Food Hydrocolloids 73 (2017) 335e343

Contents lists available at ScienceDirect

Food Hydrocolloids
journal homepage: www.elsevier.com/locate/foodhyd

Effect of the type of carbohydrate on the DVS critical relative humidity

in spray-dried fat-filled pea protein-based powders: Comparison with
monolayer coverage and Tg values
Ewa Domian a, Anna Brynda-Kopytowska a, *, Jolanta Cie
sla b, Ewa Ostrowska-Lige˛ za c
a
Department of Food Engineering and Process Management, Faculty of Food Sciences, Warsaw University of Life Sciences e SGGW, Nowoursynowska 159c
St., 02-776 Warsaw, Poland
b
Institute of Agrophysics, Polish Academy of Sciences, Do
swiadczalna 4, 20-290 Lublin, Poland
c
Department of Chemistry, Faculty of Food Sciences, Warsaw University of Life Sciences e SGGW, Nowoursynowska 159c St., 02-776 Warsaw, Poland

a r t i c l e i n f o a b s t r a c t

Article history: The objectives of this study were to: (1) determine the critical relative humidity (RHc) using DVS
Received 26 April 2017 ramping for spray-dried fat-filled pea protein-based powders formulated with different types of soluble
Received in revised form fiber as a carbohydrate component and different oil types, and (2) compare the RHc values to the
10 July 2017
monolayer coverage values obtained via DVS water vapor isotherm, as well as the MDSC Tg values. The
Accepted 12 July 2017
Available online 18 July 2017
experimental design included three compositional variants of powders, where the oil fraction was made
up of the blend of palm and rapeseed or rapeseed at different ratios (52.5% or 35% w/w total solids). Each
group of model powders was formulated with four different carbohydrates (Nutriose, inulin, poly-
dextrose, and trehalose), wherein a carbohydrate-to-protein-material ratio was at 5.3:1 in all cases.
Results demonstrated that changes of the RHc of spray-dried fat powders were significantly affected by
the type of carbohydrate. Neither fat type nor fat content had a significant effect upon the RHc of
powders. Powders with Nutriose showed the highest RHc at 25
C (70e80%), followed by powders with
inulin (RHc 30e40%) and powders with trehalose and polydextrose (RHc 30e35%). A strong positive
correlation was found between RHc and the monolayer coverage values obtained via DVS water vapor
isotherm, as well as the MDSC Tg values. In addition, a strong negative correlation was demonstrated
between RHc and specific water activity “as is” for powders after their spray drying. In turn, no corre-
lations were found between moisture content in powders after their spray drying and RHc, the mono-
layer coverage values, as well as the MDSC Tg.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction spectroscopy and X-ray diffraction constitute a basic analytical set-

Spray drying is one the most widespread methods of
manufacturing amorphous systems (Ho, Truong, & Bhandari, 2016).
The characterization of amorphous or partially amorphous food
materials has been of particular interest in recent years (Bhandari &
Roos, 2016). Modern quality standards require accurate quantifi-
cation of non-equilibrium states and glass transitions in the food
materials to minimize manufacture issues and batch-to-batch
variations of the finished products. Full characterization of any
amorphous material is paramount in the development of successful
food systems. Calorimetric methods, dynamic vapor sorption,
* Corresponding author.
E-mail address: anna_brynda@sggw.pl (A. Brynda-Kopytowska).
up used for qualitative and quantitative characterization of disor-
dered materials.
The encapsulation technology by spray drying has still a sig-
nificant potential as a novel solution to the problems associated
with formulation, fortification, stabilization and delivery of oil-
labile ingredients to foods which cannot be simply achieved by
other methods (Bakry et al., 2016; Rodríguez, Martín, Ruiz, & Clares,
2016). Various studies have shown that a wall system consisting of
protein in conjunction with carbohydrates is effective in stabilizing
encapsulated oil during long-term storage of spray-dried powder
(Domian, Sułek, Cenkier, & Kerschke, 2014; Gharsallaoui, Saurel,
Chambin, & Voilley, 2012; Jayasundera, Adhikari, Aldred, &
Ghandi, 2009; Nesterenko, Alric, Violleau, Silvestre, & Durrieu,
2014). The ratio of protein and carbohydrate and types (molecular

