Structure and Mechanical

Properties of Fat Crystal

Networks

Alejandro G. Marangoni
University of Guelph
Guelph, Ontario, Canada

What is this? How is it different from this?

margarine
olive oil

1
 So, what is a fat after all?
• To the chemist and chemical engineer: a complex mixture of high melting
point triacylglycerols in low melting point triacylglyceorls with complex
phase behavior – all can be explained from knowledge of molecular
composition and phase behavior (SFC and phase diagrams are king…)

• To the crystallographer: a metastable polycrystalline material prone to

recrystallization and fractionation – all can be explained from knowledge of
polymorphism (XRD is king and DSC is queen…)

• To the colloidal scientist: a colloidal gel (or colloidal crystal) composed of a

network of polycrystalline fat particles which trap liquid oil within – all can
be explained from knowledge of the mesoscale structure, i.e., colloidal sizes,
interactions, distribution (rheometer is king and microscope is queen…)

• To the consumer: something that makes you fat, but tastes good (cookies and
chocolate rule…)

• All agree on the last point only

Factors affecting the texture

of fats and fat-structured foods

1. Solid fat content

2. Primary crystal habit (polymorphism)

3. Nano and Microstructure

a. crystallite morphology and size
b. spatial distribution of network mass

4. Interparticle interaction forces

2
 What controls hardness?

Temperature G´ (Pa) SFC (%) Fy (N) K

(ºC) (N/mm)
5 2.19E+07a 90.5a 602a 231a

15 2.64E+07b 90.5a 593a 223a

20 1.63E+07c 80.5b 312b 90b

24 4.08E+07d 60.0c 344b 96b

Means with a common superscript within a column are not significantly different (P>0.05).

Floc

Network Bulk fat

Cluster

Domain Lamella

D d
Single
crystallite

3
 Rheology
Mechanical Strength Fat
Sensory Impressions
Macroscopic World
>0.2mm

Crystal Network

Heat, mass, momentum transfer

Solid Fat Content

Microstructure
Crystal Clusters
0.1-200μm
Heat, mass, momentum transfer

Polymorphism Crystals

Heat, mass, momentum transfer Nanostructure

0.4–100nm
Molecular Structure Triacylglycerol Molecules

Triacylglycerols– Molecular Structure

O
Glycerol
H2C-O-C-CH2-(CH2)n-CH3
Backbone

|
CH3-(CH2)n-CH2-C-O-C-H
O |
Fatty Acyl Chain
H2C-O-C-CH2-(CH2)n-CH3
O
Triacylglycerol = glycerol + 3 fatty acids

4
Fatty Acids and Triacylglycerol Types
Fatty Acids Long-Chain Saturated 14:0, 16:0, 18:0
Medium-Chain Saturated 8:0, 10:0, 12:0
Short-Chain Saturated 4:0, 6:0

Mono-Unsaturated 18:1
Poly-Unsaturated 18:2, 18:3, 22:6
Geometric Isomers (cis vs. trans)
Positional Isomers (ω-3,6,9)

Triacylglycerols Simple Homogeneous

Mixed
Complex and varied positional distribution of
fatty acids on TAG molecule.

Triacylglycerol Crystallization
Process
T>Tm Isotropic Melt

Decrease temperature below melting

point of solid into metastable region
(undercooled or supersaturated)

Crystalline Solid

T<Tm

5
 Crystallization Process – more to the
story?
Isotropic Melt Structured Liquid?

Amorphous Solid? Crystalline Solid

Gibbs-Thompson Equation
ΔGn=Anγ-Vn(Δμ/Vm)
5.0×10 -14
ΔGn =1.46⋅10-14 J/n
ΔGn (J/nucleus)

r c=3.4⋅10-7 m
0.0×10 -00

σ=30 mJ m-2
-5.0×10 -14 ΔGm =150 J mol-1
Vm s=8.5⋅10-4 m3 mol-1

rc
-1.0×10 -13
1.0×10 -08 1.0×10 -07 1.0×10 -06 0.00001
r (m)

6
 O O O O

2L
O O O O
O H O H O H O H
O O O O
O O O O
O O O O
Long Spacings O O O O
(001) O O O O
H O H O H O H O
O O O O
O O O O
O O O O

O O
O O
O H O H
O O
O O
O O
O O
O H O H
O O
O O

O O O O

3L
Long Spacings
(001)
O O O O

O O
O O
H O H O
O O
O O
O O
O O
H O H O
O O
O O

7
 Crystalline structure of fats

3L
2L

Modified from Garti & Sato: Crystallization

Processes in Fats and Lipid Systems, 2nd ed.

Polymorphism
→Same chemical composition but different solid-state structure

melt

α→β’→β
Lower melting point Higher melting point
Lower density Higher density
Lower stability Higher stability

Melt-mediate and/or solid-state phase transitions

(polymorphic transformations)

8
The Crystalline State – The Subcell
(Polymorphism)
Characteristic short spacings:
β: 4.6Å (Triclinic)
β’: 3.8Å and 4.2Å
(orthorhombic ⊥)
Triclinic

Orthrhombic α: 4.15Å (Hexagonal)

Stability, melting point and
packing density:
α < β’ < β
Also: γ (orthorhombic ⊥), δ
and several β’ polymorphs
Subforms possible
Hexagonal

