Water behaviour in processed cheese spreads: DSC and ESEM study

Abstract

The behaviour of water in two processed cheese spreads, either standard or cream-enriched, was studied by differential scanning
calorimetry (DSC) in the −50 to 45°C temperature range as a function of controlled dehydration. The results were analyzed
and related to cheese microstructures observed by environmental scanning electron microscopy (ESEM). Water freezing and subsequent
ice melting were found to be dependent on the cheese composition in hydrophilic components and on water confinement within
the micro-domains delimited by the fat droplets. Both cheeses exhibited partial water freezing from supercooling state while
ice nucleation process was shown to be tightly affected by water connectivity within the cheese matrices.

Full text

Water behaviour in processed cheese spreads
DSC and ESEM study
Hela Gliguem ÆDorra Ghorbel ÆCe
´cile Grabielle-Madelmont Æ
Benoı
ˆt Goldschmidt ÆSylviane Lesieur Æ
Hamadi Attia ÆMichel Ollivon ÆPierre Lesieur
Special Chapter dedicated to the memory of dr. Michel Ollivon
ÓAkade
´miai Kiado
´, Budapest, Hungary 2009
Abstract The behaviour of water in two processed
cheese spreads, either standard or cream-enriched, was
studied by differential scanning calorimetry (DSC) in the
-50 to 45 °C temperature range as a function of controlled
dehydration. The results were analyzed and related to
cheese microstructures observed by environmental scan-
ning electron microscopy (ESEM). Water freezing and
subsequent ice melting were found to be dependent on the
cheese composition in hydrophilic components and on
water confinement within the micro-domains delimited by
the fat droplets. Both cheeses exhibited partial water
freezing from supercooling state while ice nucleation pro-
cess was shown to be tightly affected by water connectivity
within the cheese matrices.
Keywords Dehydration DSC ESEM Ice melting 
Oil-in-water emulsion Water crystallization
Introduction
Processed cheese spreads are oil-in-water (O/W) emulsions
in a gel state consisting in a proteolipidic network includ-
ing an aqueous phase which contains salts and dispersing
fat droplets. They are sold in most countries whatever the
local climate. Thanks to their preparation recipe and pro-
cess, they are considered to be stable products from a
microbiological point of view with a reasonable shelf life.
Processed cheese spreads are produced by blending, heat-
ing and vigorous stirring of a mixture of shredded natural
cheeses of different types and degrees of maturity with
chelating (or melting) salts and fat under partial vacuum
and constant stirring until a homogeneous mass is obtained
[1]. Modern processing technologies may also use various
other ingredients to reduce costs, provide flavor or texture,
or improve the shelf life. A second type of processed
cheese spread is based on a processed cream formula. In
this case, the natural cheeses are replaced by a fresh curd
mass and cream is added. Melting salts are also added to
the blend, but may be different of those used above.
During manufacture of processed cheese spreads, some
water is added to produce a smooth and stable emulsion
[2]. Water helps in hydrating the proteins and dispersing
the components. Water is also required to achieve certain
product attributes such as softness in cheese spreads. The
amount of water added differs according to whether the
product is a full-fat or a low-fat product. Commercial full-
fat processed cheese spreads have mass contents of mois-
ture and fat in the 40–60% range and of 16–20% at least,
respectively [3–5]. However, commercial low-fat pro-
cessed cheese spreads (10–24 mass% fat) have been found
to contain as much as 73 mass% moisture [6].
Until now, few studies have been carried out on char-
acterization of water properties in this kind of food
H. Gliguem C. Grabielle-Madelmont S. Lesieur M. Ollivon
Equipe Physico-Chimie des Syste
`mes Polyphase
´s, UMR 8612 du
CNRS, 5 rue J. B. Cle
´ment, 92296 Cha
ˆtenay-Malabry, France
D. Ghorbel H. Attia
Unite
´d’Analyses Alimentaires, ENIS, BPW, 3038 Sfax, Tunisia
D. Ghorbel
De
´partement de Ge
´nie Biologique et Chimique, INSAT,
Centre Urbain Nord, B.P. 676, 1080 Tunis, Tunisia
B. Goldschmidt
De
´partement Recherche Applique
´e, Groupe - Fromageries Bel,
7 bd de l’industrie, 41100 Vendo
ˆme, France
P. Lesieur (&)
Faculte
´des Sciences, Universite
´Henri Poincare
´, UMR 7565,
54506 Vandœuvre-le
`s-Nancy Cedex, France
e-mail: pierre.lesieur@uhp-nancy.fr
123
J Therm Anal Calorim (2009) 98:73–82
DOI 10.1007/s10973-009-0376-x

structure. Most investigations on exudation or syneresis
and methods for measuring water-holding capacity have
been carried out on curd [7,8]. More recently, Marches-
seau and Cuq [9] have studied the change in water-holding
capacity during processed cheese storage by using ultra-
centrifugation methods. However, no work has been
reported on thermal behaviour of water in processed cheese
previously described as a complex system [10].
According to food freezing literature, water may be
distinguished into two types with respect to its aptitude to
freeze. On one hand, the ‘‘free’’ or ‘‘bulk’’ water, also
called ‘‘freezable’’ water, freezes at few degrees below
0°C, due to a phenomenon of nucleation, but melts either
at 0 °C in the absence of soluble substances like salts or
below 0 °C in the opposite case. This behaviour is typical
of homogeneous hydrated systems, the thermotropic phase
transition between ice and liquid aqueous solutions being
indeed affected by salts. For instance, 20 mass% NaCl
shifts the water solid-to-liquid transition temperature down
to -20 °C. In the same way, the transition enthalpy vari-
ation is known to be lowered in the presence of soluble
solutes, typically from 6 kJ mol
-1
down to 4 kJ mol
-1
when the transition temperature drops from 0 °C down to
-40 °C[11]. On the other hand, water that remains
‘‘unfrozen’’ at temperatures below the equilibrium bulk
freezing temperature, i.e., in the presence of ice, is called
‘‘unfreezable’’ and referred to as ‘‘bound’’ water according
to Wolfe et al. [12]. However, this notion of bound water
versus that of freezable water must be considered with care
as suggested by Van der Sman and Boer [13]. Otherwise, it
has been demonstrated that some food materials can
undergo a glass transition so that a fraction of bulk water
may remain unfrozen since it would evolve towards an
amorphous glass state rather than crystallized ice [12,13].
