a r (r 2a
u (r)
Z κ −κ )
+ κ
>
( )
2
B s
4 L n | Z | 2n
κ = π +
B
2
L = e /( k T) ε
B≈ 0.7nm
r a
( )
vdW eff
1 6
2a 2a
r 2a r
u (r) A
r r
, ,
−
−
− ≈
−
u(r)
u (r)
elu
vdW(r)
vdW-attraction repulsion
[ ] µ m
r
9
r / σ
1 2 3
u(r)
r 2a
u(r) 0 r
, ,
< σ =
> σ
=
∞
Model systems: colloidal
Tuning of strength and range by changing solvent and salt content
Sterically stabilized particles:
• Pairwise additive N-particle potential energy (approximation for CS, not HS)
( )
N N(
i j)
N( )
i ji j i j
U u u r
< <
= ∑ − = ∑
r r r
r
N= { r
1,..., r
N}
„hard spheres“
Phase behavior of colloidal hard - sphere dispersion
Pusey & van Megen, Nature 320, 1986
fcc
cp
π 18
φ =
sphere system
Nv φ= V
Kepler: 1611 Hales: 1998
11
2. Pair Distribution Function
- Basic properties
- Potential of mean force
- Relation to scattering & thermodynamics
Canonical NVT-ensemble: N >> 1 spherical particles in volume V at temperature T
( ) ( ) ( ) ( )
N(n)
1 n n 1 N
N
N
e U
,..., N N -1 ... N - n 1 d ...d
+ Z
−β
ρ = +
∫
r
r r r r
n out of N
probability d.f. of rN
joint d.f. for n << N
particles at
r
n= { r
1,..., r
n}
( )
NN 1 N ßU ßU
Z V, T =
∫
d ...dr r e− =∫
dr e− N1
r N
V
: average number density
N 1 2 N 1 2 N
g r r , g r r g (r)
Define pair distribution function as:
( ) ( )
( ) ( )
(2)
1 2
N 1 2 (1)N (1)
1 2
N N
g , ρ , 1
= →
ρ ρ
r r r r
r r
for r = r12 → “∞”Isotropic fluid state (no crystal or external field) :
: NVT radial distribution function
r
12r
2r
1Basic properties
13
( )
(2)( ) ( )
NN N 2 2 3 N
N
N N -1 U
(r) e
g r d ...d
Z ρ −β
= =
ρ ρ
∫
r
r r idN
( )
g r 1 1
= − N
ideal gas : U = 0 and ZN = VN
•
gN measures pair correlations relative to ideal gas( ) ( )
( )
NN 12 3 N N
N N -1 U
d g r d d ...d e−β / Z N 1
ρ = = −
∫
r ρ∫
r∫
r r r1 2
1
V
∫
d dr r•
4πr2ρ gN is average particle number in shell [r,r+dr] ( i.e. g(r) is conditional pdf )r
dr
Thermodynamic limit for macroscopic system :
General behavior of g(r) :
( )
NN,V
g r lim g (r)
=
→∞ ρ = N / V fixed• g(r) ≥ 0, g (r → ∞ ) = 1
• g(r) ≅ 0,
whenβ u ( r ) >> 1
• g(r) = exp [- β u ( r ) ]
forρ → 0
• g(r)
continuous whereu(r)
piecewise continuous2. neighbor shell
( )
g r
1
1 2
r/σ
Correlation length ξ(T)
1. neighbor shell
Radial distribution function g(r) for soft pair potential (LJ, Ar, Xe, charged colloid)
2
N(r, r r) 4 r r
g (r) + ∆ 1
ρ π ∆ →
=
ideal gas
fluid near-field order
gas liquid crystal (T>0)
T = ε2 / kB T = ε/ kB T = 0.2 / kε B 0.026
φ = φ = 0.419 φ = 0.471
r /σ g(r)
uLJ(r)
g(r) ≈ e−β
Radial distribution function of a 2D Lennard - Jones system (MD simulation)
sphere
Nv φ = V
17
-
jump atr = σ
- g (r < σ) = 0
sinceu (r < σ) = ∞
- g (r ) = exp [- β u (r)] = θ (r - σ )
forρ = 0 r/σ
1
1 2
•
g(r) for soft pair potential•
g(r) for hard-sphere dispersion