9
Rigaku Multiflex

10
 X-rays

2 2

d
x x
2 2

Crystal Planes

nλ = 2d sin θ

REFLECTION

l
r

α
2
X-RAY r
BURN-THROUGH
SAMPLE

DETECTOR

11
 γ α

β’ β

Cocoa Butter Powder XRD

3000
65.71 001
4.60 O O O O

2500 O
O
O
H
O
O
O
H O
O
O
H O
O
O
H

O O O O

O O O O

SAXS
2000 WAXS O O O O

O O O O
O O O O
H O H O H O H O
O O O O
Intensity

O O O O

1500
002
32.65
1000 3.99 3.88
46.34 3.77

004 5.43 3.71

500 005 4.23
16.01 5.18
12.87 8.08

0
0 5 10 15 20 25 30 35 40
2θ

12
Differential Scanning Calorimeter
TA Q1000

Tcrystalization

ΔH Tonset crystalization
Heat Flow

ΔH
Tonset
melting
Tmelting

0 10 20 30 40 50 60
Temperature (°C)

13
 0.0 0.0

-0.5

Heat Flow (W/g)

Heat Flow (W/g)

-0.5
-1.0
β
-1.0 -1.5 β'
α
-2.0
-1.5 β
β' -2.5
A C
-2.0 -3.0
-20 -10 0 10 20 30 40 50 60 0 10 20 30 40 50 60

Temperature (oC) Temperature ( oC)

0.00 0.00
Heat Flow (W/g)

Heat Flow (W/g)

-0.25 -0.25

γ
-0.50 -0.50

-0.75 -0.75
α β'
B D α
-1.00 -1.00
-10 0 10 20 30 40 50 60 -30 -20 -10 0 10 20 30 40 50

Temperature (oC) Temperature (oC)

0 CH2 0 C R1
Cocoa Butter
R2 C 0 CH

CH2 0 C R3 ƒPredominant fatty acids

0
include: palmitic acid (16:0),
Triacylglycerol (TAG) Molecule stearic acid (18:0) and oleic
acid (18:1)

ƒMainly composed of
symmetrical triacylglycerols
POP, SOS and POS

Double- and triple-chain-length

structures

14
Table 3: Fatty Acid Composition of Cocoa Butter
Fatty Acid Cocoa Butter (wt %)
16:0 28.1 ± 0.45a
18:0 34.4 ± 0.47a
18:1 32.6 ± 0.28a
18:2 2.81 ± 0.06a
18:3 and 20:0 1.95 ± 0.21a
a
Data represents the average of seven replicates ± standard error.

Table 4: Triacylglycerol Composition of Cocoa Butter

Triacylglycerol (carbon number) Experimental Results (wt%)
50 19.8 ± 0.27b
52 47.6 ± 0.21b
54 31.4 ± 0.33b
56 1.11 ± 0.13b
b
Data represents the average of four replicates ± standard error.

15
Table 1: Polymorphic forms of cocoa butter as determined by various research groups
Wille and Lutton (1966) Merken and Vaeck (1980) as reported by van Malssen
et al. (1996a)
Polymorphic Tm (°C) Polymorphic Tm (°C) Polymorphic Tm (°C)
Form Form Form
I 17.3 γ 16-18 γ -5-+5
II 23.3 α 20.7-24.2 α 17-22
III 25.5
IV 27.5 β’ 26-28 β’ 20-27
V 33.8 β 33.7-34.9 β 29.34
29-34
VI 36.3

Table 2 : Characteristic short spacings as determined by XRD for the various

polymorphic forms of cocoa butter (Larsson, 1994).
Polymorphic Short Spacings (Å)
form
γ (sub -α ) (I) 3.87(m), 4 .17(s)
α (II) 4.20(vs)
β 2 ’ (III) 3.87(vw), 4.20(vs)
β 1’ (IV) 3.75(m), 3.88(w), 4.13(s), 4.32(s)
β 2 (V) 3.65(s), 3.73(s), 3.87(w), 3.98(s), 4.22(w), 4.58(vs), 5.13(w), 5.38(m)

β 1 (VI) 3.67(s), 3.84(m), 4.01(w), 4.21(vw), 4.53(vs), 5.09(vw), 5.37(m)

The relative intensity: very strong (vs), strong (s), medium (m), weak (w) or very weak (vw).

16
3000
65.71
4.60
2500

2000

1500

32.65
1000 3.99 3.88
46.34 3.77

5.43 3.71
500 4.23
16.01 5.18
12.87 8.08

0
0 5 10 15 20 25 30 35 40

17
18
19
20
 Phase Behavior

Fats are mixtures of lipids, predominantly triacylglycerols,

but also contain minor polar components such as diacyl-
glycerols, monacylglycerols, free fatty acids, phospho-
lipids, sterols, etc.

Continuous solid solution (mixed crystals), eutectic,

monotectic (partial solid solution), compound formation
phase behaviors are commonly observed

Conditions for fat compatibility

1. Equivalent thermal properties
Melting points
Melting and solidification ranges
SFC
2. Similar molecular size, shape and packing
To allow isomorphous replacement or
formation of a single lattice unit in mixtures
3. Similar polymorphism
Transformation from stable to unstable forms
should occur as readily for binary mixtures as with
individual components

21
Continuous solid solution – compatible

Eutectic – incompatible (minima)

Monotectic – significant amounts of lower

melting component solubilized by higher melting
component. Eutectic point close to m.p. of
lower melting component; partial solid solution

Compound formation (maxima)

Liquidus lines predicted by Hildebrand

equation
350
End of melt (K)

345

340

335 PPP
-1 SSS
T=[(R*ln(x)/ΔH)-1/Tm ]
330
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
mol fraction

ln(xi)=ΔHi/R (1/Tm-1/T)

22
Continuous solid solution: SSS/ESS, POS/SOS

Eutectic behavior: PPP/SSS, EEE/SOS, POS/PSO,

PPP/LLL, PPP/SOS

Compound formation: SSO/SOS, POP/OPO

Monotectic behavior: SSS/OOO, SSS/LLL, PPP/POP,

SSS/SOS

Bruker Minispec pNMR

23
 Magnetization Signal Intensity

Solid – liquid
components (total Liquid
hydrogen content) component
only

Pulse tB=10 tA=70

Time (μs)

Isosolid Diagrams (Phase Diagrams?)

