CHEMISTRY/BIOCHEMISTRY

Dynamic Crystallization of
Dark Chocolate as Affected by
Temperature and Lipid Additives
C. LOISEL, G. LECQ, G. KELLER and M. OLLIVON

ABSTRACT
The dynamic crystallization was studied in a lab-scale
Scraped Surface Heat Exchanger by following variations with
time of torque applied during different isotherms between
24.9C and 31.0C. Crystallization was in two steps for T 26.2C
but it was in one step at T<26.2C.The first step corresponded
to crystallization of about 1% solid fat and was related to that
of saturated triacylglycerols (SSS) which segregated from
other cocoa butter triacylglycerols (TAGs) because of their
low solubility in TAGs. The second step, an abrupt increase of
apparent viscosity, leading to complete crystallization was
attributed mainly to the monounsaturated TAGs in the
form.
Different amounts of tristearin increased the apparent viscosity and reduced the latent time preceding the first step,
but did not influence the main crystallization. Stearic acid and
distearin additions also influenced chocolate crystallization.
Key Words: chocolate, crystallization, lipid additives,
scraped surface, heat exchanger

INTRODUCTION
COCOA BUTTER CONSTITUTES, WITH SUGAR, ONE OF THE MAIN
components of chocolate. Its polymorphism, which greatly affects the
physical aspect and properties of chocolate products (gloss, snap,
contraction and chocolate blooming during storage) through the tempering and crystallization process, has been extensively studied. Six
different polymorphic forms, (I to VI), with increasing order of melting points, are generally reported to describe the cocoa butter polymorphism (Wille and Lutton, 1966; Chapman et al, 1971; Huyghebaert and Hendrickx, 1971; Lovegren et al., 1976; Davis and Dimick,
1986; Adenier et al., 1975, 1993). However, the existence of some of
these (forms III and VI) has been debated (Merken and Vaeck, 1980;
Schlichter-Aronhime et al., 1988; Schlichter-Aronhime and Garti,
1988).
Many earlier studies assumed that cocoa butter crystallizes like a
pure compound but others (Manning, 1984; Davis and Dimick, 1989a,
b; van Malssen, 1994; Dimick, 1994, Chaiseri and Dimick, 1995a, b;
Loisel et al., 1998) have shown that cocoa butter crystallization induces segregation of different lipid families. In a first step, the nucleation
process has been postulated to start from complex lipids and trisaturated TAGs (SSS, Saturated-Saturated-Saturated), other TAGs and
simple lipids being excluded from seed crystals. Davis and Dimick,
(1989a, b) reported that cocoa butter seed crystals formed during early
crystallization contained high concentrations of glycolipids (11.1%),
phospholipids (6.6%) and triacylglycerols (67.7%). Cocoa butters
that contained higher proportions of tristearin seemed to crystallize
faster. Chaiseri and Dimick (1995a, b), studied cocoa butters of different geographical origins and found that at the early stages of crystalli-

zation under agitation at 26.5C, cocoa butters solidified into high

melting and low-melting fractions. The low melting fraction was composed of polymorphs IV and V of cocoa butter. The high melting
fraction at the latter stages of crystallization had DSC endotherms at
about 3436C (form VI). The concentration of StOSt (1,3-stearoyl2-oleoylglycerol) in the crystals during growth was higher than that in
the original cocoa butter. As crystallization progressed, the proportion
of StOSt in the TAGs fraction increased. Cocoa butters characterized
by a higher rate of nucleation and crystal growth showed higher stearic acid in their diacylglycerol (DAG) fractions and higher StOSt and
POSt (1-palmitoyl-2-oleoyl-3-stearoylglycerol) compared to others
(characterized by slower nucleation and growth rates) which showed
a higher content of polyunsaturated TAGs (StOO, 1,stearoyl-2,3oleoylglycerol and POO, 1,palmitoyl-2,3-oleoylglycerol) (Chaiseri
and Dimick, 1995a, b). No relation was found between saturated
TAGs content of the initial crystals during nucleation and the crystallization behavior of cocoa butters. This confirmed observations of
Cebula et al. (1991) that trisaturated TAGs had no effect on chocolate
temperability. However, Hachiya et al. (1989) had reported that the
seeding of a dark chocolate by the b form of tristearin accelerated
cocoa butter crystallization.
In our previous studies (Loisel et al., 1998; Loisel, 1996), monitoring of cocoa butter crystallization by thermal analysis and X-ray
diffraction vs. temperature (XRDT) showed that on cooling to different temperatures (0.5 to 5C/min.), segregation occurred among the
TAGs with the formation of high-melting crystals characterized by an
increase of SSS content (11% compared to 3% in cocoa butter). XRDT
analysis showed that the high-melting crystals, which were formed
first during cooling, melted last. They show a 2L (double chain length)
long spacing at about 44, very close to that of the b form of tristearin,
which disappears only at about 37.5C, i.e. 23C above the melting
point of the VI form of cocoa butter (3L, triple chain length).
In order to confirm the fractionated crystallization of cocoa butter
in chocolate, a dynamic crystallization test was developed in a labscale Scraped Surface Heat Exchanger (SSHE) (Loisel et al., 1997).
Chocolate crystallization could be carried out as a function of temperature and time in this SSHE and the nature and quantity of seed
crystals be optimized by different temperature cycles. A single crystallization was observed after a slow cooling of chocolate from about
40C to 26.1C. No temperature > 26.1C was directly applied to the
melted chocolate during that study. Our present objective was to study
chocolate crystallization during different defined isotherms (Tc) between 24.9C and 31.0C, reached after direct cooling from 40C to
Tc. The observation of two stages during cocoa butter crystallization
and the origin of each were focused on by addition of some minor
lipid additives such as saturated TAGs, saturated diacylglycerols and
free fatty acids in the chocolate composition. The influence of emulsifiers on chocolate crystallization was also tested.

