Journal of Food Engineering 85 (2008) 73–83

www.elsevier.com/locate/jfoodeng

Water sorption isotherms and glass transition temperature

of spray dried tomato pulp
Athanasia M. Goula a,*, Thodoris D. Karapantsios a, Dimitris S. Achilias b,
Konstantinos G. Adamopoulos c
a
Department of Chemical Technology, School of Chemistry, Aristotle University, University Box 116, 541 24 Thessaloniki, Greece
b
Laboratory of Organic Chemical Technology, School of Chemistry, Aristotle University, 541 24 Thessaloniki, Greece
c
Department of Chemical Engineering, School of Engineering, Aristotle University, 541 24 Thessaloniki, Greece

Received 12 May 2006; received in revised form 2 July 2007; accepted 4 July 2007
Available online 2 August 2007

Abstract

Adsorption isotherms of tomato pulp spray dried in dehumidified air were determined at six temperatures between 20 and 70 °C using
a gravimetric technique. The data obtained were fitted to several models including two-parameter (BET, Halsey, Smith and Oswin),
three-parameter (GAB), and four-parameter (Peleg) relationships. A non-linear least square regression analysis was used to evaluate
the models constants. The GAB model best fitted the experimental data. The isosteric heat of sorption was determined from the equi-
librium adsorption data using the Clausius–Clapeyron equation. Isosteric heats of sorption were found to decrease exponentially with
increasing moisture content. The enthalpy–entropy compensation theory was applied to the sorption isotherms and indicated an
enthalpy controlled sorption process. Glass transition temperatures (Tg) of spray dried tomato pulp conditioned at various water activ-
ities were determined by differential scanning calorimetry. A strong plasticizing effect of water on Tg was found with a large reduction of
Tg when the moisture content increased. These data were satisfactory correlated by the Gordon and Taylor model.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Glass transition; Isosteric heat; Sorption isotherm; Spray drying; Thermodynamic properties; Tomato powder

1. Introduction mono-layer of water (BET model) (Brunauer, Emmett, &

Teller, 1938), kinetic models based on a multi-layer and
For half a century scientists have realized that water condensed film (e.g. GAB model, Van den Berg & Bruin,
activity (aw) is more important to the quality and stability 1981), semi-empirical (e.g. Halsey model, Halsey, 1948)
of foods than the total amount of water present. The water and purely empirical models (e.g. Oswin and Smith models,
sorption isotherm is an important tool, especially in low Oswin, 1946; Smith, 1947).
moisture foods. It can be applied to optimize the drying The thermodynamic properties of foods provide an
or rehydration conditions and determine the stability of understanding of water properties and energy requirements
the product during storage. It is common to present sorp- associated with the sorption behavior. Differential heat of
tion isotherms by mathematical models based on empirical sorption, often referred as isosteric heat of sorption, is used
and/or theoretical criteria. In the literature there is a long as an indicator of the state of water adsorbed by the solid
list of available isotherm models, which can be divided into particles and its knowledge is important when designing
several categories; kinetic models based on an absorbed equipment for dehydration processes. The differential
entropy of a material is proportional to the number of
available sorption sites at a specific energy level, whereas
*
Corresponding author. Tel.: +30 2310 997798; fax: +30 2310 997759. Gibbs free energy is indicative of the affinity of sorbents
E-mail addresses: [email protected], [email protected] (A.M. Goula). for water and provides a criterion whether water sorption

0260-8774/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2007.07.015
74 A.M. Goula et al. / Journal of Food Engineering 85 (2008) 73–83

occurs as a spontaneous process (Telis, Gabas, Menegalli, ture including: potato, carrot, tomato, green pepper, and
& Telis-Romero, 2000). onion slices (Kiranoudis, Maroulis, Tsami, & Marinos-Kou-
Recently, the concepts related to water activity have ris, 1993), starch powders (Al-Muhtaseb, McMinn, &
been enriched by those of glass transition temperature Magee, 2004b), cereal grains (Tolaba, Suarez, & Viollaz,
(Tg), thus providing an integrated approach to the role of 1997), cassava and melon seeds (Aviara & Ajibola, 2002),
water in food. Tg defines a second-order phase change garlic (Madamba, Driscoll, & Buckle, 1996), rice (Benado
temperature at which a solid ‘‘glass” is transformed to a & Rizvi, 1985), and sugar beet root (Iglesias, Chirife, & Viol-
liquid-like ‘‘rubber”. As the temperature increases above laz, 1976). However, moisture sorption and thermodynamics
Tg various changes, such as increase of free volume, data for spray dried tomato pulp were not found in the liter-
decrease of viscosity, increase of specific heat, and increase ature. As far as glass transition temperature is concerned,
of thermal expansion, are noticed. The most important state diagrams have been reported for apple slice (Bai, Rah-
changes affecting food behavior are related to the exponen- man, Perera, Smith, & Melton, 2001), grape slice (Sa &
tial increase of molecular mobility and decrease of viscos- Sereno, 1994), strawberry (Moraga, Martinez-Navarrete,
ity. These factors govern various time-dependent and & Chiralt, 2004), freeze-dried pineapple (Telis & Sobral,
often viscosity related structural transformations, such as 2001), mango powder (Ayala, Walter, Martinez-Monzo,
stickiness, collapse, and crystallization during food pro- Fito, & Chiralt, 2002), freeze-dried persimmon (Sobral, Tel-
cessing and storage. The importance of Tg of amorphous is, Habitante, & Sereno, 2001), freeze-dried kiwifruit
food materials for processing and storage stability has been (Moraga et al., 2006), carrot (Georget, Smith, & Waldron,
recognized and emphasized by many researchers and 1999), freeze-dried onion (Sa & Sereno, 1994), freeze-dried
a wide range of potential food applications of the glass tomato cubes, and air-dried tomato halves (Telis & Sobral,
transition phenomenon have been identified (Matveev, 2002), but no data were found for spray dried tomato pulp.
Grinberg, & Tolstoguzov, 2000; Moraga, Martinez- Thus, the objective of this work was to provide experi-
Navarrete, & Chiralt, 2006; Roos, Karel, & Kokini, 1996). mental data for the sorption characteristics of spray dried
Fresh tomato and tomato products have aroused new tomato pulp in order to model the sorption isotherms using
scientific interest in the last few years. Among the various selected equations and to define the glass transition temper-
tomato products, powder enjoys a specialist market. ature, water activity and water content relationships.
Tomato solids in powder form have many advantages,
including ease of packing, transportation, and mixing. 2. Materials and methods
Tomato powder may be produced by various means but,
due to the high proportion of pulpy material in tomato 2.1. Spray drying of tomato pulp
juice, vacuum drying methods are not highly successful
and drum drying and various spray drying techniques are A Buchi mini spray dryer (Model 191) was employed for
used to produce the most powder at the present time. How- the spray drying process. A peristaltic pump pumped tomato
ever, tomato pulp is a product that is difficult to spray dry, pulp to the atomizer, and atomization was performed using a
as the powder tends to become soft and sticky while still two-fluid nozzle (inside diameter 0.5 mm), which used com-
warm, and poses a caking problem due to its hygroscopic pressed air. The modification made on the original design
and thermoplastic behavior (Goula & Adamopoulos, consisted of connecting the spray dryer inlet air intake nipple
2003). In previous work, an experimental spray dryer was to an air drying unit by a flexible plastic air duct. The com-
modified for drying tomato concentrate. The modification pressed air was also dehumidified before its supply to the
made to the original dryer design consisted of connecting two-fluid nozzle. An Ultrapac 2000 adsorption dryer (Model
the spray dryer inlet air intake to an absorption air dryer. 0005) with two desiccant cartridges was used to dry the air
The modified spray drying system proved advantageous down to 0.01 g of water per kg of dry air. The atomizer pres-
over the standard laboratory spray dryer. Preliminary air sure, the feed temperature, and the feed rate were
dehumidification improved not only product recovery, 5 ± 0.1 bar, 32.0 ± 0.5 °C, and 1.75 ± 0.05 g/min, respec-
but, also product properties (Goula & Adamopoulos, tively, whereas the feed was medium concentrated tomato
2005a; Goula & Adamopoulos, 2005b). pulp with a constant total solids mass concentration of
Both water activity and glass transition have been used 14 ± 0.05%, containing 1.40 ± 0.02% insoluble solids,
extensively in the literature to evaluate the storage stability. 5.61 ± 0.07% sugars, 1.53 ± 0.03% acid, 2.20 ± 0.02% pro-
According to Sablani, Kasapis, and Rahman (2007), a prod- tein, and 1.10 ± 0.01% salt. Tomato pulp was spray dried
uct is most stable at its monolayer moisture content, i.e. a with an air inlet temperature of 130 °C (±1 °C), drying air
water activity value of about 0.1–0.3, or at or below the cor- flow rate of 22.75 m3/h (±0.18 m3/h), and atomizing agent
responding glass transition temperature. Moisture sorption flow rate of 600 l/h (±20 l/h).
data have been reported for several foods (Akanbi, Adeyemi,
& Ojo, 2006; Chen & Jayas, 1998; Kim, Song, & Yam, 1991; 2.2. Determination of sorption isotherms
Maroulis, Tsami, Marinos-Kouris, & Saravacos, 1988;
McLaughlin & Magee, 1998). In addition, thermodynamic The equilibrium moisture content of tomato powder was
properties data for various foodstuffs are abundant in litera- determined by a gravimetric technique at seven relative
 A.M. Goula et al. / Journal of Food Engineering 85 (2008) 73–83 75