http://dx.doi.org/10.1016/j.foodhyd.2017.07.011
0268-005X/© 2017 Elsevier Ltd. All rights reserved.
336 E. Domian et al. / Food Hydrocolloids 73 (2017) 335e343
weight, conformational structure and reducing groups) of carbo-
hydrates can influence the physical properties and oxidative sta-
bility of the encapsulated oil. The primary function of proteins is to
emulsify the system rather than to encapsulate it (Vega & Roos,
2006). Carbohydrates, such as maltodextrins, corn syrup solids
and lactose, which are often referred to as secondary carriers
(fillers), have been used to improve oxygen barrier properties of the
wall matrix to enhance drying properties of the wall matrix, and to
improve physical properties of spray-dried oil powders (Sheu &
Rosenberg, 1995).
The use of pea proteins for the microencapsulation reflects the
actual “green” tendency in the food, pharmaceutical and cos-
metics industries (Nesterenko, Alric, Silvestre, & Durrieua, 2013;
Tamm, Herbst, Brodkorb, & Drusch, 2016). Pea proteins, as an
individual coating material (Bajaj, Tang, & Sablani, 2015) or in
conjunction with trehalose (Domian, Brynda-Kopytowska, &
Marzec, 2017), have recently been successfully used to encapsu-
late flaxseed oil. However, in the literature concerning microen-
capsulation uses, these proteins are generally associated with
polysaccharides (Gharsallaoui et al., 2012; Pierucci, Andrade,
Farina, Pedrosa, & Rocha-Lea ~o, 2007). Indeed, polysaccharide/
protein interactions give new functions to pea proteins without
chemical or enzymatic modification, regarding particularly solu-
bility and surfactant properties (Liu, Elmer, Low, & Nickerson,
2010). However, limited attempts have performed to examine
the oil encapsulation with oligo- and polysaccharide materials
that are relatively soluble and that have sugar-like properties in
food applications. Speciality carbohydrates have been developed
commercially that allow great flexibility when replacing sugar in
formulations (Auerbach & Dedman, 2012). The majority of these
speciality carbohydrate ingredients are non-digestible oligo- or
polysaccharides, these products are lower in calories than sugar,
have soluble dietary fibre and, often, prebiotic properties. These
include: gluco-polysaccharides (including polydextrose, resistant
starches and maltodextrins) and fructo-oligosaccharides (FOS)
(including inulin). The important physical feature of these car-
bohydrate polymers is their high glass transition temperature
(Tg), which is significantly greater than the Tg of most other sugars
and is partly a function of molecular weight (Bhandari & Hartel,
2005, chap. 11; Sablani, Kasapis, & Rahman, 2007; Sillick &
Gregson, 2010).
The presence of solid glasses in encapsulating matrices can
directly affect the storage, bioavailability, and delivery properties of
these materials. If the storage temperature of a solid food is higher
than the Tg of relevant components of the food, there is a potential
change in its physicochemical properties (Roos, 1995). Determining
the necessary threshold temperature and humidity conditions to
prevent a glass transition is critical for storage of powder products
(Bhandari & Hartel, 2005, chap. 11). Sorbed water can act as a
plasticizing agent, thus significantly lowering the Tg below the
storage temperature and cause such changes in powders such as
cohesiveness, sticking, agglomeration, lumping, and caking (Chuy
& Labuza, 1994). The water activity value that decreases the Tg to
room temperature has been identified as the critical relative hu-
midity (RHc) (Roos, 2002; Sablani et al., 2007; Sillick & Gregson,
2010).
Several studies have demonstrated that the RHc value using
automatic water vapor sorption instruments, can be used to
identify the glassy to rubbery transition, and the result was
comparable to the DSC method (Burnett, Thielmann, & Booth,
2004; Li & Schmidt, 2011). The high data resolution resulting
