Journal of Food Engineering 40 (1999) 71±79

Implication of glass transition for the drying and stability of dried

foods
B.R. Bhandari a,*, T. Howes b
a
Food Science and Technology, School of Land and Food, The University of Queensland, Gatton, QLD 4345, Australia
b
Department of Chemical Engineering, The University of Queensland, St. Lucia, QLD 4072, Australia
Received 29 January 1998; received in revised form 21 September 1998; accepted 7 December 1998

Abstract
Dry products obtained from most of the common drying processes are predominantly in a glassy amorphous form. The mobility
of the solid matrix in this state is highly limited. For the product to be stable for long periods of storage, this physical state should
not alter with time. When the temperature is above glass transition temperature (Tg ) an amorphous solid exists in a ``rubbery'' state.
In this state, the molecular mobility of the matrix and the reactants are accelerated, which results in an increased rate of physico-
chemical changes in dried products, such as sticking, collapse, caking, agglomeration, crystallisation, loss of volatiles, browning and
oxidation. These changes play an important role in the ability to both process and store dried food products. This paper reviews the
importance of Tg in relation to drying process and dried foods. Ó 1999 Elsevier Science Ltd. All rights reserved.

1. Introduction causes this structural change of the amorphous parts of

Drying is one of the major food processing opera-
tions. There are several drying techniques, among them
the most commonly applied are hot air (®xed and ¯u-
idised bed), freeze and spray drying. In most drying
conditions, a signi®cant amount of the dried product
remains in an amorphous state, mainly due to the in-
sucient time for crystallisation to occur at the given
drying conditions. Depending upon the rate of drying,
the dried product obtained can also constitute some
crystalline material. This will be however, in¯uenced by
the processing conditions, composition and property of
the individual ingredients present (Roos, Karel & Ko-
kini, 1996; Senoussi, Dumoulin & Berk, 1995; Flink,
1983). There are many product quality attributes which
are related to the physical state of the ingredients in the
dried product. The minor components such as ¯avours,
vitamins, enzymes and microorganisms in the dried
product are encapsulated into the major amorphous
ingredient matrix. Any change in the physical state of
the product has the possibility to a€ect the physico-
chemical characteristics of the minor as well as principal
ingredients of the product. One of the factors which
*
Corresponding author. Tel.: +61 7 54601196; fax: +61 7 54601171;
e-mail: Bb@burger.uqg.uq.edu.au
0260-8774/99/$ ± see front matter Ó 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 0 - 8 7 7 4 ( 9 9 ) 0 0 0 3 9 - 4
the structure is the product temperature. Above a criti-
cal value, known as glass transition temperature (Tg ), a
glassy solid structure begins to change to a ``rubbery''
state. This structural change in¯uences the process and
physico-chemical quality of the products. The purpose
of this review paper is to highlight the importance of
glass transition temperature on drying processes as well
as on the storage stability of dried food products.
2. Glass transition
The glass transition temperature is speci®cally the
property of an amorphous material. An amorphous
food is formed at non-equilibrium conditions either by
removing the dispersing medium (such as water), or
from the melt by cooling, or by rapid supercooling. This
material is not at thermodynamic equilibrium, and
therefore is unstable relative to the crystalline form
(Slade, Levine, Ievolella & Wang, 1993; Flink, 1983;
Mans®eld, 1993). It behaves as an extremely viscous
glass having a typical viscosity above 1012 Pa s
(Downton, Flores-Luna & King, 1982). In this state,
molecular movement is highly limited, which is neces-
sary for an orderly alignment of molecules to crystallise.
Nevertheless, the glass is in a meta-stable state and, will
tend to convert to the crystal eventually, with a rate
72 B.R. Bhandari, T. Howes / Journal of Food Engineering 40 (1999) 71±79
depending upon temperature and moisture content
(Flink, 1983).
A simple representative diagram of the changes to the
physical structure and form of a material is depicted in
Fig. 1. Glass transition is second-order phase transition
that occurs over a temperature range, although a single
temperature is often referred (Fox & Flory, 1950; Roos,
1995; Roos et al., 1996). As the temperature increases
above Tg , many of the physical properties of the mate-
rial suddenly change, among them, the most important
are an increase in the free molecular volume, an increase
in heat capacity (Cp ), increases in the thermal expansion
coecient (a) and dielectric coecient (e) and changes in
viscoelastic properties (Genin & Rene, 1995). The Ôfree
volumeÕ is the volume unoccupied by the ``solid matter''
of the molecules and represents the volume available for
their free movements (Flink, 1983). Accelerated changes
in the above properties as a function of temperature are
the basis for detecting the glass transition temperature.
The changes continue over a relatively narrow band of
temperatures, therefore, onset and termination Tg values
are observed. Some of the analytical methods to deter-
mine Tg are di€erential scanning calorimetry (DSC),
di€erential thermal analysis (DTA) and thermal me-
chanical analysis (TMA). The former two methods de-
tect changes in Cp , while the third detects changes in the
elastic modulus.
2.1. Viscoelastic kinetics of an amorphous material above
Tg
A glassy state can be considered as a ``solid solution''
(Roos, 1995). As stated earlier, the viscosity of a glass is
around 1012 Pa s. It should be noted that since the
measurement of viscosity of a glass is dicult and can be
erroneous, this value is derived from an extrapolation.
When the temperature of the glass is increased to Tg ,
this material changes to rubbery state with a viscosity
ranging from 106 ÿ 108 Pa s (Downton et al., 1982). The
mechanical properties of a material such as relaxation
time and viscous properties above its glass transition
temperature (T > Tg , e.g. Tg + 100 K) are observed to
follow William±Landel±Ferry (WLF) kinetics (Roos &
Karel, 1991a; Reid, Kerr & Hsu, 1994; Wallack & King;
Fig. 1. Change of physical state of an amorphous glass through rub-
bery (transitional) to crystalline state, T ± temperature, t ± time
(adopted and modi®ed from Bhandari, Datta & Howes, 1997a).