Oscillations in g(r) more pronounced for higher density and lower temperature2. neighbor shell
r/σ
Correlation length ξ(T)
( )
g r
1
1 2
1. neighbor shell
( )
N 2N
)
12 B 12 B 3 N ß U( V
w r k T ln g(r ) k T ln d ...d e− lnZ
= − = − +
∫
r r r r
12r
2r
1Potential w(r) of mean force
derivative with respect to position vector of particle 1:
( )
3( )
1,
N 1 N
1 12 1
3 N 2
ßU ßU
d ...d U e
w r U( )
d ...d e
−
−
−∇ =
∫
−∇ = −∇∫
fixedr r
r
r r
•
Potential of mean force between two particles held fixed at r12•
Reversible work (free energy) to move two particles from infinity to distance r12 :( )
12 12w r = F(r ; N −2) − F( ; N∞ −2)
w(r)→ u(r) i.e. g(r) →e− βu(r) for ρ →0
Hard-sphere colloidal fluid
Effective attraction (“depletion force”)
r r/σ
g(r)
w(r) = - kT ln g(r)
1
0 1
Born-Green integral equation determining g(r) for given u(r)
N
1 1 1j
j 1
U( ) u(r )
>
−∇ r = ∑ ∇
assuming pairwise additivity of U(3)
1 2
1 12 1 12 3 3 1 1
3 12
g ( ,
w(r ) u(r ) d 1 u(r )
g(r )
, )
−∇ = −∇ − ρ − ∇
∫ r r r r
Yvon–Born–Greenhierarchy:
2
12 1
(3)
1 2 3
g(r ) g(r ) g(r
3 23g ( , r r r , ) = ) + O( ρ )
•
Non-linear integro-differential equation relating g(r) to u(r)probability for particle 1 at r1
given that 2 at r2 and 3 at r3 force on 1 due to 2 particle
(n) (n 1)
g ↔ g
−21
0 46 . φ =
PY
Born- Green MD
BG equation gives poor predictions for dense fluids22
• Scattering in homodyne mode: λL = λvac/ns = O(σ, ρ-1/3)
λ
Lk
iϑ L
q 4 sin 2
π ϑ
= λ laser
q k
f• Assumptions in basic scattering theory (more rigorous: Maxwell equations)
→ point scatterers (Rayleigh) : σ << λL
→ single scattering (1. Born approximation)
→ no absorption
j
A 1
B
ϑ
Quasi-elastic : f i L ph
2
k k 2 / (h m M)
c ν
≈ = π λ = <<
Phase difference between j and 1 : B1 A1+ = kf ⋅rj1 −k ri ⋅ j1 = ⋅q rj1
Scattered el. field amplitude at detector : s 2 Nj 1
{ (
1) }
L
j
E ( ) 1 exp i
R =
∝ λ
∑
q r⋅ −r qr
j1Relation to scattering & thermodynamics
23
Time-averaged measured intensity : s s* T N
{ (
l j) }
l, j 1 T
I(q) E E exp i
=
= ∝
∑
q r⋅ −rErgodic system (fluid state) :
T N 1
T " "
lim ... ... >>
→ ∞ =
l j 12 12 N
N i i i ( N 2) U( )
12 N
l j
I(q) N e ⋅ N N(N 1) e ⋅ N N(N 1)V d e ⋅ d − e−β / Z
≠
∝ +
∑
q r = + − q r = + −∫
r q r∫
r r N 12 2
1 g (r )
[ ]
V{
i N}
3N 1 d e ⋅ g (r) 1 N (2 ) ( )
= + ρ
∫
r q r − + ρ π δqforward contribution (no correlation information)
Static structure factor in T-limit : S(q) 1 d ei
[
g(r) 1]
lim 1 ( ) ( ) 0N
⋅
= + ρ
∫
r q r − = ∞ δρ q δρ −q ≥l
N N
i i 3
l
l 1 l 1
( ) d e ⋅ ( ) e ⋅ (2 ) ( )
= =
δρ = δ − − ρ = − π ρδ
∑
∑
∫
q r q rq r r r q
Osmotic compressibility equation :
( )
idT Bq 0
T T
lim S q k T
→ p
χ ∂ρ
= χ = ∂
p p
s(N → ∞, V → ∞ , ρ = N/V fixed)
Point scatterers →
r
lx
[
l]
N i
s
l 1 x a
E ( ) d e ⋅ + ( )
= <
∝
∑ ∫
q r x ψq x x
i l
x a
N i
l 1
d e ⋅ ( ) e
<
⋅
=
ψ
∝