from dynamic water sorption methods makes it possible to
observe distinct inflection points in the curve that represent
changes in sorption properties that accompany the glass transi-
tion (Carter & Schmidt, 2012). Burnett et al. (2004) developed a
ramping method, using the dynamic vapor sorption (DVS) in-
strument, to identify the RHc value for the spray-dried lactose.
Yuan and others (2011) investigated the relationship between the
Tg values of polydextrose as a function of %RH, using thermal
methods (DSC and MDSC), and the RHc values of polydextrose as a
function of temperature (15e40
C), using dynamic dewpoint
isotherms (DDI). At 25
C, the RHc value was 49.7% RH, whereas
the %RH associated with the MDSC midpoint Tg value was 44.6%.
The authors attributed the differences between the two data sets
to the time dependency of the DDI obtained RHc values. Li and
Schmidt (2011) applied in their research two instrumental (DVS
ramping and equilibrium) methods and the traditional salt slurry
method to determine the RHc values for amorphous polydextrose.
The RHc DVS equilibrium method was closest to the calculated
DSC Tg onset RHc value.
In our research the dynamic water sorption method was applied
to more complex amorphous food systems. Some carbohydrates
with sugar-replacing properties as fillers in oil microencapsulation
by spray-drying were used in this study. This paper concentrates on
the characterization of the obtained model of multicomponent
powders in terms of the prediction of temperature and humidity
conditions to prevent a glass transition in the amorphous material.
Thus, the objectives of this study were to: (1) determine the critical
relative humidity RHc using DVS ramping for spray-dried fat-filled
pea protein-based powders formulated with different types of
soluble fibre as carbohydrate component and different oil types,
and (2) compare the RHc values to the monolayer coverage values
obtained via DVS water vapor isotherm, as well as the MDSC Tg
values.
2. Materials and methods
2.1. Materials
Shortening AMADA L containing palm oil, rapeseed oil and
hardened palm oil (the fatty acid composition: 42.2% SAFA, 45.5%
MUFA, 12.3% PUFA; softening point: 28e30
C) and rapeseed oil
Maestro (fatty acid composition: 8.4% SAFA; 63.3% MUFA, 27.5%
PUFA; liquid oil at 20
C) were purchased from the Bunge Company
Poland (ZT Kruszwica S.A., Poland).
Pea protein isolate NUTRALYS® S85F (compositional specifica-
tion: protein 84% db, moisture 7.6%) and wheat dextrin soluble fiber
NUTRIOSE® FB06 (compositional specification: fiber rate AOAC 83%
db, monosaccharides 0.1%, disaccharides 0.2%, moisture 3.9%) were
kindly donated by Roquette Poland Sp. z o.o. Polydextrose STA-LITE
R90 (Tate & Lyle USA) (compositional specification: average degree
of polymerisation DP 9e10, polydextrose 92% db, glucose þ sorbitol
5.6 %db, moisture 2.9%) was obtained from Brenntag, Poland. Inulin
ORAFTI GR (Orafti, Belgium) (compositional specification: average
degree of polymerisation DP  10, inulin 93% db,
glucose þ fructose þ sucrose 7% db, moisture 2.1%) and trehalose
(Hayashibara, Japan) were obtained from Hortimex, Poland. All
chemicals used for analysis were of reagent grade.
An average osmolality and cryoscopic temperature (measured
by use of an osmometer Marcel os3000, Marcel Poland) of the 20%
(w/w) carbohydrate sample solutions were: 110.7 (mOsm/kg)
and 0.205
C for Nutriose (16.3 Brix); 209.0 (mOsm/kg)
and 0.388
C for inulin (16.3 Brix); 378.0 (mOsm/kg)
and 0.702
C for polydextrose (16.7 Brix), and 704.7 (mOsm/kg)
and 1.309
C for trehalose (18.7 Brix), respectively. Based on these
colligative properties of the solution, that are solely dependent on
the particle concentration (Rong, Sillick, & Gregson, 2009), it may
be concluded that Nutriose showed the highest number average
molecular weight of the particles, followed by inulin, polydextrose
and trehalose.
 E. Domian et al. / Food Hydrocolloids 73 (2017) 335e343 337