humidities selected between 1% and 90% and at six temper- Table 1

atures between 20 and 70 °C. Sulphuric acid solutions were Isotherm models used for experimental data fitting
used to maintain the specified relative humidity inside the Model Mathematical expression
desiccators (Ruegg, 1980). GAB (Van den Berg & Bruin, X = XmCKaw/[(1  Kaw) (1)
Tomato powder samples (3 ± 0.001 g) were placed in 1981) (1  Kaw + CKaw)]
previously weighed aluminum dishes and dried at 45 °C BET (Brunauer et al., 1938) X = XmCaw/[(1  aw) (2)
(1 + (C  1)aw)]
in an air-circulated oven over silica gel for 3 days. The sam- Halsey (Halsey, 1948) X = [A/(Tlnaw)]1/B (3)
ples were subsequently kept in desiccators over sulphuric Smith (Smith, 1947) X = A + Blog(1  aw) (4)
acid solutions of known relative humidity. A test tube con- Oswin (Oswin, 1946) X = A[aw/(1  aw)]B (5)
taining thymol was placed inside the desiccators with high Peleg (Peleg, 1993) X ¼ K 1 anw1 þ K 2 anw2 (6)
RH to prevent mold growth during storage. The desicca- Nomenclature: X: moisture content (% dry basis); Xm: monolayer value (%
tors were placed in temperature-controlled cabinets main- dry basis); T: temperature (K); C, K, A, B, K1, K2, n1, n2: constants (–).
tained at 20, 30, 40, 50, 60, or 70 °C (±1 °C) and the
samples were allowed to equilibrate until there was no dis-
cernible weight change (±0.001 g). This involved a period
100 X
N
jmi  mpi j
of approximately three weeks. The total time for removal, Me ¼ ð7Þ
weighing, and putting back the samples in the desiccators N i¼1 mi
was about 30 s. This minimized the degree of atmospheric where mi is the experimental value, mpi is the predicted va-
moisture sorption during weighing. The equilibrium mois- lue, and N is the population of experimental data.
ture content was determined by drying in an oven at 70 °C The GAB model contains three constants, which have
until constant weight. All measurements were done in trip- physical meaning and are functions of temperature:
licate and additional parallels were analysed if the individ-
ual values from the triplicates deviated more than 0.6% C ¼ C 0 expðDH C =RT Þ ð8Þ
from the triplicate mean. K ¼ K 0 expðDH K =RT Þ ð9Þ
X m ¼ X m0 expðqm =RT Þ ð10Þ