Because of the frequent concomitance of these different
behaviours, the study of water state in processed cheese
including quantified identification is complex.
The main goal of the present study was to characterize
water state in processed cheese spreads in relation with the
microstructure. Few techniques allow both the discrimi-
nation and quantification of the different types of water in
materials. Among these techniques, Differential Scanning
Calorimetry (DSC) can directly measure ice melting/
freezing temperatures and enthalpies in foods and other
systems [14]. DSC easily provides evaluation of the
freezable water amount and that of water molecules which
are unable to crystallize due either to specific interaction
with the studied matrix or to other processes. Namely, the
amount of water that freezes during cooling of a food
sample is commonly and easily determined from the
melting enthalpy of ice formed under the experimental
conditions used [15,16], especially in the case of inhibited
or difficult water crystallization. Here, the aqueous phase
of two different cheese matrices obtained by varying
composition and manufacturing process was investigated
by monitoring the thermal events associated to the freezing
of water and melting of ice by DSC. This study was
complemented by Environmental Electron Scanning
Microscopy (ESEM) observations to visualize the organi-
zation as well as the distribution of fat and aqueous phase
within the emulsions constituting the processed cheese
spreads examined.
Materials and methods
Processed cheese products
Two different processed cheese spreads were provided by
the French industrial dairy company Fromageries Bel
(Vendo
ˆme, France) and selected as representative of two
important technologies differing by ingredients and com-
position of the final product. Cheese A is a processed
cheese spread with a 52 mass% of fat on dry matter basis
containing the following ingredients: cheese from cows’
milk, butter, skim milk, milk proteins and melting salts.
Cheese B is a processed cream cheese spread with a
68 mass% of fat on dry matter basis resulting from another
mixture of different ingredients (fresh lactic curd, cream,
milk proteins and emulsifiers). The determination of pH,
dry matter [17], fat content [18], total nitrogen as protein
content [19], lactose [20] and ash [21] contents for both
cheeses A and B were provided by the supplier from
analyses performed in triplicate (see data in Table 1). The
samples were stored at 4 °C prior to analyses. All analyses
were conducted on products aged between 1 and 2 months.
Table 1 Acidity and Chemical Composition (mass % of total wet matter) of the processed cheese spreads as a result of three independent series
of analysis
Cheese pH Total moisture Fat Protein Lactose Ash
A 5.52 ±0.01 56.7 ±0.1
5
22.5 ±0.2 10.5 ±0.1 6.80 ±0.07 3.70 ±0.01
B 5.40 ±0.01 52.7 ±0.2 32.2 ±0.1 9.0 ±0.1 2.10 ±0.01 2.70 ±0.01
74 H. Gliguem et al.
123

DSC measurements
Calorimetric experiments were carried out using a Perkin-
Elmer DSC-7 apparatus (St Quentin en Yvelines, France)
supported by Pyris Thermal Analyzing Systems ver 352.
Dried nitrogen and air were used to purge the thermal
analysis system (head and glove box, respectively) during
all the experiments.
Lauric acid (99.95% purity, melting temperature,
T
m
=43.7 °C; DH
m
=8.53 kcal mol
-1
) was used as
standard to calibrate the calorimeter for all cooling and
heating scans. The samples of both cheeses, in the range of
16.5–21.0 mg, were first hermetically sealed into alumi-
num pans (50 lL total volume, Perkin Elmer, France). The
whole analysis of a sample was carried out using the same
sample-holder and monitoring of water evaporation was
conducted as previously described [16,22–24]. The
decrease in water content of each sample throughout a
series of recordings was obtained thanks to a very small
hole (less than 0.2 mm diameter) made in the pan cover
after sealing by a special sharp pin prick [24]. Between two
sets of recordings and just after the heating scan, the
sample pan was kept in dry atmosphere (dried silica gel) at
room temperature over night to allow loss of water. The
hole in the sample cap was small enough to ensure that
water evaporation during the temperature ramp (from -50
to 45 °C) of a calorimetric scan was negligible compared to
that applied between two recordings. The samples were
precisely weighed between each set of recordings
(±10
-5
g). The slow dehydration process employed war-
rants homogeneous elimination of water from the samples.
An empty pan was used as a reference and a DSC
analysis of empty pans was recorded prior to the sample
measurements in the temperature range in which the ther-
mal events of interest are observed. All thermograms were
performed at a rate of ±2°C min
-1
. Before each recording
set, the samples were first cooled from 25 to -50 °C before
two successive heating and cooling runs from -50 to
45 °C and from 45 to -50 °C, respectively. The fat matter
was in the liquid state at 45 °C and crystallized upon
cooling so that the different thermal events corresponding
to lipid melting and crystallization were superimposed all
along the experiment. The reported data concerning water
crystallization correspond to those obtained during the first
cooling scan from 25 °C down to -50 °C. The water
sample (1 mg of Millipore ultra pure water) was scanned
once in the same conditions.
DSC sample analyses were performed in duplicate for
both cheeses with excellent repeatability, but only one
series of results corresponding to about 16 cycles will be
presented here.