2.2. Experimental design following protocol. An approximately 20 mg samples of pow-

The experimental design consisted of twelve unique trials
(Table 1). The spray-dried fat-filled pea protein-based powders
consisted of PPI, vegetable oil and soluble fibre as a carbohydrate
component. The experimental design included three compositional
variants, where the oil fraction was made up of blend of Palm and
Rapeseed (PR) and Rapeseed (R) at different ratios (52.5% or 35% w/
w total solids), as presented in Table 1. Each group of model fat-
filled powders was formulated with four different carbohydrates
Nutriose (N), inulin (I), polydextrose (D), and trehalose (T), wherein
a carbohydrate-to-protein-material ratio was at 5.3:1 in all cases.
2.3. Preparation of model emulsions and fat-filled pea protein-
based powders
Model oilein-water emulsions (target 30% w/w total solids,
comprising pea protein, oil and soluble fiber) were prepared as
follows. Carbohydrate powder was dissolved in hot water (60
C),
using a paddle agitator to aid reconstitution. Pea protein isolate
(PPI) was then added slowly during mixing, and the mixture was
moved to cold storage (5
C) for 12 h for the hydration of compo-
nents. The water phase was tempered to 60
C before addition of
the fat component. When blend of palm and rapeseed oils was
used, it was melted in a separate vessel before addition. The mix
was agitated using Ultra-Turrax (IKA T18 Basic, Wilmington, USA) at
13,000 rpm for 2 min, homogenised using a 2-stage homogeniser
(Panda 2K; Niro Soavi, Italy) using a first-stage pressure of 60 MPa
and a second-stage pressure of 20 MPa. The emulsions were spray-
dried in a lab-scale Niro Spray dryer (Model MOBILE MINOR™ Niro
A/S, Denmark), equipped with a rotary atomisation system and co-
current drying, with a typical water evaporation rate of 1.5e2/h. An
inlet air temperature of 150 ± 3
C and an outlet air temperature of
60 ± 2
C were selected and disk rotation was at approximately
20,000 rpm. Samples of powder were stored at 10
C in sealed foil
bags until analysed.
2.4. DVS ramping experiments
The DVS Intrinsic analyser (Surface Measurement Systems Ltd,
London, UK) equipped with the SMS UltraBalanceTM with a mass
resolution of ±0.1 mg was used to obtain the ramping and equilib-
rium water vapor sorption isotherms. The apparatus was calibrated
using saturated LiCl, MgCl2 and NaCl slurries at 25
C.
DVS ramping experiments were performed using the
ders were placed in the sample pan in the DVS apparatus and
were initially dried in a 200 sccm (standard cubic centimeters)
stream of dry air (<0.1% RH) for several hours to establish a dry
mass at a constant temperature (25 ± 0.1
C). Next, the sam-
ples were exposed to a linearly increasing relative humidity up
to 100% RH at a selected ramping rate. For each powder the
experiment was carried out at 2, 4, 6, 8 and 10 RH/h ramping
rate, respectively. Change in sample weight was converted to
moisture content (% db).
The glass transition RH was taken as the inflection point on the
sorption curve which indicates the border between surface
adsorption and absorption in the bulk structure of the material. It
was found using a simplified least squares 2nd-derivative method.
The crystallisation RH was taken as the point where the sample
mass decreased drastically due to relaxation to the crystalline state.
The RH glass transition values obtained using the DVS ramping
method are a linear function of ramping rate (Burnett et al., 2004).
The RHc values were determined using the linear extrapolation to
zero ramping rate.
Multiple experiments (n ¼ 3) were performed on the poly-
dextrose sample at the same humidity ramping rate of 6% RH/hr to
establish the reproducibility of the measurements. Only one run
was performed at each humidity ramping rate for the spray-dried
powders.
2.5. DVS water vapor isotherms
Water vapor isotherms were determined using DVS equilibrium
method at 25
C. About 20 mg samples of powders were placed into
pan of the calibrated DVS Intrinsic analyser (Surface Measurement
Systems Ltd, London, UK). The samples were dried at 25
C to
constant mass under a stream of dry nitrogen (0% RH) for 480 min.
The samples were then exposed to the following RH profile: 0e5%
RH in 0.05, 0.25, and 2.5% RH increments, 5e95% RH in 5% RH in-
crements. For each step, the changes in mass (m) were plotted
against time. Equilibrium was considered to be reached when
change in mass with time (dm/dt) was lower than 0.001 mg/min for
at least 10 min. The last one result (weight of sample) at each RH
point was used to calculate the moisture content (w).
The GuggenheimeAndersonede Boer (GAB) equation over the
aw range values between 0.05 e 0.25 and 0.05e0.5 was used to
model water sorption isotherms and to determine the apparent
monolayer coverage (water content wm (% db) and water activity
awm). The GAB model isotherm (Van den Berg, 1985), with

Table 1
Experimental design of the formulation.

Code of Powder Protein Fat Carbohydrate Carbohydrate-to-protein-material ratio Carbohydrate-to-fat-material ratio

a b c
type content (%) type content (%) type content (%)