1988, Aguilera, Levi & Karel, 1993). When the property
is viscosity, the WLF equation is described as:
l ÿC1 …T ÿ Tg †
log10 ˆ ; …1†
lg C2 ‡ …T ÿ Tg †
where l is the viscosity, lg the viscosity at glass transi-
tion temperature Tg , T the temperature, and C1 and C2
are ``universal'' constants (17.44 and 51.6 K, respec-
tively). Roos and Karel (1991d) found this relationship
using such ``universal'' constant values valid for an
amorphous sucrose and lactose. Soesanto and Williams
(1981) also validated the WLF equation for an amor-
phous mixture of sucrose and fructose. Aguilera et al.
(1993) also successfully used the WLF equation to pre-
dict the viscosity of ®sh protein hydrolyzate matrix at
the onset of collapse during freeze drying. Nevertheless,
Peleg (1992) has debated the validity of these ®xed
constants after experimenting with some polymers and
amorphous sugars. He suggested that the WLF equation
with the universal coecients cannot be used as a gen-
erally valid model. It was demonstrated that for certain
polymers and sugars this model leads to a considerable
error in the magnitude of viscosity, particularly 20±30 K
above the Tg . Peleg (1996) further reported that the
upward concavity of changes in a transitional region,
which cannot be predicted by WLF or Arrhenius
equation, can be described by a model with the structure
of FermiÕs function. All the above reports indicate that
although the WLF model has been validated for various
food systems, a general application of this model in a
food system needs to be carefully examined.
The same form of this equation can be applied to
predict the way that mechanical properties such as re-
laxation time change above Tg , replacing l with the
relevant property, keeping the same constants. Various
other physico-chemical changes that are related to mo-
lecular mobility and which occur above Tg , may also be
modelled by using the WLF equation (Roos, 1995).
However, as stated earlier an application of the WLF
equation has to be carefully examined in a complex food
system. As an example, in one experiment Roos and
Himberg (1994) reported that the rate of non-enzymatic
browning of a food model system composed of malt-
odextrin, lysine and xylose did not follow the WLF
equation.
2.2. In¯uence of molecular weight of food polymer on Tg
Low molecular weight polymers (e.g. sucrose) and
monomers (e.g. fructose, glucose) in their pure form
have a low glass transition temperature whereas longer
chain molecules have higher glass transition tempera-
tures (Table 1). With a homologous family of food
polymers ± such as maltodextrins with varying average
molecular weights, the Fox and Flory (1950) relation-
ship has been used to determine the e€ect of molecular
 B.R. Bhandari, T. Howes / Journal of Food Engineering 40 (1999) 71±79 73

Table 1 2.3. Prediction of Tg of a multicomponent system

Glass transition temperature of anhydrous sugars and carbohydrate
polymers
The principal solid components of food materials are
Food materials Molecular weight Tg (o C)a b protein, carbohydrates and fat. Carbohydrate constitu-
ents have the largest e€ect in in¯uencing the glass
Fructose 180 5 transition temperature of an amorphous dried food
Glucose 180 31
Galactose 180 32
material. The glass transition temperature of principal
Sucrose 342 62 sugars and other carbohydrate materials found in food
Maltose 342 87 are presented in Table 1. Common sugars such as
Lactose 342 101 fructose, glucose and sucrose have a very low glass
Maltodextrinsb transition temperature, so their in¯uence to depress Tg is
DEc 36 500 100
DE 25 720 121
very notable in sugar-rich foods. The e€ect of protein
DE 20 900 141 and fat on Tg is not signi®cant in common food systems
DE 10 1800 160 such as milk powders (Jouppila & Roos, 1994a,b; Shi-
DE 5 3600 188 mada, Roos & Karel, 1991).
Starch 243d Water is the major component responsible for de-
a
Roos (1993). pressing the Tg of food materials signi®cantly, as water
b
Roos and Karel (1991c). has a very low glass transition temperature of ÿ135o C
c
d
Dextrose Equivalent. (Johari, Hallbruker & Mayer, 1987). Thus, water is
Predicted. considered as a strong plasticiser in a food system.
The glass transition temperature of a mixture of
various compatible components (including water) is a
weight of a food polymer on Tg (Slade & Levine, 1994;
non-linear function of the glass transition temperature
Roos & Karel, 1991c)
of individual components. Gordon and Taylor (1952)
Kg (Eq. (4)) and Couchmann and Karasz (1978) (Eq. (5))
Tg ˆ Tg1 ÿ ; …2†
M proposed mathematical relationships for determining Tg
where Tg is glass transition temperature of the starch of a mixture. The Couchmann and Karasz equation (Eq.
derivative, Kg a constant (ÿ25 000 K), Tg1 the limiting (5)) is basically an extension of Gordon and Taylor
Tg at in®nite molecular weight (starch), and M is the equation (Eq. (4)).
molecular weight. Tg1 and Kg are material dependent. w1 Tg1 ‡ kw2 Tg2
To and Flink (1978) found this equation valid for Tgm ˆ ; …4†
w1 ‡ kw2
glucose polymers with three or more glucose chains.  
Busin, Buisson and Bimbenet (1996) found a linear re- DC
w1 Tg1 ‡ DCp2 w2 Tg2
lationship between Tg (°C) and ÔDextrose Equivalent Tgm ˆ  p1  ; …5†
DCp2
(DE)Õ for maltodextrins over a range of DE (2 ± 100) w1 ‡ DCp1 w2
with a satisfactory correlation coecient (r ˆ 0.98) (Eq.