The onset temperatures (T
Onset
) were taken at the
intersection of the baseline with the tangent to the left side
of the melting peak or to the right side of the freezing one.
The specific heats of water melting were obtained by
integrating the area under the peaks by using the DSC
apparatus software.
Lyophilization
Lyophilization was carried out as a separate experiment on
both types of samples prior to microscopy examination
with the aim to get freeze dried samples. Large cheese
samples (about 20 920 915 mm) were placed in a
freezer at -70 °C for 8 h, and then lyophilized in a freeze
dryer under vacuum for 12 h after initial weighing. The
level of residual moisture in lyophilized samples was
determined by mass differences.
Environmental scanning electron microscopy (ESEM)
Spread cheese samples (5 9593 mm) were examined
by using an Environmental Scanning Electron Microscope,
XL 30 type ESEM–Quanta (Philips/FEI), at beam voltage
of 15 kV and reduce pressure of 0.9 mm Hg. The detection
system was a Gaseous Secondary Electron Detector
(GSED). Both types of samples, wet and lyophilized ones,
were mounted onto stubs using a double-sided adhesive
tape. Samples were inserted into the microscopy chamber
at 4 °C. Only faces of sample not modified by cutting edge
were examined. The working distance between the samples
and the detector ranged from 8.8 to 11.3 mm. Air pockets,
fat droplets and water-rich areas were identified according
to the levels of grey (contrast level). Gas or void volumes
usually appear as black regions due to the lack of matter.
The fat matter areas are seen in more or less dark grey and
always lighter than trapped air. The aqueous phase is
characterized by significantly higher intensity.
Results
The composition characteristics of the two processed
cheeses studied and referred to as cheeses A and B are
reported by Table 1. As expected from initial component
mixture, the fat content in cheese B was markedly found
higher than in cheese A. While the global moisture of
cheese B slightly exceeded that of cheese A, protein, lactose
and ash levels appeared smaller. The difference in pH value
arose from the use of acid curd in cheese B preparation.
Thermotropic behaviour of water as a function
of cheese hydration level
Water crystallization and ice melting temperatures as well
as corresponding enthalpy variations were determined by
Water behaviour in processed cheese spreads 75
123

DSC from cheese A and B samples as a function of their
progressive dehydration. Each series of recording was
conducted on the same sample aliquot placed into a spe-
cially perforated capsule, the water content of which varied
from its initial value down to lower values until residual
water was no more able to crystallize. Water loss (W
L
) was
controlled by high precision weighing and expressed as
mass% of initial sample mass while hydration level was
related to the actual sample mass. The DSC profiles were
recorded in the total temperature range explored upon
heating scans or cooling scans. Whatever hydration levels
of the cheese samples, the lipids constituting the fat frac-
tion exhibited crystallization and fusion processes within
the 10–35 °C which were found fully superimposed
including onset temperatures, shape and surface area of the
thermal peaks (data not shown). This implies that thermal
events related to water state transitions did not depend on
specific lipid–water interactions as previously noticed [25,
26]. In the following, only the temperature interval corre-
sponding to water events was reported. Fat behavior being
out of the scoop of this study was recently published sep-
arately [27]. Figure 1shows representative thermograms of
cheese A. DSC cooling curves (Fig. 1a) at water losses
smaller than 30% by mass (38 mass% residual hydration)
exhibit a single narrow and exothermic peak occurring at
temperatures significantly below 0 °C. This depicts su-
percooling behaviour of the aqueous phase similar to that
observed for pure water (Fig. 1a, inset). According to such
a process, the liquid state is maintained at low temperatures
until it abruptly crystallizes due to homogenous crystal
nucleation. It can be noticed that lowering the water con-
tent of the cheese proportionally decreases the surface area
of the DSC signal as well as the crystallization onset
temperature which was found always below that of pure
water (T
Onset
=-17.2 °C). At hydration levels lower than
38%, the crystallization temperature continues to be shifted
down with decreasing hydration while peak broadening is
observed. At remaining 24.4% hydration and below, any
exothermic event was no more detected indicating that
water was not able to freeze upon the thermal treatment
imposed.
Heating curves (Fig. 1b) are characterized by a single
asymmetrical endothermic event which systematically
occurs over a range of subzero temperatures even for
W
L
=1.0% (T
Onset
=-5.8 °C). The peak widths are
larger than that for pure water (Fig. 1b, inset) and tem-
peratures characterizing the onset of the melting transition
are clearly below 0 °C. Moreover, both the T
Onset
values
and surface area of the endotherms are progressively
decreasing with decreasing water content. At the lowest
residual water level analyzed, a very weak endothermic
signal was noticed (Fig. 1b) at T
Onset
of -26 °C, whereas
no crystallization was recorded upon cooling. This suggests
the existence of a crystallization process too slow to be
observed at a cooling rate of -2°C min
-1
but fast enough
to allow nucleation at -50 °C just before recording heating
scan. Furthermore, samples containing lower water
amounts showed neither crystallization nor melting
processes.