PR53N PPI 7.5 PR 52.5 N 40 5.3: 1 0.8: 1

PR53I PPI 7.5 PR 52.5 I 40 5.3: 1 0.8: 1
PR53D PPI 7.5 PR 52.5 D 40 5.3: 1 0.8: 1
PR53T PPI 7.5 PR 52.5 T 40 5.3: 1 0.8: 1
PR35N PPI 10.3 PR 35 N 54.7 5.3: 1 1.6: 1
PR35I PPI 10.3 PR 35 I 54.7 5.3: 1 1.6: 1
PR35D PPI 10.3 PR 35 D 54.7 5.3: 1 1.6: 1
PR35T PPI 10.3 PR 35 T 54.7 5.3: 1 1.6: 1
R35N PPI 10.3 R 35 N 54.7 5.3: 1 1.6: 1
R35I PPI 10.3 R 35 I 54.7 5.3: 1 1.6: 1
R35D PPI 10.3 R 35 D 54.7 5.3: 1 1.6: 1
R35T PPI 10.3 R 35 T 54.7 5.3: 1 1.6: 1
a
PPI e Pea Protein Isolate.
b
PR e blend of Palm oil and Rapeseed oil, R - Rapeseed oil.
c
N e Nutriose, I e inuline, D e polydextrose, T - trehalose.
338 E. Domian et al. / Food Hydrocolloids 73 (2017) 335e343
constants C and k, was converted to a second-order polynomial,
giving a quadratic equation.
Accuracy of equations fitting was evaluated with mean square
error (R2) and the mean relative percentage deviation (RMS) value.
2.6. Determination of Tg using MDSC
A differential scanning calorimeter in a modulated mode (DSC
Q200, TA Instruments, New Castle, USA) equipped with a refriger-
ated cooling system (RCS) (New Castle, USA) was used to determine
the temperatures of glass transition (Tg) in the powders. Values of
Tg were determined for the spray-dried powders at their specific
water activity values “as is”. Each time, before the MDSC experi-
ments the aw of samples was determined at 25
C using the
Rotronic HC2-AW instrument (Rotronic AG, Switzerland). Samples
(8e12 mg) were crimped to attain better thermal contact, and then
hermetically sealed to examine each specimen in the presence of
the water “as is” after spray drying. The cell was purged with
50 mL min1 dry nitrogen and calibrated for the baseline on an
empty oven and for temperature using standard pure indium.
Specific heat capacity (Cp) was calibrated using a sapphire. An
empty hermetically sealed aluminum pan was used as a reference
in each test. Samples were scanned from 50 to 190
C at a con-
stant heating rate of 2
C/min with an amplitude of ±1
C and 60 s
period of modulation. Curves were analysed with respect to the
total, reversible and non-reversible heat flow (Jakubczyk,
Ostrowska-Ligeza, & Gondek, 2010; Rahman, Al-Marhubi, & Al-
Mahrouqi, 2007). Glass transition was reported with parameters
indicating its onset, midpoint and endpoint of a vertical shift in the
reversing transition curve. TA Instruments Universal analysis soft-
ware was used to analyse the glass transition temperature. The
measurements were done in three replicates for each sample.
2.7. Statistics
Statistica 10 software (Statsoft) was used to evaluate and analyse
the data obtained during this study. The Excel sheet was used for
data collection and presentation.
3. Results and discussion
3.1. DVS ramping experiments, glass transition relative humidity
The spray-dried powders were exposed to the ramping hu-
midity profile from 0% to 100% RH. Fig. 1 displays a typical net
percent change in mass with a 2.0% RH/h ramping rate versus time
plot for spray-dried R35T powder sample. Fig. 1 was annotated to
reflect the changes in sorption behavior of the powder containing
Fig. 1. Relative rampling experiment (2% RH/h) for spray-dried fat powder R35T at
25
C. Solid line shows the net change in mass while dotted line shows the RH profile.
trehalose as previously discussed in Burnett et al. (2004), Yuan,
Carter, and Schmidt (2011) and Li and Schmidt (2011). There are
several features in the mass response with linearly increasing RH.
At low RH values (below 30% RH), the water uptake is relatively low.
The moisture uptake in this region is probably dominated by sur-
face adsorption. Above 30% RH there is a sharp increase in moisture
sorption, most likely due to bulk absorption dominating the sorp-
tion mechanism. The glass transition relative humidity (RHg) was
measured at the transition point between surface adsorption and
absorption into the bulk (~890 min and 30% RH). The amorphous to
crystalline transition is clearly illustrated in Fig. 1 by the sharp loss
in mass around 1650 min. After crystallisation, the sample has a
much lower moisture sorption capacity.
Fig. 2 displays the results for similar experiments from 2% RH/h
to 10% RH/h for spray-dried PR35N, PR35I, PR35D and PR35T
powders containing 10.3% PPI, 35% palm and rapeseed oil and 54.7%
carbohydrate component (figures not shown for the other spray-
dried powders). It is worthy of notice that all features remain sta-
ble over the entire range of ramping rates. The glass transition point
was clearly noticeable in all powders containing inulin, poly-
dextrose, and trehalose. In turn, in the DVS ramping curves plotted
for the powders with Nutriose, the glass transition occurred most
likely at RH > 70%, which was also accompanied by dissolution of
components. The transition from the amorphous to the crystalline
stage was clearly manifested by a significant mass loss in the
powders containing inulin and trehalose.
The RHg ramping values (for all powders containing inulin,
polydextrose and trehalose) appeared to decrease with a
decreasing ramping rate and exhibited a linear trend (figures not
shown).
A linear regression equation of the plot of the RHg values versus
the flow rates between 2% RH/h to 10% RH/h used to obtain the RHg
at a zero flow rate value as the critical relative humidity RHc
(Table 2). The statistical analysis (Kruskal-Wallis ANOVA, p < 0.05)
demonstrated that changes in the RHc of spray-dried fat powders
were significantly affected by the type of carbohydrate (C), however
there were no statistically significant differences between the
samples with polydextrose and trehalose. Neither the type of fat (F)
Fig. 2. Net change in mass for spray-dried fat powders PR35N, PR35I, PR35D and
PR35T at 25
C for different ramping rates. Solid line shows the net change in mass
while dotted line shows the RH profile.
 E. Domian et al. / Food Hydrocolloids 73 (2017) 335e343 339