MATERIALS & METHODS

Author Lecq is with the Centre Jean Thves, groupe DANONE, Branche Biscuits, 6
rue Edouard Vaillant, BP 16, 91207 Athis-Mons Cedex, France. Authors Keller and
Ollivon are with the Laboratoire de Physico-Chimie des Systmes Polyphass, CNRS
URA 1218, Universit Paris-Sud, 92296 Chtenay-Malabry, France. Author Loisel is
affiliated with both the Centre Jean Thves and Univ. Paris-Sud. Direct inquiries to
Dr. C. Loisel at Centre Jean Thves.

Materials

The chocolate was a mixture of a low-fat dark chocolate (30.0% fat

with 0.4% lecithin), and pure cocoa butter (CB) of the same origin in
order to provide a 32.4% final fat content. Each tempering process
was conducted using 828g of enriched chocolate (800g of initial choc-

Volume 63, No. 1, 1998JOURNAL OF FOOD SCIENCE

73

Dynamic Crystallization of Dark Chocolate . . .

Table 1Percent concentration of triacylglycerols in cocoa buttera
TAGs

OLO

PLO

PLP

OOO

POO

PLSt

StOO

StLSt

POP

POSt

StOSt

StOAr

StPP

PstSt

StStSt

Total(%)

Cocoa
butter

0.4

0.5

1.6

0.4

2.8

1.4

3.4

17

36.6

27.3

1.9

1.4

1.1

0.5

99.3

TAGs
fractions

polyunsaturated = 13.6

monounsaturated = 83.4

trisaturated =3.0

100

a Ar: arachidic acid (C 20:0), L: linoleic acid (C 18:2), O: oleic acid (C 18:1), P: palmitic acid (C 16:0), St: stearic acid (C 18:0).

olate + 28g of CB). The TAGs composition of cocoa butter had been
determined by HPLC analysis (Loisel et al., 1998; Loisel, 1996) (Table 1). The composition was calculated assuming that i) unsaturated
fatty acids are distributed preferentially in the sn2 position on the
glycerol, ii) linoleic and oleic acids are distributed in this position of
TAGs having the same HPLC partition number in the proportion
given by the fatty acid analysis reported earlier (Loisel et al., 1997).
The composition in minor lipids has been determined by GC and
reported (Loisel et al., 1997; Loisel el al., 1998). We found
TAGs = 97%, diacylglycerols = 1.1%, monoacylglycerols = 0.2%,
free fatty acids = 1.6%; phosphatides = 0.15% and others 0.25%.
The influence of minor components on chocolate crystallization
was studied by replacing the original cocoa butter by the same amount
of tristearin (Sigma, Grade II, purity >90%), distearin or stearic acid
(Sigma, Grade I, purity >99%). Weights added were 0.05g for lowfat chocolate and cocoa butter and 0.1 mg for tristearin, distearin and
stearic acid. Lecithins (Chocotop 50, 100 and 320 and Lecimulthine
from Lucas-Meyer) and PRPG (Admul Wol 1403, Quest) were added
to increase the emulsifier content from 0.4% to 0.75% (chocolate fat =
32.4%).
Lab-scale scraped surface heat exchanger

Chocolate crystallization was followed in a lab-scale Scraped Surface Heat Exchanger (SSHE) in which torque variation was monitored as a function of chocolate temperature and time (Loisel et al..,
1997; Loisel, 1996). The rotation speed of the SSHE shaft was fixed
at 20 rpm during all experiments. The rapid decrease of chocolate
temperature from 40C to the desired temperature (Tc) in the range
24.931.0 0.1C was achieved in about 5 min by switching between
two temperature-controlled refrigerated baths (the first bath at 40C,
the second at Tc) which had not been possible during our previous
studies. The resisting torque was related to an equivalent viscosity
(Pas) determined from one standard silicone oil. As the torque-viscosity correlation was found to depend on oil type, we preferred the
use of torque units (Nm). The standardization of the SSHE was
carried out using a torquemeter (0.1-5Nm 10%). The following
linear relationship was found between the SSHE signal, expressed in
arbitrary units (a.u.), and the torque (Nm) (r2 = 0.987)

blade friction on chocolate crystallization was confirmed by using the

SSHE shaft with no blade. A delay of the chocolate crystallization of
about 30 min. was observed. For this reason, the initial torque value
was imposed at 0.9 0.02 Nm at 40C at the beginning of each
crystallization test in order to standardize the friction of the blades in
the SSHE. In addition, viscosity and yield value of chocolates were
determined at 40C at the end of each experiment to verify the absence
of degradation. A Rheomat 115 (r1 = 13.6 mm, r2 = 20 mm) was
used and Casson-Steiner viscosity and yield value were calculated
from the data between 220 s-1 and 1 s-1 on the decreasing curve
(Casson-Steiner Model). The influence of lipid additives on viscosity
and yield value of chocolate was also determined. The conditions of
the crystallization tests with additives were similar except that Tc was
fixed to 28.2C as explained below.
Thermal analysis

Thermal analysis of chocolate was carried out using a PerkinElmer DSC-7. Samples were taken from the SSHE at several stages of
crystallization. They were rapidly put into an aluminium pan (40 L)
stored on a metallic block heated at the same temperature as the chocolate (24.9<Tc<31C). The pan with about 30 mg of chocolate was
rapidly placed in the DSC-7 at 21C and heated at 10C/min to 50C.
The chocolate mass was accurately determined by the difference between the empty pan weighed before, and the pan with chocolate
weighed after thermal analysis. This protocol, by minimizing the cooling of the chocolate before thermal analysis which inevitably induced
formation of unstable crystals and made the thermal profile more
complex (Manning, 1984), allowed the melting of seed crystals present
in the tempered product to be recorded separately from unstable forms.
The apparatus was calibrated with lauric acid (purity > 99.9%) at
10C/min (Grabielle-Madelmont, 1983) (Loisel, 1996). The peak temperatures were taken at the maxima of the endotherms (Tm , indicated
by a mark on DSC recordings) instead of the onsets, because of peak

