Journal Engineering 43 (2017) 31±40

ofFood

www.elsevier.com/locate/jfoodeng

RaoultÕs law based food water sorption isotherm

Piotr Lewicki P.

Department of Food Engineering, Warsaw Agricultural University, Nowoursynowska 166, 02-787 Warsaw, Poland
Received 22 March 2017; accepted 6 September 2017

Abstract
new model water sorption isotherm is developed on the basis RaoultÕs law. is assumed that water present in food occurs states, as free water with
A of of It intwo

properties the bulk water and as water hydration. Hydrated molecules are considered as entities with molecular weights larger than those non-
of of new of

hydrated molecules. Hydration reduces the free concentration water thus a€ects water activity of and insolution.

Application the developed equation sorption data showed gives approximation sorption isotherms better than that o€ered by the model.
of to food that it of much GAB

Moreover, predicts in®nite adsorption at a ˆ 1, the property which is not o€ered the equation. The new equation makes possible interpolate
it
w
by GAB it to

isotherms at high water activities close one. The probability the new equation will ®t the food isotherm small is higher than 90% and substantially
to that with RMS

exceeds found model. a two-parameter model makes substantial improvement over three-parameter equation. Ó 2017
that fortheGAB As it the GAB

Elsevier Science rights reserved. Ltd.Al

Notation 1. Introduction
a activity

A constant Raoult's law is a law initiated by Francois M. van Raoult

b constant
constant the model constants
(1830-1901) to study the properties of elements containing
c in GAB

constant the model mass nonvolatile solvents, and and also learn about water. The sound
F, G, H in GAB (g)

k molecular weight (g/m

ole)
from Raoult's law is: "the ideal vapor pressure of the solution is
m number molecules of influenced by the solvent vapor pressure and the mole fraction of
number experimental points the solute contained in the solution". Raoult's Law in Ideal Mixes
M
of

residuals (%)
n is a mixture that obeys Raoult's law. Actually there is no mixture
mean square (%)
that can be considered ideal. But some mixtures of conditions are
root

N water content H2O/g (g d.m

.)

R variance really close to ideal conditions.

RMS
In a solution, some high-energy molecules can use their
u energy to defeat the intermolecular attraction of the surface of the
v
liquid and release themselves to then become steam. The smaller
the intermolecular power, the more molecules can escape at a
Greek symbols certain temperature. At certain temperatures, some of the
a constant
constant
molecules will have enough energy to escape from the surface of
v
the solution. Raoult's Law of Water Activities
explain the size of the degree of attachment to water. Based
Subscripts
c com
plex
on its attachment, water is divided into free water, physically
e experimental bound water, and chemical-bound water. The sound of Raoult's
f h free solid, hydrationm
onolayerpredicted law of water activity is: "Water activity is directly proportional to
m solute total
the number of molecules in the solvent and inversely proportional
s
p water

to the molecule in the solution".

t
Not all mixes are ideal. These nonideal mixtures experience
w

deviations from Raoult's law. There are two kinds of Raoult's law
deviations, namely:
a. Positive deviation
Positive deviations of Raoult's law occur when interactions in
each substance are stronger than interactions in a mixture of
substances (A - A, B - B> A - B). This deviation results in a
0260-8774/00/$ see matter Ó 2017 Elsevier Science rights reserved.
- front Ltd.Al
positive enthalpy (mHmix) (endothermic) and results in an
PII: S0260-8774(99)001 3 0 - 2 increase in The state equilibrium have been mathe-matically
of

described by numerous models (van den Berg

& Bruin, 1981). Models, most commonly used des- cribe to

sorption water foods, are based surface chemistry adsorption

of by on
32 P.P. Lewicki / Journal of Food Engineering 43 (2000) 31±40

is formed water molecules (Franks, 1975)

ion byfour essary annealing small amylose chains. for of The

while egg phosphatidylcholine is hydrated 12.4 ‹ 1 mole water with of transition B-type heating occurs water content above of the to the A-type upon at 18%

(Hauser, 1975). 10 water molecules are associated About with (Buleon, Bail, Colonna Bizot, 1998). Starches consist crys-
Le & of both

phosphatidylcholine headgroups (Crowe, Clegg Crowe, & talline and amorphous regions. crystalline regions typically The exhibit

1998). Urea is hydrated seven moles water and guanidinium with of resistance solvent penetration. Hence, water a€ects the to

ion binds twelve moles water (Nandel, Verma, Singh Jain, of & structure acting as a plasticizer the amorphous regions. of At

1998). su•cient water contents, mobility amorphous regions is of

measurement unfreezable showed substantial amount

A of water that of initiated and swelling
water interacts so strongly with the polymer is observed.
of

other constituentsthat is not able crystallise during it to such a complex system as ¯our dough, unfreezable water is
In

cooling. amount unfreezable water plasma proteins is equal

The of in 23±24% at moisture resistance mixing increases rapidly.
and 28% to

to0.47 g/g, collagen 0.26 g/g matos, Faure, Bonjour Couach, in (Si- & This is due possible interactions between gliadin ®brils.
e€ect to The

1975). egg white, bound water is 0.55 g/g, cornstarch is 0.35

In in it lubricating fect water is present 35% moisture (Daniels, 1975). ef- of at

g/g cellulose is 0.16 g/g (Steinberg Leung,

andin it & 1975). Atthe same time starch shows water protons water content no at 2%

There is no single temperature freezing water foods. of of in The and gluten moisture lower than at 7.5%

thermodynamic freezing of waterin the

(Chinachoti, 1998). Since stailing bread is accompa- of

soybean protein occurs over a range as broad as 40 K nied the shift a mobile immobile water fraction is suggested
by from to an it

(Johari Sartor, 1998). This is mainly due bations the

& to pertur- of that water induces reorganisation amorphous areas releases of and

ice±water equilibrium by the interactions water of withprotein. constraints present leading further crystallisation (Slade to &

internal dynamics proteins recovers dration level 0.1±0.2

The of at hy- Levine, 1991).
g/g protein. Monoclinic lysozyme is active at level hydration 0.2 of of Water exerts its solvent properties water contents higher at

g/g protein (Nagendra, Sukumar Vijayan, 1998). This value & than the monolayer value. Mobilisation point sodium chloride for in

is much lower NaCl±casein system occurs

the at

monolayer capacity (Gregory,

than the am
ountof aw 0.3 1975; Kinsella 1986). starch sucrose glucose system
> (Gal, & Fox, The or