Table 2
The critical relative humidity RHc values determined at 25
C for moisture-induced
glassy to rubbery transition from the DVS ramping of spray-dried fat powders and
raw carbohydrate samples. RHc values obtained using the RHg values versus the
flow rates linear regression equation.

Powder RHc (%) R2 of linear fit

PR53N 85.06 0.9551

PR53I 41.91 0.9711
PR53D 34.13 0.9630
PR53T 38.64 0.9570
PR35N 83.72 0.9834
PR35I 44.25 0.9383
PR35D 28.53 0.9862
PR35T 36.36 0.9926
R35N 84.46 0.9524
R35I 27.67 0.9967
R35D 26.67 0.9636
R35T 29.12 0.9775
N e e
I 32.837 0.8070
D 36.493 0.9537
T e e
a
CV (%) 3 ÷ 17 e
b
C, p 0.033 e
b
F, p NS e
b
C:F ratio, p NS e

Explanation:
a
CV e the coefficient of variation.
b Fig. 3. DVS moisture sorption isotherms at 25
C (white symbols) and moisture con-
Kruskal-Wallis ANOVA (p < 0.05): C - type of Carbohydrate, F - type of Fat, C:F
tent vs. water activity (aw) of the “as is” sample obtained by spray drying (black
ratio - Carbohydrate-to-fat-material ratio; NS ¼ non-significant.
symbols) for powders PR35N, PR35I, PR35D and PR35T.

nor the content of fat expressed by the carbohydrate-to-fat-

material ratio (F:C ratio) caused significant differences in the crit-
ical relative humidity (RHc) of the powders. The powders with
Nutriose showed the highest RHc, followed by the powders con-
taining inulin and these containing trehalose and polydextrose. It
may, therefore, be concluded that - regardless of fat type and
content - the powders with Nutriose should not undergo undesir-
able caking as a result of verification even at high relative humidity
of the air (up to RH 70%) at 25
C. In turn, the caking of powder at
25
C, as a result of the glassy transition of amorphous carbohy-
drates contained in them is likely to occur at RH above 30e40% for
the inulin-containing powders and at RH above 30e35% for pow-
ders containing polydextrose and trehalose.
3.2. DVS water vapor isotherms, monolayer coverage
Fig. 3 depicts a comparison of the course of water adsorption
isotherms obtained at a temperature of 25
C using the DVS ana-
lyser for exemplary powders PR35N, PR35I, PR35D and PR35T
(isotherms not shown for the other spray-dried powders). The
points marked on the isotherms represent mean values, whereas
variability coefficients reached up to 7%. The shape of adsorption
isotherms itself indicates that the type of carbohydrate determines
the mechanism of adsorption and state of water in a given powder,
irrespective of fat type and content. All obtained isotherms may be
divided into three zones. In the first zone (aw < 0.25), wherein
water is most probably retained around available polar groups of
the material in the form of a monolayer (Iglesias & Chirife, 1995;
Lewicki, 2004), the highest equilibrium content of water was
found in the powders with Nutriose (N), followed by these with
inulin (I), polydextrose (D), and trehalose (T). In the second zone
(0.25 < aw < 0.70), the course of isotherms plotted for the powders
with trehalose, inulin, and polydextrose reveals cracks formed,
most likely, as a result of moisture-induced phase transitions, i.e.
glassy transitions, as well as the successive crystallisation (notice-
able especially in the powders with trehalose). In the third zone
(aw > 0.7), wherein the material probably swells and softens and
where saccharides dissolve, the highest water content was deter-
mined in the powders with polydextrose (D), followed by these
containing Nutriose (N), inulin (I), trehalose (T), respectively.
The monolayer coverage was determined with the use of the
GAB model which takes account of the modified properties of the
adsorbent concerning the multilayer adsorption (Wolf et al. 1984,
Van den Berg, 1985) and is commonly applied for description of
isotherms of many food products (Iglesias & Chirife, 1995; Lewicki,
2004). Due to the phase transitions proceeding in the powders in
the intermediate water activity range, the description of sorption
data was limited with the range of the low water activity. Table 3
summarises parameters of the applied GAB model in two aw
ranges and provides the goodness of fit of this model expressed by
the coefficient of determination (r2) and mean square error (RMSE).
Assuming the values of RMSE at a level lower than 10% as a good fit
of the model to sorption data, it needs to be concluded that this
assumption was met by the GAB model in both ranges of aw
(Boquet, Chirife, & Iglesias, 1978; Lomauro, Bakshi, & Labuza, 1985).
When comparing water content wm and water activity awm in
the monolayer (Table 3), it may be noticed that parameters awm for
aw 0.05e0.5 had higher vales than the awm for aw 0.05e0.25. Only
the type of carbohydrate (considering powders division into three
groups: Nutriose, inulin and polydextrose þ trehalose) caused
significant differences in wm and awm (Kruskal-Wallis ANOVA test,
p < 0.05). The values of wm and awm obtained for the GAB model in
aw range of 0.05e0.5 were as follows: 2.3e4.7 %db and 0.2e0.3 for
the powders with Nutriose (N), 1.5e3.0 %db and 0.15e0.19 for the
powders with inulin (I), as well as 1.2e1.6 %db and 0.10e0.23 for
the powders with polydextrose (D) and trehalose (T).
The moisture content (w) and water activity (aw) of the “as is”
sample in the powders directly after spray-drying well corre-
sponded with their sorption data, which was show in Fig. 3. In
addition, it may be observed that in the case of the powders with
inulin, polydextrose and trehalose, their moisture content (w:
1.4e2.8 %db) and water activity (aw: 0.13e0.20) were at a similar
level in respect of the obtained values of monolayer coverage
340 E. Domian et al. / Food Hydrocolloids 73 (2017) 335e343