SFC vs. temperature profile for
AMF/Cocoa Butter
100
0% AMF
Solid Fat Content (%)

90
10% AMF
80 20% AMF
70 30% AMF
60 40% AMF
50 50% AMF
40 60% AMF
30 70% AMF
80% AMF
20
90% AMF
10 100% AMF
0
0 5 10 15 20 25 30 35 40 45 50
o
Temperature ( C)

24
 40
35
Temperature (oC) 30
25
20
15
10
5
0
0 10 20 30 40 50 60 70 80 90 100
Composition (%w/w MMF)

60
Temperature (oC)

50

40

30

20

10
0 10 20 30 40 50 60 70 80 90 100
Composition (% HMF)

25
 60
HMF
55

Dropping Point (oC)

MMF
50 AMF
45
40
35
30
25
20
0 10 20 30 40 50 60 70 80 90 100
Composition (%w/w milkfat fraction)

SFC influences hardness…

150
* * * Cocoa Butter
135
Penetration Depth

120 MMF
AMF
(1/10 mm)

105
40% HMF
90
75 40% AMF
40% MMF
60
10% MMF
45
10% HMF
30
10% AMF
15
0 *<150
24 hr Tempering 48 hr Tempering
Tempering Time (hours)

26
Ideal Solubility – Fact or fiction?

ln(xi)=ΔHi/R (1/Tm-1/T)

Polymorphism and
Microstructure

27
 Time –Temperature State Diagram for the
Polymorphism of Statically Crystallized Cocoa
Butter
γ SOS β’ β
30
25 α+β’
20
Temperature (oC)

15 β’
10
5 β’+ β
0 α
-5
-10
-15 γ+α
-20
-25
-30
1 10 100 1000 10000 100000
Time (min)

The α form
A C

B D

28
 The β’ form
A C

B D

The β form A D

B E

C F

29
 Cocoa butter crystallized at 0ºC
A D

α β’
(1 day) (14 days)

B E

α →β’ β’
(5 days) (21 days)

C F

α→ β’ β’
(28 days)
(7 days)

Cocoa butter crystallized at 20°C

A

β’
E
(1 day)
β
B (35 days)
β’→β
F
(7 days)
β
C
(35 days)

β’→β
(21 days) G
β
(35 days)
D
β’→β
(21 days)

30
Cocoa butter crystallized at 26°C
A D

γ SOS β
(1 day) (7 days)

B E

β’
β
(3 days)
(28 days)

C F

β β
(7 days) (28 days)

Growth of a Polycrystalline
Fat Crystal Network

31
 Camera

Microscope

Heating
Freezing
Stage

32
Growth of a fat crystal network

33
 Effect of cooling rate
1.2oC/min

0.4oC/min

6.0oC/min

Effect of Surfactant Addition

0.1% vs. 0.5% Tween 60

34
 Temperature Effect

5oC 25oC

Effect of film size on slide

170:m 20:m

35
Cocoa Butter Fat Crystal Network
Visualized using AFM

Cryo-TEM Isobutanol-extracted SSS

36
37
 Relationship between structure and
crystallization behavior
1. Growth mode (Avrami model)
SFC( t ) n
k - rate constant of crystallization
= 1 − e − kt n - dimensionality of growth and
SFC(∞) the type of nucleation

2. Nucleation (τ, ΔGc- Fisher-Turnbull model)

(
−
ΔGc + ΔGd
) ΔGc - activation free energy
J= NkT
h e kT
of nucleation
Δ Gd - activation free
energy of diffusion

Table 6: Avrami exponent values for the different types of growth and nucleation
(Sharples, 1966)
Avrami Exponent Various types of growth and nucleation
3+1=4 Spherulitic growth fromsporadic nuclei
3+0=3 Spherulitic growth frominstantaneous nuclei
2+1=3 Disc-like growth fromsporadic nuclei
2+0=2 Disc-like growth frominstantaneous nuclei
1+1=2 Rod-like growth fromsporadic nuclei
1+0=1 Rod-like growth fromsporadic nuclei

38
 n=4 n=1

75
SFC (%)

50

25
k=0.00103min-1.5
n=1.50
SFC∞=67.7%
0
0 50 100 150 200 250 300 350
Time (min)

39
Avrami Exponent vs. Crystallization
Temperature

Avrami Exponent 4

0
-30 -20 -10 0 10 20 30
o
Crystallization Temperature ( C)

Phase Technology Cloud Point

Analyzer

40
 75

Signal (a.u.)
50

25 τnucleation-A
τnucleation-B

0
0 100 200 300 400
Time (min)

15.0 7.5
A B
ΔGc (kJ/mol)

12.5 5.0
ln(τT)

10.0 2.5
Slope = 221100
R2=0.99
7.5 0.0
0.00000 0.00001 0.00002 0.00003 0 10 20 30
1/T(ΔT)2 Crystallization Temperature ( °C)

41
Avrami Exponent vs. Crystallization
Temperature

Avrami Exponent
4

0
-30 -20 -10 0 10 20 30
o
Crystallization Temperature ( C)

Induction Times as Determined from

Crystallization Curves at Various
Temperatures
10.0
α α +β' β' β
7.5

5.0
log τ

2.5

0.0

-2.5

-5.0
-30 -20 -10 0 10 20 30
o
Crystallization Temperature ( C)