1998). Monolayer coverage lysozyme is completed for at shows mobilisation around a ˆ 0:8 (Duckworth, 1981) in the point
w

300 molecules molecule enzyme. However, complete of water per of the starch and b- amylase mixture, the appearance maltose begins of to

hydration occurs 360±400 water molecules/molecule at of

be noticed a reaches 0.65±0.7. Experiments dextran and when
w
with

lysozyme (Gregory, 1998). tended chain lysozyme needs Ex- of

nitroxide probes showed that reaction ascorbic mechanical with acid in

some 1000±1400 water molecules hydration (Gregory, 1998). for ful

mixture was initiated a 0.75 and was strong at a 0.9 (Simatos, at
w
>
w
>

Collagen contains 0.26 g water/g protein which is unfreezable, but

Le Meste, Halphen, Petro€& 1981).

the water acquires the same heat fusion as bulk water of

Caseinate containing nitroxide radicals showed slow motion
the probe water content below 0.3 g/g (Le Meste Duckworth,
hydration level 0.60 g/g protein. bovine serum bumin the
of at d.m. &

1988). water content higher than 0.25±0.30 g/g d.m. caseinate,

at In al-

At

partial entalphy water becomes identical free water above 0.54 of tothatof

both solute di€usivity and protein ¯exibility is observed. the At

g/g (Simatos et al., 1975). hydration lysozyme is estimated Ful of at

water content 1 g/g protein, mobility more mobile spin- of dry the of the

0.45±0.50 g/g protein, but the reduction water content below of

labelled side chains and probes was about 10 times lower those than of

0.75 g increases denaturation temperature enzyme (Gregory,

g/ of the
same groups molecules dilute solution Meste Duckworth,
the of in (Le & 1988).

1998). The above data are evidence thattwo

liquid homogenoussolution in the hydration range
In

di€erent hydration phases occur. By analogy withion

0±3 g/g d.m. rotational di€usivity increases linearly increasing with

hydration, these phases are denoted as and shell. shell water A B TheA

water content Meste Voilley, 1988). mobilisation was (Le & The point

is similar fraction freezing water. The shell water di€ers to the of non- B from
below 0.01 g/g In dextran, gelatine, casein and starch d.m.

the water by melting temperature and heat and entropies

bulk of
mobilisation was approximately 0.25 g/g caseinates, point d.m. In the

fusion. continues a hydration level about 1.4 g/g which

It to of
paramagnetic probes were immobile water content at

corresponds the amount water whose tional correlation to of m

o-
below 0.25 g/g d.m. water content larger than 1 g/g d.m. all At

times are di€erent from

thatof bulk
probes were mobile Meste, Viguier, (Le Lorient

water (Gregory, 1998). &Simatos, 1990).

is a necessary component crystallo- graphic until cell in
Water of the
suggested
Gilbert (1986) availability water is de®ned by the free that the of

both the and B-type structures starch. and 36 water A- of Four

energy the transfer system. free energy
for fromfood matrix to reaction This

molecules are present most recent models and types models inthe ofA B of

starch, respectively (Imberty Perez, 1988; Imberty, Buleon, & Trau&

Perez, 1991). than 10% water is nec- More of

 P.P. Lewicki / Journal of Food Engineering 43 (2000) 31±40 33

corresponds liquefaction water/water bond to the of isotherms foods wholerange

tion of in the of water

formation. water becomes available chemical reactions when The for activities.
liquid-like cluster water molecules is formed. of

sorption isotherms
Water mechanical mixtures could be of not

predicted basis sorption iso- therms individual components on the of of

2. The model
1997b). Swelling, conformational changes, polymer±polymer
(Lewicki,

interactions, binding ions, crosslinking and plastici- of

is assumed hydration components
It that of food

sation amorphous regions were proposed as the

of rea-
reduces the concentration water with properties water. of of the bulk

sons the observed di€erences. Simply the mobility molecules

for of
Moreover, is assumed hydration molecule forms a complex it that of

was responsible super- imposability experimental isotherms for the lack of of

with properties di€erent those a molecule and the water. than of bulk

those
with pre-dicted.
According this model, food consists in free water with to

above presented examples show some water is so

All the that
properties the bulk water, and hydrated dry matter. Hence, of

strongly bound components water protons give signals by food that

water is divided fractions: free water water hydration
in food into two and in

representing the rigid phase. shells. Thus,

most hydration water promotes molecular

The of mo-

bility conformational changes polymers. ever hydration

mˆ m‡ m …1†
and of How- the

water has still properties water. Hence, hydration process di€erent thanthebulk the re-
t f h

duces the free concentration water e€ectively. this of In the light of and the hydrated dry matter is present the amount in

evidence, surface adsorption theory becomes questionable. the

mc ˆ ms ‡ m …2†
h

Once molecules can move and polymers can change their

the

conformation and state molecular weight:

with

should the process still be envisaged as the surface ad- Mc ˆ Ms ‡ 18:016n: …3†
sorption water molecules shall be considered as the dilutionof or it

process already existing solution? With these assumptions water activity can be expressed
RaoultÕs law the following
of

with in form
:

