You are on page 1of 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/320961368

Determination of chocolate melting properties by capacitance based thermal


analysis (CTA)

Article  in  Journal of Food Measurement and Characterization · March 2018


DOI: 10.1007/s11694-017-9677-0

CITATIONS READS

15 5,616

2 authors:

Juzhong Tan William L Kerr


Florida A&M University University of Georgia
29 PUBLICATIONS   556 CITATIONS    140 PUBLICATIONS   3,994 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Predictive Modeling of Pecan Quality during Commercial Storage and Distribution View project

Chocolate properties measurement and control View project

All content following this page was uploaded by Juzhong Tan on 09 November 2017.

The user has requested enhancement of the downloaded file.


Food Measure
DOI 10.1007/s11694-017-9677-0

ORIGINAL PAPER

Determination of chocolate melting properties by capacitance


based thermal analysis (CTA)
Juzhong Tan1 · William L. Kerr1   

Received: 22 March 2017 / Accepted: 31 October 2017


© Springer Science+Business Media, LLC 2017

Abstract  In this study, a capacitance thermal analyzer ingredients in a continuous fat phase [1]. Chocolates are
(CTA) was designed and tested for measuring the melting categorized as dark, white, or milk chocolate based on the
properties of chocolates, and compared with those measured content of cocoa mass, dairy ingredients and cocoa butter.
by DSC and dynamic rheology. Chocolates with different The production of chocolate starts from raw cocoa beans,
fat content and particle size distribution (PSD) were placed which are then subjected to fermentation, roasting, milling,
between stainless steel plates, while capacitance and temper- winnowing, refining, and tempering. The melting properties
ature were monitored between 20 and 60 °C. The PSD did of chocolates are critical to their quality because they greatly
not influence the ­Tonset (~ 25 °C) and ­Tpeak (33 °C) measured influence consumer acceptability, appearance and storage
by DSC. However, samples with finer particles had lower stability of the product. Improper processing can lead to
­Tend than those with coarser particles (36.59–37.28 °C). Var- undesirable melting temperatures and crystal states, along
ying fat content did not result in differences in the DSC melt- with loss of glossy appearance or formation of fat bloom
ing curves. Samples with smaller particle sizes had lower during storage [2–4].
temperatures at peak capacitance than those with larger par- One important factor that influences melting properties
ticles, with peak temperatures ranging from 30.8 to 39.3 °C, is the crystal form of the fat phase in chocolate. The fat
while higher peak capacitance values (2.61–2.84 ­10− 11 F) crystals may exhibit polymorphism, that is they may have
were measured by CTA. Samples with higher fat content had more than two distinct crystalline forms in the same sam-
lower peak temperatures (range 34.7–39.71 °C) but higher ple [2]. At least six polymorphic forms have been identi-
peak capacitance values (range 3.29–4.3 1­ 0− 11F). Values fied, designated Forms I through VI in the Roman numeral
from the CTA were best correlated with results determined system [5]. The forms have different melting temperatures:
by dynamic thermal rheometry. 16–18 °C for Form I (γ), 22–24 °C for Form II (α), 24–26 °C
for Form III (β′), 26–28 °C for Form IV(β1′), 32–34 °C for
Keywords  Chocolate · Melting properties · Capacitance · Form V (β1), and 34–36 °C for form VI (β2). The Greek
Thermal analysis letter designations refer to the way in which triglycerides are
packed into a unit cell in the crystal structure. The preferred
polymorph is Form V in most cases. Thus, most chocolate
Introduction undergoes tempering to ensure that mostly Form V exists in
the finished product. Typically, the chocolate is heated just
Chocolate is a complex semi-solid suspension of fine enough (~ 50 °C) to melt and remove all crystal structures
particles containing cocoa, sugar, and sometimes dairy from the cocoa butter. The chocolate is cooled to ~  27 °C
allowing Form V, and some Form IV, crystals to grow. The
product temperature is then raised (~  29–31 °C) to melt off
* William L. Kerr Form IV crystals, and Form V seed crystals may be added
wlkerr@uga.edu
to encourage the process.
1
Department of Food Science and Technology, University Particle size distribution and fat content of the chocolate
of Georgia, Athens, GA 30602, USA are also important factors influencing appearance, consumer