(3)). They also suggested that the measurement of Tg where Tgm is the glass transition temperature (K) of the
could also be a useful tool to determine the degree of mixture, wi the mole fraction of component i, DCpi the
hydrolysis of starch. change in heat capacity of component i between glassy
and rubbery states and Tgi is the Glass transition tem-
Tg ˆ ÿ1:4…DE† ‡ 449:5: …3† perature (K) of component i.
The Tg of very high molecular weight food polymers The Gordon and Taylor equation is typically applied
such as starches and proteins cannot be determined ex- either to predict the in¯uence of moisture content on the
perimentally as they start decomposing before reaching Tg or to estimate the Tg of a binary polymer mixture e.g.
Tg . In this case, the Tg can only be predicted (Busin et low and high molecular weight carbohydrates (Orford,
al., 1996; Roos, 1995), as an intercept (Tg1 ) of a plot of Parker, Ring & Smith, 1989; Aguilera et al., 1993; Roos,
Tg vs 1/M (Eq. (2)). Alternatively, the multicomponent 1995). In the Gordon and Taylor equation the ratio of
analysis in Section 2.3 could be used, by adding di€ering change in heat capacity can be replaced by a constant (k)
amounts of a plasticising agent such as water and again for a binary mixture (water and solute) to be similar in
looking at the intercept as water content tends to zero. form as Couchmann and Karasz equation. The
Slade and Levine (1994) mentioned that although a Couchmann and Karasz equation (Eq. (5)) can be ex-
general correlation between molecular weight and Tg of tended and applied for multicomponent mixture systems
carbohydrates has been widely accepted, one should be such as water, glucose and sucrose. (Roos, 1995, Ar-
aware that the Tg can vary substantially, even within a vanitoyannis, Blanshard, Ablett, Izzard & Lillford,
series of compounds of the same molecular weight due 1993). For an n component system, the Couchmann and
to their chemical structure. Karasz equation can be expanded as:
74 B.R. Bhandari, T. Howes / Journal of Food Engineering 40 (1999) 71±79
Pn starch derivatives (glucose syrups/maltodextrin at higher
iˆ1 wi DCpi Tgi
Tgm ˆ P n : …6† dextrose equivalent values) has been well recognised
iˆ1 wi DCpi
(Lazar, Brows, Smith, Wang & Lindquist, 1956; Bren-
nan, Herrera & Jowitt, 1971; Karatas & Esin 1990;
Bhandari, Senoussi, Dumoulin & Lebert 1993, Bhandari
3. In¯uence of glass transition temperature on drying
et al., 1997a; Bhandari, Datta, Crooks, Howes & Rigby,
processes
1997b). During the drying of sugar-rich products, they
may either remain as syrup or stick on the drier chamber
Among various food processing operations, drying is
wall. There is also the problem of unwanted agglomer-
the most commonly used method to preserve and reduce
ation in the drier chamber and conveying system. This
the volume of the food materials. The ``collapse'' of the
can lead to lower product yields and operating prob-
dehydrating food during freeze drying, stickiness of the
lems.
product during spray drying, caking and agglomeration
In recent years, the stickiness problem of sugar
of powders during processing and storage are some of
products has been related to their low Tg (Roos et al.,
the properties which are related to the glass transition
1996). Fig. 2 depicts the major changes taking place
temperature. This section illustrates the signi®cance of
during the spray drying of a sugar-rich product. The ®rst
Tg in various types of drying situations.
step of spray drying is the atomisation of feed. Normally
3.1. Freeze drying a concentrate solution is atomised into ®ne droplets of
the range of 20±150 lm diameter (Masters, 1991). As
In the freeze drying process, the liquid product is ®rst water evaporates the solutes in the droplet become more
frozen and then the water is removed by sublimation. concentrated. The physical state of the product changes
This is a low temperature process, and has been con- from liquid as it passes through the drier from solution
sidered superior than other drying methods in terms of to syrup and ®nally to solid form. Based on the report
product quality (taste and aroma) though being rela- published by Bellows and King (1973) and Downton et
tively expensive. While evaporating the moisture, the al. (1982), if the viscosity of the product is below a
product becomes porous in nature and the solid network critical level of 107 Pa s, it may stay as syrup even at low
should be able to hold this porous structure. If the moisture levels. Depending upon the product charac-
temperature of the dehydrating porous product is above teristics and composition, and drying conditions, the
Tg , the viscosity of the solid material may not be enough surface of the drying droplets may remain plastic, re-
to support the structure and ``collapse'' or shrinkage
occurs (To & Flink, 1978; Tsourou¯is, Flink & Karel,
1976). To maintain the original volume, the product
temperature should not be above Tg . This means that
the collapse rate increases as the viscosity of the drying
matrix decreases below 107 Pa s above its Tg . This
structural shrinkage leads to poor aroma retention, poor
rehydration characteristics and uneven dryness. For
various food liquids during freeze drying, the collapse
temperature can vary between ÿ5o C and ÿ60o C (Bel-
lows & King, 1973) depending upon their composition.
Foods which are high in sugars, such as fruit juices, will
have lower collapse temperature. Therefore, it is im-
portant to know the collapse temperature so that the
drying proceeds without any loss of product quality. In
the freeze drying experiments of sugar solutions at var-
ious solids concentration, Tsourous¯is et al. (1976)
showed that the collapse temperatures can be raised by
the addition of high molecular weight materials. It
should be noted here that the increase in collapse tem-
perature is directly related to the increase in glass tran-
sition temperature as stated earlier.
