Home >
Documents >
Soft glassy rheology: modelling & measuring strains in ... · SGR Virtual strain Distributions...

Share this document with a friend

Embed Size (px)

of 44
/44

Transcript

SGR Virtual strain Distributions Dynamics Summary

Soft glassy rheology:modelling & measuring strains in amorphous flows

Michel Tsamados1, Adriano Barra2, Peter Sollich2

1 Centre for Polar Observation and Modelling, University College London2 Disordered Systems Group, King’s College London

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Outline

1 Soft glasses: Phenomenology and SGR model

2 Virtual strain analysis

3 Shear flow: steady state distributions

4 Shear flow: dynamics

5 Summary and outlook

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Outline

1 Soft glasses: Phenomenology and SGR model

2 Virtual strain analysis

3 Shear flow: steady state distributions

4 Shear flow: dynamics

5 Summary and outlook

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Soft glasses: phenomenology

Foams, dense emulsions, onion phases, colloidal glasses, clays,pastes, . . .

Common rheological features:

flow curves σ(γ)− σY ∼ γp (0 < p < 1),Herschel-Bulkley (if yield stress σY 6= 0) or power-lawNearly ‘flat’ viscoelastic spectra G′(ω), G′′(ω) for lowfrequencies ω (also in cytoskeleton?)Rheological aging

Suggests common underlying features: arrangements ofparticles/droplets etc are disordered and metastable

Analogy with glasses

Soft glassy rheology approach exploits this; minimal model(based on Bouchaud’s trap model)

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Soft glasses: Linear rheology

Complex modulus for dense emulsions (Mason Bibette Weitz 1995)

Almost flat G′′(ω): broad relaxation time spectrum, glassy

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Colloidal hard sphere glassesMason Weitz 1995

G′′(ω) again becomes flat as volume fraction increases

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Onion phasePanizza et al 1996

Vesicles formed out of lamellar surfactant phase

Again nearly flat moduli

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Microgel particlesPurnomo van den Ende Vanapalli Mugele 2008

G′′(ω) flat but with upturn at low frequencies

Aging: Results depend on time elapsed since preparation,typical of glasses

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

SGR model

Divide sample conceptually into mesoscopic elements

Each has local shear strain l, which increments withmacroscopic shear γ

But when strain energy 12kl2 gets close to yield energy E,

element can yield

Yielding resets l = 0, and element acquires new E from somedistribution ρ(E) ∼ e−E

Yielding is activated by an effective temperature x; modelsinteractions between elements (also: thermodynamic interpretation)

l

E

l

E

l

E’

l

E

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

SGR model

Divide sample conceptually into mesoscopic elements

Each has local shear strain l, which increments withmacroscopic shear γ

But when strain energy 12kl2 gets close to yield energy E,

element can yield

Yielding resets l = 0, and element acquires new E from somedistribution ρ(E) ∼ e−E

Yielding is activated by an effective temperature x; modelsinteractions between elements (also: thermodynamic interpretation)

l

E

l

E

l

E’

l

E

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Equation of motion

In dimensionless units (for time, energy)

P (E, l, t) = −γ∂P

∂l− e−(E−kl2/2)/xP + Γ(t)ρ(E)δ(l)

Γ(t) = 〈e−(E−kl2/2)/x〉 = average yielding rate

Macroscopic stress σ(t) = k 〈l〉Without shear, P (E, t) approaches equilibriumPeq(E) ∝ exp(E/x)ρ(E) for long t

Get glass transition if ρ(E) has exponential tail; happens atx = 1 if ρ(E) = e−E

(possible justification from extreme value statistics)

For x < 1, system is in glass phase; never equilibrates ⇒ aging

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

SGR predictions

Flow curves: Find both Herschel-Bulkley (x < 1) andpower-law (1 < x < 2)

Viscoelastic spectra G′, G′′ ∼ ωx−1 are flat near x = 1In glass phase (x < 1) find rheological aging,loss modulus G′′ ∼ (ωt)x−1 decreases with age t

Steady shear always ‘interrupts’ aging,restores stationary state

Stress overshoots in shear startup, nonlinear G′ and G′′,linear and nonlinear creep, normal stresses (in tensorialversion). . .

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

A broader issue: Defining local strains

Model assigns local strain for any single configuration

Harder than coarse graining change of strain between twosuccessive configurations

Problem: no reference configuration, as in a crystal

Aim

Develop method for assigning local strains and yield energies tomaterial elements, from single snapshots of simulation data

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Outline

1 Soft glasses: Phenomenology and SGR model

2 Virtual strain analysis

3 Shear flow: steady state distributions

4 Shear flow: dynamics

5 Summary and outlook

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Defining elements

Focus on d = 2 (d = 3 can be done but more complicated)

Make elements circular to minimize boundary effects

Position circle centres on square lattice to cover all of thesample (with some overlap)

Once defined, element is co-moving with strain:always contains same particles (“material element”)

Avoids sudden change of element properties when particlesleave/enter, but makes sense only up to moderate ∆γ

Measuring average stress in an element is easy but how do weassign strain l, yield energy etc for a given snapshot?