13
Vol.:(0123456789)
J. Tan, W. L. Kerr

acceptance, and processing properties. The influence of property to measure as it does not require determination of
these factors on the rheological properties, texture and both the real and imaginary components of ε*. In addition,
consumer evaluation of chocolates have been well studied. many capacitance measuring instruments determine values
Afoakwa et al. [6] found that higher fat content and smaller at discrete frequencies. Thus, they are much simpler to build
particle size decreased the plastic viscosity of molten dark and less expensive than full-scale network analyzers. In this
chocolate. Some researchers have noted that large particle study, a capacitance based thermal analysis (CTA) system was
size of the cocoa mass can lead to sensations of grittiness proposed to study the melting properties of chocolate. The
or coarseness in the mouth [7]. Do et al. [8] studied how fat apparatus was constructed at the University of Georgia using
content and particle distribution of chocolate liquor influ- metal plates and an inexpensive capacitance meter. Capaci-
ence the hardness and heat resistance of solid chocolates. tance is the ability of a subject to store electrical charge, and
They also found that chocolates with higher fat content need the capacitance of a parallel-plate capacitor can be determined
longer heating times to be completely melted. In addition, as:
chocolates with finer particles exhibited longer melting
A
times than those with larger particles. However, the influ- C = ε� ε0 (1)
d
ence of fat content and particle size distribution on melting
temperatures has been barely studied. where ε′ is dielectric constant or permittivity of the medium
Differential scanning calorimetry (DSC) is one of the between the parallel-plates, ε0 is the dielectric constant when
most commonly used methods to determine the melting the medium is air, Ais the overlap area (completely and
properties of chocolates. The DSC compares the flow of exclusively filled with medium) of the two parallel plates,
heat between a sample and empty reference pan during a and d is the distant between the two plates. The value of
constantly changing temperature. Large changes in heat flow ε0 varies only by 0.005% as temperature increases from 0
are indicative of increases in specific heat related to phase to 50 °C [13]. A medium such as food, particularly when
transitions. Often the heat flow change during a phase transi- undergoing a phase transition, will have an ε′ and capaci-
tion is not a specific point, but associated with a temperature tance that changes with temperature.
range. Therefore, there are several ways to characterize melt- A few studies have used capacitance sensors similar to the
ing properties. That is, starch gelatinization, fat melting and CTA to study the glass transition and melting temperatures of
other first-order transitions can be characterized by an onset frozen carbohydrate and NaCl solutions [14]. They found tran-
temperature ­(Tonset), peak temperature (­ Tpeak), ending tem- sition temperatures similar to those measured by differential
perature ­(Tend) and enthalpy of melting (∆Hmelt) [9]. scanning calorimetry. A similar system has been successful
The melting properties of chocolates have also been stud- for measuring glass transitions in soft and hard sugar candies
ied by dynamic thermal rheology [10]. These instruments [15]. In this study, a system is described to study the melt-
measure the storage modulus (G′) or loss modulus (G″). As ing properties of chocolates with different fat content, particle
cocoa butter melts, the firm chocolate becomes softer and size distribution and cocoa mass content. DSC and viscoelastic
may begin to flow. Thus, G′ is a good indicator to show the properties were also measured for comparison.
beginning and the end of chocolate melting. Generally, G′ Frozen or glassy aqueous systems containing salts or
is on the order of 1­ 03–104 Pa for the solid and diminishes to low molecular weight compounds such as sugars and acids
­100–101 Pa once melted. undergo a large change in ε′ when transitioning from solid to
Dielectric thermal analysis (DETA) is another method to liquid. In contrast, high fat systems have comparatively smaller
study phase transitions of food materials. DETA is carried differences in ε′ between solid and liquid states. Thus, the cell
out by a dielectric analyzer, which measures the change of was modified from previous versions to enhance the measured
dielectric constant ε′ and dielectric loss ε″ of a food mate- differences between states. The modified CTA consists of one
rial as a function of temperature or AC frequency. The ε′ smaller upper plate and one larger base plate. During a tem-
is the ability of a material to store energy in response to an perature scan, the chocolate sample (with larger area than the
applied electric field. In contrast, ε″ describes the ability of upper plate but smaller than the base plate) melts and collapses
one material to dissipate energy in response to an applied when the temperature reaches its melting point. The composi-
electric field, and that typically results in heat generation tion of the medium between the parallel plates changes from
[11]. The dielectric loss of a food material reaches a peak chocolate only to layers of chocolate and air. If the layers are
value when it reaches the second-order transition tempera- separate, the capacitance of the medium is given by:
ture, as with a glass transition. On the other hand, when
A
a food material undergoes a first-order transition such as C = (𝜃(T)ε� + (1 − 𝜃(T))ε0 )ε0 (2)
d
melting, the dielectric constant will increase greatly [12].
Capacitance is an electrical property dependent upon the where 𝜃(T) is the volume fraction of chocolate in the
dielectric properties of the medium. However, it is an easier medium at a specific temperature.