Fig. 2. Schematic representation of physical changes on droplets
3.2. Spray drying
during spray drying process (Dehydration I, II, III represent only the
arbitrary stages of dehydration, l ± viscosity, Tg ± glass transition
The problems of stickiness during spray drying of temperature, Tsurface ± surface temperature of drying particle (Bhandari
sugar-rich foods such as fruit juices, honey and some et al., 1997a).
 B.R. Bhandari, T. Howes / Journal of Food Engineering 40 (1999) 71±79
sulting in sticking on the drier wall or among them-
selves. The amorphous product obtained at the end of
the drying process could, therefore, be either syrup, a
sticky powder or a relatively free-¯owing powder. In a
practical sense, the critical viscosity is reached at tem-
peratures 10±20o C above Tg (Roos & Karel, 1991b). It
can be therefore, assumed that the temperature of the
surface of the particle during spray drying should not
reach 10±20o C above Tg . The glass transition tempera-
ture of sugar-rich products is so low that spray drying of
the pure product is generally not economically feasible.
Some speci®c driers (Birs process) have also been men-
tioned elsewhere to dry fruit juices at very low temper-
ature conditions (Hayashi, 1989). The most common
approach to dry such products has been to add high
molecular weight additives. This has the e€ect of raising
Tg . However, few attempts have been made to develop a
direct relationship between Tg and stickiness in a spray
drying process. In a recent report, Busin et al. (1996)
studied the role of Tg on the stickiness behaviour of the
maltodextrin-sugar (fructose, glucose, and sucrose)
mixtures during spray drying. They used product re-
covery as a measure of the ease of spray drying. Though
the powder recoveries were correlated to Tg in the case
of individual sugar-maltodextrin mixtures, there was no
consistency on recoveries between three sugar-malt-
odextrin mixtures with the function of Tg . Bhandari et
al. (1997b) found a correlation between Tg and the spray
drying behaviour of sugars and fruit juices with a
maltodextrin additive (Fig. 3). They also described an
empirical approach (Eq. (7)) to optimise the amount of
additive needed to be added with the sugar-rich feeds
before spray drying, by assigning drying index values to
individual components according to their ability to be
dried successfully in a spray drying system.
Xn
a i Xi ˆ Y ; …7†
iˆ1
Fig. 3. Drying index versus glass transition temperature for anhydrous
maltodextrin, sucrose, glucose and fructose; and pineapple juice
powder at various pineapple solids:maltodextrin ratios (water activity,
aw ˆ 0.085 ÿ 0.12) (Bhandari et al., 1997b).
75
where Xi is the fractional weight of a component i (e.g.
sucrose, glucose..); ai the index value assigned for that
particular component and Y is the overall index. The
index values were calculated so that when Y < 1 drying is
dicult, Y ˆ 1 indicates a marginally successful drying,
and Y > 1 a successful drying. Powder recovery was
again used as the criterion for spray drying perfor-
mance. Eq. (7) was found to be useful in determining
optimum additive requirements for honey and pineapple
juice drying (Bhandari et al., 1997b).
3.3. Hot air drying
There are no published reports available on the
consequences of Tg on hot air drying, nevertheless the
knowledge gained in other types of drying can also be
logically applied in this process. Hot air drying (®xed or
¯uidised bed) is normally employed to dry solid foods.
There exists glass transition related problems during
drying of sugar-rich products such as fruit leathers and
other products which are extremely sticky. These prod-
ucts appear wet even at very low moisture content. The
leathery texture of these products even at lower moisture
content is due to the plasticising e€ect of low molecular
weight sugars which depress the Tg . Drying of other
fruits may pose problems, particularly if there is an ex-
cessive disruption of skins or tissues resulting in leaching
of sugars out on the surface.
Sticky products will not ¯uidise well in a ¯uidised bed
system (Papadakis & Bahu, 1992). This problem may
also be encountered while dehydrating thick pasty
products and cut fruit pieces. The shrinkage and
toughening of the solids product may also be related to
Tg (Karathanos, Anglea & Karel, 1993). The shrinking
phenomenon is dependent upon the physical pliability of
the material which should be in¯uenced by Tg and other
factors. Karathanos et al. (1993) demonstrated a con-
tinuous shrinkage of celery during drying. This was at-
tributed to the drying air temperature which was well
above Tg for celery undergoing drying. In high sugar
products, if the dehydration temperature is higher than
Tg , the product may be soft during drying. Upon cool-
ing they will harden, as the product temperature drops
below Tg .
4. In¯uence of Tg on physico-chemical changes in the dried
products
To maintain product quality during storage, the
physico-chemical state of the dried product has to be
maintained for as long as possible. Any changes in its
structure not only changes the physico-sensorial attri-
butes but also it may have some other deteriorating
consequences such as increases in the rate of chemical
reactions, structural damage and microbial quality
76 B.R. Bhandari, T. Howes / Journal of Food Engineering 40 (1999) 71±79
changes etc. An overview of these changes is described
below.
4.1. Crystallisation, agglomeration and caking
All amorphous products are metastable and, there-
fore can crystallise over time during storage. The rate of
crystallisation is a function of T ÿ Tg , with increasing
crystallisation rates for higher temperatures (Senoussi et
al., 1995; Roos & Karel, 1991b). Moreover, the amor-
phous form of low molecular weight carbohydrates and
protein hydrolyzates are very hygroscopic (Aguilera, de
Valle & Karel, 1995). If a local portion of the product in
a packaging picks up moisture, the glass transition
temperature is locally depressed for that particular
portion and crystallisation rate will be accelerated. Tight
and orderly molecular packing during crystallisation
cannot generally accommodate excess moisture which
results in the loss of adsorbed water and absorption of
this ejected moisture at the surface of neighbouring
particles creating interparticulate liquid bridges resulting
in ÔcakingÕ (Jouppila & Roos, 1994a,b; Peleg & Hollen-
bach, 1984). Surrounding particles, which absorb
moisture, will also be crystallised and the crystallisation
can proceed as a chain phenomenon.