42
 Rheology of Fats

Small Deformation
Dynamic Controlled
Stress Rheometer

TA AR2000

43
Viscoelasticity of Fats

G”/G’ ~ 0.1

Shear strains of ~ 0.1% or less

G’ frequency-independent

44
 30000000 0.05

Storage Modulus (Pa) 25000000 0.04

Strain (%)
20000000
0.03
15000000
0.02
10000000

5000000 0.01

0 0.00
0 2000 4000 6000 8000 10000
Stress (Pa)

20000000 1600000
A B
Shear Modulus (Pa)
Shear Modulus (Pa)

16000000 1200000

12000000
800000
8000000

400000
4000000

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Frequency (Hz) Frequency (Hz)

10000000 60000000
C D
Shear Modulus (Pa)

Shear Modulus (Pa)

8000000
45000000

6000000
30000000
4000000

15000000
2000000

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Frequency (Hz) Frequency (Hz)

45
 100000000 0.5
A
Shear Modulus (Pa)

0.4

Strain (%)
0.3
10000000
0.2

0.1

1000000 0.0
0 5000 10000 15000 20000 25000 30000 35000
Stress (Pa)

Large Deformation Mechanical Tests

46
 2 7 .5
2 5 .0 Y ie ld F o r c e ( N )

2 2 .5

2 0 .0

Force (N) 1 7 .5
1 5 .0 Y ie ld W o r k ( N m m )
1 2 .5
1 0 .0

7 .5

5 .0
2 .5 Y ie ld d e fo r m a tio n ( m m )
0 .0
0 .0 0 .5 1 .0 1 .5 2 .0

D is p la c e m e n t (m m )

Modeling the microstructure of a fat

47
 5 AD 0.5
G= Φ
24πH o3.5

Nederveen, C.J. Dynamic mechanical behavior of

suspensions of fat particles in oil. J. Colloid Science 18:
276 (1963)

A
E= Φ
2πd o
3

For small deformations only

There also exists a modification of van den Tempel’s model by Sherman (1968)

48
 G~Φ
10.0 25
A B
7.5 20

G' (MPa)
G' (MPa)
milkfat 15
tallow
5.0

2.5 10

0.0 5
30 35 40 45 50 55 60 40 50 60 70 80
Φ (%) Φ (%)
15 1.5
C D
G' (MPa)

G' (MPa)
palm 10 1.0
lard
oil 5 0.5

0 0.0
20 30 40 50 60 70 20 25 30 35 40 45
Φ (%) Φ (%)

Network model needs to take

into account the presence of
particle (crystal) aggregates

49
G~Φ for single particle network does
not provide satisfactory description of
system’s behavior

Rather: G~Φμ
Several parameters that determine
the state of aggregation are lumped
together into one dimensionless
‘scaling factor’

He also suggests that μ~N

Good agreement with experiments

About 106 particles per

aggregate

More agitation leads to

less aggregation

50
 G' ~ Φ μ
d+x
μ=
d−D
4 .3
μ≈
3− D

Heertje, I. (1993)

Microstructural Studies in
Fat Research

Food Structure 12, 77-94

51
Support for the view of fats as
crystallite fractal gels grows

Johansson also shows that:

ξ~Φ1/(D-d)

52
Weak-link vs. Strong Link

High SFC vs. Low SFC

Fractal analysis explains the

softening observed
in milkfat upon chemical
interesterification

Narine, S.S. and Marangoni, A.G. (1999)

Fractal nature of fat crystal networks
Physical Review E 59: 1908

53
Structural view – a colloidal gel?

Fractal Floc

Link

Particle

Fractality and the

Fractal Dimension

54
Characterization of the Spatial
Distribution of Mass within the
Fat Crystal Network

Fractal Geometry

Fat crystal networks display

statistical self-similarity
and scaling behavior
characteristic of stochastic
fractal systems

55
Statistical Self similarity – you guess!

56
57
 4x

10x
Self-similarity?
40x

100x

Db=1.60

Scaling Behavior and Fractional

Dimensionality

Scaling behavior G~Φμ Scaling behavior N~LD

20 6

19
5
log (N)

18
ln G'

17 4
16
r 2=0.88 3 D=2.01 (0.06)
15
D=2.45 r 2=0.994
14 2
3.50 3.75 4.00 4.25 4.50
3.5 4.0 4.5 5.0 5.5 6.0
ln Φ
log (length)

58
 μ
ξ ~Φ

Marangoni and Ollivon, 2007, Chemical Physics Letters 442: 360-364

59
 ξ~t1/D
2.25
A B
Log10 Diameter (μm)

slope = 0.67 ± 0.009

2.00 slope = 0.72 ± 0.055
D = 1.49 ± 0.019
D = 1.40 ± 0.106
1.75

1.50 SFC linear with time

SFC linear with time
1.25
2.00 2.25 2.50 2.75 3.00 3.25 2.00 2.25 2.50 2.75 3.00 3.25
Log10 Time (seconds) Log 10 Time (seconds)

Milkfat Spherulites at 20oC

for M~t
Batte and Marangoni, 2005, Crystal Growth and Design 5: 1703-1705

Fracal Structural-mechanical model

The view

60
Structure-mechanical model
Thermodynamic arguments

6δ
1 1
A
E = * Φ 3− D = Φ d −D
aε 2πaε d o
* 2

G = 13 E

Marangoni, 2000, Phys. Rev. B 62, 13951

Marangoni and Rogers, 2003, Appl. Phys. Lett. 82, 3239

Structure-mechanical model
Particle-spring arguments
1
A
G= Φ 3− D
6πcad o3

Narine and Marangoni, 1999, Phys. Rev. E 60, 6991

61
The Yield Stress (Compression)

σ * = E ⋅ε *
6δ 3− D
1
σ =
*
Φ
a
Marangoni and Rogers, 2003, Appl. Phys. Lett. 82, 3239