(1950) proposed application solution modynamics the

Hill of ther- to

adsorption process. The thermody- namic system considered … mf =18 :016 †

aˆ : …4†
consisted non-adsorbed gas being in equilibrium with gas …mf=18:016† ‡ …mc=Mc†
w

in

adsorbed on the adsorbent interface placed somewhere and the di€use

above equation ful®ls boundary

between
The the following

conditions:

adsorbed non-adsorbed gas. Hill's was and theory

modi®ed Maguer applied sorption isotherm potato by Le (1985) and to the of

a ˆ 0; w
m ˆ 0; f

starch. Further application solution thermodynamics water of to a ˆ 1; w

m ! 1: f

sorption isotherms foods was done Kumagai, Iwase, of by

Assuming
Kumagai, Mi- (1994). application solution
mh
zuno and Yano The of

thermodynamics food water sorption isotherms treats the

ˆv …5†
to

condensed phase as the solution. m t

these circumstances seems appropriate

and combining the amount free
Under it to

Eq.(5)withEq.(1) of

use Raoult's express water activity system. approach was lawto in thefood This

water is
used Bone (1973) formulate intermediate moisture foods.
by to The

solution approach water state foods was taken Ross (1975) to in by

mf ˆ …1 ÿ v†m :t …6†
toestimate water activity intermediate mois- ture foods. in
Substitution yields: ofEq.(6)intoEq.(4)

Palnitkar and Heldman (1970) proposed use RaoultÕs law

mt …1 ÿv†=18 :016
the of for

high molecular weight compo- aˆ : …7†

…mt…1ÿ v†=18:016† ‡ …mc =Mc†
w

nents introduction e€ective molecular of food by of

weight. Chen and Karmas (1980) used e€ective molec- ular Rearranging gives: Eq.(7)

weight its use was ambiguous and the method use describe
but foundno to

m
a ˆ …8†
t

relationship between water activity and water

mt ‡ …18:016m c=……1ÿ v†M ††:
concentration. w

c
The objective this paper was develop an equa- based of to tion on

RaoultÕs law and describing water sorp- Since mc and Mcare known, is written as not Eq.(8)

m
aˆ ; …9†
t

m‡ a
w

where
18 :016 mc
aˆ …10†
…1ÿ v†M c
Eq.(9) solved a yields for
34 P.P. Lewicki / Journal of Food Engineering 43 (2000) 31±40

· the parameters the Oswin equation are found in by

plotting log u vs log……1=aw † ÿ 1† (Oswin, 1946). procedure This

excludes point a ˆ 1, hence there is possibility extrapolate at

w
no to

data high water activities. Parameters A and b (15) are to in Eq.

obtained proximate solution the criterion highest by ap- of Eq. (14) with ei- ther the coeient of

determination least standard error estimation. This or the of

procedure lows include the point a ˆ 0; a ˆ 1 calcula- al- to

w
into

tions, hence the data high water activities can be at interpolated.

1. Relationship between a and water activity wheat ¯our

Fig. for and
3. Application to sorption data
sodium caseinate.
sorption isotherms 38 products 31
Water of and

1 model mechanical mixtures (Table were described derived 1) by the

a ˆm ÿ1 : t
…11† equation. comparison, three-pa- For two

a w

rameter models were used analysis. model developed in the The by

Water content is usually expressed as Anderson (1946), (1953) Guggenheim (1966) as model De Boer and know
n theGAB

mt is scribed by the following

uˆ : …12†
de- equation:

ms um kcaw
For ms ˆ 1, (11) can be written as
Eq.

uˆ : …16†
1 …1 ÿ ka †‰1‡ …c ÿ 1 ka
† Š
a ˆu ÿ1 : …13† w w

a w

Having water sorption isotherm a a given water at ac-

Equation developed by Lewicki (1998) is also a three-
tivity can be calculated and the relationship a ˆ f …aw † can parameter model
be found a selected food product. is important notice, a ˆ 1; " #

1 1
for It to that for
w

a ˆ 0. This is agreement with the boundary conditions and uˆF ÿ : …17†

‡
point in

G
should be included calculations. Example a relationship into of
…1ÿ aw† 1 aH w

between a and a is presented 1. w

inFig.

relationship between a and a is very well proximated by

The
w
ap- The Table Curve software (curve ®tting software 2D

the following equation user-de®ned equations, Jandel Sci.) was used ®t tested
with built-in and to

equations experimental iso- therms. The goodness ®t was to of

b measured by calculation residuals R and root mean square of ( ) (RMS)

1 expressed in per cent. Statistical measures were calculated from

aˆA ÿ1 : …14† the

a w following equations:
Combining Eq. (14) with Eq. (13) and solving water for

ue ÿ up
content the following is obtained

bÿ1 Rˆ 100; …18†

1 ue
uˆA ÿ1 : …15†
a s•••••••••••••••
w
Pÿ
This is the food water sorption isotherm derived …ue ÿup †=ue
•••••••••••••••••
onthe
RMSˆ 100 ••••• : …19†
basis RaoultÕs of law.
2
N
Derived equation is like the Oswin (1946) equation, Residuals were calculated using Excel andRMS
soft-

however substantial di€erences analysis occurin of

ware (Microsoft).

sorption data. These are Considering the application analysed equations of to

· the Oswin equation is an empirical one, and is a quency it fre-

sorption data the following assumptions were made

curve derived by Pearson. Equation presented this paper is in

· ®ts P 25% were considered as loaded withRMS

derived the basis the RaoultÕ s law and parameter a has a

on of

large error be accepted,

withtoo to

physical meaning. When amount hydration water an of

· the model being a modi®cation the tilayer adsorption

for GAB of mul-

increases a also creases, and the system which there is in- for in no
developed Brunauer, Emmett by

free Teller (1938) some limits were applied. value k 1 is

and other The of > not

water a approaches in®nity. Moreover, the system in

feasible the thermo- dynamic as as mathematical from well point of

in which hydration doesnot occur (no solutes) a ˆ 0. view.