The DSC cooling curves recorded from cheese B appear
very comparable to those obtained with cheese A (Fig. 2a),
except that no broadening was noticed even at 45.7 mass%
of water loss (13.2% of water content). The DSC heating
curves are characterized by asymmetrical peaks which are
decreased in surface area and shifted down lower temper-
atures with increasing dehydration (Fig. 2b) as similarly
described for cheese A. Below 13.2 mass% water, both
water crystallization and ice melting processes were van-
ished. It can be pointed out that, for a given hydration level,
water crystallization and ice melting temperatures in the
case of cheese B were found systematically higher than
those found in the case of cheese A.
x 5
x 5
-42 -34 -26 -18 -10 -40 -30 -20 -10 0 10
-50
Pure water
Heat flow (mW.g–1) Exo
Pure water
(a)
WL (%)
1.0
22.9
29.3
36.9
41.3
42.6
42.7 (24.4%)
(25.0%)
(26.3%)
(31.4%)
(38.9%)
(43.9%)
(56.7%)
WL (%)
1.0
22.9
29.3
36.9
41.3
42.6
42.7 (24.4%)
(25.0%)
(26.3%)
(31.4%)
(38.9%)
(43.9%)
(56.7%)
(b)
-50 -42 -34 -26
Temperature (°C)
Tem
p
erature/°C Tem
p
erature/°C
Temperature (°C)
-18 -10
-40 -30 -20 -10 0 10
T Onset
Fig. 1 DSC curves recorded
during cooling (a) and
subsequent heating (b)of
cheese A at 2 °C min
-1
as a
function of water loss (W
L
)
expressed in % of the total
sample mass. Numbers in
parentheses are hydration levels
of the samples (% by total
sample actual mass). Insets:
DSC curves of pure water
freezing (a) and pure ice
melting (b). Onset temperature
T
Onset
of ice melting were
determined as indicated in (b)
76 H. Gliguem et al.
123

Figure 3gathers the evolution of the experimental
crystallization and melting temperatures determined for
both cheeses as a function of residual water level. It is
worth noting that ice formation takes place at temperatures
significantly below the melting ones, confirming the su-
percooling behaviour of water. The temperatures charac-
terizing the solid-to-liquid state transition of water
gradually decrease with decreasing hydration level in a
close manner for both series of cheese samples. More
precisely, looking for instance at the melting temperatures,
these are first very slightly decreased with decreasing water
content down to 40 and 30 mass% for cheese A and B,
respectively. Beyond, successive dehydration steps lead to
increasingly significant lowering of the onset temperature.
The variation of the specific heat (DHin J g
-1
) asso-
ciated with the transition of ice to liquid aqueous solution
and measured upon ice melting was plotted in Fig. 4as a
function of the water mass remaining in the samples. As a
whole, DHvalues found for water in cheeses are signifi-
cantly lower than the enthalpy variation of 334.5 J g
-1
for
pure water melting in similar DSC recording conditions. At
a given water amount, the value measured for cheese A
was systematically below that found for cheese B. For both
cheeses and by decreasing hydration, the specific heat is
first slightly before abruptly decreasing, thus depicting a
change in water ability to freeze. Both parts of the DH
versus water curves can be rationally assimilated to straight
lines. The intercept of these lines gives an estimate of the
sample hydration threshold that delimits the two water
behaviors and found around 40 or 37 mass% of initial
water (32 or 28% mass of total sample) for cheese A or B.
In principle and as already experimented [24], the extrap-
olation to zero ordinate should give the amount of water
which is unable to freeze. Indeed, zero DHvalue indicates
that no more water could crystallize in the samples during
cooling so that no melting event could be observed. By
fitting the experimental points by an empiric polynomial
law and determining the intercept at zero ordinate, esti-
mates of the critical water contents were found nearly at 19
and 15% by mass of the initial water constituting cheeses A
and B, respectively. The impediment to crystallization may
be due to interactions with the hydrophilic components of
the cheeses, i.e., lactose, proteins and mineral salts (ashes)
the proportions of which are given in Table 1. Calculations
indeed lead to global hydrophilic substances amount of
4.3 mg in cheese A against only 2.0 mg for cheese B, in
agreement with the greater proportion of water that can not
freeze in cheese A.
Cheese microstructure characterization
Environmental scanning electron microscopy micrographs
of cheeses A and B in their initial hydration state are shown
in Fig. 5. Fat droplets (FD) are mainly seen as individual
Heat flow (mW. g–1) Exo
-50 -42 -34 -26 -18
Temperature/°C Temperature/°C
-10 -40
45.5
37.7
35.9
31.7
21.6
4.8 (50.3%)
(39.7%)
(30.7%)
(26.2%)
(24.0%)
(13.2%)
WL (%)
(a) (b)
45.5
37.7
35.9
31.7
21.6
4.8 (50.3%)
(39.7%)
(30.7%)
(26.2%)
(24.0%)
(13.2%)
WL (%)
-30 -20 -10 0 10
Fig. 2 DSC curves recorded
during cooling (a) and
subsequent heating (b)of
cheese B at 2 °C min
-1
as a
function of water loss (W
L
)
expressed in % of the total
sample mass. Numbers in
parentheses are hydration levels
of the samples (% by total
sample actual mass)
0
-8
-16
A
B
-24
T onset/°C
-32
-4001020304050
Hydration/mass%
60
Fig. 3 Plots of onset freezing (open symbols) and onset melting
(filled symbols) temperatures (T
Onset
;±0.8 °C precision) characteris-
tic of water contained in cheese A (diamond symbols) and cheese B
(square symbols) versus water fraction expressed as mass% of total
sample. Vertical lines indicate the hydration levels from which DSC
signal was no more detected
Water behaviour in processed cheese spreads 77
123

spheroids appearing as dark grey spots of variable size.
More precisely, in cheese A, the fat droplets typically do
not exceed 1 lm in mean diameter (Fig. 5a) while in
cheese B, they present a larger size distribution and tend to
form clusters (AFD, Fig. 5b, c) which can fuse into non-
globular fat domains (NGF) also called ‘‘free’’ fat domains
(Fig. 5c). These last are no more protected by the milk
original fat globule membrane and correspond to the ulti-
mate step of emulsion destabilization [28].
According to these observations, cheese A shows a
degree of fat emulsification higher than that of cheese B.