2.3. Measurement of glass transition temperature where T is the absolute temperature, R is the universal gas
constant, C0, K0, Xm0, and qm are adjustable constants for
Tomato powder samples of about 1 g (±0.01 g) were the temperature effect and DHC and DHK are functions of
conditioned at 25 °C using sulphuric acid solutions to the heat of water sorption (Maroulis et al., 1988):
maintain the water activity level between 0.04 and 0.55, DH C ¼ H m  H n ð11Þ
according to sorption isotherm methodology. After equi-
DH K ¼ k  H n ð12Þ
librium was reached, samples of about 10 mg were taken
for differential scanning calorimetry (DSC) analysis and where Hm and Hn are the heats of sorption of the mono-
the remaining material was analyzed for moisture content layer and the multilayer of the water respectively, and k
by drying at 70 °C. The glass transition temperature was is the heat of condensation of pure water.
determined by DSC, with a Perkin–Elmer Pyris 1 differen- There are two methods of regression analysis for esti-
tial scanning calorimeter supplied with proper software. mating the constants of the GAB equation from experi-
The samples were heated at 10 °C/min between 60 °C mental moisture sorption data; the direct and the indirect
and 80 °C in an inert atmosphere. Preliminary runs with method. In the direct regression method the six GAB con-
2 and 5 °C/min heating rates gave virtually the same stants Xm0, C0, K0, qm, DHC, DHK are estimated by substi-
results, so the 10 °C/min rate was chosen for convenience. tuting Eqs. (8)–(10) into Eq. (1). All the experimental data
The reference was an empty pan, while liquid nitrogen was are used directly with the same weight. In the indirect or
used for sample cooling before the runs. The mid point of successive method, the three GAB constants Xm, C and K
the glass transition was considered as the characteristic are estimated at each temperature by regression analysis
temperature of the transition. All measurements were done of Eq. (1) and then the constants Xm0, C0, K0, qm, DHC,
in triplicate. DHK are estimated by regression analysis of Eqs. (8)–
(10). The experimental data are grouped at various temper-
atures for the first regression, and they are subsequently
2.4. Data analysis represented by derived values. The derived values may
introduce significant uncertainties into the second, third
2.4.1. Sorption isotherms and fourth regressions. Thus, the application of the indirect
The isotherm models used to fit the data are presented in regression method depends very much on the confidence
Table 1. A non-linear least square regression analysis was limits and regions of the constants Xm, C, and K that are
used to determine the model parameters and the goodness obtained from the first approximation. According to Mar-
of the fit of each model was evaluated by calculating the oulis et al. (1988), the indirect non-linear regression
mean relative percentage deviation modulus, Me, defined as method is not recommended since the basic three constants
76 A.M. Goula et al. / Journal of Food Engineering 85 (2008) 73–83

of the GAB equation are interrelated. Thus, in this work A plot of DH versus DS from values of Eq. (16) can be
the fit of the GAB equation to the water vapor adsorption evaluated for each set of sorption data. These values can be
data of the spray dried tomato pulp was studied using the correlated according to the following equation:
direct regression. DH ¼ T B DS þ DGB ð18Þ
where TB denotes the isokinetic temperature, which repre-
2.4.2. Thermodynamic properties sents the temperature at which all reactions in the sorption
The net isosteric heat of sorption can be determined series proceed at the same rate. Since there is a high degree
from Eq. (13), which is derived from the Clausius–Clapey- of linear correlation between enthalpy and entropy, the
ron equation, applied to the system and pure water with compensation theory can be assumed to be valid for sorp-
the following assumptions: (1) moisture content of the sys- tion. Krug, Hunter, and Grieger (1976) recommended a
tem remains constant and (2) heat of vaporization of pure test for the compensation theory, which involves compar-
water and excess heat of sorption do not change with tem- ing the isokinetic temperature with the harmonic mean
perature (Al-Muhtaseb et al., 2004b) temperature, Thm, which is defined as
 n
o lnðaw Þ q T hm ¼ Pn ð19Þ
¼  st ð13Þ i¼1 ð1=T Þ
oð1=T Þ X R
where n is the number of isotherms.
where qst is the net isosteric heat of sorption in J/mol, T is A linear chemical compensation pattern exists only if
the absolute temperature in K, X is the equilibrium mois- TB 6¼ Thm. If TB > Thm the process is enthalpy driven, while
ture content in kg/kg dry solid, and R is the universal gas if the opposite condition is observed, the process is consid-
constant (8.314 J/mol K). ered to be entropy-controlled.
The change in molar differential entropy can be calcu-
lated from Gibbs–Helmholtz equation (Telis et al., 2000): 2.4.3. Water plasticization behavior
To predict the plasticization effect of water, glass transi-
DH  DG tion temperature and moisture content data were fitted to
DS ¼ ð14Þ
T the model proposed by Gordon and Taylor (1952):
where DS is the change in specific entropy in J/mol K, DH ð1  xw ÞT gs þ kxw T gw
Tg ¼ ð20Þ
is the change in enthalpy in J/mol, and DG is the free en- ð1  xw Þ þ kxw
ergy in J/mol calculated as
where Tg, Tgs, Tgw are the glass transition temperatures of
DG ¼ RT ln aw ð15Þ the mixture, solids, and water, respectively, xw is the mass
Substituting Eq. (15) in Eq. (14) and rearranging: fraction of water and k is the Gordon–Taylor parameter.
The relationship between glass transition temperature
DH DS and water activity was modeled by an empirical model
ln aw ¼  ð16Þ
RT R developed by Khalloufi, El-Maslouhi, and Ratti (2000):
The enthalpy and entropy changes can be calculated from Ca2w þ Daw þ E
Tg ¼ ð21Þ
Eq. (16) by plotting the sorption isotherm in the form of ca2w þ daw þ 1
ln(aw) versus 1/T for certain values of the material moisture
where Tg is the glass transition temperature in K and C, D,
content and then determining the slope (DH/R) and the
E, c, and d are model parameters.
intercept (DS/R). The net isosteric heat of sorption is deter-
mined by applying Eq. (13). This approach assumes that qst
is invariant with temperature, with the application of the 3. Results and discussion
method requiring measurement of sorption isotherms at
more than two temperatures. An empirical exponential 3.1. Sorption isotherms
relationship between the isosteric heat of sorption and
the moisture content has been proposed (Tsami, Mari- The experimental moisture sorption data for tomato
nos-Kouris, & Maroulis, 1990): powder at 20, 30, 40, 50, 60, and 70 °C are shown in
Fig. 1. The sorption isotherms demonstrate an increase in
qst ¼ q0 expðX =X 0 Þ ð17Þ
equilibrium moisture content with increasing water activ-
where q0 is the isosteric heat of sorption of the first mole- ity, at constant temperature, and are sigmoid in shape for
cule of water in the food in J/mol, and X0 is a characteristic most of the examined temperatures, characteristic of amor-
moisture content of the food material in kg/kg dry solid, at phous materials rich in hydrophilic components (Al-Muh-
which this isosteric heat of sorption has been reduced by taseb, McMinn, & Magee, 2004a). At low and intermediate
63%. The constants q0 and X0 are estimated by fitting water activities, the so-called multilayer sorption region,
Eq. (17) to the values of qst, obtained by applying Eq. moisture content increases linearly with aw, whereas at high
(13) to experimental isotherms. water activity levels, the so-called capillary condensation
 A.M. Goula et al. / Journal of Food Engineering 85 (2008) 73–83 77

0.8
T (°C)
for water binding as a result of physical and/or chemical
20 30 40 50 60 70
changes induced by temperature. According to Palipane
0.7 and Driscoll (1992), at increased temperatures water mole-
cules get activated to higher energy levels and break away
Moisture content, X (kg/kg dry solids)