 P.P. Lewicki / Journal of Food Engineering 43 (2000) 31±40 35

Table 1
Products tested the goodness ®t predicted sorption isotherms for of of

Product Water activity range Source

Apple cellular ®ber 0.11±0.98 Labuza (1985)

Avicel 101 MCC Beef

atgeenan
Carro
PH 0.11±0.90 Wolf et al. (1984) Ow
n

Caseinate, sodium salt 0.03±0.85 results Labuza (1985)Ow

n

Caseinate, sodium salt 0.11±0.98 results

Caseinate, sodium salt 0.022±0.90 results
Ow
n

Casein 0.022±0.865 results results

Ow
n Ow
n

Cellulose, powdered 0.015±0.813 Bizot results

(1983)Ow
n

Cellulose, powdered 0.022±0.865 Labuza (1985) Ow

n

Cellulose, powdered 0.20±0.90 results Labuza

Citrus pulp ®ber Co€ee 0.022±0.865 (1985) Bizot (1983)

bran
Corn ¯our 0.11±0.98 Labuza (1985)
pectin
GelatinLM Macaroni 0.015±0.813 Labuza (1985)
Mushroom Boletus edulis ( )Oat 0.11±0.98 Labuza (1985) Ow
n

bran ¯our 0.05±0.95 results results

Ow
n

Onion 0.11±0.98 Labuza (1985) Ow

n

Pea ¯our 0.11±0.98 results results

Ow
n

Polyglicine 0.11±0.98 Bizot results

(1983)Ow
n Ow
n

Potato ¯akes Rice 0.015±0.813 results Labuza

Rice bran Soy bran ¯our 0.03±0.85 (1985) Labuza
Starch gel, potato
¯our 0.11±0.98 (1985) Bizot,
0.03±0.85 Buleon,
0.03±0.85 Mouhouo-Riou and

0.1±0.9 Multon (1985) Ow

n

0.03±0.85 results Bizot (1983)Ow

n

0.015±0.813 results results

Ow
n Ow
n

0.11±0.98 results results

Ow
n

0.11±0.98 Labuza (1985) Ow

n

0.033±0.980 results Bull (1944)Ow

n

results

Starch gel, potato 0.055±0.973

Starch, Starch, potato potato 0.112±0.903
Starch, Starch, potato potato 0.015±0.813
Tomato Wheat pow
der 0.022±0.865
bran Wheat ¯our ¯our 0.022±0.865
Wool 0.05±0.80
Yeast Saccharomyces cerevisiae
( ) 0.11±0.98
0.015±0.851
Model 0±0.95
Caseinate: Celluloseˆ 1:1 0.03±0.748
Caseinate: Celluloseˆ 1:2
Caseinate: Celluloseˆ 2:1
Starch:Caseinate:Cellulose:Glucose:Citric acid:Saltsˆ 1:1:1:0.5:0.3:0.2
Starch:Caseinate:Celluloseˆ 1:1:1 0.022±0.865 results
Ow
n

Starch:Caseinate:Glucose ˆ 1:1:1 0.022±0.865 results

Ow
n Ow
nresults
Starch:Caseinate:NaCl ˆ 50:33:17 0.022±0.865 results
Ow
n Ow
nresults
Starch:Caseinate ˆ 1:1 0.022±0.865 results
Ow
n Ow
nresults
Starch:Caseinate ˆ 1:1 0.022±0.865 results
Ow
n Ow
nresults
Starch:Caseinate ˆ 1:3 0.022±0.865 results
Ow
n Ow
nresults
Starch:Caseinate ˆ 2:3 0.022±0.865 results
Ow
n Ow
nresults
Starch:Caseinate ˆ 2:3 0.022±0.865 results
Ow
n Ow
nresults
Starch:Caseinate ˆ 2:8 0.022±0.865 results
Ow
n Ow
nresults
Starch:Caseinate ˆ 2:8 0.022±0.865 results
Ow
n Ow
nresults
Starch:Caseinate ˆ 3:1 0.015±0.813 results
Ow
n results
Ow
n

Starch:Caseinate ˆ 3:2 0.022±0.865

Starch:Caseinate ˆ 3:2 0.015±0.813
Starch:Caseinate ˆ 8:2 0.022±0.865
Starch:Caseinate ˆ 8:2 0.022±0.865
Starch:Caseinate ˆ 8:2 0.015±0.813
Starch:Cellulose:Glucose ˆ 1:1:1 0.022±0.865
0.015±0.813
0.022±0.865
0.022±0.865
0.022±0.865
36 P.P. Lewicki / Journal of Food Engineering 43 (2000) 31±40

Table 1 (Continued)

Product Water activity range Source

Starch:Celluloseˆ 1:1 0.022±0.865 results

Ow
n

Starch:Celluloseˆ 1:2 0.022±0.865 results

Ow
n results
Ow
n

Starch:Celluloseˆ 2:3 0.015±0.813 results

Ow
n results
Ow
n

Starch:Celluloseˆ 2:8 0.015±0.813 results

Ow
n results
Ow
n

Starch:Celluloseˆ 3:2 0.015±0.813 results

Ow
n results
Ow
n

Starch:Celluloseˆ 8:2 0.015±0.813 results

Ow
n

Starch:Citric acidˆ 1:1 0.022±0.865

Starch:Glucose ˆ 1:1 0.022±0.865
Starch:Glutamic acidˆ 4:1 0.022±0.865
Starch:NaCl ˆ 4:1 0.022±0.865

Moreover, keep the error u within‹15% to of

m
the highest probability
The ®t experimental to datawith

value c be greater 5.5 (Lewicki,

of must than small gives LewickiÕs equation (Table Smaller
RMS 4).

1997a). Hence, ®ts k 1 c 5.5 were consid- ered as not with > or < probability is quaranted when the new equation is used. The
ful®lling theoretical requirements, regard- less the calculated of least probability is expected when the model is applied. Since GAB

RMS. RaoultÕs law based equation has only parameters its applicability two

Results curve ®tting are collected Table 2. mato powder

of in To- water sorption iso- therms seems be very good, and makes
to food to

sorption isotherm could be described by any the tested not of substantial improvement over the equation. GAB

equations su•cient accuracy. with Results presented this paper are in di€erentfrom

Dried onion and mechanical mixture starch and so- of those published Bakshi Labuza, (1985a,b). Analysis over by Lomauro, and of

dium chloride water sorption isotherms could be described only

300 isotherms showed the model gave the highest percentage that GAB of

by the developed equation. Neither LewickiÕs equations could GAB nor ®ts mean relative deviation modulus less than ®ve. That was
with

describe these isotherms su•cient accuracy. with from 42.9% meats 100% nuts and oilseeds. Oswin equation, for to for The

following number ®ts ful®lling

The above-men- tioned of which was also analysed Lomauro and co-workers, was as by not

assumptions was obtained 69 analysed isotherms: the model out of GAB good as the equation. GAB