SSHE signal (a.u.) = 24.7 x Torque (Nm) - 15.2

Torque (Nm) = 0.04 x SSHE signal (a.u.) + 0.62
On recordings, the onset of chocolate crystallization was determined graphically as the point (time) at which the tangent to the crystallization curve intercepted the base line (Fig. 1).
Before each crystallization run, chocolate was heated at a temperature between 45C and 77C (according to the melting temperature of
the additives) for about 2h in order to eliminate any influence of the
samples thermal history. Each experiment was performed in duplicate. With the minor component-containing chocolate, a new sample
was prepared for each experiment, we had observed that long exposure to high temperature with blade rotation (e.g., 77C during 90
min) led to a delay in nucleation of about 20 min, and thus poor
repeatability. This delay could be explained by the practice of adjusting the SSHE blades to provide a constant torque value at the beginning of each run. After long periods of stirring at high temperature, the
viscosity of the chocolate increased, probably because of degradation
of the lecithin, and the blades had to be adjusted. The effect of the
74JOURNAL OF FOOD SCIENCEVolume 63, No. 1, 1998

Fig. 1Change in resisting torque of the SSHE as related to time at

temperature (Tc). The first torque increase recorded during the very
first minutes, until temperature equilibration, corresponds to the
cooling of chocolate from 40C to T c. The second and third torque
increases resulted from chocolate crystallization. The induction times
t1 and t2 corresponding to onsets of crystallization (Tc27.2C) and
amplitude of the torque jump, h, recorded for the first crystallization were determined graphically from tangents to the curve. T0
corresponds to the onset of crystallization for Tc<27.2C. (a) and (b)
arrows show sampling points referred to in Fig.3.

overlapping. Therefore, melting points observed for the different polymorphic forms at peak maximum were systematically higher than the
published values or those measured at onset. This lag, which was 1 or
2 Kelvin for pure TAGs, reached 4 to 6 K for mixtures such as cocoa
butter (e.g. a 10 mg sample of polymorphic form V heated at 5K/min.)
displayed onset temperature of 29.1 and maximum of 34.8C.

RESULTS & DISCUSSION

Standard dark chocolate crystallization

Standard dark chocolate was melted at 45C and rapidly cooled to

24.9, 25.4, 26.3, 27.2, 28.2, 29.1, 30.0 or 31.0C (0.1C). The
torque had to be adjusted (Fig.1) to maintain constant rotation speed
(20 rpm) of the SSHE shaft at these temperatures. The torque increase
recorded at the beginning of each experiment corresponded to the
change in chocolate viscosity observed during cooling from 45C to
the desired temperature (Tc). A time-delay (about 10 min) was necessary to reach this temperature and reach thermal equilibrium as well as
constant torque. Although this delay depended on Tc, it was noted that
after 5 min., the temperature of the chocolate mass was very close to
the desired value (about Tc + 0.3C). Since nucleation processes,
whatever the polymorphic form considered, were expected to start as
soon as chocolate temperature was below its melting point, the time of
chocolate crystallization reported here was taken from that point. Practically, it has been calculated as the total time from the beginning of the
cooling (40C) to the relevant crystallization onset minus 5 min.
After a constant torque value was reached (Fig. 1), one or two
domains of torque variation with time were observed, depending on
chocolate temperature. These variations have been interpreted as the
development of chocolate crystallization. A single sharp crystallization was observed for Tc=24.9, 25.4 and 26.3C, while a two-step
crystallization clearly appeared for Tc  27.2C. In most cases, before
the onset of crystallization, a first torque plateau was observed, the
duration of which was temperature-dependent.
For Tc 27.2C, the sharp crystallization recorded at lower temperature was no longer observed and was replaced by a more gradual
process. The main torque increase observed at the end of each experiment was preceded by a limited torque jump leading to a second
plateau for Tc  27.2C. The mean value of the torque measured
during this second plateau was between 1.3 and 1.6 Nm., and tended
to decrease when the chocolate temperature increased from 27.2 to
31.0C. This evolution may have been due to a decrease of chocolate
viscosity which is expected when chocolate is heated (Loisel et al.,
1997). The dependence of the chocolate viscosity on temperature,
determined in our previous study, allowed us to evaluate the torque
decrease (about 0.1Nm) expected for a temperature increase of 2.8C
(from 28.2 to 31.0C). This calculated mean torque variation was
close to the decrease observed (0.15Nm) over this range of temperature. However, this torque decrease was not entirely related to chocolate viscosity changes.
A first interpretation of this plateau could be that a polymorphic
transition, such as a

transition, occurred during cocoa butter
crystallization when Tc  27.2C. However, the temperature at which
the initial (less stable) crystals were formed (Tc  28C) would rule
out the hypothesis of the formation of
 crystals, which were characterized by a melting point between 25.1 and 27.5C (Wille and Lutton,
1966; Huyghebaert and Hendrickx, 1971).
The appearance of a second plateau might indicate that no further
crystallization occurred and that an equilibrium stage was attained, for
instance when crystal growth was exactly compensated by their solubilization in the liquid phase. Another hypothesis is that a fractionated
crystallization, corresponding to the first torque jump, may occur in
the liquid phase when Tc  27.2C. The fact that a plateau was attained
at all temperatures Tc  27.2C indicates that a defined amount of
crystals was formed at each temperature. The decrease of the torque
jump as the temperature increased shows that the amount decreased
with temperature. This variation could be attributed to increasing sol-