For both matrices, the fat droplets are dispersed and
entrapped in a protein network revealed as a light grey
background surrounding the fat globules and mainly
composed of caseins [29]. The proteins are found close or
within the aqueous phase domain that also contains water-
soluble salts and lactose (Table 1) and do not totally seem
to be uniformly distributed but partly aggregated at a
micrometer scale as depicted by the presence of dark small
grains beside areas almost white (Fig. 5a, b). The water-
rich fraction then appears unlike a homogenous aqueous
phase. Moreover, the very light domains can be attributed
to water-rich areas which then are clearly localized as
larger surface domains in cheese B than seen in cheese A.
Micrographs shown in Fig. 6exemplify the microstruc-
ture of cheeses A and B after freeze-drying and then con-
taining residual water contents of 7.1 and 3.5% by mass,
respectively. It is worth noting that this dehydration pro-
cedure avoided sample heating. According to DSC analysis
performed on both lyophilized samples (data not shown), no
water crystallization and subsequent fusion could be
detected upon repeated cooling and heating scans in the
-50 to 40 °C range. For both samples, dehydration leads to
the formation of cracks similar in shape but larger in the
case of cheese B (Fig. 6a, c), with lengths not exceeding a
100 lm for cheese A and reaching several hundreds of
micrometers for cheese B. The structures of the cheese
matrices seen at higher magnification (Fig. 6b, d) closely
resemble those characterizing the hydrated samples
320
240
160
80
00 10203040
Water/mass%
Enthalpy variation /J g–1
50 60 70 80 90 100
Fig. 4 Enthalpy variation (DHin Joules per gram of water, ±8
to ±13 J g
-1
precision) measured upon ice melting as a function of
water expressed as the percentage of initial water mass for cheese A
(diamond symbols) and B (square symbols). Dashed curves corre-
spond to the best polynomial fitting of the experimental points
(R=0.99)
Fig. 5 Scanning electron micrographs of cheeses A (a) and B (b,c).
FD fat droplets, Aair pockets, PA protein aggregates, AFD
aggregated fat droplets, NGF non globular fat. band cmicrographs
visualize different zones of the same sample
Fig. 6 Scanning electron micrographs of cheeses A (a,b) and B (c,
d). FD fat droplets, PA protein aggregates, CL clusters of aggregated
fat globule. Water contents are 7.1 mass% (a,b) and 3.5 mass% (c,d)
78 H. Gliguem et al.
123

(Fig. 5a, b) except that the surface density of the fat droplets
tends to increase in both cheeses and more fat coalescence is
observed in cheese B. In summary, before dehydration, fat
globules have a smaller size in cheese A while after dehy-
dration they appear more coalesced in cheese B.
Discussion
According to above-described chemical analysis and
microscopy results, the two processed cheese spreads here
studied present a commonly observed matter organization
consisting in emulsion fat droplets distributed within an
aqueous phase including proteins, lactose and mineral salts.
Precise characterization of the aqueous compartment can
be achieved through the continuous exploration of water
thermal behaviour as a function of controlled cheese dehy-
dration. This has been indeed successfully demonstrated
for the study of surfactant-water or polymer-water systems
[23,24].
From the structural point of view, ESEM investigation
reveals significant differences between cheeses A and B.
Both of them show fat droplets embedded in a hydrated
matrix rich in proteins. On the basis of previous reports
[4,29,30], proteins in milk are principally caseins which
localize at the fat/water interface or form individual
micelles, and serum proteins which are water-soluble and
then mixed with lactose and salts. The fat droplets are
surrounded by protein shells directly in contact with the
bulk aqueous solution. From images of native samples seen
in Fig. 5, cheese A presents particles with smaller sizes and
lower polydispersity than those observed for cheese B. It
can be reasonably predicted that the network of the water
molecules extends into micro-domains with volumes of
local dimensions lower in cheese A than in cheese B.
Moreover, the latter is partially composed of non-dispersed
fat then with the risk of water demixing. Upon gentle
freeze-drying that preserves the overall fat organization,
Fig. 6clearly indicates that water loss induces tighter
packing of the fat droplets without drastically changing
their number and size. As a consequence, cheese dehy-
dration should essentially act on the extent of the aqueous
spaces and therefore on the local concentration not only of
proteins but also of lactose and mineral salts. In this way,
smaller and rather calibrated droplets like those of cheese
A should delimit numerous small pockets of aqueous phase
which could be separated from each other by eliminating
water. In the case of cheese B, the presence of larger
globules should lead to wider aqueous areas and may shift
the occurrence of restricted domain formation. When
cooling, ice crystal growth from a seed can drive larger
amounts of surrounding water in cheese B than in cheese
A, in good agreement with the distribution of the cracks
visualized by ESEM after ice sublimation process. Indeed,
these cracks are provoked by sample lyophilization and
cannot only coincide with ice localization.
The DSC results provide further information on the
aqueous phase. In the operating conditions used, pure water
underwent a typical supercooling process. Ice crystalliza-
tion was indeed observed at -18 °C, i.e., well below the
standard temperature of 0 °C, and characterized by the
vertical right side and narrowness of the thermal peak. As
expected, the melting of the ice crystals in equilibrium with
pure liquid water was observed at 0 °C[11]. Both cheeses
similarly showed supercooling phenomenon and ice crys-
tallization was shifted at subzero temperatures which were
all the lower as the water content is decreased and far
below the crystallization temperature observed for pure
water. The downshift of ice freezing temperature compared
to ice melting one is also an evidence of supercooling. On
one hand, this behaviour suggests that within the cheeses,
ice forms according to a mechanism of homogenous
nucleation, a seed crystal being required to allow the
crystal structure to develop in the whole aqueous com-
partment. On the other hand, the fact that the depressions of
the freezing point as well as that of the melting point
depend on the hydration level or, more precisely, on the
solute concentration, refers to colligative properties of
water. This is a direct consequence of the Raoult law which
predicts that the addition of a liquid-soluble but solid-
insoluble solute to water only lowers the chemical potential
of the solvent in the liquid solution. Thus, the temperature
of the solid to liquid solution transition is decreased.