± 56; LewickiÕsequation ± 65 percentage ®ts means relative deviation

The of with

and the developed equation ± 68, which is 81%, 94% and modulus less than 5 were 0 products from for milk to

99%, respectively. Hence, the probability receiving ®t with < of RMS

93.8% nuts and oilseeds. for

25% is around 80% the model, and higher than 90% other for GAB for two
Equation derived in this paper and based on RaoultÕs law
analysed equations. takes into account, besides the experimental points, a a ˆ 0; point

analysis LewickiÕs new equation was done only those

Further of GAB, and the for
a ˆ 1. This makes a substantial ence over the procedure
w
di€er-

proposed Oswin (1946) Lomauro et al. (1985a) and probably

isotherms, which
by or

increases goodness ®t analysed isotherms.

ful-

®lled the assumed limits.

of to

Equations analysed in this paper contain two three

Residuals calculated by Eq. (18) express the di€erence
or

between experimental and predicted values a at point.

parameters. was interesting see precision parameters It to how of

estimation in¯uences variance culated water content. was

frequency residuals is presented 2. relationship between
total of cal- It

assumed standard deviation each parameter is equal the es-

The of in Fig. The

that of to‹5% of

frequency and the level residuals is well described by the of

timated value. The principle variance additivity was used in of

Pearson distribution. The most frequent residuals are collected this analysis (see Appendix and the results are presented 6. is A) in Fig. It

Table 3.
in
evident model needs a high precision the parameters
that the GAB of

Analysis cumulative frequency residuals (Fig. shows that the

of of 3)
estimation a if

probability obtain ®t water sorption isotherm within a given

to of
®t small is expected. Standard deviation predicted water
with RMS of

range residuals is very of dif-

content is strongly dependent water activity and a 0.8 is larger on at
w
> it than

analysed equations.
ferent for For Eq. (17) and
10%. Eq.(17)

ÿ56 R 6 ‡ the probability is 65.2%, the GAB 5% for

model is 42.6% and the developed equation is

it for it

water activity, and approaches ‹ 10% water activities close at to

range residuals is extended

43.1%. If the of to

one. developed equation yields results similar those obtained

ÿ106 R 6 ‡ 10% the respective probabilities are 84.0%,
The to

LewickiÕs equation a 0.9 estimated water content is loaded

for butat > with

74.0% and 70.5%.

w

larger error than calculated that forEq(1

. 7).

frequency analysed equations is presented in Fig. 4. The

The of RMS for

most frequent RMS Eq. for (17)

yields standard deviation, which is little dependent on

is 2:24%, the model is 9:47% and for GAB it forthe

new equation is 4:66%. Cumulative frequencies it of

the analysed equations are presented in Fig. 5.

RMSfor
 P.P. Lewicki / Journal of Food Engineering 43 (2000) 31±40 37

Table 2
Constants analyzed equations in

Product Equation

GAB Eq. (17) Eq(1

. 5)

u m
k c F G H A b

Apple cellular ®ber solution

No 0.1989 0.5149 4.3814 0.0955 0.2611
Avicel 101 MCC Beef
Carro
ag
teenan
PH 0.0390 0.7895 9.47 0.0975 0.2698 0.7333 0.0540 0.5657
Caseinate, sodium salt 0.0377 0.9812 14.47 0.0602 0.7576 0.6169 0.0758 0.4942
Caseinate, sodium salt 0.1162 0.9035 42.12 0.3850 0.2985 0.7215 0.2220 0.5838
Caseinate, sodium salt solution
No 0.2115 0.4483 3.0198 0.0971 0.1955
Casein 0.0607 0.8110 6.86 0.1249 0.3738 0.7095 0.0807 0.4908
Cellulose, pow
dered 0.0510 0.8853 15.56 0.1047 0.4724 0.4360 0.0873 0.5850
Cellulose, powdered 0.07801 0.7541 7.63 0.1820 0.2622 0.7571 0.0958 0.5048
Cellulose, powdered 0.0858 0.6540 7.12 0.2276 0.1345 0.9299 0.0959 0.6060
Citrus pulp ®ber Co€ee 0.0342 0.8489 11.91 0.1057 0.2236 0.8127 0.0549 0.5826
bran
Corn ¯our 0.0285 0.7294 12.30 0.0715 0.2234 0.6476 0.0373 0.5830
pectin
GelatinLM Macaroni 0.0187 0.8750 15.26 0.0399 0.4424 0.4812 0.0311 0.5739
Mushroom Boletus edulis ( )Oat 0.0435 0.9422 19.35 0.1238 0.4170 0.6027 0.0902 0.5252
bran ¯our solution
No 0.1509 0.3776 9.6308 0.0440 0.2284
Onion Pea ¯our 0.0432 0.1338 0.3642 0.5131 0.0933 0.5701
Polyglicine 0.0923 0.9237 56.18 0.2752 0.3981 0.5763 0.1958 0.5456
Potato ¯akes Rice 0.0828 0.9380 114.06 solution
No 0.1427 0.4377
Rice bran Soy bran ¯our 0.0426 0.9541 7.26 0.0859 0.0715 0.4802
Starch gel, potato
¯our solution
No 0.9071 10.61 0.0653 0.5098 0.5628 0.0799 0.5421
Starch gel, potato 0.0354 0.1156 0.8038 0.2979 0.0854 0.6070
Starch gel, potato solution
No