Model simulations
3000
a=200 nm (a)
D=2.8
σ* (Pa)

2000 δ (J m )→var
-2 0.03

0.01
1000

0.005
0
0.15 0.20 0.25 0.30 0.35
3000
δ=0.01J m-2 (b)
D=2.8
σ* (Pa)

2000 a (nm)→var
100
200
1000
500

0
0.15 0.20 0.25 0.30 0.35
3000
δ=0.01J m -2
(c)
a=200 nm
σ*(Pa)

2000 D→var
2.7
2.75
1000
2.8

0
0.15 0.20 0.25 0.30 0.35
Φ

62
 3000
δ=0.01J m-2
a=130 nm

σ*(Pa)
2000 D=2.79

1000

(a)
0
0.15 0.20 0.25 0.30 0.35

8000
δ=0.01J m-2
a=810 nm
6000

σ*(Pa)
D=2.67
4000

2000
(b)
0
0.1 0.2 0.3 0.4 0.5

15000
δ=0.01J m-2
a=740 nm

σ*(Pa)
10000 D=2.69

5000

(c)
0
0.1 0.2 0.3 0.4 0.5 0.6
Φ

Methods of Determination

63
 Processing procedure for calculation of
scaling fractal dimensions of fat crystal
network
Original
Polarized light
Original polarized light microscopy image microscopy
image

Adjust contrast of the images

Thresholded
image

Threshold it to binary image

Calculate Db by Benoit 1.3, Df by a particleCounting.m,

and DFT by Photoshop

Fractal Dimension by Microscopy - particle

counting

PLM image Threshold Particle Count

• Df is related to the spatial

distribution of mass
• higher Df Æ
more orderly distribution

64
Mass Fractal Dimension (D) by particle
counting (sensitive to order)

N = cLD
N = number of particles
L = length of ROI
c = constant of proportionality
Log N = D Log R + Log c
Y = m X + b

Narine and Marangoni, 1999, PRE 59:1908

Fractal Dimension Determination by

Particle Counting
6

5
log (N)

3 D=2.01 (0.06)
r2=0.994
2
3.5 4.0 4.5 5.0 5.5 6.0
log (length)

65
 Artifacts!
4
r2>0.99
Dinc=2.05
log N 3
Dexc=2.18
2

1 Dinc =2.17
Dexc =2.49
0
-0.25 0.00 0.25 0.50 0.75 1.00
log L

Mass Fractal Dimension by Box

Counting (sensitive to degree of
occupancy of embedding space)

Benoit 1.3 (Truesoft Int’l Inc.)

66
 Low (1.5) High (1.9)

Effect of crystallization temperature on cocoa

butter microstructure

67
Effect of crystallization temperature on milkfat
microstructure

2.0

1.9

1.8
Dbox

1.7

1.6

1.5

1.4
0 5 10 15 20 25 30
o
Temperature ( C)

Cooling rate effects on milkfat microstructure

1.95

1.90

1.85
D (d=2)

1.80
0 days
1.75
1 day
1.70 7 days
14 days
1.65
0 1 2 3 4 5 6

Cooling rate ( oC/min)

68
 Fractal Dimension by Microscopy:
box counting

• Db is sensitive to the degree of fill in the network:

higher Db Æ higher degree of fill

What is the fractal dimension?

> >

N~>d Np~>D

Φ = NpVp/NV ~ (>/a)D/(>/a)d
Φ ~ (>/a)D/(> /a)d = (> /a)(D-d) > ~ aΦ1/(D-d)

69
Fractal dimension from the characteristic
slope of FFT power spectrum
1
S= -0.811 (A)

Log (Magnitude 2)
0
| slope |
-1 S= -2.14
D = 2+
-2

-3
2
0.0 0.5 1.0 1.5 2.0 2.5
1
(B)
Log (Magnitude 2)

0 S= 0
S= -0.76
-1
S= -2.11
-2

-3
-0.5 0.0 0.5 1.0 1.5 2.0 2.5
1
(C)
Log (Magnitude 2)

0
S= 0
-1
S= -1.16
-2
S= -3.42
-3

-4
0.0 0.5 1.0 1.5 2.0 2.5
Log (Frequency)

Fractal dimension - rheology

μ
⎛ SFC ⎞
G' = λ⎜ ⎟
⎝ 100 ⎠

⎛ SFC ⎞
ln G ' = ln λ + μ ln ⎜ ⎟
⎝ 100 ⎠

70
Weak-link Theory - Fat Crystal Networks
Developed for colloidal gels and adapted to
fat crystal networks:

G’ ~ Φμ ~ λΦ1/(d-D)
G’ = dynamic shear elastic modulus (storage modulus)
λ = constant which depends on particle properties and
particle-particle interactions
Φ = solids’ volume fraction of fat sample

Narine and Marangoni, 1999, PRE 59:1908

Using the Weak-Link Theory to Calculate

Fractal Dimensions Rheologically

1
G ' ~ λΦ d −D

Log G’ = [1/(3-D)] log Φ + logfcλ

71
 10.0 25
A B
7.5 20

G' (MPa)
G' (MPa) 5.0 15

2.5 10

0.0 5
30 35 40 45 50 55 60 40 50 60 70 80
Φ (%) Φ (%)
15 1.5
C D
G' (MPa)

G' (MPa)
10 1.0

5 0.5

0 0.0
20 30 40 50 60 70 20 25 30 35 40 45
Φ (%) Φ (%)

8 6
A A
Storage Modulus

4
6
2
(MPa)

ln G'