ubility of the crystals in the liquid as the temperature was raised. The
weak amplitude of this jump, compared to the torque value recorded at
the end of each experiment when the shaft blocked (2.2Nm, Fig. 1),
tended to indicate that the crystals formed were made up of a minor
lipid fraction of cocoa butter.
For Tc < 27.2C, the single sharp crystallization observed was
similar to those reported for a slower cooling to 26.1C at 0.5C/min
(Loisel et al., 1997). The two-step crystallization and the two plateaus
had not been previously observed. The reason for this was that the
chocolate was always directly cooled to 26.1C, which was below the
temperature (Ti = 26.2C), corresponding to the linear curve intercept
(Fig. 2) delimiting the two regimes of crystallization, i.e. a single- or
double-step. In our previous study, the fast crystallization observed at
26.1C was attributed to the growth of an unstable form of cocoa
butter (form IV). On heating above that temperature, a polymorphic


transition was observed, accompanied by rapid crystal growth
at Tc>Ti. The rate of this growth was only measured between 30.5 and
about 33.0C, since below that temperature it was too fast to be determined. This fast growth was not observed when chocolate was directly cooled to Tc>Ti , without passing below Ti (Fig.1). This confirmed
that starting in both cases from liquid chocolate, stable crystals were
obtained more quickly at a certain temperature after melting of unstable varieties than by direct cooling to that same temperature. Note that
this process is commonly used in the fat industry for crystallization of
final products by SSHE.
The induction times, t0, t1 and t2, at which onsets of crystallization
were observed, were determined as shown (Fig.1). The logarithm of
the induction time (log(t)) for each of the different steps was a function of the chocolate temperature (Tc) (Fig. 2). The three different
linear relations were observed, corresponding to different crystallization mechanisms, which confirmed the formation of three types of
crystals. Below Ti (26.2C) only one type of crystals appeared whereas two types formed above the critical limit. The critical temperature
limit was probably very close to this value since slight rise was seen
on the 26.3C recording (Fig. 1) when it was observed on an enlarged
time scale (not visible on graph at scale shown).
In order to confirm this hypothesis, thermal analyzes were carried
out with chocolate samples taken at different times during the crystallization process. The first series concerned samples taken at the beginning of each single crystallization at torque values between 1.3 and 1.6
Nm (at 24.9, 25.4 and 26.3C). The second series corresponded to
samples taken during the first and second steps of crystallization (at
27.2, 28.2, 29.1, 30.0 or 31.0C). Samples corresponding to the first

Fig 2Change in the logarithm of induction times vs temperatures

of chocolate crystallization (Tc). single crystallization which appeared in recordings for Tc 27.2C and first step of the crystallization curves for recordings at T c  27.2C; the second step of
crystallization. Lines are the corresponding linear fits.

Volume 63, No. 1, 1998JOURNAL OF FOOD SCIENCE75

Dynamic Crystallization of Dark Chocolate . . .

and second step were taken at torque values 1.3-1.6Nm (as for previous samples taken below 27.2C) and 2 Nm (arrows, Fig. 1). The
DSC profiles (Fig. 3) show the melting recordings of samples of
tempered chocolate taken at 31.0C during the first (a) and second (b)
step of crystallization. Crystals appearing during the first step were
characterized by a higher melting point (36.6C) than those during the
second (34.6C). All DSC profiles at Tc>Ti were similar to those
shown (Fig. 3) and thus confirmed that the crystals which were formed
first, melted last (Fig. 4).
The melting point of the crystals appearing during the first step for
Tc  27.2C increased regularly from 34.0 to 36.6C as a function of
Tc. Those of the two forms which appeared during the second step
were systematically lower and in the ranges 30.532.3C and 34.0
34.6C, respectively. The occurrence of a series of peaks, the maximum melting point of which was about 31C, indicated the presence
of
 crystals, probably formed during the cooling of the sample in the
DSC apparatus (21C) (Fig. 3, 4). The melting of these crystals was
not considered further, since they were not representative of the solid

Fig. 3 Melting DSC recordings of the chocolate samples taken

during the first (a) and the second (b) steps of the crystallization
process (a and b also refers to Fig.1 arrows) for Tc = 31.0C (sample
weight was about 30 mg and heating rate 10C/min.)

Fig. 4Melting points observed by DSC of seeds taken during the

first and the second steps of the crystallization process at points
corresponding to a and b of Fig. 1 as a function of temperature at
which chocolate has been crystallized (Tc). crystals which appeared
during the first step of the crystallization curves for Tc  27.2C. ,
crystals which appeared during the second step of these same curves
and V during the single crystallization which occurred at Tc<27.2C.
The dotted line is a linear fit of the melting temperatures found for
the high-melting fraction.

76JOURNAL OF FOOD SCIENCEVolume 63, No. 1, 1998

phase sampled at Tc. In contrast, the presence of two types of crystals,

the melting temperatures of which ranged from 34.0C to 34.6C and
from 34.0 to 36.6C, indicates the probable existence of two
varieties characterized by different TAG compositions.
An estimation of the amount of crystalline phase formed during the
first torque jump was calculated from the DSC recordings during the
second plateau using the following formula:
% (high melting crystals/cocoa butter) = 100 H / (H0 32.4)
where: H = experimental melting enthalpy of crystals (J/g); H0 =
130J/g = mean value of the melting enthalpy of the
forms of cocoa
butter determined by DSC from two samples crystallized as forms V
and VI, the structures of which have been confirmed by X-ray diffraction in our laboratory. 32.4 = total fat content of chocolate (%).
DSC recordings of the chocolate samples taken during the plateaus
(Fig.1) at 27.2, 28.2, 29.1, 30.0 or 31.0C indicated that only 0.90.3%
of the cocoa butter was crystallized. Thus, these thermal analyses
confirmed the fractionated crystallization of a minor lipid fraction of
cocoa butter in the SSHE for Ti  26.2C.
This fractionation during chocolate crystallization could be attributed to the phase separation of SSS TAGs in cocoa butter as reported
by Davis and Dimick (1989a, b) and Chaiseri and Dimick (1995a, b).
They have shown that during cocoa butter crystallization, high melting crystals, with a high content of complex lipids and saturated TAGs,
may separate from the liquid. This fractionation of the cocoa butter
was confirmed by use of X-ray diffraction as a function of temperature and DSC analysis, and may be explained by the low solubility of
SSS in mono and polyunsaturated TAGs (Loisel et al., Submitted).
The TAG analysis of the high melting fraction has shown that it was
enriched in SSS (by four times). In our current results, the highmelting crystal content estimated by DSC (0.90.3%) represented
only about one third of the SSS content of the cocoa butter (3.0%,
Table 1), confirming the partial solubility of saturated TAGs in the
chocolate liquid phase.
Taking into account the results of these studies and the fact that this
value of about 1% was very close to the content of diacylglycerols
(DAGs) (1.1%) and free fatty acids (FFAs) (1.6%) in cocoa butter
(Table 2), we assumed that this minor fraction could be mainly composed of such high-melting lipids and we considered their influence
on chocolate crystallization.
Influence of lipid additives on chocolate crystallization