0.9288 43.20 solution

No
0.3507 0.3852 0.0982 0.2990
Starch gel, potato 0.0539 0.1223 0.0978 0.6679
Starch gel, potato 0.0883 0.2153 0.0835 0.5745
0.8489 53.18 0.3869 0.2825
Starch gel, potato 0.0545 0.1163 0.1042 0.6308
0.6268 5.55 0.1328 0.9649
Wheat bran Wheat ¯our 0.0571 0.2116 0.1085 0.5001
¯our
0.8937 43.78 0.4712 0.2971
Wool 0.0378 0.0724 0.0960 0.4696
0.6770 9.20 0.1923 0.7401
Yeast Saccharomyces cere- ( 0.0392 0.1013 0.0881 0.5461
0.9843 42.89 0.7012 0.1515
visiae ) 0.0633 0.2067 0.1263 0.6020
0.9488 123.06 0.4592 0.3120
0.0643 0.2079 0.1177 0.6299
Model 0.8976 42.27 0.2894 0.5150
0.0986 0.2521 0.1249 0.5785
C:Celˆ 1:1 0.8484 29.13 0.2351 0.6069
0.0942 0.2213 0.1222 0.5656
C:Celˆ 1:2 0.7567 8.37 0.2100 0.7940
0.0871 0.2199 0.1234 0.6233
C:Celˆ 2:1 0.7293 10.75 0.2500 0.6484
S:C:Cel:G:CA:Salts- 0.0961 0.2436 0.1283 0.5840
ˆ 1:1:1:0.5:0.3:0.2 0.7453 17.42 0.2376 0.5572
solution
No 0.0801 0.0960 0.4696
S:C:Celˆ 1:1:1 0.7306 13.61 0.2216 0.6287
0.0810 0.2041 0.1144 0.6054
S:C:Gˆ 1:1:1 0.6860 0.2209
0.0733 0.2087 0.1056 0.6157
S:C:NaClˆ 50:33:17 0.2322 035634
solution
No 0.0592 0.0878 0.5851
S:Cˆ 1:1 0.7358 17.40 0.2121 0.6456
S:Cˆ 1:1 0.7819 16.13 0.9699 0.1393
S:Cˆ 1:3
S:Cˆ 2:3
S:Cˆ 2:3
S:Cˆ 2:8
S:Cˆ 2:8 0.0445 0.8146 17.20 0.0877 0.3063 0.5341 0.0556 0.5832
S:Cˆ 3:1 0.0361 0.8042 17.20 0.0877 0.3063 0.5341 0.0556 0.5832
S:Cˆ 3:2 0.0500 0.8232 13.90 0.1167 0.3400 0.5633 0.0767 0.5624
S:Cˆ 3:2 solution
No 0.0832 0.6083 1.2530 0.0638 0.2666
S:Cˆ 8:2
0.0553 0.7970 16.07 0.1348 0.2974 0.5549 0.0834 0.5804
solution
No 0.0446 0.9028 0.0555 0.0725 0.5648
solution
No 0.0556 1.2602 0.1110 0.0835 0.6248
0.0703 0.7779 18.96 0.1753 0.2757 0.4788 0.1089 0.6405
0.0747 0.7827 11.04 0.1846 0.2843 0.6525 0.1063 0.5335
0.0654 0.8036 20.50 0.1587 0.3077 0.4903 0.1029 06190
0.0686 0.7828 14.14 0.1614 0.3005 0.5631 0.1007 0.5805
0.0693 0.8154 16.51 0.1637 0.3288 0.5281 0.1076 0.5932
0.0662 0.8143 14.15 0.1498 0.3486 0.5359 0.1011 0.5749
0.0621 0.8515 17.11 0.1403 0.3875 0.4903 0.1025 0.5810
0.0844 0.7374 20.45 0.2180 0.2243 0.5390 0.1210 0.6250
0.0707 0.7929 30.34 0.1747 0.2915 0.4295 0.1139 0.6589
0.0744 0.7645 10.43 0.1734 0.2836 0.6521 0.1012 0.5306
0.0808 0.7453 14.78 0.1985 0.2502 0.5882 0.1135 035769
38 P.P. Lewicki / Journal of Food Engineering 43 (2000) 31±40

Table 2 (Continued)

Product Equation

GAB Eq. (17) Eq(1

. 5)

u m
k c F G H A b

S:Cˆ 8:2 0.0793 0.7742 25.11 0.1986 0.2682 0.4726 0.1224 0.6459
S:Cˆ 8:2 0.0725 0.7876 8.65 0.1686 0.2982 0.7029 0.0964 0.5074
S:Cel:Gˆ 1:1:1 0.0217 0.9842 18.48 0.0379 0.7240 0.3072 0.0466 0.4632
S:Celˆ 1:1 0.0525 0.7704 14.27 0.1297 0.2658 0.5956 0.0747 0.5762
S:Celˆ 1:2 0.0397 0.8182 16.20 0.0949 0.3265 0.5418 0.0618 0.5643
S:Celˆ 2:3 0.0425 0.7891 20.70 0.1094 0.2977 0.4974 0.0698 0.6088
S:Celˆ 2:8 0.0319 0.8281 16.63 0.732 0.3585 0.5113 0.0507 0.5760
S:Celˆ 3:2 0.0544 0.8481 16.96 0.1223 0.3892 0.4949 0.0890 0.5713
S:Celˆ 8:2 0.0652 0.8234 19.83 0.1519 0.3475 0.4830 0.1047 0.5964
S:CAˆ 1:1 solution
No 0.0444 1.1183 0.3103 0.0579 0.5776
S:Gˆ 1:1 0.0329 0.0544 0.6526 0.1937 0.0602 0.6170
S:Glut. acidˆ 4:1 0.0582 0.9382 20.00 0.1365 0.3645 0.3505 0.1034 0.6461
S:Naclˆ 4:1 solution
No
0.8422 38.71 solution
No 0.10165 0.5687

C ± caseinate, ± citric acid, Cel ± cellulose, ± glucose, acid ± glutamic acid, and S ± potato
CA G Glut. starch.

Fig. 3. Cumulative frequency residuals analysed equations.

of for

Fig. 2. Frequency residuals analysed equations.

of for

Table 3
Frequency residuals of

Equation Residuals (%) Frequency (%)

Eq.(15) 3.02 24.2

GAB
Eq.(17) 0.33 56.3
0.90 40.2

4. Conclusions

Equation derived this and based on RaoultÕs in work

gives approximation water sorption iso- therms better

law of food much than that o€ered by the

GAB Moreover, predicts in®nite adsorption a ˆ 1, the

model. it at
w

property which is o€ered the GAB equation. The equation

not by
Fig. 4. Frequency analysed equations.
ofRMS
for

allows predict water adsorp- tion a close one, because a point

to at
w
to

a ˆ 0, a ˆ 1 is included calculations. is a two-parameter

w
into This

a€ectsestimated water content practically independently

equation and the error the parameters are loaded
water activity. a two-parameter equation, the rived equation
with

of As de-

makes substantial improvement over three-parameter the GAB model.