0.6 from the water-binding sites of the food, thus decreasing

the equilibrium moisture content. However, many
0.5 researchers reported that the presence of large quantities
of crystalline sugar, such as glucose, in dry foods can cause
0.4 a different effect of temperature at high aw levels. Myhara
and Sablani (2001) reported that the equilibrium moisture
0.3 content of high glucose raisin or date at aw up to 0.55
decreases as the temperature increases, whereas at higher
0.2 water activity levels increases sharply as the temperature
increases. However, in the sorption isotherms presented
0.1 in Fig. 1, the temperature effect is not dependent on aw
level. This may be due to the fact that tomato powder con-
0.0 tains high levels of protein and insoluble solids.
0.0 0.2 0.4 0.6 0.8 1.0 The typical sigmoid curves of sorption isotherms pre-
aw
dicted by the GAB model are also shown in Fig. 1. The
Fig. 1. Sorption isotherms of spray dried tomato pulp at various fit of the GAB equation to the water vapor adsorption data
temperatures – experimental (symbols) and predicted by the GAB model of the spray dried tomato pulp was performed using the
(lines).
direct regression approach. The results are shown in Table
2. Considering the values for R2 and Me the GAB equation
can be considered satisfactory for fitting the adsorption of
region, water content rapidly increases with aw. The expla- tomato powder. Maximum, minimum, and average values
nation for the nature of the isotherms may be that, at low for Xm, C and K were calculated using the estimated con-
water activities, physical sorption on strongly active sites of stants Xm0, C0, K0, qm, DHC, DHK (Table 2). Estimated
proteins occurs, since water can be sorbed only to surface – values for Xm are within the reported values for vegetables,
OH sites of crystalline sugar. Moreover, according to which vary between 18% and 22% (dry basis) and were
Falade and Aworh (2004), at low aw it is possible to have found to be greater than those for fruits, which vary
a local dissolution of sugar alcohols, a swelling of proteins between 10% and 15% (dry basis) (Kiranoudis et al.,
and appearance of new active sites. In the intermediate aw 1993). The monolayer moisture content Xm is recognized
range, sorption takes place at less active sites and, from this as the moisture content affording the longest time period
point on, there is gradual dissolution of sugars resulting in with minimum quality loss at a given temperature. Below
a complete exudation of sugars in solution at high water it, rates of deteriorative reactions, except oxidation of
activities. However, sorption isotherm of tomato powder unsaturated fats, are minimal. Therefore, at a given tem-
follows the type II BET classification shape, and not the perature, the safest water activity level is that correspond-
type I, which is usual for high sugar foods. This observa- ing to Xm or lower. The safest aw values corresponded to
tion may be attributed to the high protein content of spray
dried tomato pulp.
Table 2
The isotherms of many foods at room temperature are Fitting of the GAB model to the water vapor adsorption for the spray
known. However, few investigations have been conducted dried tomato pulp
to determine isotherms at temperatures higher than room Mean coefficient of determination (R2) 0.996
temperature, and in many of these works the effect of tem- Mean relative percentage deviation modulus (Me), % 3.49
perature on the aw of the saturated salt solutions, used to Constant Xm0, % dry basis 13.82
control the relative humidity, is not accounted, so errors Constant qm, kJ mol1 1.06
Maximum Xm, % dry basis 21.32
in subsequent computations are introduced. Generally,
Minimum Xm, % dry basis 20.04
the effect of temperature on the sorption isotherm is of Average Xm, % dry basis 20.64
great importance given that foods are exposed to a range Constant C0 6.10  106
of temperatures during storage and processing, and water Heat related to monolayer water (DHC), kJ mol1 36.08
activity changes with temperature for the same moisture Maximum C 18.83
Minimum C 1.95
content. Temperature affects the mobility of water mole-
Average C 7.09
cules and the dynamic equilibrium between vapor and Constant K0 1.062
adsorbed phases. Experimental data indicate that the equi- Heat related to multilayer water (DHK), kJ mol1 0.641
librium moisture content decreases with increasing temper- Maximum K 0.848
ature, at a constant water activity. This trend may be Minimum K 0.816
Average K 0.833
attributed to a reduction in the total number of active sites
78 A.M. Goula et al. / Journal of Food Engineering 85 (2008) 73–83

Xm for spray dried tomato pulp were predicted using data uble solids, such as glucose and fructose, by the centrifuga-
in Table 2 and are tabulated in Table 3. tion used to separate the tomato serum. According to
The values for C are also within the reported limits for Giovanelli et al. (2002), this product requires higher water
vegetables and values for K are smaller than 1 as dictated removal during drying to reduce water activity and is more
by the GAB equation. Parameter DHC, is the difference sensitive to moisture variations during storage. In addition,
in enthalpy between monolayer and multilayer sorption, at low and medium water activities the water contents of
which is expected to have a large positive value, due to freeze dried pulp and air dried halves were lower than that
the strong exothermic interaction of water vapor with pri- of spray dried pulp, whereas at high aw levels freeze and air
mary sorption sites of the material. The estimated value of dried products showed a rapid increase in their water con-
DHC is also in agreement with values reported for various tent. Generally, monosaccharides exhibit a long flat seg-
vegetables, which were found to be greater than those for ment in their sorption isotherm within the aw range of
fruits (Maroulis et al., 1988). Parameter DHK is the differ- 0.1–0.6 and rapid increases in their water content at aw lev-
ence between the heat of condensation of water and the els above 0.6–0.8 (Biliaderis, Lazaridou, & Arvanitoyannis,
heat of sorption of a multimolecular layer. The estimated 1999). In contrast, the other tomato constituents absorb
value was found to be positive but small which is in agree- water even at low aw levels. Thus, a feasible explanation
ment with the literature (Samaniego-Esguerra, Boag, & for the lower water uptake of freeze and air dried products
Robertson, 1991; Tsami et al., 1990). at low water activity is that the modified spray drying sys-
Giovanelli, Zanoni, Lavelli, and Nani (2002) also used tem achieved extremely low drying temperatures (Goula &
the GAB equation to model experimental sorption data Adamopoulos, 2005a) preventing thermal degradation of
reported for freeze dried tomato pulp, freeze dried insolu- various tomato constituents that adsorb water. Conversely,
ble solids-rich tomato, and air dried tomato halves. The at high aw levels, where primarily sugars adsorb water,
relevant sorption isotherms at 20 °C are shown in Fig. 2 freeze dried pulp and air dried halves were more hygro-
along with those for spray-dried tomato. Freeze-dried scopic than spray dried powder, as they were characterized
insoluble solids-rich tomato was the least hygroscopic of by a higher sugar content level, as a result of their produc-
all dried products. This may be due to the removal of sol- tion method.
The parameters for other sorption models are presented
in Table 4, along with the mean relative Me. The GAB
Table 3 model gives Me values ranging from 3.1% to 4.0% with
Monolayer moisture content (Xm) and corresponding aw for spray dried an average value of 3.5% and best describes the experimen-
tomato pulp tal adsorption data throughout the entire range of water
Temperature (°C) Monolayer moisture content, Xm (% dry basis) aw activity. This observation is similar to that obtained by
20 21.3 0.23 other researchers, who studied sorption isotherms of vari-
30 21.1 0.30 ous fruits and vegetables. Kiranoudis et al. (1993) deter-
40 20.7 0.35 mined the equilibrium moisture content of potato, carrot,
50 20.5 0.39 tomato, green pepper, and onion within the range of 10–
60 20.2 0.44
90% relative humidity and at three different temperatures
70 20.0 0.49
(30, 45, and 60 °C) and found the GAB model satisfactory
for the prediction of the experimental data obtained.
McLaughlin and Magee (1998) determined the sorption
isotherms for potatoes at temperatures of 30, 45, and
spray dried tomato pulp 60 °C and, among the models tested, the GAB model gave
1.0 freeze dried tomato pulp the best fits. According to Al-Muhtaseb et al. (2004a), the
Moisture content, X (kg/kg dry solids)