13
Determination of chocolate melting properties by capacitance based thermal analysis (CTA)

The objective of this research was to evaluate the use of isopropanol in a 50 ml centrifuge tube, then mixed with a
a modified CTA system to measure melting transitions in vortex mixer for 2 min. The samples were then introduced
chocolate. These were compared to results measured by DSC into the Mastersizer dispersion mixer until an obscuration of
and rheometry. In addition, samples with differing cocoa 20% was reached. Between samples, the detector and laser
mass particle size and fat content were tested to determine were aligned and backgrounds were calibrated. The size dis-
if these had an influence on the measured properties. tribution was expressed as the relative volume of particles
in each size range (Malvern MasterSizer Micro Software
v 2.19). Several particle size distribution parameters were
Materials and methods determined including d­ 10, ­d50 and d­ 90, that is the particle size
below which 10, 50 or 90% of the sample lies. In addition,
Materials the Sauter mean diameter ­(d3,2) and volume mean diameter
­(d4,3) were determined.
Cocoa mass containing 70.4% cocoa solids was purchased
from Callebaut (Wieze, Belgium). Raw organic cocoa butter Capacitance based thermal analysis (CTA)
was purchased from Saaqin (Hicksville, New York). Raw
cocoa nibs were provided by Cocoatown (Roswell, Atlanta). The parallel-plate capacitor cell (Fig. 1) was built at the Uni-
Commercial chocolates were purchased from a Kroger gro- versity of Georgia Instrument Design and Fabrication Shop
cery in Athens, Georgia. (Athens, GA). It consisted of a 100 mm diameter stainless
base plate, a 60 mm diameter stainless upper plate, a cylin-
Preparation of chocolate samples drical Teflon side wall, and a threaded Teflon cap through
which a 12.5 mm diameter stainless screw could turn, and
Chocolate samples with different fat content were produced which was welded to the upper plate. This allowed the upper
by mixing raw cocoa butter with cocoa mass in percentages plate to be spaced a specified distance from the base plate.
of 10, 20, 30, 40, and 50% by weight. Each mix was treated The base plate and upper knob had 4 mm holes drilled to
in a tempering machine (Model Revolution 2, ChocoVision provide connection with a banana or needle plug. A cylinder
Corp., Poughkeepsie, NY). The program heating schedule spacer was made from Teflon with outer diameter 100 mm,
included an increase to 46 °C (over 10 min) followed by inner diameter 80 and 100 mm in height. The lower side was
a decrease to 27 °C (over 12 min). At that point Form V fixed to the base plate by a groove on the base plate.
chocolate seed crystals were added followed by an increase
in temperature to 31 °C with a hold for 5 min. The well-
tempered chocolate was poured into to 5 cm × 5 cm × 1.5 cm Plug jack
chocolate molds (Wilton INC. Woodridge, IL) and held
overnight at 4 °C .
Particle size of the chocolate samples was controlled by
Knurled knob
varying the grinding time. Raw cocoa butter was mixed with
cocoa nibs in a percentage of 10% and processed by a dou-
ble-conical stone grinding concher (Model ECGC05, Cocoa-
town, Roswell, Atlanta). About 5 ml cocoa liquid was taken
from the mix for particle size analysis every 15 min for the
first 2 h, and every 1 h thereafter. Five different batches of Teflon
cocoa butter and mass were processed in the concher for 15, Adjustable
screw
30 min, 1 h, 2, 4, and 12 h, respectively. The final chocolate
liquor from each batch was used to make chocolate samples
Stainless
using the tempering method described above. plates Upper plate

Particle size distribution Plug jack Base plate


Thermocouple
The particle size distribution was determined by laser dif-
fraction using a Malvern Mastersizer (Model MSS, Mal- Heating plate
vern Instruments Ltd., Malvern, England). The presenta-
tion was selected as “Custom” with refractive indices of
n(CH3 )2 CHOHn = 1.378 and ­nparticle = 1.590 [16]. About 2 g Fig. 1  Capacitance-based thermal analysis (CTA) cell showing stain-
of chocolate sample was melted and dispersed with 30 ml less conducting plates, Teflon container and hating plate

13
J. Tan, W. L. Kerr

Each chocolate sample was placed on the base plate and Statistical methods
the upper plate lowered until it contacted the sample. The
base plate was then heated by a Thermolyne Nuova hot- All tests were repeated at least three times and results pre-
plate at constant temperature (80 °C). The measurement of sented as the mean and standard deviation. The results were
capacitance was done with an LCR meter (NI PXI 4072, compared by one-way ANOVA using SAS 9.3 (SAS Institute
National Instrument Corporation, Austin, TX) attached to Inc., Cary NC) to determine the effects of the fat content
a PXI chassis (NI PXI 1042, National Instrument Corpo- and the particle size distribution of the chocolate sample on
ration, Austin, TX). A T-type thermocouple (Probe 1/16′′ the melting properties measured by DSC and CTA. Tukey′s
diameter, OMEGA Engineering, Stamford, CT) penetrated HSD was used to determine significant differences amongst
the base plate to just make contact with the sample on one treatments at the 95% level of confidence.
side. Temperature data was collected with a data acqui-
sition board (NI 9219, National Instrument Corporation,
Austin, TX). The virtual instrument interface and data col- Results and discussion
lection were accomplished with LabVIEW software (Ver-
sion 2015, National Instrument Corporation, Austin, TX). Particle size reduction and particle size distribution