Crystallisation process rejects impurities including
volatiles. Senoussi et al. (1995) found a loss of diacetyl
as a function of rate of crystallisation of lactose during
storage. They found that when the lactose was stored
20o C above Tg , the amorphous product went through
immediate crystallisation and practically all diacetyl was
lost after 6 days. Levi and Karel (1995) also found in-
creased rate of loss of a volatile 1-n-propanol in an
amorphous sucrose system as a result of crystallisation.
In an enclosed condition, crystallisation increases the
water activity of the system, which may in¯uence the
deteriorating reaction rates and microbial quality of the
product.
4.1.1. Crystallisation kinetics as a function of Tg
The crystallisation rate of various amorphous sugars
(lactose, sucrose, sucrose/fructose) may also be deter-
mined by using WLF kinetics (Eq. (8)), rather than
common Arrhenius theory (Peleg, 1996; Levi & Karel
1995; Roos & Karel, 1991b)
 
t ÿC1 …T ÿ Tg †
log10 ˆ …8†
tg C2 ‡ …T ÿ Tg †
where t is the time of crystallisation, tg the time to
crystallise at glass transition temperature Tg , C1 and C2
are constants (17.44 and 51.6 K, respectively), and T is
temperature (K). As stated earlier in Section 2.1, the
WLF relationship has been found applicable to predict
the viscosity of the glass as a function of temperature.
As crystallisation is directly related to the viscosity and
the molecular mobility of the glass, Roos and Karel
(1991b) found the applicability of Eq. (8) for the crys-
tallisation of amorphous lactose and sucrose. Karel
(1991) also suggested that the use of the WLF equation
for estimation of changes in quality based on the mo-
lecular mobility theory is an oversimpli®cation which
can lead to erroneous results. Once again, the validity of
Eq. (8) which is directly derived from the WLF equation
should also be carefully examined for various complex
food systems.
4.1.2. Caking
The caking of an amorphous powder is a time de-
pendent phenomenon. The rate of caking due to viscous
¯ow property of the material is a function of T ÿ Tg , and
the kinetics of caking can again be determined by using
a WLF-type relationship (Aguilera, et al., 1995).
Downton et al. (1982) studied the following mechanistic
model (Eq. (9)) for estimating the critical viscosity (lc ,
Pa s)
jcs
lc ˆ ; …9†
KD
where j is the dimensionless proportionality constant of
order unity; s the contact time (s) c the surface tension,
for interstitial concentrate 0.07 N/m, and K is the frac-
tion of the particle diameter required as bridge width for
a sucient strong inter-particle bond, usually 0.01±
0.001, D particle diameter (m).
This model relationship is particularly important in
relation to the stability of the dried product during
drying process and storage, as sticking and structural
collapse phenomenon have been found to take place at a
critical viscosity range of 106 ±108 Pa s. It is important to
note that the caking event is time dependent while the
sticking is relatively an instantaneous phenomenon (say
s ˆ 1 s). Longer contact times increase the tendency to-
ward sticking and caking, all other things being equal.
So, the dried product recovered during drying with a
relatively free ¯owing property could also cake in a
collection or packaging container over a period of time
if the surface viscosity is still relatively low due to higher
temperature or moisture levels (Bhandari et al., 1997a).
For this reason, the dried product needs to be cooled
immediately to an appropriate temperature before
packaging. The interesting consequence of this is that
the viscosity (and thus temperature) requirements for
storage are much more stringent than those for pro-
cessing, however, it is still possible to estimate critical
values for both.
Lloyd, Chen and Hargreaves (1996), found in a recent
study that the caking of spray dried lactose was related
to onset temperature of Tg . In their study, they com-
pressed an amorphous lactose powder at various tem-
peratures in a compaction apparatus under a given
pressure and found that the packing density of the
compacted product increased above onset temperature
 B.R. Bhandari, T. Howes / Journal of Food Engineering 40 (1999) 71±79
of Tg , con®rming the viscous ¯ow of the product above
Tg . The ®nal shrinkage of the product under compaction
increased by increasing T ÿ Tg .
4.1.3. Clumping and agglomeration
Agglomeration is itself a process to instantise several
food powders to improve their wettability property. This
is achieved by two means: by applying heat or rewetting
or combining both (Peleg, 1993). If the dried product
with relatively higher moisture content is warmed to
above its glass transition, its surface becomes viscous
and can form a bridge between adjacent particles which
results in an agglomeration (Peleg, 1993; Jouppila &
Roos, 1994a,b; Chuy & Labuza, 1994). The viscosity
should be low enough to cause virtually instantaneous
sticking of the particles as described by Downton et al.
(1982). No direct study on this process with relation to
Tg has been published yet.
The storage of sugar-rich dried product like fruit
pieces or whole fruits such as sultanas in a bulk bin can
cause clumping and the formation of a hard block. This
may be caused by several reasons. The free sugar can
leach out onto the surface due to skin damage. Storing
this product at low temperature, the viscosity of highly
concentrated sugar syrup can increase substantially
which results in cementing the pieces together into a
solid mass. There could also be a partial crystallisation
of sugars and coalescence. This phenomenon can be
more prominent when the products are packed into a
big mass which reduces the interparticulate distance and
can also cause further leaching of the syrup from in-
ternal tissues at the surface. To minimise this problem,
the product is normally surface coated with oil to avoid
a direct contact between sticky surfaces or further de-
hydrated (which will elevate Tg ) (Patil, Chakrawar,
Narwadkar & Shinde, 1995).