Crystallization test determination. Crystallization test were carried out to study the influence of the lipid additives tristearin (StStSt),
distearin and stearic acid. The chocolate temperature was fixed at
28.2C so as to observe the fractionated crystallization of chocolate
in 3 h to perform three different experiments within one day. Another reason for using this temperature was that cocoa butter cannot
crystallize under its unstable forms (
 ) at 28.2C. This allowed us to
quantify the effects of lipid additives on both steps of chocolate crystallization (induction time as well as torque). The latter step corresponded to the formation of a
form of the monounsaturated TAGs of
cocoa butter.
Before starting crystallization tests, we verified that TAG, DAG
and FFA incorporation into chocolate did not sharply modify the
rheological properties of chocolate, and thus prevent the standardization of blade friction in the SSHE. The effect of rheological properties
on the blade adjustment was quantified by recording, at 28.2C, the
dynamic crystallization of three chocolates containing 35%; 32.4%
and 30.0% fat. The Casson-Steiner viscosities and the yield values of
the chocolates were 1.650.02 Pas and 6.18 1 Pa (35%), 2.600.02
Pas and 12.4 1 Pa (32.4%), 4.520.02 Pas and 17.251 Pa (30.0%)
respectively. No significant difference was found between the induction times of the different chocolates, regardless of crystallization
step. However, we observed that a lower fat content tended to increase
the amplitude of the first torque jump. The torque variation decreased

Table 2Lipid composition of chocolates

Lipid additives
in chocolate

Chocolate with
no lipid additive
(800/28/0)

Cocoa
Butter
(g)

Lipid additive Lipid additive

amount
(%/chocolate fat)
(g)

% of lipid
fraction in
final
chocolateb

28

0.0

0.00

SSS: 3.0
DAGs: 1.1
FFAs: 1.6

Tristearin
(800/23/5)

23

5.0

1.7

SSS: 4.6

Tristearin
(800/25/3)

25

3.0

1.0

SSS: 4.0

Tristearin
(800/27/1)

27

1.0

0.3

Table 3Main values of the crystallization curves reported in Fig. 5

and 8
First crystallization

Second crystallization

Experiments

t1
(min)

h
(N.m)

Slope
(N.m/hr)

t2
(Min)

Dark chocolate

48.57.2

0.350.09

1.40.2

138.83.2

SSS TAGs
3.3%
4.0%
4.6%

221
6.50.5
2.50.5

0.430.05
0.730.05
0.930.05

2.620.04
4.8 0.4
8.60.3

136
138 1
146

DAGs
1.4%

552

0.330.05

1.230.02

1893

FFAs
1.8%
2.1%

552
552

0.330.05
0.330.05

1.230.02
1.230.02

1743
1891

SSS: 3.3

Distearin
(800/27.25/0.75)

27.25

0.75

0.3

DAGs: 1.4

Stearic acid
(800/26.5/1.5)

26.5

1.50

0.6

FFAs: 2.1

Stearic acid
(800/27.250.75)

27.25

0.75

0.3

FFAs: 1.8

aThe abbreviation 800/X/Y corresponds to the weight ratio of low fat chocolate/cocoa butter/
lipid additive.
b SSS = trisaturated triacylglycerols; DAGs = diacylglycerols; FFAs = free fatty acids. In this
column, except for the first line, only changed compositions have been reported.

from h = 0.520.05Nm for chocolate with fat content 30% to h =

0.180.02Nm for chocolate with fat content 35%. The slope of the
first crystallization jump (torque variation versus time) decreased similarly from 1.640.05Nm/hr (30%) to 1.230.02Nm/hr (35%). The
higher torque jump observed for chocolate containing only 30% of
cocoa butter could be explained by the appearance of high-melting
crystals during the chocolate crystallization with more effect on viscosity in a low-fat, than in a higher-fat chocolate since the mean distance between solid particles would be larger in the higher fat sample
(in which the solid-solid interactions are decreased). Then, it appeared
necessary to keep a constant value for the chocolate fat content for
each crystallization test, in order not to change its viscosity and to be
able to compare the specific effects of the lipid additions. For this
purpose, a defined quantity of cocoa butter was systematically substituted by the lipid fractions to keep a total fat content of 32.4% (Table 2).
The viscosity and yield value of the different additive-enriched
chocolates were checked at the end of each experiment to verify that
no deterioration had occurred. For all crystallization tests, the measured values of chocolate viscosities and yield values remained between 2.61 and 2.77 Pas and 12.1 and 15.1 Pa, respectively. These
values were very close to those of the dark chocolate (2.60 0.02 Pas
and 12.4 1 Pa). Therefore, no significant effects of rheological parameters on crystallization changes were expected.
The repeatability of the crystallization test was reported (Table 3)
as standard deviations of the parameters measured for three different
samples corresponding to the same chocolate without additives (32.4%
fat) (pure dark chocolate). All results with different lipid additives
were compared to this control.
Influence of tristearin addition