Consequently, the melting point of the solid in the presence
of liquid supplemented by a solute is less than that of the
pure compound. Moreover, at-equilibrium freezing point
depression is proportional to the solute concentration, at
low ones at least. The cheeses here studied contain pro-
teins, lactose and salts which can be, at least partly, dis-
solved in water so that on cooling the constitutive aqueous
solutions do not freeze at 0 °C but at lower temperatures
which fit the melting point line referred to as liquidus line
in the solute-water phase diagram. Ice formation consumes
pure water and results in a more concentrated solution
that will freeze at an even lower temperature. In the
absence of supercooling, melting and freezing pathways
are undistinguishable and crystallization should appear as a
non-isothermal transition until complete solidification of
water. This generally takes place through a eutectic reac-
tion at a very low temperature. The shape of the temper-
ature variation seen in Fig. 3strongly suggest the existence
of such an eutectic behaviour which should occur below
-50 °C in agreement with the state diagram of water milk
previously established [31]. As both cheeses exhibit super-
cooling, DSC cooling recordings under-estimate the
effective transition temperature and yield exothermic
Water behaviour in processed cheese spreads 79
123

events out of equilibrium. The crystallization exotherms
remain narrow as if isothermal transitions were present
although this is due to the sudden transformation of the
metastable liquid into ice. Only the heating pathway, which
is not affected by any nucleation process, must be con-
sidered to closely describe the state diagram of the aqueous
fraction of the cheeses. This last clearly shows expected
non-isothermal melting since the peak width significantly
exceeds that observed for pure water. In this respect, both
cheeses behave similarly and present eutectic-like behav-
iour, the limit of the liquidus line at low water content
being symbolized by the vertical dashed lines in Fig. 3.
The shift between the two cheese samples can be first
explained by the different amount of hydrophilic sub-
stances they contain which is almost twice higher for
cheese A (Table 1). The influence of the proportions of the
hydrophilic substances, namely the lactose proportion, and
possibly of the nature of the constitutive proteins and/or
mineral salts cannot be either excluded.
By reconsidering more carefully the shape of the exo-
thermic peaks characteristic of ice formation, those at low
water contents deserve attention. They indeed reveal a dis-
crepancy between cheeses A and B. As exemplified by
Fig. 7, cheese A shows progressive water freezing along a
rather wide temperature range as if crystals would not
cooperatively grow from initial nuclei what is exactly the
case for cheese B. By taking into account that water in both
cheeses undergo supercooling and in agreement with ESEM
analysis, a reasonable explanation could be that the decrease
in the distance between fat droplets upon water elimination
creates, in cheese A, aqueous zones of likely variable con-
centrations and/or compositions which do not communicate
anymore and then behave independently. Each small
domain then would freeze from its own supercooled state at
a temperature different from the other ones, however, close
enough to be not discriminated by the DSC apparatus.
Moreover, water crystallization in cheese A is system-
atically delayed compared to cheese B. This may be again
related to the existence of the more reduced water spaces
within the matrix of cheese A, in agreement with previous
observations concerning water-in-oil emulsions, for which
the smaller the size of water droplets and the higher their
salt concentration, the lower the freezing temperature
[29,32]. Another superimposed phenomenon may be the
reduction of ice crystal growth in cheese A, due to very
small volumes of liquid water available in the vicinity of
the nuclei [33].
The evolution of the specific heat measured during
heating DSC scans and associated with ice melting shows
two-step behaviour of water in both cheeses depending on
the hydration level. The overall DHvalue per gram of
water remains lower than 240 or 290 J g
-1
for cheese A or
B, respectively. These values are plainly below the
enthalpy variation of ice melting for pure water. It is worth
noting that melting heat of 334 J g
-1
corresponds to the
ice-to-liquid water transition occurring at 0 °C so that it
could be expected lower at temperatures significantly
colder due to differences in the heat capacities of ice (close
to 2 J g
-1
K
-1
) and liquid water (close to 4 J g
-1
K
-1
)
versus temperature. Another factor is the presence of the
water-soluble compounds lactose, proteins and mineral
salts. Indeed, upon heating, the energy exchange by water
in the cheeses is the sum of two contributions: the energy
required by the ice-to-liquid transition and the energy
corresponding to the solubilisation of the hydrophilic
substances into the liquid water. Ice melting is an endo-
thermic event (positive enthalpy variation) and in the
hypothesis of pure ice crystallization, it should be equal
to that of pure water (334 J g
-1
) and the related heat
exchange directly proportional to the water mass. Solute
solubilisation is spontaneous and probably contributes to
lower overall enthalpy variation for both cheeses. Pres-
ently, cheese A which contains more soluble species
exhibits lower enthalpy changes. This is in favour of an
effect associated to the solubilisation of lactose, proteins
and salts. Solute solubilization corresponds to negative free
energy variation. However, depending on the nature of the
solutes, the event is either exothermic (negative enthalpy
variation) or endothermic (positive enthalpy variation). In
the latter case the enthalpy variation is compensated by
entropy variation. Whatever the solubilisation behaviour,
the exchanged heat is proportional to the solute concen-
tration and then varies with the water content. The fact
that the specific melting heat was found below that of
pure water and progressively decreased with decreasing
water mass suggests that solute solubilisation may be
exothermic rather than endothermic. An alternative
explanation may be partial water freezing upon sample
cooling so that the recovered melting enthalpy variation
-45 -40 -35 -30
Heat flow (mW g–1) Endo
-25
A
B
-20
Tem
p
erature/°C
Fig. 7 DSC curves recorded at 2 °C min
-1
during cooling of cheeses
Aand Bcontaining 26.3 and 26.2% water by mass of total sample,
respectively
80 H. Gliguem et al.
123

reduced to the total water mass may be found systemat-
ically lower than that expected if all the water would have
crystallized.