 P.P. Lewicki / Journal of Food Engineering 43 (2000) 31±40 39

where
ou 1 1
ˆ ÿ ;
oF …1ÿ aw† G 1 ‡ aH w

ou F ln …1 ÿaw †
ˆÿ ;
oG …1ÿ aw† G
ou FaH ln aw
ˆÿ 2
: w

oH 1 ‡ aH w

Forthe model the variance u is given GAB of by

2 2 2
ou ou ou
v…u† ˆ v…um† ‡ v…c† ‡ v…k†;
ou m
oc ok
Fig. 5. Cumulative frequency analysed equations.
ofRMS
for
where
ou cka
ˆ ;
w

ou …1ÿ kaw†‰1‡ …c ÿ 1†kawŠ

m

ou um ka
ˆ ; w

Table 4 oc ‰1‡ …c ÿ 1†kawŠ2

Probability ®t water sorption isotherm assumed
um …c ÿ1 †aw
to with RMS

ou um aw
ˆ 2
‡ 2
:
Equation Probability (%) ok …1ÿ kaw† ‰1 ‡ …c ÿ 1†kawŠ
ÿ2:56 RMS ÿ56 RMS ÿ106 RMS Eq. (15) developed this the variance u is
For in work of

6 2:5 65 6 10 expressed by the following equation

2 2
Eq.(15) 12.4 43.9 80.8 ou ou
GAB
Eq.(17) 40.4 62.1 82.3 v…u† ˆ v…A† ‡ v…b†;
8.9 24.2 67.6 oA ob
where
bÿ1
ou 1
ˆ ÿ 1 ;
oA a w

bÿ1
ou 1 1
ˆA ÿ1 ln ÿ1 :
ob a w
a w

References

Anderson, (1946). Modi®cation the Brunauer, Emmett R.B. of and

Teller equations. Journal of the American Chemical Society 68 , ,

686±691.
Bone, (1973). Water activity intermediate moisture foods. Food
D. in

Technology27 71±76. , (4),

construct sorption isotherms. Jowit, Escher, B. Hallstro• m,

Bizot,H. (1983).Usingthe GAB m
odetlo InR. F. H.F.T.Me€ert,

W. Spiess Vos, Physical properties of foods (pp. 43±53). London:

E. L. & G.

Applied Sci.
6. Relationship between predicted water content and water
Fig. ac-

Bizot, Buleon, Mouhouo-Riou, J. (1985).

tivity analysed equations. Parameters estimated with‹5%
H., A., N. & Multon, L.

for error.

Some facts concerning water vapour sorption hysteresis on starch. potato n

I D.

Simatos J. Multon, Properties of water in foods in relation to quality

& L.

and stability 83±93). Dordrecht: Martinus Nijho€. (pp.

Brunauer, S., Emmett, P. Teller, (1938). Adsorption gases multimolecular H., & E. of on

layers. Journal of the American Chemical Society 60309±319. , ,

Appendix A Buleon, Bail, Colonna, Bizot, (1998). Phase polymorphic transitions

A., Le P., P., & H. and of

starches low and intermediate water contents. S. Reid, The properties of

at n
ID.

Considering equation developed the by Lewicki water in foodsISOPOW ,

(1997b) the variance u is equal of to

6 (pp. 160±178). London: Blackie.

2 2 2
ou ou ou
v…u† ˆ v…F † ‡ v…G† ‡ v…H †;
oF oG oH
40 P.P. Lewicki / Journal of Food Engineering 43 (2000) 31±40

Bull, B. (1944). Adsorption water vapor by proteins. Journal of

H. of Meste, Voilley, (1988). In¯uence hydration rotational
M.,& A. of on

the American Chemical Society 661499±1507. , , solutes systems. Journal of Physical

di€usivityof inm
odel

Chen, Karmas, (1980). Solute activity e€ect water activity. Lebensmittel

A. C. C., & E. on Chemistry 921612±1616. , ,

Wissenschaft und Technologie14101±104. , , Meste, Viguier, Lorient, Simatos, (1990). Rotational di€usuvity solutes
M., L., D., & D. of in

Chinachoti, (1998). dynamics properties water relation thermal characteristics

P. NMR of n
i to concentrated caseinates: in¯uence glycosilation. Journal of Food of

in bread. S. Reid, The properties of water in foods, ISOPOW 6 (pp. InD. Science55724±727. , ,

139±159). London: Blackie. Lewicki, (1997a). The applicability the model water sorption isotherms.
P. P. of GAB to food

Crowe, J. Clegg, J. S., Crowe, (1998). Anhydrobiosis: water replacement

H., & L. M. the International Journal of Food Science and Technology32553±557. , ,

hypothesis. S. Reid, The properties of water in foods, ISOPOW 6 In D. Lewicki, P. P. (1997b). Water sorption isotherms and their estimation model n
i food

(pp. 440±455). London: Blackie. mechanical mixtures. Journal of Food Engineering,

Daniels, W. (1975). Some e€ects water in wheat ¯our doughs.
N. R. of 3247±68. ,

Duckworth, Water relations in foods 573±586).

InR.B. (pp. New Lewicki, (1998). three parameter equation moisture sorption isotherms.
P. P. A for food

Academic Press.
York: Journal of Food Process Engineering 21 127± , ,

Boer, J. (1953). The dynamic character of adsorption Clarendon Press.

De M. . Oxford: 144.
Duckworth, B. (1981). Solute mobility relation water content water activity. R. in to and In Lomauro, J. Bakshi, S., Labuza, P. (1985a). Evaluation food moistureC. A. & T. of

Rockland Stewart, Water acivity: in¯uences on food quality

L. B. & G. F. (pp. sorption isotherm equations. Part vegetable and meat products. I. Fruit,

295±317). Academic Press. New York: Lebensmittel Wissenschaft und Technologie 18 ,

Franks, (1975). Water, ice and solutions simple molecules.