4.2. Di€usion of volatiles and permeability to oxygen
The glass transition temperature is directly related to
the molecular mobility of not only the matrix, but also
to water and other components entrapped in the matrix
(Karel, Buera & Roos, 1993; Shimada et al., 1991).
Volatile components in a food system have very limited
mobility in a glassy matrix. Di€usion through the matrix
is primarily through the pores of the matrix (Whorton,
1995). Volatile di€usivity is greatly increased, when the
temperature exceeds the glass transition and continues
to increase with an increase in temperature (Roos &
Karel, 1991a; Whorton & Reineccius, 1995). The rub-
bery matrix above the glass transition temperature ac-
celerates di€usion due to the increased free volume and
change in transport and solution behaviour of volatiles
(Whorton, 1995). Senoussi et al. (1995) studied the re-
tention of diacetyl in spray dried lactose and skim milk
systems and found that the rate of crystallisation and
77
loss of volatiles were related to T ÿ Tg . In a previous
study, To and Flink (1978) found that the loss of vola-
tiles were very high at collapse temperatures during
freeze drying or during storage of the dried product.
Levi and Karel (1995) found an increase in the rate of
release of volatile 1-n-propanol in an amorphous su-
crose:ranose system stored at above Tg .
Permeability to oxygen of glass matrices is also lim-
ited which protects sensitive ¯avours against oxidation.
Gejl-Hansen and Flink (1977) reported that the oxygen
absorption into a freeze dried lipid-carbohydrate system
was increased signi®cantly at the collapse temperature,
which is known to be related to Tg .
4.3. Crispness and plasticity
Crispness and brittleness are some of the important
attributes of dried products. Increased amount of low
molecular weight sugars and water reduce the Tg of the
product and these qualities are lost. The water acts a
plasticiser in a food system. Addition of low molecular
weight sugars and polyols as humectants can maintain
the plastic or chewy texture of intermediate dried
product even at relatively low moisture content by
lowering the Tg of the product. The common humect-
ants used in food are glucose, fructose, sucrose, invert
sugar and glycerol. Loss of crispness due to lowering of
Tg of extruded products has also been reported (Katz &
Labuza, 1981; Nelson & Labuza, 1993; Roos, 1995).
4.4. Chemical changes
The chemical changes in dry products are very slow.
But, because of the increase in internal mobility of re-
actants and di€usivity to oxygen, various chemical re-
actions are also accelerated in dried products if stored at
above Tg . Karmas, Buera and Karel (1992) and Buera
and Karel (1993) found an increased rate of non-enzy-
mic reactions in dried vegetables stored above Tg . The
rate of oxidation of sensitive compounds in a dry matrix
is also enhanced above Tg due to crystallisation which
forces the encapsulated materials from the system to the
surface. Increase in di€usivity of oxygen through the
matrix above Tg also accelerates deteriorative reactions,
but the consequence of crystallisation on oxidation is
signi®cantly higher because of the release of encapsu-
lated sensitive compounds (Gejl-Hansen & Flink, 1977;
Karel, 1985; Shimada et al., 1991; Moreau & Rosen-
berg, 1996). Shimada et al. (1991) reported that an un-
encapsulated methyl linoleate underwent rapid
oxidation while encapsulated oil remained unoxidised,
in a lactose-based food model. Labrousse, Roos & Karel
(1992) reported a complete release of encapsulated
methyl linoleate oil in amorphous sugars due to crys-
tallisation (above Tg ) which resulted into oxidation,
whereas upon collapse there was only a partial release of
78 B.R. Bhandari, T. Howes / Journal of Food Engineering 40 (1999) 71±79
the remained methyl linoleate into the dense matrix.
Moreau and Rosenberg (1996) also found a rapid oxi-
dation of milk fat migrated on the surface as a result of
lactose crystallisation from a whey proteins-milk fat
matrix. In other study, Saltmarch, Vaginini-Ferrari &
Labuza (1981) found a highest rate of Maillard reaction
in a whey powder at 0.44 critical water activity, where
the amorphous lactose began to crystallise releasing
water, consequently mobilising reactants for the
browning reaction.
5. Conclusion
The understanding of glass transition temperature
has advanced signi®cantly in very recent years. Its im-
portance in drying and dried foodstu€s has now been
recognised. Though there are quite a signi®cant number
of publications appearing day by day in this area, not
many have dealt directly with the process in the real
situation, where there are added complexities associated
with variations with time of temperature and composi-
tion, as well as the presence of a range of major and
minor components. Now, the basic understanding of
glass transition can be applied to various food opera-
tions to optimise processing and storage conditions.
References
Aguilera, J. M., de Valle, J. M., & Karel, M. (1995). Caking
phenomenon in amorphous food powders. Trends in Food Science
and Technololgy, 6, 149±155.
Aguilera, J. M., Levi, G., & Karel, M. (1993). E€ect of water content
of the glass transition and caking of ®sh protein hydrolyzates.
Biotechnology Progress, 9, 651±654.
Arvanitoyannis, I., Blanshard, J. M. V., Ablett, S., Izzard, M. J., &
Lillford, P. J. (1993). Journal of the Science of Food and Agriculture,
63, 177±188.
Bellows, R. J., & King, C. J. (1973). Product collapse during freeze-
drying of liquid foods. AICHE Symposium Series, 69, 33±41.
Bhandari, B. R., Datta, N., & Howes, T. (1997a). Problems associated
with spray drying of sugar-rich foods. Drying Technology, 15(2),
671±684.
Bhandari, B. R., Datta, N., Crooks, R., Howes, T., & Rigby, S.
(1997b). A semi-empirical approach to optimise the quantity
required to spray dry sugar-rich foods. Drying Technology, 15(10),
2509±2525.
Bhandari, B. R., Senoussi, A., Dumoulin, E. D., & Lebert, A. (1993).