At high dehydration, below around 30 mass% total
sample (from around 5 mg and 3 mg water in cheese A and
B, respectively), the significant decrease in enthalpy vari-
ation measured during cheese heating process confirms that
part of the constitutive water molecules did not crystallize
upon the preceding cooling down to -50 °C. The critical
water contents deduced from extrapolation to zero enthalpy
(Fig. 4), reduced to the hydrophilic components of cheeses,
amount to very close mass fractions of either 34% (cheese
A) or 36% (cheese B), i.e., a quite similar solute mass
fraction of 65%. Logically this composition can be
assimilated to that of the expected eutectic-like mixture
corresponding to the coexistence between liquid aqueous
solution, ice and a glassy solid state as supported by pre-
vious works [12,31]. While ice is, theoretically and at
equilibrium, constituted by pure water, the glass state is
issued from a viscous aqueous liquid composed of water
and hydrophilic species. Upon dehydration strong solute–
water interactions develop to the detriment of water–water
interactions and hinder water crystallization. Another
possible reason is the increase in viscosity catalyzed by
dehydration and cooling which impedes the diffusion and
gathering of the water molecules required for the freezing
process. It should be pointed out that in the present study,
no glass transition could be recorded during either cooling
or heating DSC measurements. This may be explained first
by a very low temperature of such a glass transition that
have been already found either at -80 °C in cheese curds
[34] or in the -64 to -69 °C range for fresh cheeses [35].
Then the glassy solid formation in cheeses A or B would
not occur at -50 °C. The fact would be that the water
molecules ‘‘bounded’’ to the hydrophilic substances would
remain in a liquid-like state unable to freeze. The second
explanation may be related to the low glass-to-liquid
relaxation enthalpy [34] that would be not measurable by
the DSC apparatus we used. From a practical point of view,
these results as a whole show that the moisture in both
cheeses splits into two water states: free water molecules
which act as a solvent of part of the hydrophilic solutes
(proteins, lactose and mineral salts) and hydration water
which tightly interacts with the hydrophilic solutes to form
hydrate-like structures. At high hydrations, free water
freezes at relatively moderate temperatures then at mod-
erate viscosities and ice formation, which is energetically
favorable, manages to draw the hydration water so that
most or all water crystallizes. At low content, water
becomes more confined and the viscosity increases so that
free water molecules hardly diffuse. Freezing is delayed
and the hydration water is retained in the vicinity of its
neighboring molecules and cannot organize to produce ice.
This confinement effect is especially ascertained for cheese
A. Moreover, ESEM experiments performed after sample
heating for different periods of time (data not shown)
showed that while cheese B underwent partial coalescence
of the fat droplets, cheese A was not significantly modified.
Trend of cheese B to demixing, at the opposite of cheese A
that conserves its microstructure, strengthens thermal
behavior interpretation in terms of fragmentation of the
aqueous phase and network tortuosity.
Conclusions
Very little information about water behavior in processed
cheese spreads had been previously reported likely due to
the complexity of their composition and structural organi-
zation. This study is a first attempt to characterize the
moisten fraction of these food materials by combining
ESEM and DSC. Despite rather middle resolution, ESEM
technique revealed convenient to describe the microstruc-
ture of cheese matrices in their native hydration state. The
images clearly reported a dispersion of fat globules dis-
tributed within an aqueous network composed of partially
crystallized proteins. DSC analysis on samples submitted
to gentle and progressive dehydration was particularly
informative. It highlighted that the moisten continuum
corresponds to an aqueous solution of the cheeses hydro-
philic components, i.e., lactose, salts and proteins, a part of
the water molecules being unable to freeze even at -50 °C.
The thermal behavior of water was sensitive to the con-
nectivity of the aqueous network and permits to distinguish
cheese emulsions differing by the size distribution of the
fat globules. As indeed confirmed by direct ESEM visu-
alization, the more significant the delay of water freezing
and the less cooperative the nucleation-growth process of
ice crystals, the smaller the fat droplets and the more
developed the array of water channels apt to form isolated
micro-domains. At this stage of interpretation, it would be
interesting to complete the structural approach by NMR
analyses very recently reviewed as a powerful tool to
deepen food microstructure [36]. This will be the subject of
a further work. DSC coupled to controlled dehydration thus
provides an interesting and simply implementable method
to identify the micro-organization of processed cheese
spread, mainly implicated in the textural properties of these
products.
Acknowledgements We gratefully acknowledge Mr. Abdessalem
Harrabi and Ms. Fatma Masmoudi for assistance in the use of ESEM
technique (service commun d’analyses de l’INRST, Borj Cedria,
Tunisia).
Water behaviour in processed cheese spreads 81
123

References
1. Paquet D. Processed cheeses: physico-chemical aspects. In:
Lorient D, Colas B, Le Mestre M, editors. Functional properties
of food macromolecules. Les cahiers de l’ENSBANA. Paris:
Technique & Documentation Lavoisier; 1988. p. 227–41.
2. Berger W, Klostermeyer H, Merkenich K, Uhlmann G. Processed
cheese manufacture. Ladenburg: BK Ladenburg GmbH; 1993.
3. Chang PK. Process cheese containing a modified whey solids. US
Patent 4166142 1979.
4. Caric M, Kalab M. Processed cheese products. In: Fox PF, editor.
Cheese: chemistry, physics and microbiology, vol. 2. London:
Chapman & Hall; 1993. p. 467–505.
5. Hockenberry TM, Evanston AJM (1994) Method for continuous
manufacture of process cheese-type products, US Patent 5350595.
6. Lee SK, Klostermeyer H. The effect of pH on the rheological
properties of reduced-fat model processed cheese spreads. LWT-
Food Sci Technol. 2001;34:288–92.