F. of InR. 111±117.
Duckworth, Water relations in foods 3±22). Academic Press.
B. (pp. NewYork: Lomauro, J. Bakshi, S., Labuza, P. (1985b). Evaluation moisture sorption
C. A. & T. of food

S. (1975). Solvent versus non-solvent water casein±sodium chloride±water

Gal, in isotherm equations. tea, nuts, oilseeds, spices starchy foods. Part I. Milk, co€ee, and

system. Duckworth, Water relations in foods (pp. 183±191). New In R. B. York: Lebensmittel Wi- ssenschaft und Technologie 18118±124. ,

Academic Press. Nagendra, Sukumar, Vijayan, (1998). Role water HG

. ., N.,& M. of in

S. (1986). concepts water activity storage stability. Charalambous,

Gilbert, G. New on and n
I G. plasticity, stability and action proteins: the crystal structures lysozyme very of of at

Proceedings of the fourth interna- tional ¯avor conference, the levels hydration. Proteins: Structure, Functions and Genetics 32
low of , (2),

shelf life of foods and beverages (pp. 229±240.

791±802). Amsterdam: Elsevier. S., S., Singh,
Nandel,F. S. (1998).
Verm
a, B., & JainD
, . V.

Gregory, B. (1998). Protein hydration and glass transition. S.

R. InD. Mechanism hydration urea guanidinum a model study deneturation of of and ion: of of

Reid, The properties of water in foods, ISOPOW 6 57±99). (pp. proteins. Pure Applied Chemistry 70 , (8),

London: Blackie. 659±664.

Guggenheim, E. (1966). Application of statistical mechanics Clarendon A. . Oxford: Oswin, (1946). The kinetics package life. The isotherm.
C. R. of II.

Press. Journal of the Society of Chemical Industry 65419±421. Palnitkar, , , M. P., &

Hauser, (1975). Water/phospholipid interactions.

H. In R. BD
.uck- Heldman, (1970). prediction D. R. Direct of

worth, Water relations in foods (pp. 37±71). New Academic York: heat sorption dry moisture equilibrium data. In of for foodfrom

Press. Chen, Karmas (1980). Solute activity activity. Lebensmittel

A.C. C. & E. e€ectonw
ater

Hill, (1950). Statistical mechanics adsorption. Adsorption thermodynamics

T. L. of IX. Wissenschaft und Technologie 14 ,

and solution thermodynamics. Journal of Physical Chemistry 18 , (3), 101±104.

246±256. Ross, (1975). Estimation water activity in intermediate
K.D. of foods.

Imberty, Perez, S. (1988). revisit three dimensional structure the B-type

A., & A to the of Food Technology29 26±34. , (3),

amylose. Biopolymers 27 1205±1221. , (8), Simatos, Faure, Bonjour, Couach, (1975). Di€er- ential thermal analysis
D., M., E.,& M.

Imberty, Buleon, Trau, Perez, S. (1991). Recent advances

A., A., V.,& and di€erential scanning calorimetry study water foods. Duckworth, n
i the of in In R. B.

knowledge strach structure. Starch/StaÈrke 43 (10), 375±384. Johari,

in of , G.P., & SartorG
,. Water relations in foods (pp. 193±209). New Academic Press. York:

(1998). Thermodynamic kinetic features vitri®cation and phase and of Simatos, Le Meste, Petro€, Halphen, (1981). Use electron spin resonance
D., M., D.,& G. of for

transformations proteins and other constituents hydrated soybean, a high of of dryand the study solute mobility relation moisture content in model food systems. of n
i to InL.

protein cereal. S. Reid, The properties of water in foodsISOPOW 6 InD. , B. Rockland G. Stewart, Water acivity: in¯uences on food quality (pp. & F.

(pp. 103±138). London: Blackie Academic. 319±346). New Academic Press. York:

Kinsella, J. Fox, (1986). Water sorption by proteins: whey protein. CRC

E.,& P.F. milk and Slade, Levine, (1991). Beyond water activity: recent advances
L.,& H.

Critical Reviews in Food Science and Nutrition 24 91±139. , (2), based on an alternative approach the assessment food quality and safety. to of

Kumagai, Iwase, Kumagai, Mizuno, T. (1994). solution

H., M., H., A., & Yano, Application of CRC Critical Reviews in Food Science and Nutrition 30 , ,

thermodynamics water sorption isotherms materials. Bioscience, to the of food 115±359.

Biotechnology, Biochemistry58 475±481. , (3), Steinberg, Leung, (1975). Some applications wide-line pulsed M. P. & H. of and NMR n
i

Labuza, P. (1985). Water binding humectants. Simatos J.

T. of n
ID. & investigations water foods. Duckworth, Water relations in foods of in In R. B. (pp.

Properties of water in foods in relation to quality and stability

L. Multon, (pp. 233±248). Academic Press. New
York:

421±445). Dordrecht: Martinus Nijho€. van den Berg, Bruin, S. (1981). Water activity and its estimation C.,&

Maguer, (1985). Solution thermodynamics starch system. Simatos J.

Le M. and the In D. & L. systems: theoretical aspects. Rockland F. Stewart, Water acivity:
in food In L. B. & G.

Multon, Properties of water in foods in relation to quality and in¯uences on food quality 1±61). Academic Press. (pp. NewYork:

stability 133±151). Dordrecht: Martinus Nijho€. (pp. Spiess,

Wolf, W., Jung, Weisser, Bizot, & Duckworth, B. (1984). The water-
W. E. L., G., H., H., R.

Meste, Duckworth, B. (1988). Dynamic behaviour probes spin labelled side

M.,& R. of and vapour sorption isotherms microcrystalline cellulose puri®ed starch. of (MCC) and of potato

chains during dehydration caseinates. International Journal of Food of Results collaborative study. Journal of Food Engineering 351± of , ,

Science and Technology23457±466. , , 73.