Spray drying of concentrated fruit juices. Drying Technology, 11(5),
33±41.
Brennan, J. G., Herrera, J., & Jowitt, R. (1971). A study of some of the
factors a€ecting the spray drying of concentrated orange juice on a
laboratory scale. Journal of Food Technology, 6, 295±307.
Buera, M. P., & Karel, M. (1993). Application of the WLF equation to
describe the combined e€ects of moisture and temperature on
nonenzymatic browning rates in food systems. Journal of Food
Processing and Preservation, 17(1), 31±45.
Busin, L., Buisson, P., & Bimbenet, J. J. (1996). Notion de transition
vitreuse appliquee au sechage par pulverisation de solutions
glucidiques. Sciences des Aliments, 16, 443±459.
Chuy, L. E., & Labuza, T. P. (1994). Caking and stickiness of dairy-
based food powders as related to glass transition. Journal of Food
Science, 59(1), 43±46.
Couchman, P. R., & Karasz, F. E. (1978). A classical thermodynamic
discussion of the e€ect of composition on glass-transition temper-
atures. Macromolecules, 11(1), 117±119.
Downton, G. E., Flores-Luna, J. L., & King, C. J. (1982). Mechanism
of stickiness in hygroscopic amorphous powders. Industrial and
Engineering Chemistry Fundamentals, 21, 447±451.
Flink, J. M. (1983). Structure and structure transitions in dried
carbohydrate materials. In M. Peleg, E. B. Bagley (Eds.), Physical
Properties of Foods (pp. 473±521). AVI, Connecticut.
Fox, T. G., & Flory, P. J. (1950). Second-order transition temperature
and related properties of polystyrene I in¯uence of molecular
weight. Journal of Applied Physics, 21, 581±591.
Gejl-Hansen, F., & Flink, J. M. (1977). Freeze-dried carbohydrate
containing oil-in-water emulsions: microstructure and fat distribu-
tion. Journal of Food Science, 42, 1049±1055.
Genin, N., & Rene, F. (1995). Analyse du role de la transition vitreause
dans les procedes de conservation agro-alimentaire. Journal of Food
Engineering, 26, 391±407.
Gordon, M., & Taylor, J. S. (1952). I Non- crystalline copolymers.
Journal of Applied Chemistry, 2, 493±500.
Hayashi, H. (1989). Drying technologies of foods their history and
future. Drying Technology, 7(2), 315±369.
Johari, G. P., Hallbruker, A., & Mayer, E. (1987). The glass liquid
transition of hyperquenched water. Nature, 330, 552±553.
Jouppila, K., & Roos, Y. H. (1994a). Water sorption and time-
dependent phenomenon of milk powders. Journal of Dairy Science,
77(7), 1799±1807.
Jouppila, K., & Roos, Y. H. (1994b). Glass transitions and crystalli-
zation in milk powders. Journal of Dairy Science, 77(7), 2907±2915.
Karatas, S., & Esin, A. (1990). A laboratory scraped surface drying
chamber for spray drying of tomato paste. Lebensmittel-Wissensc-
haft und-Technologie, 23, 354±357.
Karathanos, V., Anglea, S., & Karel, M. (1993). Collapse of structure
during drying of celery. Drying Technology, 11(5), 1005±1023.
Karel, M. (1985). Mobility of food components and phase transitions.
In D. Simatos, J. L. Multon (Eds.), Properties of water in foods (pp.
153±169). Martinus Nijho€, The Netherlands.
Karel, M. (1991). Mobility of food components and food stability. In
A.S. Muzumdar, I. Filkova (Eds.), Drying'91 (pp. 27±35). Inter-
national Drying Symposium, Prague, 1990. Elsevier, Amsterdam.
Karel, M., Buera, M.P., & Roos, Y. (1993). E€ects of glass transitions
on processing and storage. In J. M. V. Blanshard, P. J. Lillford
(Eds.), The glassy state in foods (pp. 13±34). Nottingham University
Press, Leicestershire.
Karmas, R., Buera, M. P., & Karel, M. (1992). E€ect of glass
transition on rates of nonenzymatic browning in food systems.
Journal of Agriculture and Food Chemistry, 40, 873±879.
Katz, E. E., & Labuza, T. P. (1981). E€ect of water activity on the
sensory crispness and mechanical deformation of snack food
products. Journal of Food Science, 46, 403±409.
Labrousse, S., Roos, Y., & Karel, M. (1992). Collapse and crystalli-
zation in amorphous matrices with encapsulated compounds.
Sciences des Aliments, 12(4), 757±769.
Lazar, M. E., Brown, A. H., Smith, G. S., Wang, F. F., & Lindquist,
F. E. (1956). Experimental production of tomato powder by spray
drying. Food Technology, March, 129±134.
Levi, G., & Karel, M. (1995). The e€ect of phase transitions on release
of n-propanol entrapped in carbohydrate glasses. Journal of Food
Engineeering, 24(1), 1±13.
Lloyd, R. J., Chen, D. X., & Hargreaves, J. B. (1996). Glass transition
and caking of spray dried lactose. International Journal of Food
Science and Technology, 31, 305±311.
Mans®eld, M. L. (1993). An overview of theories of the glass
transition. In J. M. V. Blanshard, P. J. Lillford (Eds.), The glassy
 B.R. Bhandari, T. Howes / Journal of Food Engineering 40 (1999) 71±79
state in foods (pp. 103±122). Nottingham University Press,
Leicestershire.
Masters, K. (1991). Spray Drying Handbook (5th ed.). Longman
Scienti®c and Technical, England.
Moreau, D. L., & Rosenberg, M. (1996). Oxidative stability of
anhydrous milkfat microencapsulated in whey proteins. Journal of
Food Science, 61(1), 39±43.