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Virtual strain analysis

Cannot “cut” an element out of sample and then strain untilyield – unrealistic boundary condition

Idea: Use rest of sample as a frame

Deform the frame affinely to impose a virtual strain γ

Particles inside element relax non-affinely to minimize energy

Gives energy landscape ε(γ) of element

Yield points are determined (for γ > 0 and < 0) by checkingfor reversibility for each small ∆γ (adaptive steps)

Local analysis effectively at T = 0 to avoid stochastic effects;for consistency, do steepest descent to nearest global energyminimum of entire configuration first

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Example: Virtual strain sequence 1

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Example: Virtual strain sequence 2

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Example: Virtual strain sequence 3

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Example: Virtual strain sequence 4

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Example: Virtual strain sequence 5

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Example: Virtual strain sequence 6

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Example: Virtual strain sequence 7

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Example: Virtual strain sequence 8

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Example: Virtual strain sequence 9

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Element energy landscape

0.5

0.55

0.6

0.65

0.7

0.75

0.8

-0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03

ε

γ˜

l

∆E+

γ+

E+

εmin

Extract: minimum energy εmin, strain away from local minimuml = −γmin, yield strains γ±, yield barriers E±

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

So what is the reference configuration?

Locally determined

Stress-free point on virtual energy landscape

Local strain = virtual strain difference between originalconfiguration and reference

Doesn’t presuppose specific structure for referenceconfiguration (cf. Graner et al’s texture tensor)

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Local modulus

Quadratic fit of energy near minimum, or linear fit of stress,gives local modulus k

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

-0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03

σ

˜γ

l

k(˜γ+l)

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Outline

1 Soft glasses: Phenomenology and SGR model

2 Virtual strain analysis

3 Shear flow: steady state distributions

4 Shear flow: dynamics

5 Summary and outlook

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Systems studied

Polydisperse Lennard-Jones mixtures (Tanguy et al), quenchedto low temperatures (T = 10−4 � Tg)

Low shear rates γ = 10−4; N = 104 particles at ρ = 0.925Steady shear driven from the walls (created by “freezing”particles in top/bottom 5% some time after quench)

Check for stationarity & affine shape of velocity profile beforetaking data

Each element contains ≈ 40 particles (diameter ≈ 7)

Large enough to have near-parabolic energy landscape,small enough to avoid multiple local yield events inside oneelement (Tanguy, Tsamados et al)

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Simulation lengthscales

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Yield energy distribution

0 0.5 1E

+/A

0

0.5

1

1.5

2

2.5

Roughly exponential tail as SGR model would postulateSymmetric: E−/A has same distribution within error bars

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Yield strain distributions

-0.5 0 0.5−γ−, γ+

0

1

2

3

4

5

6

Symmetric as assumed in SGRPower-law approach towards small yield strains?

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Modulus distribution

10 20 30 40 50k = K/A

0

0.02

0.04

0.06

0.08

Clear spread; not constant as assumed in model

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Yield energies remain controlled by yield strains

0 10 20 30 40 50k

0

0.5

1

1.5

E+/A

0 0.1 0.2 0.3 0.4 0.5 0.6γ+

Dominant effect on variation of E+ is from yield strain γ+,not from modulus k

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Local strain distribution

-0.05 0 0.05 0.1l

0

5

10

15

20

25

30

Negative l, would need to extend SGR to allow frustration,l 6= 0 after yield (δ(l) → ρ(l|E) ∝ (1− kl2/2E)b)

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Outline

1 Soft glasses: Phenomenology and SGR model

2 Virtual strain analysis

3 Shear flow: steady state distributions

4 Shear flow: dynamics

5 Summary and outlook

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Evolution of local strain with time

0.2 0.21 0.22 0.23 0.24 0.25 0.26γ

-0.01

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

l

Some evidence for sawtooth shape assumed by SGRRearrangement events can perturb many elements at a time

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Population picture of l-dynamics

0 0.05l0

0

0.05

l1

∆γ = 0.005

0 0.05l0

∆γ = 0.01

0 0.05l0

∆γ = 0.02

Scatter plot of l1 = l(after ∆γ) vs l0 = l(initial)Separation into strain convection and yield events?

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Change in other landscape propertiesExample of modulus

0.2 0.21 0.22 0.23 0.24 0.25 0.26γ

15

20

25

30

35

40

k

0.2 0.21 0.22 0.23 0.24 0.25 0.26γ

-0.01

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

l

Stays largely constant between yields as expected;same for yield barriers etc

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Strain maps

-0.02

0

0.02

0.04

0.06

0.08

0.1

"datayield_noblank.dat" u 16:17:($11) every 1::1357::1469

0 10 20 30 40 50 60 70 80 90 100 0

10

20

30

40

50

60

70

80

90

100

Significant correlations along principal strain axes ±45o

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Outline

1 Soft glasses: Phenomenology and SGR model

2 Virtual strain analysis

3 Shear flow: steady state distributions

4 Shear flow: dynamics

5 Summary and outlook

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Summary and outlook

Virtual strain method for assigning local strains, yield energies

Generic: can be used on configurations produced by any(low-T ) simulation

Also for experimental particle positions, given model ofinteraction?

Steady state distributions in shear flow broadly in line withSGR though e.g. local modulus 6= const

Dynamics of local strain has typical sawtooth shape; localstrain rate is of same order as global one but not identical

To be done: effect of varying γ, T , ρ

Also: analysis of induced yield events – well modelled byeffective temperature?

Peter Sollich Modelling and measuring local strains

SGR Virtual strain Distributions Dynamics Summary

Yield events

Reversibility check: increment virtual strain, minimize energy,reduce virtual strain again, minimize energy

Compare original and final configuration via largest particledisplacement ∆∆ = 0: reversible, ∆ > 0: irreversible

Surprisingly, find no obvious lower limit on ∆ > 0In practice ignore irreversibility if ∆ < 0.01Robust: using ∆ < 0.1 gives qualitatively same results

Peter Sollich Modelling and measuring local strains

Recommended