freeze dried insoluble solids-rich tomato GAB model adequately represented the sorption isotherms
air dried tomato halves of potato and wheat starch, whereas Akanbi et al. (2006)
0.8
concluded that equilibrium moisture contents of tomato
0.6 slices obey closely the GAB equation.
Only experimental data with aw < 0.50 were fitted to the
0.4 BET model, because above that value the model hypothesis
fails and the equation is not able to predict sorption behav-
0.2 ior accurately. One of the two BET constants, Xm, is, as in
the GAB equation, the monolayer capacity. The values of
0.0 Xm ranging from 14.6% to 19.0% (dry basis) were within
0.0 0.2 0.4 0.6 0.8 1.0 the reported values for vegetables (Tsami et al., 1990)
aw and lower than that obtained by the GAB model. Despite
Fig. 2. Water sorption isotherms (20 °C) modeled by the GAB equation the theoretical limitations of the BET adsorption analysis,
for various dry tomato products. The dotted lines are based on data by the monolayer value was found to be a reasonable guide
Giovanelli et al. (2002). with respect to various aspects of interest in dried foods.
 A.M. Goula et al. / Journal of Food Engineering 85 (2008) 73–83 79

Table 4
Parameters of the isotherm models for spray dried tomato pulp
Model Constant Temperature (°C)
20 30 40 50 60 70
GAB Xm (% dry basis) 21.31 21.02 20.71 20.53 20.21 20.04
C 18.83 8.85 5.76 4.30 2.85 1.95
K 0.816 0.823 0.830 0.836 0.842 0.848
Me (%) 3.82 3.95 3.13 3.21 3.44 3.38
Overall Me (%) 3.49
BET Xm (% dry basis) 18.96 16.24 15.58 14.92 14.56 14.78
C 19.04 14.06 8.18 6.11 3.37 2.03
Me (%) 3.40 10.40 9.43 8.75 19.93 17.22
Overall Me (%) 11.52
Halsey A 23.39 23.12 22.85 22.60 22.31 22.10
B 1.80 1.69 1.57 1.50 1.40 1.33
Me (%) 86.14 61.55 50.78 45.90 39.63 36.31
Overall Me (%) 53.39
Smith A 0.067 0.049 0.032 0.020 0.006 0.005
B 0.763 0.754 0.756 0.759 0.746 0.722
Me (%) 58.30 30.11 23.61 18.91 30.06 42.91
Overall Me (%) 33.98
Oswin A 0.335 0.303 0.277 0.258 0.227 0.202
B 0.402 0.441 0.483 0.515 0.567 0.603
Me (%) 24.87 13.78 8.58 10.03 18.25 25.56
Overall Me (%) 16.85
Peleg K1 0.600 0.572 0.585 0.615 0.680 0.657
K2 0.465 0.416 0.436 0.440 0.351 0.541
n1 5.28 4.27 4.87 5.22 4.49 8.66
n2 0.54 0.60 0.75 0.85 0.89 1.40
Me (%) 8.07 1.89 7.12 6.98 11.36 9.34
Overall Me (%) 7.46

The second parameter, C, is also an energy constant as the and sugars. The different polar groups of the polymers pro-
GAB constant, but with slightly different physical meaning vide energetically different sites for sorption. Besides, as a
(Moraga et al., 2006). It allows us to classify sorption iso- polymer sorbs water it undergoes changes of constitution,
therms according to Brunauer’s classification. As the C val- dimensions and other properties. Water sorption also leads
ues were higher than 2, the obtained sorption isotherms can to phase transformations of the sugars contained in the
be classified as type II. The Peleg model also adequately fit- food. All these facts make the interpretation of the differen-
ted the results with an average Me value of 7.5%. Al-Muh- tial heat curves difficult; there are, however, some observa-
taseb et al. (2004a) also reported that the Peleg model tions that could be made from the results. The net isosteric
provides a good description of the isotherms of potato heat of water sorption at each moisture content was
starch, peas and cured beef and Lewicki (1998) determined obtained by fitting Eq. (13) to the data of equilibrium water
the sorption isotherms of 31 food materials and found the activity versus temperature, obtained from sorption iso-
Peleg model to show the best fit. The Halsey model gave therms. The results are shown in Fig. 3 and are similar to
the least accurate predictions, giving an average Me value those presented by Kiranoudis et al. (1993), who reported
of 53.4%. Wang and Brennan (1991) and Al-Muhtaseb a value of 42.9 kJ mol1 for tomato slices at a moisture
et al. (2004a) also reported that the Halsey model is inade- content of 12% (dry basis) falling to 7.4 kJ mol1 at 36%
quate for representing the sorption isotherms for various (dry basis). In addition, the net isosteric heat of sorption
foods such as potato and starch powders. The Smith and remained significant for a wide range of values of material
Oswin models also failed to describe the experimental data moisture content. This observation is in agreement with
accurately giving average Me values of 34.0% and 16.9%, Tsami (1991) and Kiranoudis et al. (1993), who stated that
respectively. in the case of vegetables, the net isosteric heat of desorption
was large for moisture contents between 6% and 36% (dry
3.2. Sorption thermodynamic properties basis), while in the case of fruits, it was found negligible
over this range of moisture content, and only increased
Water sorption in foods is a complex phenomenon. The dramatically when the material moisture approached zero.
main water-adsorbing constituents in foods are various Moreover, in the case of fruits the heat of sorption was
polymers (proteins, starch, cellulose, hemicellulose, etc.) found to be slightly negative at high moisture content,
80 A.M. Goula et al. / Journal of Food Engineering 85 (2008) 73–83