The effects of the addition of 0.3, 1.0 and 1.6% tristearin (StStSt)
on chocolate crystallization at 28.2C were compared (Fig. 5). The
main parameters determined from these crystallization curves (Table
3) showed the addition of tristearin reduced t1 and increased h. The
crystallization process was very sensitive to this effect and even a
small increase of saturated TAGs from 3.0 to 3.3% sharply accelerated the onset of fractionated crystallization (t1 was reduced by half).
This confirmed that the first jump observed (Fig.1) was related to the
fractionated crystallization of SSS. Changes in the induction times, t1

and t2 were also compared (Fig. 6) as a function of total SSS concentration (noted [SSS]) in the chocolate fat (Table 2). A small increase of
[SSS] decreased t1 without affecting t2 significantly. Two asymptotic
curves were observed for the plot of t1 vs [SSS]. At high [SSS], t1
was so short that it indicated a spontaneous crystallization of the
minor lipid fraction. At low [SSS], t1 was so high that we could
assume [SSS] asymptotically tended towards its limit of solubility in
the undercooled liquid phase. This solubility limit in chocolate (about
2.1%) could be evaluated as: [SSS fat content] (3.0%) - [high-melting
fraction](0.9%) at 28.2C.
The solubility of SSS in the liquid chocolate at 28.2C was determined by plotting the torque jump amplitude (h) vs the amount of
StStSt added (Fig. 7). The torque jump was linearly dependent on the
amount of tristearin crystallized in the domain in which the additions
were made. Assuming that this linearity extended to the null concentration of SSS, the extrapolation to null torque jump (h = 0) yields a
negative value (about 1.0%) which corresponds to the amount of
crystallized SSS (such a plot is commonly used for evaluation of
concentrations of minor components in a complex mixture by analytical techniques).
The amount of crystallized SSS in the dark chocolate at 28.2C
represents about 1% of the total fat content of chocolate or about one
third of the trisaturated TAGs. This is very close to the quantity of
solid phase which had been determined by thermal analysis (0.90.3%).
Therefore the amount of solubilized SSS was about 2.0%. This is
comparable to the concentration found by HPLC analysis (2.3%) for
a sample in which the phase crystallized at 300.5C had been removed prior to analysis (Loisel et al., Submitted). Taking into account
the slight increase in solubility expected at the higher temperature,
there was good agreement between all these determinations. These
values also confirmed results of Cebula and Smith (1992), who observed by thermal analysis that addition of trisaturated TAGs to
Coberine (Loders Croklaans, Netherland) (a cocoa butter substitute with close TAGs composition) accelerated the crystallization and
also raised the temperature of nucleation of the fat.
The fractionated crystallization we observed for chocolate, as well
as that previously found in pure cocoa butter, may be partly explained
by the very low solubility of the trisaturated TAGs in the liquid phase,
which is mainly composed of monounsaturated TAGs. This result is
similar to that for fractionation of pure cocoa butter (Loisel et al.,
Submitted). This fractionation, which has been studied by different
techniques, and its dependence on temperature, can be explained in the
same way. The coexistence and phase separation of the two fractions
is similar to the behavior of TAGs representative of the two fractions,
such as StStSt and StOSt for the SSS and SUS respectively. The
StStSt/StOSt phase diagram, (reported by Rossell, 1967), illustrated
this behavior. Note that the phase diagram PPP/POP is similar to that

Volume 63, No. 1, 1998JOURNAL OF FOOD SCIENCE77

Dynamic Crystallization of Dark Chocolate . . .

of SSS/StOSt except for a temperature shift (Rossell, 1967; Ollivon,
1992). Then, we assumed that the behavior of mixtures of the two
series of TAGs would not be very different from that of the pure
compounds. These phase diagrams have in common that the limit of
solubility in the solid state of SSS in SUS is extremely low, while that
of SUS in SSS is quite high. The StStSt/StOSt phase diagram indicated that for [StOSt]  80% and T  45C, the StStSt composition of
the solid phase (and thus, its melting temperature) in equilibrium with
the liquid phase, mainly composed of StOSt, was highly temperaturedependent. A small increase of temperature at which crystals are obtained would result in a large change in solid phase composition. The
amount of solid phase formed is increased and the initial and maximum melting points are shifted towards higher temperatures when Tc
is decreased (Fig. 4).
In addition, our previous results showed that the high-melting
crystals from the liquid phase after storage for 3 wk at 300.5C were
composed of about 11% of SSS (compared to the initial SSS content
of cocoa butter, 3.0%). The enrichment of the crystals in SSS confirmed the probable homology of the SSS/SUS phase diagram of
cocoa butter (unknown) with StStSt/StOSt (Rossell, 1967).

Also, in the range studied, whatever their concentration, trisaturated TAG crystallization had no effect (Fig. 5) on the crystallization of
the monounsaturated TAGs of cocoa butter which occurred later, since
t2 could be considered as constant (Table 3). Above 27.2C, during
this dynamic process, both trisaturated and monounsaturated TAGs
crystallized directly under their
form, since Tc was above the melting
point of the
form of cocoa butter and that the  
transition of
trisaturated TAGs in cocoa butter occurs below 28C (Loisel et. al,
1998). These results show that the crystallization of SSS in a
form
(2L) did not accelerate the formation of
crystals of the monounsaturated TAGs (3L) of cocoa butter. This confirmed results of Cebula
and al., (1991) but not those of Hachiya et al., (1989) who observed
that dark chocolate crystallization was slightly accelerated by adding
StStSt (form
).
The SSS effect we observed could be related to our previous
results (Loisel et al., 1998) which showed DSC and X-ray diffraction
data as a function of temperature. The  form of SSS (d001 = about 50
; d001 refers to first order spacing of the dhkl = 001 planes) did not
accelerate the crystallization of the monounsaturated TAGs of cocoa
butter under their unstable forms (sub, d001 = 53 and  d001 = 49).