The largest particle size ­(d90) for a typical cocoa mix (fat and
DSC nibs) is shown in Fig. 2 as a function of grinding time. A
compilation of the size and distribution measurements (­ d10,
The melting temperature of each sample was determined ­d50, ­d90, ­d32, ­d4,3) is shown in Table 1. Most of the particle
by differential scanning calorimetery (Model DSC 1, Met- size reduction was achieved within the first 2 h of grind-
tler-Toledo International Inc., Greifensee, Switzerland). ing. The initial ­d90 was 77 ± 22 µm, which was reduced to
Approximately 12 mg of each sample was sealed into a 15.5 ± 5.3 µm within 4 h, to 11.7 ± 4.1 µm within 8 h, and
40 μl aluminum pan. An empty pan was used as a refer- finally reached a constant value of ~  8 µm (Fig. 2). As shown
ence sample. Dry nitrogen gas was used to minimize water in Table 1, the ­d90 and ­d4,3 were significantly smaller after
condensation in the measuring cell. The temperature was 15 min, 30 min, 4 h and 8 h. The smallest diameter particle
scanned from 0 to 60  °C at a heating rate of 5  °C per ­(d10), however, did not decrease after 15 min. This indicates
minute. The onset temperature ­(Tonset), peak temperature that the stone grinding is most effective at reducing the size
­( T peak), end temperature ­( Tend) and enthalpy of melting of large particles, but a point is reached at which no further
(∆Hmelt) were calculated using the STARe Thermal Analy- reduction in particle size is achieved.
sis Software. The melting index ­(Tindex) was computed as A few researchers have studied the effects of particle size
­(Tend − Tonset). Each sample was analyzed in triplicate and on the properties of chocolate products. Bolenz and Man-
mean values and standard deviations reported. ske [17] researched the impact of fat content during grind-
ing on particle size distribution in chocolate milk. Afoakwa
et al. [18] reviewed the role that particle size distribution
Dynamic rheological analyses
100
The viscoelastic properties of the samples as a function
of temperature were measured with a dynamic rheometer
(Model Discovery HR-2, TA Instrument Inc., New Cas- 80
Particle Size d90 (µm)

tle DE). Before the tests were initiated, the samples were
melted and tempered in the rheometer. Approximately 4 g 60
of each chocolate sample was placed on the temperature-
controlled stage below a 40  mm cone and plate probe 40
(1.997°), then raised to 48 °C at a rate of 5 °C/min. The
sample was then cooled to 26.7 °C at a rate of 2 °C/min. 20
After a 5 min rest period, the temperature was raised to
31.6 °C at a rate of 2 °C/min. After the solid chocolate
0
was formed, the sample was then cooled to 10 °C at a rate 0 5 10 15 20 25
of 15 °C/min. The instrument was run in the small-strain Time (h)
oscillatory mode, with dynamic strains set to 1%. The tem-
perature ramp was set at 10–60 °C at a rate of 5 °C/min. Fig. 2  Change in average particle size (­d90) of chocolate over time
during refining

13
Determination of chocolate melting properties by capacitance based thermal analysis (CTA)

Table 1  Particle size d10 (μm) d50 (μm) d90 (μm) d(3,2) (μm) d(4,3) (μm)
distribution of samples as a
function of refining time 15 min b
1.13  ± 0.14 c
11.63  ± 2.1 d
46.41  ± 19 b
2.94  ± 0.27 39.46d ± 5.4
30 min 0.84a ± 0.11 7.64b ± 1.3 30.67c ± 7.9 2.73b ± 0.18 15.98c ± 3.3
4 h 0.71a ± 0.07 5.93a ± 1.6 15.50b ± 5.3 2.06a ± 0.22 7.25b ± 1.2
8 h 0.68a ± 0.07 5.16a ± 1.1 11.75a ± 4.1 1.95a ± 0.25 5.85a ± 0.71

Mean values  ± 95% confidence intervals


Values in a column not followed by the same superscript letter are significantly different (p < 0.05)

10 Table 2  DSC melting temperatures as a function of refining time

15min Tonset (°C) Tend (°C) Tpeak (°C) Tindex (°C)


8 30min
15 min 24.60a ± 0.24 37.28b ± 0.31 32.97a ± 0.11 12.68b ± 0.19
4h 30 min 25.04ab ± 0.18 37.05b ± 0.13 32.86a ± 0.08 12.01b ± 0.23
Volume (%)

6 8h 4 h 25.50b ± 0.14 36.62ab ± 0.22 32.94a ± 0.21 11.12a ± 0.26


8 h 25.46b ± 0.27 36.59a ± 0.17 32.83a ± 0.14 11.13a ± 0.22
4 Mean values  ± 95% confidence intervals
Values in a column not followed by the same superscript letter are
significantly different (p < 0.05)
2