40 Eq (23) 0.09 tion was found to be TB= 396 K. The harmonic mean
isosteric heat of sorption temperature was calculated as Thm = 317 K, a value signif-
Net isosteric heat of sorption qst (kJ/mol)

35 entropy 0.08
icantly different from the value of TB, confirming the suit-
0.07 ability of the isokinetic theory for water sorption in spray
30
dried tomato pulp and since TB > Thm, the process can be

Entropy, ΔS (kJ/mol K)
0.06
25 characterized as enthalpy driven.
0.05
20 3.3. Water plasticization behavior
0.04
15
0.03 Representative DSC traces obtained with the sample
10 0.02
conditioned at aw = 0.25 are shown in Fig. 4. Similar
curves were obtained for the samples conditioned at the
5 0.01 other aw levels. These curves are similar to those deter-
0 0 mined by other authors in other fruits and vegetables in
4 8 10 15 20 25 30 40 50 60 70 80 the same domain of aw (Telis & Sobral, 2001; Telis &
Moisture content, X (%, dry basis) Sobral, 2002). Generally, the thermograms obtained
showed the typical second-order transition, the glass tran-
Fig. 3. Isosteric heat of sorption and entropy for spray dried tomato pulp
at various moisture contents.
sition of the amorphous materials that produce a step
change in the heat flow due to changes in heat capacity
due to the contribution of the endothermic dissolution of at the temperature of phase transition. Values of Tg
sugars in the sorbed water, a phenomenon that was not obtained for the different samples as a function of moisture
observed in the case of spray dried tomato pulp. content are shown in Fig. 5. Tg values were slightly lower
Fig. 3 clearly illustrates a progressive increase in the heat than that obtained by Telis and Sobral (2002) for air dried
of sorption with decreasing moisture content. At high and freeze dried tomato halves, which had lower NaCl and
moisture content the heat of sorption approaches zero, citric acid content. Glass transition temperature is known
meaning that the heat of sorption is equal to the heat of to decrease with decreasing molecular weight (Roos
vaporization of water. The effect of moisture content on et al., 1996; Slade & Levine, 1995), in such a way that
heat of sorption was fitted by: higher NaCl and citric acid could have caused the observed
decrease in Tg. According to Telis and Sobral (2002), Tg
qst ¼ 48:25 expðX =15:36ÞðR2 ¼ 0:982Þ ð22Þ values of air dried tomato halves, especially those corre-
where X is the moisture content in % dry basis and qst is the sponding to osmotically pre-treated ones, were lower than
isosteric heat of sorption in kJ mol1. Similar effects of that observed for freeze dried material at the same moisture
moisture content on the heat of sorption were reported content. In that case the lower Tg values were attributed to
by Mazza and LeMaguer (1978) for yellow globe onion, gained solutes, NaCl and sucrose.
Saravacos, Tsiourvas, and Tsami (1986) for sultana raisins, Fig. 5 shows that the water plasticization effect causes a
Tsami (1991) for dried currants, figs, prunes, and apricots, significant decrease of the glass transition temperature with
Telis et al. (2000) for persimmon skin and pulp, and Al- increase in moisture content. This observation is similar to
Muhtaseb et al. (2004b) for starch powders. The physical that obtained by other researchers, who studied glass tran-
explanation for the rapid increase in qst at low moisture sitions and state diagrams for various fruits and vegetables
contents may be that in the initial stages of sorption there
are highly active polar sites on the surface of the food
material, which are covered with water molecules to form 0
a monomolecular layer. -0.02
From the straight lines calculated by Eq. (16) at different -0.04
moisture contents the values of DS were obtained (Fig. 3).
-0.06
The results showed a strong dependence of differential
-0.08
Exothermal heat flow

entropy on moisture content with an exponential trend

similar to that exhibited by differential enthalpy. This -0.1
observation is similar to that obtained by Al-Muhtaseb -0.12
et al. (2004b), who studied the water sorption thermody- -0.14
namic properties of starch powder. A plot of enthalpy ver- -0.16 0.05 W/g
sus entropy showed a linear correlation:
-0.18
DH ¼ 395:670  DS þ 0:331 ðR2 ¼ 0:998Þ ð23Þ -0.2
-60 -50 -40 -30 -20 -10 0 10 20 30
As the law of compensation could be applied in the range
Temperature (°C)
of moistures studied, the results were correlated in accor-
dance with Eq. (24) and the isokinetic temperature of sorp- Fig. 4. DSC profile for tomato powder conditioned at aw of 0.25.
 A.M. Goula et al. / Journal of Food Engineering 85 (2008) 73–83 81

350 experimental data 350

experimental data
Gordon & Taylor model
Khalloufi et al. model
300 300
Tg (K)

Tg (K)
250
250

200
200

150
0 0.1 0.2 0.3 0.4 150
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Moisture content, X (kg/kg dry solids)
aw
Fig. 5. Experimental (symbols) and predicted by Gordon and Taylor
model (line) data for the relationship between glass transition temperature Fig. 6. Experimental (symbols) and predicted by Khalloufi et al. model
(Tg) and moisture content (X) of spray dried tomato pulp. (line) data for the relationship between glass transition temperature (Tg)
and aw for spray dried tomato pulp.