4 0

-2
2
-4

0 -6
15 20 25 30 35 40 3.00 3.25 3.50 3.75 4.00 4.25

Solid Fat Content (%) ln φ

12 5.0
B B
Storage Modulus

10 4.5

8 4.0
(MPa)

ln G'

6 3.5

4 3.0

2 2.5

0 2.0
20 25 30 35 40 45 50 55 60 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4
Solid Fat Content (%) ln φ

72
Dilute fat with vegetable oil under conditions where
fat is not dissolved significantly in the oil?

100
AMF
CB
80
PO
SFC (%)

60

40

20

0
0 20 40 60 80 100
Fraction of Fat in Oil (% w/w)

Milkfat

18

S=6.01
16 I=22.3
r 2 =0.88 S=2.20
I=18.1
14 r 2 =0.79
ln G'

12
S=8.40
I=25.9
10 r 2 =0.93

8
-2.0 -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6
ln (SFC/100)

73
 Palm Oil

19

S=1.80
18 I=18.2
r 2 =0.88
17
S=3.36
I=18.3
ln G'

16 r 2 =0.96

15

14 S=3.77
I=21.1
r 2 =0.98
13

12
-2.2 -2.0 -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0
ln (SFC/100)

Cocoa Butter
20

18

16 S=1.56
ln G'

I=I=17.2
r 2 =0.39
14
S=3.98
I=I=20.3
12 r 2 =0.96

10
-2.4 -2.0 -1.6 -1.2 -0.8 -0.4 -0.0
ln (SFC/100)

74
 10
(A)
8

ln (σo)
S=1.97
6 r 2 =0.83
S=4.9
r 2 =0.89
4 S=8.83
r 2 =0.96
2
Strain at the limit of linearity always -2.0 -1.5 -1.0 -0.5
10
Increases with SFC, thus fats are (B)
9
Always in the weak-link rheological
8

ln (σo)
Regime and thus:
7 S=2.054
6 r 2 =0.93
:=1/(3-D)
5

4
-2.5 -2.0 -1.5 -1.0 -0.5 0.0
10
(C)
9

ln (σo)
7

Awad, Rogers, Marangoni, 2004, J. Phys. 6 S=2.96

r 2 =0.84
Chem. B 108, 171 5

4
-2.5 -2.0 -1.5 -1.0 -0.5 0.0
ln (SFC/100)

Effect of D on the Elastic Modulus

0.6
G'/λ (φ =0.5)

0.4

0.2

0.0
1.8 2.0 2.2 2.4 2.6 2.8 3.0
Df

75
 Relationship to Large Deformation
Rheological Behavior?
4000
Yield Force (g force)

3000

2000

1000
r2=0.80
0
4 6 8 10 12
E' (MPa)

Effects of shear on Dr (AMF)

3.0

2.8

2.6
Dr

2.4

2.2 26.5o C
21o C
2.0
0 100 200 300 400
Shear Rate (RPM)

VANHOUTTE, 2002 (Ph.D. thesis, Gent University)

76
 3.0 10
26.5o C
2.8 21o C 5
lnλ=:5.5-0.018⋅SR
2.6

ln λ
Dr

-1 0
D=2.8-29⋅SR
2.4
-5
2.2 21o C
D=3.0-50⋅SR-1 lnλ=:8.1-0.045⋅SR
26.5o C
2.0 -10
0.000 0.005 0.010 0.015 0.020 0.025 0 100 200 300 400
1/Shear Rate (RPM-1) Shear Rate (RPM)

β λ
Dmax − D = ln = −α ( SR )
SR λo

Crystallization behavior and structure

2.0
D=0.023⋅ln(τ-1) -1.81 A
1.9
r2=0.91
1.8
D

1.7

1.6

1.5
-10 -8 -6 -4 -2 0 2 4 6 8 D = D * + β ln J
ln (τ-1)
2.0
B
1.9

1.8
D

1.7 Marangoni and McGauley, 2003,

1.6 D=1.92-0.08⋅n Crystal Growth and Design 3, 95
r2=0.93
1.5
0 1 2 3 4
n

77
 1
6δ 6δ
1
σ =
*
Φ d −D
= Φ d − D*− β ln J

a a

Thermomechanical Method for the

Determination of the Rheology
Fractal Dimension

78
 AMF SFC vs Temperature

60

50

40
SFC (%)

30

20

10

0
0 5 10 15 20 25 30 35 40
Temperature (deg Celcius)

AMF G' vs Temperature

2.00E+07

1.80E+07

1.60E+07

1.40E+07

1.20E+07
G' (Pa)

1.00E+07

8.00E+06

6.00E+06

4.00E+06

2.00E+06

0.00E+00
4 7 10 13 16 19 22 25
Temperature (deg Celcius)

79
 AMF G' vs SFC

2.00E+07

1.50E+07
G' (Pa)

1.00E+07

5.00E+06

0.00E+00
0 10 20 30 40 50 60
SFC (%)

18
Dilution
17 r2=0.938
D=2.70
16
ln G'

15
Thermo
14 r2=0.996
D=2.68
13
-1.4 -1.2 -1.0 -0.8 -0.6
Φ (SFC/100)

80
 CB SFC vs. Temperature vs.

100
90
80

70
60
SFC (%)

50
40
30
20
10

0
0 5 10 15 20 25 30 35 40
Temperature (deg Celcius)

CB G' vs. Temperature

4.50E+07
4.00E+07
3.50E+07
3.00E+07
G' (Pa)

2.50E+07
2.00E+07
1.50E+07
1.00E+07
5.00E+06
0.00E+00
15 17 19 21 23 25 27 29
Temperature (deg Celcius)