Fig. 5Influence on chocolate crystallization recorded at Tc=28.2C

of the addition of different amounts of tristearin. Percentages correspond to the content of trisaturated TAGs in chocolate fat after
tristearin incorporation. The base line drawn shows the reference
from which the amplitude of the torque jump h was graphically
determined as in Fig. 1.

Fig. 6Influence of addition of trisaturated TAGs (% of chocolate fat)

on induction time of the first jump (t1) (3% correspond to the pure
dark chocolate with no addition of tristearin) ( , experimental points).
The curve fit ( solid line) tends towards both axis asymptotically.

Fig. 7Relationship between amplitude of the torque first jump

observed during chocolate crystallization (Fig. 5) and the tristearin
amount added (g/g of chocolate fat). The value of 1% corresponds to
the quantity of tristearin which crystallizes at 28.2C in the dark
chocolate (with no tristearin addition).

Fig. 8Effects of stearic acid or distearin addition on chocolate

crystallization at 28.2C. (1) corresponds to chocolate repeatability
with the pure dark chocolate, (2) and (3) to chocolate with addition of
0.3 and 0.6% of stearic acid, respectively, and (4) to chocolate with
0.3% of distearin (Table 2).

78JOURNAL OF FOOD SCIENCEVolume 63, No. 1, 1998

Fast crystallization of the SSS observed during that same study for a
rapidly cooled (-2C/min) cocoa butter sample, as well as a short t1
found at low Tc (Fig.6) tend to show that during this dynamic process,
the crystallization of trisaturated TAGs occurs even for Tc<27.2C,
but the rapid crystallization of the monounsaturated TAGs under their

form did not enable their formation to be detected (Fig.1). The
crystallization of the minor SSS was masked by that of the main
TAGs.
Influence of distearin or stearic acid addition

The effects of adding stearic acid or distearin on the dynamic

crystallization of dark chocolate at 28.2C were compared (Fig. 8)
The influence of these additives on the crystallization parameters, t1, t2
and h, (Table 3) indicated that the first jump amplitude was not modified by addition of free fatty acid (FFA) and diacylglycerol (DAG)
and that t1 was only slightly modified. Thus, the trisaturated TAGs
crystallization was neither directly affected by the presence of these
compounds nor related to their concentration in chocolate. On the
contrary, results showed that FFAs and DAGs slowed the second step
of chocolate fat crystallization. These effects confirmed the observations of Cebula and Smith (1992) by thermal analysis that diacylglycerols slowed the subsequent velocity of growth and retard the TAG
polymorphic transitions on heating.
Influence of emulsifiers additions

Emulsifiers, such as lecithins and PRPG (polyricinopolyglyceroesters), were added to the pure dark chocolate to study their influence
on crystallization. However, this could not be studied in the SSHE
because their introduction in the cocoa butter modified both the overall
rheological behavior of chocolate (Casson-Steiner viscosity was reduced and yield value was increased for lecithins; while the reverse
was observed for PRPG) and, as a result, the blade friction onto the
SSHE surface. Due to a large decrease of blade friction, it was necessary to increase considerably their pressure on the SSHE surface.
Thus, results recorded with emulsifiers were not directly comparable
with those recorded with pure dark chocolate, although excellent repeatability of the recordings was found. However, no significant difference was observed for t1, saturated TAGs tending to crystallize in
the same way as with the pure dark chocolate. Moreover, a faster
crystallization of the monounsaturated TAGs of cocoa butter was
never observed, t2, for the enriched-chocolate, even when friction was
increased, which would indicate that emulsifier addition always delayed crystallization.
Concerning the influence of additives on chocolate crystallization
in general, our observations confirmed those of Davis and Dimick
(1989a, b). They reported, during cocoa butter crystallization, some
minor components, as trisaturated TAGs, crystallized separately from
the other TAGs. Moreover, assuming reliable comparisons between
cocoa butter and chocolate, the absence of connection between crystallization of the high-melting TAGs and that of the monounsaturated
TAGs of the chocolate fat explains why Chaiseri and Dimick (1989a,
b) reported that the behavior of cocoa butter from different geographical origins was not related to SSS, but rather depended on StOSt,
POSt, StOO and POO concentrations.

CONCLUSION
CHOCOLATE CRYSTALLIZATION DURING A DYNAMIC PROCESS IS A
two-step process for Tc>26.2C but it is a one step process at
Tc<26.2C. The first step was related to crystallization of part of the
saturated (SSS) TAGs, about 1% of the whole fat chocolate (1/3 of
SSS), while the second or main crystallization, mainly corresponded
to the monounsaturated TAGs (SUS). The SSS solubility in the liquid
phase of chocolate (about 2%) could be calculated from data by DSC
and dynamic crystallizations in SSHE of chocolate with different
amounts of tristearin. This technique made it possible to study the
behavior and effects of chocolate additives. No relationship occurred

between the amount of saturated TAGs and the rate of crystallization

in the
form of monounsaturated TAGs. Incorporation of stearic acid
and distearin slightly decreased the growth rate of SSS crystals and
delayed their formation. The induction time of crystallization of monounsaturated triacylglycerols under their
form increased almost
linearly with concentration. The influence of emulsifiers such as lecithin could not be reliably studied in this way since their introduction in
the cocoa butter modifies blade friction and overall rheological behavior. Cocoa butter crystallization cannot be considered as that of a pure
compound, although it is largely influenced by its three main monounsaturated triacylglycerols (POP, POSt and StOSt). Fat fractionation,
which was observed for the first time in chocolate, occurs during
cooling and results from low solubility of saturated triacylglycerols in
the monounsaturated major fraction. This confirms, for chocolate,
results obtained recently with pure cocoa butter by synchrotron X-ray
diffraction and DSC (Loisel et al., 1998).