Another relationship of interest in the prediction of the

(Sa & Sereno, 1994). Telis and Sobral (2001) determined physical state in food materials is given in Fig. 6, which
the glass transition temperatures of freeze dried pineapple presents a plot of Tg data against aw. This relationship
conditioned at various water activities and concluded that was modeled by the empirical model developed by Kha-
for water activities lower than 0.90, the plasticizing effect of lloufi et al. (2000):
water on the Tg was evident, with a great reduction of Tg 744:17a2w þ 516:38aw þ 335:25
caused by increasing moisture content. Sobral et al. Tg ¼ ðKÞ ðR2 ¼ 0:993Þ
2:44a2w þ 3:71aw þ 1
(2001) studied the phase diagram for freeze dried persim-
ð24Þ
mon and observed that in the hygroscopic domain
(aw < 0.90), the plasticizing effect is clear, whereas at water This equation allows for a rapid and reliable method for
activities higher than 0.90, the glass transition curve exhib- calculating the Tg for spray dried tomato pulp stored under
its a discontinuity, with a sudden increase of Tg that specific relative humidity conditions. Fig. 5 allows us to ob-
approaches a constant value. This plasticizing activity of tain the critical water content at which, at a specific storage
water may be based on the weakening of hydrogen bonds temperature, the glass transition occurs. If the product is
and dipole–dipole intra- and inter-macromolecular interac- commercialized at 25 °C, the critical value of water content
tions due to the shielding of these mainly attractive forces at which the product ‘‘glass” is transformed to a liquid-like
by water molecules. One more contributory factor is the ‘‘rubber” is 10.7% dry basis. Above this value, tomato
low Tg of amorphous water, since this difference in Tg powder becomes sticky and the crystallization of amor-
between water and food macromolecules decreases the vis- phous compounds could take place. This critical value is
cosity of biopolymer–water mixtures over a wide range of lower than the monolayer moisture content, thus indicating
temperatures. that the Xm is not a water content that assures quality pres-
The Gordon and Taylor model has proved to be a reli- ervation during storage of spray dried tomato pulp.
able predictor of glass transition temperatures of sugars at
various water contents and has been used in several fruit 4. Conclusions
and vegetable samples that can be considered as binary
mixtures of solids and water. It was fitted to experimental On the basis of this work the following conclusions can
points using Tgw = 138 K, with the following parameters be drawn:
calculated by non-linear regression: Tgs = 412.8 K and
k = 6.69, with R2 = 0.985. Fig. 5 shows the curve predicted  The moisture sorption isotherms of tomato pulp spray
by Gordon and Taylor model. The calculated k value was dried in dehumidified air follows a sigmoid isotherm
similar to those reported for fruits and sugar solutions curve typical of the type II BET classification shape.
(Khalloufi et al., 2000; Moraga et al., 2004). Telis and  Temperature affects the sorption behaviour; the equilib-
Sobral (2002) fitted glass transition temperature-moisture rium moisture content decreases with increasing temper-
content data for freeze dried and air dried tomato to the ature, at constant water activity.
model proposed by Gordon and Taylor and found Tgs  Within the temperature range investigated, the kinetic
and k values of 603.9 K and 9.35, respectively. This differ- three-parameter GAB model and the empirical four-
ence in the values of Gordon and Taylor parameters can be parameter Peleg equation were found to best represent
attributed to differences in composition (mean molecular the experimental data throughout the entire range of
weight of the solutes) of the tomato liquid phase. water activity.
82 A.M. Goula et al. / Journal of Food Engineering 85 (2008) 73–83

 An exponential relation was found to adequately Goula, A. M., & Adamopoulos, K. G. (2005a). Spray drying of tomato
describe the dependence of isosteric heat of sorption pulp in dehumidified air: I. The effect on product recovery. Journal of
Food Engineering, 66, 25–34.
on the equilibrium moisture content. Goula, A. M., & Adamopoulos, K. G. (2005b). Spray drying of tomato
 Water acts as a plasticizer in spray dried tomato pulp pulp in dehumidified air: II. The effect on powder properties. Journal of
decreasing its glass transition temperature and the Gor- Food Engineering, 66, 35–42.
don and Taylor model could adequately represent the Halsey, G. (1948). Physical adsorption on non-uniform surfaces. Journal
sugar matrix glass transition curve. of Chemistry Physics, 16, 931–937.
Iglesias, H. A., Chirife, J., & Viollaz, P. (1976). Thermodynamics of water
vapour sorption by sugar beet root. Journal of Food Engineering, 11,
91–101.
Acknowledgement Khalloufi, S., El-Maslouhi, Y., & Ratti, C. (2000). Mathematical model
for prediction of glass transition temperature of fruit powders. Journal
This research was supported in part by Greek State of Food Science, 65, 842–845.
Scholarship Foundation. The authors thank Dr. George Kim, K. H., Song, Y., & Yam, L. K. (1991). Water sorption characteristics
Z. Papageorgiou for conducting the DSC measurements. of dried red peppers (Capsicum annum L). International Journal of Food
Science and Technology, 29(3), 339–345.
Kiranoudis, C. T., Maroulis, Z. B., Tsami, E., & Marinos-Kouris, D.
References (1993). Equilibrium moisture content and heat of desorption of some
vegetables. Journal of Food Engineering, 20, 55–74.
Akanbi, C. T., Adeyemi, R. S., & Ojo, A. (2006). Drying characteristics Krug, R. R., Hunter, W. G., & Grieger, R. A. (1976). Enthalpy–entropy
and sorption isotherm of tomato slices. Journal of Food Engineering, compensation. 1. Some fundamental statistical problems associated
73, 157–163. with the analysis of van’t Hoff and Arrhenius data. Journal of Physical
Al-Muhtaseb, A. H., McMinn, W. A. M., & Magee, T. R. A. (2004a). Chemistry, 80, 2335–2341.
Water sorption isotherms of starch powders. Part 1: Mathematical Lewicki, P. P. (1998). A three parameter equation for food moisture
description of experimental data. Journal of Food Engineering, 61, isotherms. Journal of Food Process Engineering, 21, 127–144.
297–307. Madamba, P. S., Driscoll, R. H., & Buckle, K. A. (1996). Enthalpy–
Al-Muhtaseb, A. H., McMinn, W. A. M., & Magee, T. R. A. (2004b). entropy compensation models for sorption and browning of garlic.
Water sorption isotherms of starch powders. Part 2: Thermodynamic Journal of Food Engineering, 28, 109–119.
characteristics. Journal of Food Engineering, 62, 135–142. Maroulis, Z. B., Tsami, E., Marinos-Kouris, D., & Saravacos, G. D.
Aviara, N. A., & Ajibola, O. O. (2002). Thermodynamics of moisture (1988). Application of the GAB model to the moisture sorption
sorption in melon seed and cassava. Journal of Food Engineering, 55, isotherms for dried fruits. Journal of Food Engineering, 7, 63–78.
107–113. Matveev, Y. I., Grinberg, V. Y., & Tolstoguzov, V. B. (2000). The
Ayala, A., Walter, D., Martinez-Monzo, J., Fito, P., & Chiralt, A. (2002). plasticizing effect of water on proteins, polysaccharides and their
Transiciones de fase en funcion del contenido de humedad en mango mixtures. Glassy state of biopolymers, food and seeds. Food Hydro-
var. Kent. In P. Fito, A. Mulet, A. Chiralt, & A. Andres (Eds.), colloids, 14, 425–437.
Ingenieria de alimentos. Nuevas fronteras en el siglo XXI (pp. 1–7). Mazza, G., & LeMaguer, M. (1978). Water sorption properties of yellow
Valencia: Editorial de la UPV. globe onion. Canadian Institute of Food Science and Technology
Bai, Y., Rahman, M. S., Perera, C. O., Smith, B., & Melton, L. D. (2001). Journal, 11, 189–193.
State diagram of apple slices: Glass transition and freezing curves. McLaughlin, C. P., & Magee, T. R. A. (1998). The determination of
Food Research International, 34, 89–95. sorption isotherms and the isosteric heats of sorption for potatoes.
Benado, A. L., & Rizvi, S. S. H. (1985). Thermodynamic properties of Journal of Food Engineering, 35, 267–280.
water in rice as calculated from reversible and irreversible isotherms. Moraga, G., Martinez-Navarrete, N., & Chiralt, A. (2004). Water
Journal of Food Science, 50, 101–105. sorption isotherms and glass transition in strawberries: Influence of
Brunauer, S., Emmett, P. H., & Teller, E. (1938). Adsorption of gases in pretreatment. Journal of Food Engineering, 62, 315–321.
multimolecular layers. Journal of American Chemistry Society, 60, Moraga, G., Martinez-Navarrete, N., & Chiralt, A. (2006). Water
309–320. sorption isotherms and phase transitions in kiwifruit. Journal of Food
Biliaderis, C. G., Lazaridou, A., & Arvanitoyannis, I. (1999). Glass Engineering, 72, 147–156.
transition and physical properties of polyol-plasticized pullulan-starch Myhara, R. M., & Sablani, S. (2001). Unification of fruit water sorption
blends at low moisture. Carbohydrate Polymers, 40, 29–47. isotherms using artificial neural networks. Drying Technology, 19(8),
Chen, C., & Jayas, D. S. (1998). Dynamic equilibrium moisture content 1543–1554.
for grain drying. Canadian Agricultural Engineering, 40(4), 299–303. Oswin, C. R. (1946). The kinetics of package life III. The isotherm. Journal
Falade, K. O., & Aworh, O. C. (2004). Adsorption isotherms of osmo- of Chemical Industry, 65, 419–421.
oven dried African star apple (Chrysophyllum albidum) and african Palipane, K. B., & Driscoll, R. H. (1992). Moisture sorption character-
mango (Irvingia gabonensis) slices. European Food Research and istics of inshell macadamia nuts. Journal of Food Engineering, 18,
Technology, 218, 278–283. 63–76.
Georget, D. M. R., Smith, A. C., & Waldron, K. W. (1999). Thermal Peleg, M. (1993). Assessment of a semi-empirical four parameter general
transitions in freeze dried carrot and its cell wall components. model for sigmoid moisture sorption isotherms. Journal of Food
Thermochimica Acta, 332, 203–210. Process Engineering, 16, 21–37.
Giovanelli, G., Zanoni, B., Lavelli, V., & Nani, R. (2002). Water sorption, Roos, Y. H., Karel, M., & Kokini, J. L. (1996). Glass transitions in low
drying and antioxidant properties of dried tomato products. Journal of moisture and frozen foods: Effects on shelf life and quality. Scientific
Food Engineering, 52, 135–141. status summary. Food Technology, 50, 95–108.
Gordon, M., & Taylor, J. S. (1952). Ideal copolymers and the second Ruegg, M. (1980). Calculation of the activity of water in sulphuric acid
order transitions of synthetic rubbers. I. Non-crystalline copolymers. solutions at various temperatures. Lebensmittel-Wissenschaft und
Journal of Applied Chemistry, 2, 493–500. Technologie, 13, 22–24.
Goula, A. M., & Adamopoulos, K. G. (2003). Spray drying performance Sa, M. M., & Sereno, A. M. (1994). Glass transitions and state diagrams
of a laboratory spray dryer for tomato powder preparation. Drying for typical natural fruits and vegetables. Thermochimica Acta, 246,
Technology, 21(7), 1273–1289. 285–297.
 A.M. Goula et al. / Journal of Food Engineering 85 (2008) 73–83 83