81
 CB G' vs. SFC

4.50E+07
4.00E+07
3.50E+07
3.00E+07
G' (Pa)

2.50E+07
2.00E+07
1.50E+07
1.00E+07
5.00E+06
0.00E+00
0 10 20 30 40 50 60 70 80 90
SFC (%)

19
Dilution
18
r2=0.92
17 D=2.43
ln G'

16

15 Thermo
14 r2=0.98
D=2.40
13
-2.0 -1.5 -1.0 -0.5 0.0
Φ (SFC/100)

82
 Dongming Tang’s project
• Determine the physical basis of the scaling fractal
dimensions used to quantify the microstructure of
fat crystal networks in 2D and 3D

• Compare the scaling fractal dimensions of fat

crystal networks in 2D and 3D space

• Use fractal dimensions in 3-dimensional space to

explain the rheological properties of fat crystal
networks

Common methods to determine fractal

dimensions

• Db
• Df
• DFT

N ~ R-Db
N ~ RDf
DFT in 2D = (4+slope)/2

83
 2D simulation images of fat crystal
networks
Sample images generated
by computer simulation.
The shape, crystal size,
area fraction and the
distribution orderliness for
the images are (A) Line, 13
pixels, AF=10%, 100%
randomly distributed; (B)
disk, 29 pixels, AF=10%,
100% evenly distributed;
(C) square, 49 pixels,
AF=15%, 100% randomly
distributed; (D) diamond,
25 pixels, AF=8%, 30%
evenly distributed; (E)
diamond, 25 pixels,
AF=10%, 50% evenly
distributed; (F) diamond,
25 pixels, AF=8%, 70%
evenly distributed.

Tang D. and A.G. Marangoni (2006) Journal of American Oil Chemists Society 83: 309-314.

2D simulation on Db
2.0 1.8
AF=20%
1.8 1.6

1.6 1.4
Db
Db

R=2 AF=10%
1.4 R=6 1.2
R=14
1.2 R=32 1.0
AF=2%
R=42
1.0 0.8
0 5 10 15 20 25 30 35 40 0 10 20 30 40 50
Area fraction of crystals (%) Crystal radius

The effects of area fraction on Db for

The effects of crystal size on Db. Solid symbols are
different crystal sizes. R is the radius of fat
for 100% randomly distributed crystals, and open
crystals
symbols are for 100% evenly distributed crystals.

84
 2D simulation on Df

Radial distribution pattern of simulation images with different Df. (A) Df = 0.998; (B) Df = 3.052.

Tang and Marangoni (2006) Journal of American Oil Chemists Society 83: 309-314.

2D simulation on DFT

1.8 1.68
Area fraction
1.7 of crystals 1.66 Distribution order
of crystals
2%
1.6 1.64 0%
D FT
D FT

6%
30%
1.5 10% 1.62
50%
20%
70%
1.4 30% 1.60
90%

1.3 1.58
1 2 3 4 5 6 7 AF2% AF4% AF6% AF8%AF10%
Radius of crystals Area fraction of particles

85
 2D simulation summary
Affecting
Rheology D Db Df DFT
Factors

Crystal shape ? √ X √
Crystal size ? √ X √
Area fraction
of crystals
X √ X X

Distribution
pattern
? X √ X

Different measures of fractality

cannot agree since they are sensitive
to different structural features within
the network.

Cannot use the argument that all

values of the fractal dimension must
be the same for the system to be
fractal…….

86
 2D fractal dimension vs 3D fractal
dimension

Tang D. and A.G. Marangoni (2006) Chemical Physics Letters 433: 248-252.

Acquiring images of fat samples in 2-

and 3- dimensional space
um
20 um

2-3

2D image taking method

Cover slip
200 um

Glass slide

Fat sample
3D image taking method

87
3-Dimensional Imaging of Bulk Fats
by Wide-field Deconvolution
Polarized Light Microscopy

Novel Approach:
Wide-field Deconvolution
• traditional uses (confocal, fluorescence)
– adapted to TLB (Holmes & O’Conner)
• allows for thicker samples (true volumes)
• 3-dimensional data sets (x, y, z)
• employs a deconvolution algorithm
(removes haze from unfocussed portions of
the specimen)
• 2D projection, 3D renderings

88
 Z-series • image acquisition

bottom

top y
• 50x LWD plano
• N.A. = 0.55 x

Deconvolution Algorithm
• Openlab: Volume Deconvolution
– theoretical point spread function
– uses 8 nearest neighbors (NN)
– removes unfocussed haze from each plane

• Autoquant: Autodeblur
– Estimates the point spread function
– Uses the image data (Blind)
– Removes unfocussed haze from each plane

• sample:
high-melting fraction of milk fat (HMF) diluted in triolein
(70:30), crystallized isothermally at 28ºC for 15 minutes

89
Comparisons: NN

2D Projections
Raw

Blind

NN
Comparisons:
XZ and YZ
Raw

Blind

90
 3D-FD

The main window of 3D-FD after the calculation of particle-counting dimension,

Df. Note that the first data points on the left calculation graph was manually
checked off.

© Marangoni, A.G., Tang, D. and TruSoft Inc., 2006

3D simulation study on Db
2.7 2.7

2.5 2.5

2.3 2.3
Db

Db

r=3 r=10
2.1 r=4 r=14 2.1
AF=2%
r=5 r=18
1.9 1.9 AF=10%
r=6 r=22
AF=20%
1.7 1.7
0 5 10 15 20 25 0 5 10 15 20 25 30
Area fraction of crystals (%) Crystal radius (pixels)

The effects of AF on Db for different crystal The effects of crystal size on Db of fat crystal
sizes. r is the radius of fat crystals networks with evenly distributed crystals. AF is
area fraction of crystals.