∫
x q x x ∑
q rp s
( ) ( ) ψ x = ε x − ε
P(q
I(q) ∝ N ) S(q )
a 2 i 0
a
0
d e ( )
d ( P( )
) q
⋅ ψ
=
ψ
∫
∫
x q x x
x x
local
scattering strength
Single-particle form factor
→
isotropic spheres→
Rayleigh-Gans-Debye (RGD) regime :no phase difference of light through particle and solvent, respectively, i.e
Inherent assumptions :
2 nπ −n a /λ <0.1
P(q =0) =1 P(q = ∞ =) 0
I(q) ∝ N S(q)
Weakly scattering colloidal spheres with a = O(λ) :
25
Example: optically uniform colloidal spheres with Ψ = const
( )
2 1 2
3j (qa)
P(q) b(q)
qa
= =
q σ ≈ 9
•
Large particles scatter more strongly in forward direction•
q > 4.5/a not accessible by light scattering FormamplitudeInterpretation: diffusive Bragg scattering from particle density waves
4 ns
q sin
2
π ϑ
= λ
s
2d sin
n 2
λ = ϑ
1. order constructive interference
d 2
q
≈ π
typical spatial resolution for selected qϑ
/ 2ϑ
/ 2λ
d
27
S(q) and g(r) for a charge-stabilized dispersion
sphere
NV
φ = V
m m
q ≈ π 2 / r r
mlow compressibility
particle cage
r
mSummary: relation to scattering & thermodynamics
2 / qπ
i (t) r
photomultiplier
k
ik
flaser
q ϑ
{ } ( ) [ ]
N
, 1
S(q) 1 cos (0) (0) 1 d cos g(r) 1
N =
=
∑
⋅ i − j = + ρ∫
⋅ −i j
q r r r q r
P(q
I(q) ∝ N ) S(q )
: for single and quasi-elastic scatteringstatic
structure factor
1 S(q) 1 g(r) 1 FT− −
ρ
= +
4
q π sin ϑ2 λ
=
( )
idT Bq 0 T T
lim S q k T
→ p
χ ∂ρ
= χ = ∂
29
Thermodynamic properties
•
Energy equation for internal energy E•
Pressure equation for particle pressure p( )
N N( )
i j N 3(
i j i k)
i j i j k
U u r u , ...
< < <
= ∑ + ∑ +
r r r r
dr
2 (3)
B 3
3
2 2! 3!
E
N
= k T +
ρ∫ d
rg(r)u(r) +
ρ∫ ∫ d
rd ' g
r( , ') u ( , ')
r r r r+ ...
2 (3) 3
id 3 2! 3 3!
p p
u ( , ')
1 d
(r)u(r) d d ' g ( , ') .. .
r r
ρ g ρ
⋅ ⋅
β∂
= − β∂ − +
∂ ∂
∫
r∫ ∫
r r r r r r2 0
N
U
2u(r)
g(r) 4 r dr∞
ρ
= ∫
π2-body
Valid for state-independent, i.e., (ρ
,T) – independent U only (one-comp.)# particles in shell
Thermodynamic properties
•
(Osmotic) compressibility equation (Kirkwood and Buff, 1951)( ) [ ]
{ }
B idT
q 0 T,N, ' T
lim S q 1 d g(r) 1 k T
→ µ
∂ρ χ
= + ρ ∫ r − = ∂Π = χ
Π
: osmotic pressure of N large particlesrelative to reservoir of small particles & solvent
Valid also for non-pairwise additive & state-dependent U
T id T
0, 0, , χ χ ≈ ∞
fluid near triple point ideal crystal
fluid at critical point
31
3. Ornstein - Zernike Integral Equations
- Direct correlations
- Critical opalescence
- Various closure relations
12 12 3 13 23
h(r ) = c(r ) + ρ ∫ d r c(r ) h(r )
•
Introduce total correlation function :h(r ) : g(r ) 1
12=
12− h(r )
12→ 0 ,
forr
12→ ∞
•
Define direct correlation function c(r) through Ornstein-Zernike equation :total correlations of 1 and 2
direct correlations
indirect correlations of 1 and 2 through particles 3,4,…
+ +
= + ...