0 affect the melting temperature as measured be either ­Tonset


0.01 0.1 1 10 100 1000
and ­Tpeak. Samples with smaller particles had slightly higher
Particle size (um)
­Tonset and slightly lower ­Tend (and therefore smaller ­Tindex).
The differences were small, however, with ­Tend rang-
Fig. 3  Particle size distribution in chocolate at different times during
ing from 36.59 ± 0.17 to 37.28 ± 0.31 °C and T ­ index from
refining
11.13 ± 0.22 to 12.68 ± 0.19 °C. Other researchers have seen
similar trends [2, 19], suggesting that in chocolate, particle
and ingredients have in determining rheological and sensory size distribution (PSD) has only a small influence on melting
properties of dark chocolate. In these cases, the researchers properties. They concluded that chocolates with finer parti-
used a three-roller refiner followed by sifting to control par- cles would take longer time to melt than their corresponding
ticle size distribution, thus quickly creating fairly uniform products with larger particles. There is no particular reason
particle size distributions with one peak. In our case, particle that the lipid phase would have differing properties due to
properties were determined solely by the grinding time. This the cocoa mass particles. Thus, any measured differences
is more reflective of industry practices, where stone grinding in melting behavior might be attributed to different crystal
is often used to create small and uniform particle sizes after sizes or polymorphs, or more likely due to differences in
long grinding times. heat transfer rates in samples with different size particles.
Figure 3 shows how the particle size distribution changed Differences in fat content (10–50%) resulted in slightly
during refining. Initially, there is a trimodal distribution with different ­Tonset, ­Tpeak and T
­ end values. ­Tonset varied from
sizes centered around 1, 8 and 190 µm. During grinding, 25.70 ± 0.14 to 26.11 ± 0.23 °C with increasing fat content,
the peak with largest particles diminished, and disappeared while ­Tend varied from 36.33 ± 0.25 to 37.47 ± 0.31 °C. Thus,
within 4 h. In conjunction, the middle peak shifted to smaller ­Tindex (reflective of peak width) increased from 10.22 ± 0.07
sizes, with the peak value going from 9 to 5 µm. The volume to 11.77 ± 0.16 °C. Again, this may reflect differences of
fraction of the medium size particles also increased with heat transfer rates in the samples. In addition, samples with
time, suggesting that this fraction grew at the expense of more fat will take longer to melt, and this plays out even as
the larger particles. the calorimeter temperature is constantly increasing. Previ-
ous studies have shown that the melting properties can be
DSC measurements of chocolate melting influence by PSD, crystallization form [20, 21], lipid types,
additives, and composition [22] of chocolates. Others have
The melting temperatures of chocolate samples measured by suggested that the amount of fat in dark chocolates influ-
DSC are shown as a function of the refining time (Table 2) or ences the degree of crystallinity and crystal size distribution
fat content (Table 3). In general, particle size did not greatly (CSD) of the samples [23].

13
J. Tan, W. L. Kerr

Table 3  DSC melting Fat content (%) Tonset (°C) Tend (°C) Tpeak (°C) Tindex (°C)
temperatures of chocolate with
different fat content 10 b
26.11  ± 0.23 a
36.33  ± 0.25 a
33.19  ± 0.25 10.22a ± 0.07
20 25.90ab ± 0.37 37.02b ± 0.17 33.40a ± 0.34 11.12b ± 0.11
30 25.68a ± 0.18 37.19b ± 0.36 33.80a ± 0.27 11.51c ± 0.06
40 25.61a ± 0.21 37.23b ± 0.41 33.57a ± 0.22 11.62cd ±
0.09
50 25.70a ± 0.14 37.47b ± 0.31 34.28a ± 0.40 11.77d ± 0.16

Mean values  ± 95% confidence intervals


Values in a column not followed by the same superscript letter are significantly different (p < 0.05)

Table 4  Rheometric melting temperatures as a function of refining Table 5  Rheometric melting temperatures of chocolate with different
time fat content
Tonset (°C) Tend (°C) Tindex (°C) Fat content Tonset (°C) Tend (°C) Tindex (°C)
(%)
15 min 28.4d ± 0.3 37.1a ± 0.2 9.08a ± 0.1
30 min 26.2c ± 0.4 37.9ab ± 0.3 11.7b ± 0.1 10 28.1d ± 0.3 37.2a ± 0.4 9.08a ± 0.2
4 h 25.6ab ± 0.3 38.3b ± 0.2 12.7c ± 0.03 20 27.2c ± 0.3 37.7ab ± 0.5 10.5b ± 0.2
8 h 25.1a ± 0.4 38.9c ± 0.2 13.8d ± 0.1 30 26.2bc ± 0.3 37.9b ± 0.3 11.7c ± 0.2
40 26.1b ± 0.2 38.2c ± 0.4 12.1d ± 0.1
Mean values  ± 95% confidence intervals 50 25.3a ± 0.4 38.9d ± 0.2 13.6e ± 0.1
Values in a column not followed by the same superscript letter are
significantly different (p < 0.05) Mean values  ± 95% confidence intervals
Values in a column not followed by the same superscript letter are
significantly different (p < 0.05)
Dynamic rheological analyses of chocolate melting
29
Changes in the dynamic rheological properties during tem-
perature scans of chocolate were best characterized by the 28.5
storage modulus (G′) (Table 4). In general, G′ in the solid
Capacitance (pF)