REFERENCES
Adenier, H., Chaveron, H. et Ollivon, M. 1993. Mechanism of fat bloom development
on chocolate. In Shelf life studies of foods and beverages; Chemical, biological,
physical and nutritional aspects, G. Charalambous (Ed.). Elsevier Science Publishers B.V.
Adenier, H., Ollivon, M., Perron, R., and Chaveron, H. 1975. Le blanchiment gras.I.
observations et commentaires. Chocolaterie Confiserie de France. 315: 7.
Arruda, D.H. and Dimick P.S. 1991. Phospholipid composition of lipid seed crystal
isolates from Ivory Coast butter. J. Am. Oil Chem. Soc. 68: 385.
Cebula, D.J., Dilley, K.M., and Smith, K.W. 1991. Continuous tempering studies on
model. confectionery systems. 45th P.M.C.A. Production Conference: 42.
Cebula D.J. and Smith K.W. 1992. Differential scanning calorimetry of confectionery
fats: Part IIEffects of blends and minor components. J. Am. Oil Chem. Soc. 69:992.
Chaiseri, S. and Dimick, P.S. 1995a. Dynamic crystallization of cocoa butter. I. Characterization of simple lipids in rapid- and slow-nucleating cocoa butters and their
seed crystals. J. Am. Oil Chem. Soc. 72: 1491.
Chaiseri, S. and Dimick, P.S. 1995b. Dynamic crystallization of cocoa butter. II. Morphological, thermal and chemical characteristics during crystal growth.. J. Am. Oil
Chem. Soc. 72: 1497.
Chapman, G.M., Akehurst, E.E., and Wright, W.B. 1971. Cocoa butter and confectionery
fats-studies using programmed temperature X-ray diffraction and differential scanning calorimetry. J. Am. Oil Chem. Soc. 48: 824.
Davis, T.R and Dimick, P.S. 1986. Solidification of cocoa butter. Proc. PMCA Prod.
Conf. 40: 104.
Davis, T.R. and Dimick, P.S. 1989a. Crystals formed during cocoa butter solidification.
J. Am. Oil Chem. Soc. 66(10): 1488-1493.
Davis, T.R. and Dimick, P.S. 1989b. Lipid composition of high-melting seed crystals
formed during cocoa butter solidification. J. Am. Oil Chem. Soc. 66(10): 14941498.
Dimick, P.S. 1994. Influence de la composition sur la cristallisation du beurre de cacao.
La cristallisation - applications au sucre et aux produits sucrs (Alliance 7 et le
CEDUS) nov. Paris.
Grabielle-Madelmont, C. and Perron, R. 1983. Calorimetric studies on phospholipidwater systems. J. Colloid Interface Sci. 95: 471.
Hachiya I., Koyano, T., and Sato, K. 1989. Seeding effects on solidification behavior of
cocoa butter and dark chocolate. I. Kinetics of solidification. J. Am. Oil Chem. Soc.
66: 1763.
Huyghebaert, A. and Hendrickx, H. 1971. Polymorphism of cocoa butter, shown by
differential scanning calorimetry. Lebensm.- Wiss. U. Technol. 4: 59.
Loisel, C., Keller, G., Lecq, G., Launay, B., and Ollivon, M. 1997. Tempering of chocolate
in a scraped-surface heat exchanger. J. Food Sci. 62: 4.
Loisel, C., G., Lecq, G., Keller G, Bourgaux C., and Ollivon, M. 1998. Phase transitions
and polymorphism of cocoa Butter, J. Am. Oil Chem. Soc. In press.
Loisel C. 1996. PhD thesis, Physico-Chimie du chocolat : Cristallisation du beurre de
cacao et proprits structurales, E.N.S.I.A., Massy, France.
Lovegren, N.V., Gray, M.S., and Feuge, R.O. 1976. Polymorphic changes in mixtures of
confectionery fats. J. Am. Oil Chem. Soc. 53: 83.
Manning D.M. 1984. Thermal and compositional properties of cocoa butter during
static and dynamic crystallization. Ph.D thesis, The Pennsylvania University, State
College, PA.
Merken, G.V. and Vaeck, S.V. 1980. Etude du polymorphisme du beurre de cacao par
calorimtrie DSC. Lebensm.- Wiss. U. Technol. 13: 314.
Ollivon, M. 1992. Manuel des Corps Gras, in Mlange de triglycrides, A. Karleskind
(Ed.), p. 491-503. Tec. et Doc, Lavoisier.
Rossell, J.B. 1967. Advances in Lipids Research. (5): 353-408.
Schlichter-Aronhime, J. and Garti, N. 1988. Solidification and polymorphism in cocoa
butter and the blooming problems. Ch. 9. In Crystallization and Polymorphism of
fats and fatty acids, Surfactant Science Series, vol 31, N. Garti and K. Sato (Ed.), p.
363-393. Marcel Dekker Inc., New York.
Schlichter-Aronhime J., Sarig S., and Garti N. 1988. Reconsideration of polymorphism
transformations in cocoa butter using DSC. J. Am. Oil Chem. Soc. 65(7): 1140-1143.
Sidisky, L.M. and Dimick, P.S. 1993. Further characterization of steryl glucosides and
esterified steryl glucosides in cocoa butter and seed crystals nuclei. Inform 4: 504.
van Malssen K.F. 1994. Real-time X-ray powder diffraction applied to cocoa butter and
graphite intercalates. Ph.D thesis, Amsterdam University
Wille, R.L. and Lutton, E.S. 1966. Polymorphism of cocoa butter. J. Am. Oil Chem. Soc.
43: 491.
Ms received 10/22/96; revised 6/30/97; accepted 8/6/97.
C. Loisel was supported for PhD preparation by a grant from Association Nationale de la Recherche
Technique (A.N.R.T., Convention CIFRE) and Danone Group. We thank G. Barratt for reading the
manuscript.

Volume 63, No. 1, 1998JOURNAL OF FOOD SCIENCE79