2 1
3 4
c
1 2
3 h c
1 2
c 2 1
i
ρ d
i• = ∫ r
indirect correlations of 1 and 2 through two intermediate particles 3 and 4•
General properties of c(r) :c )
(r= h(r) = e
−βu(r)− + 1 O( ) ρ
Direct correlations
2 4
12 12 3 13 23 3 4 13 24 34
h(r ) = c(r ) + ρ ∫ d r c(r ) c(r ) + ρ ∫ ∫ d r d r c(r ) c(r )c(r ) + O(c )
33
•
Fourier-transformed OZ equation from convolution theorem :S(q) = + ρ 1 ∫ d e r
iq r⋅h(r) = + ρ 1 h(q)
i0
h(q) d e h(r) 4 dr r sin(qr) h(r) q
⋅
π
∞= ∫ r
q r= ∫
h(q) = c(q) + ρ c(q) h(q)
•
Solving for Fourier-transformed c(r) gives S(q) in terms of c(q) :S(q) 1 0
1 c(q)
= ≥
− ρ
•
c(r) is more amenable to approximations than g(r)ρ T
fluid crystal
2-phase coex.
ρl
ρg ρl
critical point
triple point
Critical opalescence
c
c
T
T
1 p
∂ρ ρ ∂
χ = = ∞
1
1 c(q 0)
S(q 0) 1 d h(r)
− ρ →
→ = + ρ ∫
r= → ∞
c(q 0) d c(r) 1
→ =
∫
r →ρ•
System with attractive part in u(r)•
Critical opalescence near Tc→ fluid turns cloudy, opaque
→ slow dynamics
remains short-ranged
35
Hexane – Methanol mixture (0.67 mol C
6H
14: 0.44 mol CH
3OH)
≈
oT 18 C T ≈ 46 C
o=
c≈
οT T 42 C
hexane–rich phase
demixed 2-phase state mixed fluid state
critical opalescence
http://www.physicsofmatter.com/NotTheBook/CriticalOpal/Frame_Index.html
strong scattering near forward direction
•
Near Tc, assume c(q) expandablein truncated series at q = 0 (Ornstein & Zernike, 1914)2 4
0 2
c(q;T T )c c c q O(q )
ρ ≈ = − +
2 2 2
0 2 2
1 1 1
1 c c q c q
S(q) = −
− + ξ +
≈
( )
1/ 2 2 T id
T
c
c 0.62
(T)
χT T
−χ
ξ = ∝ − → ∞
( )
3[ ]
2
i 1 r /
2 c
1 e
2 r
h(r) d e S(q) 1
⋅ − ξ
π ρ πρ
=
∫
r q r − ≈2 2
2
1 c
S(q)
1 q
I(q)
= −
∝ ζ +
Correlation length
ξ
(T)( )
u uq R -1 r R
valid for and
Deviations from OZ - mean field close to Tc:
Tc 2
1 q S(q) ∝ −η
slope:
Intercept:
c2
2
c /2 ξ
η = 0.04 (3D-Ising)
∝ 1 I(q)
→∝q2 Tc =150.8 K
Argon
Tc
37 Deviations from OZ - mean field behavior very close to Tc:
Tc 2
1 q
S(q) ∝ −η η =0.04 (3D-Ising)
∝ 1 I(q)
→∝ q2
Various closure relations
Ornstein - Zernike equation
h(r) = c(r) ρ d 'c(| - '|)h(r') + ∫ r r r
+ approximate closure relation : + zero overlap condition (HC) :
c(r) =
Functional ofu(r)
&h(r) g(r < σ = ) 0
Closed integral equation for
g(r) = h(r) 1 −
g(r) S(q) u(r)
p, E,
⇒
micro-structure thermodynamics
&
…
39
Examples of important closure relations
•
Percus-Yevick approximation (PY):( )
ind1 d ' c(| ' |) g(r ') 1
c(r ) = g(r) − + ρ ∫ r r − r − = g(r) − g (r )
g(r) = e
−βw(r)[
w(r) u]
nd
(r)
i
e
g (r) ≈
−β −c(r) ≈ g(r) 1 e −
βu(r) = g(r) − y(r)
y(r) = g(r) e