state was ~ 103–104 Pa and decreased over a short tempera-


ture range corresponding to melting, to values near 1­ 01 Pa 28

in the molten state. The beginning of melting was character-


ized as ­Tonset, where the gradient of this temperature point is 27.5
at least 3 times greater than the previous three data points.
The end of melting was characterized as T ­ end, where the
27
gradient after T ­ end was less than 10% less than its previous 20 30 40 50 60
three data points. The T ­ index of each sample was calculated Temperature (°C)
as ­Tend − ­Tonset.
As found with the DSC results, increasing fat content Fig. 4  Capacitance as a function of temperature during chocolate
resulted in larger ­Tindex values, indicating samples with melting. (circle) raw data, (-) processed through low-pass Butterworth
filter
higher fat content had a broader temperature range over
which they melted (Table 5). In addition, the T ­ onset and ­Tendset
measured by the rheometer were similar to DSC measure- CTA measurements of chocolate melting
ments, but did have slightly greater ­Tonset values. The differ-
ence may be attributed to the differing sample sizes (~ 3 g Figure 4 shows a representative figure of how capacitance
vs. ~ 10 mg) required by the rheometer and DSC, respec- changes with temperature for chocolate samples in the CTA
tively, and that different physical phenomena are being device. In general, capacitance values were in the picofarad
measured by the two instruments. range for samples in the solid state. As the temperature
However, rheometric measurements did show a greater increased there was a slight increase in the capacitance,
range in ­Tonset (25.3 ± 0.4 to 28.1 ± 0.3 °C) than measured followed by a decrease after melting had commenced. The
by DSC, and greater statistical significance amongst values initial increase is likely due to an increase in the dielec-
measured for samples with differing fat levels. tric constant ε′ as temperature increased, stemming from

13
Determination of chocolate melting properties by capacitance based thermal analysis (CTA)

increased molecular mobility. However, changes would only Table 7  CTA transitions for samples with different fat content
be slight for molecules in the solid phase. There are different Fat content (%) Capacitance peak tem- Peak capacitance (pF)
lipid fractions in cocoa butter, and some may begin to melt at perature (°C)
lower temperatures. For example, Torbica et al. [24] showed
10 39.7c ± 0.5 3.29a ± 0.15
that cocoa butter had a solid fat content of 82, 76, 51 and 8%
20 37.5b ± 0.4 3.70b ± 0.02
at temperatures of 20, 25, 30 and 35 °C, respectively.
30 36.4b ± 0.5 4.11c ± 0.05
At a critical temperature, the capacitance began to drop.
40 35.5ab ± 0.3 4.16c ± 0.02
This coincided with melting of the chocolate, and its flow
50 34.7a ± 0.3 4.30d ± 0.07
into the outer chamber, which began to reduce the volume
of chocolate between the plates. The receding chocolate was Mean values  ± 95% confidence intervals
replaced by air, which has a much lower dielectric constant Values in a column not followed by the same superscript letter are
(1.00059 at 20 °C), and thus the measured capacitance began significantly different (p < 0.05)
to decrease.
The particle size distribution (PSD) of the samples sig-
nificantly influenced the temperature at peak capacitance, to flow [26, 27]. These observations concurred with results
which ranged from 30.8 ± 0.6 to 39.3 ± 0.4 °C (Table 6), from the dynamic rheology, which also indicated that higher
while peak capacitance values ranged from 26.1 ± 0.4 to fat content in chocolate samples produced lower ­Tonset.
28.4 ± 0.3 pF. That is, finer PSD resulted in lower peak Overall, samples with finer particles or lower fat content
capacitance temperature and greater peak capacitance had lower peak capacitance values than those with larger
value. As previously noted, larger cocoa mass particle size particles and more fat. One of the critical factors that influ-
can produce chocolate that is more difficult to melt. This ences capacitance is the dielectric constant ε′ of the medium.
likely resulted in higher peak temperatures for the CTA. In The ε′ of cocoa butter (ε′ ~ 2.5) is higher than cocoa mass
addition, and by design, the heating rate in the CTA was not (ε′ ~ 1.5), so that higher fat content in chocolate samples will
carefully regulated, and this also resulted in slower melting. lead to higher absolute capacitance values. The way in which
It has also been observed that PSD influences the rheo- particles influence the dielectric properties of inhomogene-
logical properties of chocolate. Samples with larger PSD ous media is complex. Several models have attempted to
yielded lower apparent viscosity and lower yield stress [25]. predict the complex permittivity of such media [28]. These
Changes in such properties may influence how the melted models typically incorporate the ε′ and ε” values for each
chocolate spreads away from the capacitance plates, and phase, the volume fraction (α), and in some cases the shape
thus affects the measured transition temperature. However, or size factors of suspended media. In general, increasing
this would be expected to give lower measured transition the volume fraction would tend to emphasize the properties
temperatures. Thus, the influence of particle size on heat of the suspended phase. Experimental studies suggest that
transfer and the resulting rate of melting are likely to be the these models perform better at lower α, and fail at some criti-
predominant factors. cal value at which percolation behavior becomes dominant.
Table 7 shows how the peak capacitance temperature var- This critical value is thought to be dependent on particle
ied with fat content in the chocolate. Samples with greater size and shape.
fat content resulted in lower peak capacitance temperature
(range 34.7 ± 0.3 to 39.7 ± 0.5 °C). This may be attributed to
the fact that chocolates with greater fat content have lower Comparison of CTA, DSC and rheological data
apparent viscosity and yield stress, again making them easier
The fat content and particles size had only a limited influ-
ence on melting temperatures as measured by DSC. In con-
Table 6  CTA transitions as a function refining time trast, CTA measurements were more sensitive to variations
in fat content and PSD. Rheological assessments of melting
Capacitance peak tem- Peak capacitance (pF)
perature (°C) behavior were also more dependent on fat content and PSD.
These observations reflect that melting has a thermodynamic
15 min 39.3d ± 0.4 28.4d ± 0.3 component as well as a time-dependent element depend-
30 min 36.1c ± 0.4 27.5c ± 1.1 ent on rates of heat transfer. Thus, the rheometer and CTA
4 h 32.4b ± 0.4 26.6b ± 0.8 both require samples of greater mass as compared to that for
8 h 30.8a ± 0.6 26.1a ± 0.4 DSC. Thus, DSC might be expected to give transition tem-
Mean values  ± 95% confidence intervals peratures with greater precision, but may not totally reflect
Values in a column not followed by the same superscript letter are melting behavior as experienced by real world users of a
significantly different (p < 0.05) product.