Table 3
Parameters and statistical results determined for the GAB model for DVS moisture sorption isotherms of spray-dried fat powders at 25
C.

Powder GAB range aw 0.05e0.25 GAB range aw 0.05e0.5

parameters statistical parameters statistical

2
C k wm (% db) awm () RMSE r C k wm (% db) awm () RMSE r2

PR53N 14.660 0.945 2.478 0.22 1.27 0.9998 14.751 1.046 2.344 0.21 0.60 0.9998
PR53I 23.472 0.950 1.822 0.12 0.21 0.9999 32.429 1.292 1.495 0.19 3.75 0.9886
PR53D 19.261 0.724 1.927 0.11 3.69 0.9905 29.025 1.512 1.211 0.23 8.02 0.9845
PR53T 8.196 1.153 1.897 0.22 2.36 0.9996 9.787 1.394 1.572 0.18 1.52 0.9986
PR35N 15.905 1.116 3.157 0.18 1.15 0.9989 12.820 0.926 3.749 0.24 3.22 0.9969
PR35I 20.371 1.047 2.947 0.18 0.52 0.9997 20.686 0.998 3.008 0.19 2.09 0.9927
PR35D 15.829 1.873 1.121 0.11 0.93 0.9991 11.648 1.673 1.292 0.14 7.12 0.9661
PR35T 27.975 1.452 1.597 0.09 0.25 0.9999 34.139 1.641 1.464 0.12 1.17 0.9826
R35N 10.186 0.838 4.151 0.28 5.28 0.9881 8.747 0.817 4.469 0.30 4.02 0.9918
R35I 20.374 1.301 2.465 0.11 0.35 0.9998 50.564 1.228 2.170 0.15 8.15 0.9571
R35D 31.137 1.548 1.298 0.10 0.40 0.9998 25.244 1.620 1.304 0.10 7.22 0.9577
R35T 12.935 1.650 1.347 0.13 0.81 0.9991 10.019 1.636 1.455 0.15 8.30 0.9524
a
C, p e e 0.022 NS e e e e 0.016 0.048 e e
a
F, p e e NS NS e e e e NS NS e e
a
C:F ratio, p e e NS NS e e e e NS NS e e

Explanation:
a
Kruskal-Wallis ANOVA (p < 0.05): C - type of Carbohydrate, F - type of Fat, C:F ratio - Carbohydrate-to-fat-material ratio; NS ¼ non-significant.