βu(r)u(| |) u(r )
y(r) = + ρ 1 ∫ d r' e
− β r r'−y(| r − r' |) 1 − e
− β '− 1 y(r ) '
Analytic solution for c(r) and S(q) in the case of hard spheres Good approximation for short - range pair potentials only
is exact
PY approximation for c(r)
cavity function, continuous at r = σ
integral equation for y(r) or c(r < σ) = - y(r) :
OZ equation
σ
Analytic PY solution for the c(r) of a hard-sphere fluid
1 2
3 3
r
rc(r ) ( ) 1 ( )
2
σ
< σ = − λ φ + φ + λ φ
σ
•
See my lecture notes for details:FFT
c
PY(r > σ = ) 0
=0.49
φ
0.4 0.3
0.1
0.3
r / σ
1 2 3
u(r)
PY hard - sphere g(r) and S(q) excellent for
φ
< 0.4φ
0.4
φ = 0.49
0.3
0.1
Analytic PY solution for the S(q) of a hard-sphere fluid
Additional approximate closure relations
•
Hypernetted chain approximation (HNC)•
Rescaled mean spherical approximation[ ] [ ]
PY
w(r) u(r)
c (r) = g(r) e −
−β −≈ g(r) − − β 1 w(r) + β u(r)
[ ]
c
HNC(r) = − β u(r) ln g(r) − + g(r) 1 −
RMSA eff
c (r) = − β u(r) , r > σ > σ g(r = σ
eff+) = 0
Good for soft, long-ranged pair potential such as Yukawa & Coulomb
Analytic solution for screened Coulomb potential
Accuracy of OZ integral schemes depends on u(r) and dimensionality43
Thermodynamic inconsistency of integral equations
•
PY, HNC, RMSA etc. approximate the exact g(r):
Hybrid methods partially reinstall thermodynamic consistency Virial Compr. Energyp ≠ p ≠ p
Example: hard spheres
id
p - 1 p
φ
Rogers-Young (RY) scheme: mixing of PY and HNC
[ ]
( )
RY
h(r) c(r)
u(r) f (r)
1
1
c (r) e e 1
f (r)
−β
+ −
= ⋅ −
Compr Virial
T T
χ = χ
determines α
, r r
: 0 : ,
α α
→ ∞
→
HNC PY
f (r) = − 1 e
−αrScreened Coulomb potential:
Hard spheres near freezing:
45
• Rescaled mean spherical approximation (RMSA) :
u(r) ,
effc(r)
= −β r > σ > σ g(r = σ
eff+) = 0
• Hypernetted chain approximation (HNC) u(r) (r)
c(r)
= e
−β⋅ e
γ− γ (r) 1 − γ (r) : h(r) c(r) = −
• Percus - Yevick approximation (PY)
[ ]
e
u(r)1 (r) (r)
c(r)=
−β⋅ + γ − γ − 1
• Rogers - Young mixing scheme (RY): thermodynamically partially self-consistent
{ }
u(r) exp (r)f (r) f (r) 1 1
c(r)
= e
−β⋅
+ γ − − γ (r ) − 1
f (r) = − 1 e
−αr ComprVirial
T T
χ = χ
determines α
: 0 :
α→ ∞
α→
HNC PY
Summary: Various closure relations
4. Literature
1. G. Nägele. Theories of Fluid Microstructures, appeared in Soft Matter: From Synthetic to Biological Materials, 39th IFF Spring School 2008, FZ Jülich Verlag (online verfügbar)
2. J.-P. Hansen and I.R. McDonald, Theory of Simple Liquids, 3rd Edition, Elsevier Academic Press London (2006)