13
J. Tan, W. L. Kerr

The transition temperatures measured by CTA were gen- cocoa butter, melt over a broader range of temperature. This
erally greater than those measured by DSC or rheometric occurs as various distributions of fatty acids create fats with
techniques. However, a correlation between CTA values and different melting fractions. In addition, fats may exist in one
the other methods was achieved. Several regression models or more polymorphic forms that have different melting tem-
were tested but the best fit, without overfitting, between T
­m peratures. These materials are best analyzed by scanning
measured rheological and CTA methods was achieved with the temperature. The change in signal with time/temperature
a quadratic model: depends on the device. Thus the way that heat flux changes
(in DSC) with solid fat ratio (as different fractions melt) is
not identical to the way that molecular relaxations respond
)2
(3)
( ( )
Tm,rhe = a Tm,CTA + b Tm,CTA + c
during thermal rheology.
where a, b and c are fitted constants; ­Tm,CTA the melting Sample size and heating rate are other factors that skew
temperature determined by the CTA device; and T ­ m,rhe the the results. Samples with more mass take longer to melt,
melting temperature determined by rheological methods. thus spread the melting process over a wider temperature
Equation 3 was used to model all data, that is, where particle range. However, overly small samples may hamper sensitiv-
size was varied or where fat content was varied. In addition, ity. In general, lower heating rates should give truer results,
the model was applied using both the rheological onset and but may compromise sensitivity and cause unacceptably
endset transition temperatures. long measurement times. All of the methods discussed are
Figure 5 shows the results of the regression analyses. only as good as the temperature measurement. Thus, DSC,
Both onset and endset temperatures were correlated to the rheological or dielectric thermal devices need to be cali-
CTA measured values. In the case of the onset temperature, brated with materials of known melting properties.
the relationship was negative (­ r2 = 0.952) while for the end- The above factors also influence the various transition
set temperature it was positive (­ r2 = 0.950). This does show temperatures that are recorded. For example, the peak tem-
that, if desired, the CTA method could be correlated with perature often reported for DSC depends on both the sample
another technique. size and heating rate. The CTA device was more sensitive
This does raise the issue as to which approach provides to measuring the onset melting temperature than the other
the truest measure of melting point. For a pure substance, the techniques studied, thus we expect it to be most beneficial
­Tm is more easily determined as melting occurs over a very to determine where melting begins.
sharp range. Chemists have often placed crystalline materi-
als in capillary tubes and denoted where a sample visually
melts as the temperature is increased slowly. In theory, a
pure substance melts at one temperature, however there is Conclusions
always the need to impose a higher external temperature
to allow heat transfer. Mixed crystalline systems, such as The CTA device introduced in this study measured melt-
ing temperatures in chocolate with good repeatability.
The device also showed some sensitivity to the effects of
40 cocoa particle size and fat content on melting behavior. The
instrument was inexpensive as it was built from metal plates
connected to a capacitance meter, and did not require pre-
35
Tm,on = -0.0169 (Tm,CTA)2 + 0.998 (Tm,CTA) + 24.14 cise control of temperature ramping as used with DSC or
r2 = 0.952 dynamic thermal rheometers. While the measured tempera-
tures were not identical to those from DSC or rheometric
Rheo Tm (°C)

techniques, they could be easily correlated with those values


30
if required.

Acknowledgements  We would like to thank Dr. M. Balu and the


25 CocoaTown Company in Atlanta, Georgia for their help with funding
Tm,en = 0.0468 (Tm,CTA)2 - 2.937 (Tm,CTA) + 71.34 and acquisition of supplies.

r2 = 0.950

20
30 32 34 36 38 40
References
CTA Tm (°C)
1. E.O. Afoakwa, A. Paterson, M. Fowler, Factors influencing rheo-
Fig. 5  Comparison of T­ m values determined by change in storage logical and textural qualities in chocolate—a review. Trends Food
modulus (G′) during temperature scanning, and by CTA Sci. Technol. 18, 290–298 (2007)