Nelson, K. A., & Labuza, T. P. (1993). Glass transition theory and the
texture of cereal foods. In J. M. V. Blanshard, & P. J. Lillford
(Eds.), The glassy state in foods (pp. 513±517). Nottingham
University Press, Leicestershire.
Orford, P. D., Parker, R., Ring, S. G., & Smith, A. C. (1989). E€ect of
water as a diluent on the glass transition behavior of malto-
oligosaccharides, amylose and amylopectin. International Journal
of Biological Macromolecules, 11, 91±96.
Papadakis, S. E., & Bahu, R. E. (1992). The sticky issues of drying.
Drying Technology, 10(4), 817±837.
Patil, V. K., Chakrawar, V. R., Narwadkar, P. R., & Shinde, G. S.
(1995). Grape. In D. K. Salunkhe, & S. S. Kadam (Eds.),
Handbook of fruit science and technology (pp. 7±38). Marcel
Dekker, New York.
Peleg, M., & Hollenbach, A. M. (1984). Flow conditioners and
anticaking agents. Food Technology, March, 93±99.
Peleg, M. (1993). Glass transitions and the physical stability of food
powders. In J. M. V. Blanshard, & P. J. Lillford (Eds.), The glassy state
in foods (pp. 435±452). Nottingham University Press, Leicestershire.
Peleg, M. (1996). On modeling changes in food and biosolids at and
around their glass transition temperature range. Critical Reviews in
Food Science and Nutrition, 36 (1/2), 49±67.
Peleg, M. (1992). On the use of the WLF model in polymers and foods.
Critical Reviews in Food Science and Nutrition, 32(1), 59±66.
Reid, D. S., Kerr, W., & Hsu, J. (1994). The glass transition in the
freezing process. Journal of Food Engineering, 22, 483±494.
Roos, Y., & Himberg, M. (1994). Nonenzymatic browning behaviour,
as related to glass transition, of a food model at chilling
temperatures. Journal of Agricultural and Food Chemistry, 42,
893±898.
Roos, Y., & Karel, M. (1991a). Phase transitions of mixtures of
amorphous polysachharides and sugars. Biotecnology Progress, 7,
49±53.
Roos, Y., & Karel, M. (1991b). Plasticizing e€ect of water on thermal
behavior and crystallization of amorphous food models. Journal of
Food Science, 56(1), 38±43.
Roos, Y., & Karel, M. (1991c). Water and molecular weight e€ects on
glass transitions in amorphous carbohydrates and carbohydrate
solutions. Journal of Food Science, 56(1), 1676±1681.
Roos, Y., & Karel, M. (1991d). Amorphous state and delayed ice
formation in sucrose solutions. International Journal of Food
Science and Technology, 26, 553±566.
79
Roos, Y. (1993). Melting and glass transition of low molecular weight
carbohydrates. Carbohydrate Research, 238, 39±48.
Roos, Y. H. (1995). Glass transition-related physicochemical changes
in foods. Food Technology, October, 97±102.
Roos, Y. H., Karel, M., & Kokini, J. L. (1996). Glass transitions in
low moisture and frozen foods: e€ect on shelf life and quality. Food
Technology, November, 95±108.
Saltmarch, M., Vaginini-Ferrari, M., & Labuza, T. P. (1981).
Theoritical basis and application of kinetics to browning in
spray-dried whey food systems. Progress in Food and Nutrition
Science, 5, 331±344.
Senoussi, A., Dumoulin, E. D., & Berk, Z. (1995). Rentention of
diacetyl in milk during spray-drying and storage. Journal of Food
Science, 60(5), 894±905.
Shimada, Y., Roos, Y., & Karel, M. (1991). Oxidation of methyl
linoleate encapsulated in amorphous lactose-based food model.
Journal of Agricultural and Food Chemistry, 39, 637±641.
Slade, L., & Levine, H. (1994). Water and the glass transition-
dependence of the glass tranasition on composition and chemical
structure: special implications for ¯our functionality in cookie
baking. Journal of Food Engineering, 22, 143±188.
Slade, L., Levine, H., Ievolella, J., & Wang, M. (1993). The glassy state
phenomenon in applications for the food industry: application of
the food polimer science approach to structure - function relation-
ships of sucrose in cookie and cracker systems. Journal of the
Science of Food and Agriculture, 63, 133±176.
Soesanto, T., & Williams, M. C. (1981). Volumetric interpretation of
viscosity for concentrated and dilute sugar solutions. Journal of
Physical Chemistry, 85, 3338±3341.
To, E. C., & Flink, J. M. (1978). Collapse, a structural transition in
freeze dried carbohydrates II. E€ect of solute composition. Journal
of Food Technology, 13, 567±581.
Tsourou¯is, S., Flink, J. M., & Karel, M. (1976). Loss of structure in
freeze-dried carbohydrates solutions: E€ect of temperature, mois-
ture content and composition. Journal of the Science of Food and
Agriculture, 27, 509±519.
Wallack, D. A., & King, C. J. (1988). Sticking and agglomeration of
hygroscopic, amorphous carbohydrate and food powders.
Biotecnology Progress, 4, 31±35.
Whorton, C., & Reineccius, G. A. (1995). Evaluation of the mecha-
nisms associated with the release of encapsulated ¯avor materials
from maltodextrin matrices. In S. J. Risch, & G. A. Reineccius
(Eds.), Encapsulation and controlled release of food ingredients (pp.
142±159). ACS Symposium Series 590.
Whorton, C. (1995). Factors in¯uencing volatiles release from encap-
sulation matrices. In: S. J. Risch, & G. A. Reineccius (Eds.),
Encapsulation and controlled release of food ingredients (pp. 135±
141). ACS Symposium Series 590.