7. Marshall RJ. An improved method for measurement of the syn-
eresis of curd formed by rennet action on milk. J Dairy Res. 1982;
49:329–36.
8. Kocher PN, Foegeding EA. Microcentrifuge-based method for
measuring water-holding of protein gels. J Food Sci. 1993;58:
1040–6.
9. Marchesseau S, Cuq JL. Water-holding capacity and character-
ization of protein interactions in processed cheese. J Dairy Res.
1995;62:479–89.
10. Meyer A. Processed cheese manufacture. London: Food Trade
Press; 1973.
11. Franks F. The properties of aqueous solutions at subzero tem-
peratures. In Franks F editor. Water, a comprehensive treatise.
New York: Plenum Press; 1982. p. 215–338.
12. Wolfe J, Bryant G, Koster KL. What is ’unfreezable water’, how
unfreezable is it, and how much is there? CryoLetters. 2002;
23:157–66.
13. Van der Sman RGM, Boer E. Predicting the initial freezing point
and water activity of meat products from composition data.
J Food Eng. 2005;66:469–75.
14. Farkas J, Mohacsi-Farkas C. Application of differential scanning
calorimetry in food research and food quality assurance. J Therm
Anal Calorim. 1996;47:1787–803.
15. Pouchly J, Biros J, Benes S. Heat capacities of water swollen
hydrophilic polymers above and below 0 °C. Makromolekulare
Chemie. 1979;180:745–60.
16. Ollivon M. Calorimetric and thermodielectrical measurements of
water interactions with some food materials. In: Levine H, Slade
L, editors. Water relationships in foods. Advances in the 1980s
and trends for the 1990s. New York & London: Plenum Press;
1991. p. 175–89.
17. ISO 5534. Fromages et fromages fondus—De
´termination de la
teneur totale en matie
`re se
`che (Me
´thode de re
´fe
´rence), Interna-
tional Organization of Standardization; 2004.
18. ISO 3433. Fromages—De
´termination de la teneur en matie
`re
grasse—Me
´thode acido-butyrome
´trique, International Organiza-
tion of Standardization; 2002.
19. IDF-FIL 20-1. Lait. De
´termination de la teneur en azote. Partie 1:
Me
´thode Kjeldahl, International Dairy Federation; 2001.
20. ISO 22662. Lait et produits laitiers De
´termination de la teneur
en lactose par chromatographie liquide haute performance.
Me
´thode de reference, International Organization of Standard-
ization; 2007.
21. NF V04-208. Lait- De
´termination des cendres. Me
´thode de
re
´fe
´rence. French Association of Normalization; 1989.
22. Seras M, Courtois B, Quinquenet S, Ollivon M. Measurement of
the complex permittivity of dielectrics during microwave heating:
study of flours and starches. In: Jowitt R, Escher F, Kent M,
McKenna B, Roques M, editors. Physical properties of foods-2,
New York & London; 1987. p. 217–23.
23. Kekicheff P, Grabielle-Madelmont C, Ollivon M. Phase diagram
of sodium dodecyl sulfate-water system: 1. A calorimetric study.
J Colloid Interface Sci. 1989;131:112–32.
24. Cansell F, Grabielle-Madelmont C, Ollivon M. Characterization
of the aqueous phase and the water-polymer interface in latex
suspensions by differential scanning calorimetry. J Colloid
Interface Sci. 1991;144:1–17.
25. Bazmi A, Relikn P. Thermal transitions and fat droplet stability in
ice-cream mixmodel systems: effect of milk fat fractions. J
Therm Anal Calorim. 2006;84:99–104.
26. Thanasukarn P, Pongsawatmanit R, McLement D. Impact of fat
and water crystallization on the stability of hydrogenated palm
oil-in-water emulsions stabilized by whey protein isolate. J Col-
loids Surf A. 2009;246:49–59.
27. Gliguem H, Ghorbel D, Lopez C, Michon C, Ollivon M,
Lesieur P. Crystallization and polymorphism of triacylglycerols
contribute to the rheological properties of processed cheese.
J Agr Food Chem. 2009;18:3195–203.
28. Lopez C. Focus on the supramolecular structure of milk fat in
dairy products. Reprod Nutr Dev. 2005;45:497–511.
29. Fox PF, Guinee TP, Cogan TM, McSweeney PH. Fundamentals
of cheese science. Gaithersburg, Maryland: ASPEN Publishers;
2000.
30. Marchesseau S, Gastaldi E, Lagaude A, Cuq JL. Influence of pH
on protein interactions and microstructure of process cheese.
J Dairy Sci. 1997;80:1483–9.
31. Vuataz G. The phase diagram of milk: a new tool for optimising
the drying process. Lait. 2002;82:485–500.
32. Clausse D, Gomez F, Pezron I, Komunjer L, Dalmazzone C.
Morphology characterization of emulsions by differential scan-
ning calorimetry. Adv Colloid Interface Sci. 2005;117:59–74.
33. Aguerd M, Babin L, Clausse D. Contribution a
`l’e
´tude de la
nucle
´ation spontane
´e et induite des phases aqueuses disperse
´es.
J Rech Atmos. 1984;18:131–4.
34. Be
´al C, Fonseca F, Thomas A, Marin M. Caracte
´risation et lyo-
philisation de matrices fromage
`res. Sci Alim. 2006;26:89–102.
35. Andersen AB, Hansen E, Jorgensen U, Skibsted LH. Glass
transition in frozen fresh cheese. Milchwissenschaft-Milk Sci Int.
2001;56:441–3.
36. Mariette F. Investigations of food colloids by NMR and MRI.
Curr Opin Coll Int Sci. 2009;14:203–11.
82 H. Gliguem et al.
123