13
Determination of chocolate melting properties by capacitance based thermal analysis (CTA)

2. E.O. Afoakwa, A. Paterson, M. Fowler, M.J. Vieira, Influence of 16. H.G. Merkus, G.M.H. Meesters, Particulate products: tailoring
tempering and fat crystallization behaviours on microstructural products for optimal performance (Springer, New York, 2013),
and melting properties in dark chocolate systems. Food Res. Int pp. 253–273
42, 200–209 (2009) 17. S. Bolenz, A. Manske, Impact of fat content during grinding on
3. A.G. Stapley, H. Tewkesbury, P.J. Fryer, The effects of shear and particle size distribution and flow properties of milk chocolate.
temperature history on the crystallization of chocolate. J. Am. Oil Eur. Food Res. Technol. 236, 863–872 (2013)
Chem. Soc. 76, 677–685 (1999) 18. E.O. Afoakwa, A. Paterson, M. Fowler, J. Vieira, Characterization
4. D. Dhonsi, A.G.F. Stapley, The effect of shear rate, temperature, of melting properties in dark chocolates from varying particle size
sugar and emulsifier on the tempering of cocoa butter. J. Food distribution and composition using differential scanning calorim-
Eng. 77, 936–942 (2006) etry. Food Res. Int. 41, 751–757 (2008)
5. H. Schenk, R. Peschar, Understanding the structure of chocolate. 19. S.T. Beckett, The Science of Chocolate, vol. 22 (Royal Society of
Radiat. Phys. Chem. 71, 829–835 (2004) Chemistry, Cambridge, 2000). pp. 81–101
6. E.O. Afoakwa, A. Paterson, M. Fowler, Effects of particle size 20. L. Svanberg, L. Ahrné, N. Lorén, E. Windhab, Impact of pre-
distribution and composition on rheological properties of dark crystallization process on structure and product properties in dark
chocolate. Eur. Food Res. Technol. 226, 1259–1268 (2008) chocolate. J. Food Eng. 114, 90–98 (2013)
7. C. Servais, R. Jones, I. Roberts, The influence of particle size 21. C. Loisel, G. Keller, G. Lecq, B. Launay, M. Ollivon, Tempering
distribution on the processing of food. J. Food Eng. 51, 201–208 of chocolate in a scraped surface heat exchanger. J. Food Sci. 62,
(2002) 773–780 (1997)
8. T.A. Do, J.M. Hargreaves, B. Wolf, J. Hort, J.R. Mitchell, Impact 22. C. Loisel, G. Lecq, G. Keller, M. Ollivon, Dynamic crystallization
of particle size distribution on rheological and textural properties of dark chocolate as affected by temperature and lipid additives.
of chocolate models with reduced fat content. J. Food Sci. 72, J. Food Sci. 63, 73–79 (1998)
E541-E552 (2007) 23. P. Lonchampt, R.W. Hartel, Fat bloom in chocolate and compound
9. Y. Roos, Thermal analysis, state transitions and food quality. J. coatings. J. Lip. Sci. Technol. 106, 241–274 (2004)
Therm. Anal. Calorim. 7, 197–203 (2003) 24. A. Torbica, O. Jovanovic, B. Pajin, The advantages of solid fat
10. V. Glicerina, F. Balestra, M. Dalla Rosa, S. Romani, Rheological, content determination in cocoa butter and cocoa butter equivalents
textural and calorimetric modifications of dark chocolate during by the Karlshamns method. J. Food Res. Technol. 222, 385–391
process. J. Food Eng. 119, 173–179 (2013) (2006)
11. Y. Wang, T.D. Wig, J. Tang, L.M. Hallberg, Dielectric properties 25. G. Mongia, G.,G.R. Ziegler, The role of particle size distribution
of foods relevant to RF and microwave pasteurization and sterili- of suspended solids in defining the flow properties of milk choco-
zation. J. Food Eng. 57, 257–268 (2003) late. Int. J. Food Prop. 3, 137–147 (2000)
12. T.J. Laaksonen, Y.H. Roos, Thermal, dynamic-mechanical, and 26. M. Yanes, L. Durán, E. Costell, Rheological and optical proper-
dielectric analysis of phase and state transitions of frozen wheat ties of commercial chocolate milk beverages. J. Food Eng. 51,
doughs. J. Cereal Sci 32, 281–292 (2000) 229–234 (2002)
13. D. R. Lide ed., Permitivitty (dielectric constant) of gases in CRC 27. A. Sokmen, G. Gunes, Influence of some bulk sweeteners on
Handbook of Chemistry and Physics, 86th edn. (CRC Press, Boca rheological properties of chocolate. LWT Food Sci. Technol. 39,
Raton, 2005), ISBN 0-8493-0486-5 1053–1058 (2006)
14. P.A. Kilmartin, D.S. Reid, I. Samson, Dielectric properties of 28. E.M. Kiley, V.V. Yakovlev, K. Ishizaki, S. Vaucher, Applicability
frozen maltodextrin solutions with added NaCl across the glass study of classical and contemporary models for effective complex
transition. J. Sci. Food Agric. 84, 1277–1284 (2004) permittivity of metal powders. J. Microw. Power Electromagn.
15. J. Tan, W.L. Kerr, Determination of glass transitions in boiled Energy 46, 26–38 (2012)
candies by capacitance based thermal analysis (CTA) and genetic
algorithm (GA). J. Food Eng. 193, 68–75 (2017)

13

View publication stats

You might also like