91
 2D vs 3D fractal dimension

(a) Db of 2D slices at different depths of a thick sample of the high melting fraction of milk fat in
canola oil at a 7% solids’ mass fraction (b) Comparison between the 2D and 3D box counting
dimensions of samples of high melting fraction of milk fat in canola oil at 20°C at different mass
fractions of solids (Φ).

Tang and Marangoni, 2006, Chemical Physics Letters 433: 248-252.

Microscopy fractal dimensions of

HMF
SFC Low SFC range High SFC range
(<10%) (>10%)
2D Db average 1.34 1.65

3D Db average 2.06 2.37

2D Df average 1.99 1.90

3D Df average 2.76 2.67

92
 Rheology and microscopy fractal
dimensions of HMF

SFC Low SFC range High SFC range

(<10%) (>10%)
3D Db average 2.06 2.37

3D Df average 2.76 2.67

3D Dr average 2.74

Modified fractal model

Heterogeneous Stress Distribution

Principle

93
 Homogeneous factor and dynamic
microstructure

Spatial distribution of strain energy in a perfectly connected DLCA network

Ma, H.S. et al. (2000). Journal of Non-Crystal Solids, 277, 127-141.

Fat crystal network at high SFC

Force

94
 1
− kΦ
G ' = c ∗ (1 − e
b
3− D
)

Tang and Marangoni, 2008, JCIS 318: 202

95
Modified fractal model of colloidal fat
crystal networks

What about intermolecular

interactions?
Van der Waals are the end-all?

96
 Fat Crystal Network Model
Hamaker Constant

1 A 1
G' = λ φ 3- D
= φ 3− D

6 a πγ d 2
0

SFC
Variables Explanation
λ Experimental value, obtained
from a ln G’ vs. ln Φ plot
a diameter of particles
γ Strain at the limit of linearity
d0 separation between the “flocs”

Hamaker constant
Material:
Dispersed phase: Fully hydrogenated canola oil

Continuous phase: High oleic sunflower oil

• McLachlan and Lifshitz approach

• Semi-classical approach

• Fat Crystal Network Model

97
 McLachlan and Lifshitz
approach
A = Av =0 + Av >0
3 ⎛ ε − ε ⎞ 3hν e n12 − n22
= kT ⎜⎜ 1 2 ⎟⎟ +
2
( ) 2

4 ⎝ ε 2 + ε 2 ⎠ 16 2 n12 + n22 ( ) 3
2

J.Israelachvili –Intermolecular & Surface Forces, 2002 page 184

k Boltzmann’s constant 1.38 10-23 J/K

h Planck’s constant 6.626 x 10-34 J s
T Temperature (K )
ε dielectric permittivity
n index of refraction
νe electronic absorption frequency in the UV region 3 x 1015 s-1

Index 1 and 2 are for dispersed and continuous phases

Permanent dipole-permanent dipole

interaction are very important in fat crystal
interactions!

Av=o % Av=o Av>o % Av>o Total A

Svaikauskas 3.5 10-22 J 67 % 1.70 10-22J 33% 5.2x10-22 J

(30°C)
Svaikauskas 9.1 10-22 J 82% 1.77 10-22 J 18% 1.1x10-21 J
(20°C)
Johannson 3.8 10-23 J 12% 1.77 10-22 J 88% 2.0x10-22 J

(JAOC, Vol.69,
8 1992)
Kloek 3.8 10-23 J 8.5% 1.83 10-21 J 91.5% 2.0x10-21 J
(Ph.D Thesis,
1998)

98
 Hamaker from nSSS nOOO εSSS εOOO A

5.2x10-22 J
Svaikauskas 1.503867 1.469903 1.68* 2.7 (30°C)
(tristearin in sunflower
oil) 1.51299 1.466420 1.926 3.17 1.1x10-21 J
(20°C)

Johannson (JAOC,
Vol.69,no8,1992) 1.44711 1.47351 23 2.53 2x10-22 J
(tristearin in soybean
oil)
Kloek (Ph.D 1.562 1.47351 23 2.53 2x10-21 J
Thesis,1998) (tristearin
in soybean oil)

1- Bailey’s Industrial Oil and fats Products (measured at 60°C)

* Estimated from 20 °C data.
2- Meeussen, Personal communication
3 Handbook of Chemistry and Physics, R.C East, 56th edn.

Semiclassical approach
Hiemenz and Rajagopalan, Principles of Colloid and Surface  Chemistry, 1997

A = 24 π δ d 02
δ crystal-melt interfacial tension (30°C)
d 0 equilibrium space between molecules

d0 (nm) δ (J/m2 ) A (J)

0.2 0.01 * 3.0x10-20
0.2 0.005** 1.5x10-20

* L.W.Phipps, Trans. Faraday Soc. 60, 1873 (1964).

* * Latifeh Ahmadi’s work

99
 Fat Crystal Network Model
A
A = λ 6 aπγ d02
1 1
G' = λ φ 3-D = φ 3− D

6 a πγ d 0 2

Variables Explanation Experimental values

λ Experimental value, obtained from the 1.17 107 Pa
ln G’ vs. ln Φ graph
a diameter of particles 2µm,1 µm, 0.4µm,0.2 µm
γ Strain at the limit of linearity 1.9 10-4
d0 separation between the “flocs” 0.2 nm

A= 3.35 10-21 J when a = 2 μm

A= 1.68 10-21 J when a = 1 μm
A= 6.7 10-22 J when a = 0.4 μm
A= 3.35 10-22 J when a = 0.2 μm

Summary of Hamaker values

100
Thank You!

101