Sablani, S. S., Kasapis, S., & Rahman, M. S. (2007). Evaluating water Telis, V. R. N., & Sobral, P. J. A. (2001). Glass transitions and state
activity and glass transition concepts for food stability. Journal of Food diagram for freeze-dried pineapple. Lebensmittel-Wissenschaft und
Engineering., 78, 266–271. Technologie, 34, 199–205.
Samaniego-Esguerra, C. M., Boag, I. F., & Robertson, G. L. (1991). Telis, V. R. N., & Sobral, P. J. A. (2002). Glass transitions for freeze-dried
Comparison of regression methods for fitting the GAB model to the and air-dried tomato. Food Research International, 35, 435–443.
moisture isotherms of some dried fruits and vegetables. Journal of Food Tolaba, M. P., Suarez, C., & Viollaz, P. (1997). Heats and entropies of
Engineering, 13, 115–133. sorption of cereal grains: A comparison between integral and
Saravacos, G. D., Tsiourvas, D. A., & Tsami, E. (1986). Effect of differential quantities. Drying Technology, 15, 137–150.
temperature on the water adsorption isotherms of Sultana raisins. Tsami, E. (1991). Net isosteric heat of sorption in dried fruits. Journal of
Journal of Food Science, 51(2), 381–383, 387. Food Engineering, 14, 327–335.
Slade, L., & Levine, H. (1995). Glass transitions and water-food structure Tsami, E., Marinos-Kouris, D., & Maroulis, Z. B. (1990). Water sorption
interactions. Advances in Food and Nutrition Research, 38, 103–269. isotherms of raisins, currants, figs, prunes and apricots. Journal of
Smith, S. E. (1947). The sorption of water vapour by high polymers. Food Science, 55, 1594–1597.
Journal of the American Chemical Society, 69, 646. Van den Berg, C., & Bruin, S. (1981). Water activity and its estimation in
Sobral, P. J. A., Telis, V. R. N., Habitante, A. M. Q. B., & Sereno, A. food systems. In L. B. Rockland & G. F. Stewart (Eds.), Water
(2001). Phase diagram for freeze-dried persimmon. Thermochimica activity: influences on food quality (pp. 147–177). New York: Academic
Acta, 376, 83–89. Press.
Telis, V. R. N., Gabas, A. L., Menegalli, F. C., & Telis-Romero, J. (2000). Wang, N., & Brennan, J. G. (1991). Moisture isotherm characteristics of
Water sorption thermodynamic properties applied to persimmon skin potatoes at four temperatures. Journal of Food Engineering, 14,
and pulp. Thermochimica Acta, 343, 49–56. 269–287.