(Tables 3 and 4). Different observations were made for the powders
with Nutriose whose moisture content (w: 1.7e2.0 %db) and water
activity (aw: ~0.047) were significantly lower from the value of the
monolayer coverage (Tables 3 and 4).
3.3. MDSC, glass transition temperature
follows: 92e128
C for the powders with Nutriose (N), 63e74
C for
the powders with inulin (I), and 55e74
C for the powders with
polydextrose (D) and trehalose (T). It may, therefore, be concluded
that all obtained spray-dried fat-filled pea protein-based powders
should be stable and avoid glassy transition during storage at room
temperature in hermetic packages.

The onset temperature of the calorimetric glass transition in
heating should always be taken as the glass transition temperature,
(Tg) (Bhandari & Roos, 2016), although in the literature midpoint Tg
is also widely reported (Li & Schmidt, 2011; Yuan et al., 2011).
Burnett and others (2004) also concluded that the DVS ramping
RHc value represented the onset of the glassy to rubbery transition.
Average Tg onset, midpoint, and endpoint values for MDSC
thermograms are given in Table 4. These glass transition temper-
atures (Tg) were measured at the moisture content of the “as is”
sample. The statistical analysis (Kruskal-Wallis ANOVA test,
p < 0.05) demonstrated that only the type of carbohydrate (with
powders divided into three groups: Nutriose, inulin, and
polydextrose þ trehalose) caused significant differences in the glass
transition temperatures (Tg). The obtained Tg onset values were as
3.4. Comparison between RHc and monolayer coverage values and
Tg values
The critical relative humidity RHc ramping values of powders
were compared to the DVS equilibrium isotherm method mono-
layer coverage values, as well as the MDSC Tg values. The Spear-
man’s rank correlation (rs) (a non-parametric correlation of non-
normal distributions) was used to determine correlations be-
tween the above parameters. A strong positive correlation was
demonstrated between RHc and water content of the monolayer
wm (rS ¼ 0.75), water activity of the monolayer awm (rS ¼ 0.78), Tg
onset (rS ¼ 0.63), Tg midpoint (rS ¼ 0.65), and Tg endpoint
(rS ¼ 0.65). In addition, a strong negative correlation was detected
between RHc and water activity of the “as is” powder. Significant

Table 4
Average glass transition temperatures (Tg) of spray-dried fat powders, determined with MDSC, measured at moisture content (w) and water activity (aw) of the “as is” sample.

Powder w (% db) aw () Tg Onset (
C) Tg Midpoint (
C) Tg Endpoint (
C)

PR53N 2.06 0.046 91.8 101.9 111.5

PR53I 2.08 0.137 66.3 73.6 80.8
PR53D 1.66 0.201 55.5 68.1 80.8
PR53T 1.45 0.147 68.5 72.3 76.1
PR35N 2.04 0.045 128.3 134.4 140.5
PR35I 3.13 0.131 63.5 75.0 86.5
PR35D 1.70 0.181 66.5 75.0 83.5
PR35T 1.90 0.141 74.4 77.8 81.3
R35N 1.73 0.047 120.0 126.2 132.7
R35I 2.79 0.146 74.1 79.5 84.9
R35D 2.00 0.185 55.4 65.3 75.1
R35T 1.80 0.165 62.9 66.6 70.4
a
CV (%) 1 ÷ 28 1÷6 1 ÷ 19 0 ÷ 17 0 ÷ 25
b
C, p 0.022 0.013 0.041 0.022 0.022
b
F, p NS NS NS NS NS
b
C:F ratio, p NS NS NS NS NS

Explanation:
a
CV e the coefficient of variation.
b
Kruskal-Wallis ANOVA (p < 0.05): C - type of Carbohydrate, F - type of Fat, C:F ratio - Carbohydrate-to-fat-material ratio; NS ¼ non-significant.
 E. Domian et al. / Food Hydrocolloids 73 (2017) 335e343 341

correlations (rS 0.59÷0.78) were also noticed between glass tran-
sition temperatures (Tg) and water content in the monolayer wm. In
contrast, no correlation was demonstrated between Tg and water
activity of the monolayer awm.
The conducted Principal Component Analysis (PCA) allowed
identifying parameters that characterise and ipso facto
differentiate, to the greatest extent, the obtained spray-dried fat-
filled powders in relation to the critical relative humidity RHc
(Fig. 4A). The PCA was conducted with 11 active variables and
additional variables including: osmolality of carbohydrate solu-
tions, type of carbohydrate (C), type of fat (F), and F:C ratio. Two
principal components were typed which in total explained 87.5% of

A
w
1,0

0,5 wm GAB_0.25 *C:F ratio

wm GAB_0.5

*F
F2 : 11,99%

0,0
Tg on aw
awm GAB_0.25
Tg mid *C
awm GAB_0.5
Tg end
*Osmol
-0,5 RHc

-1,0

-1,0 -0,5 0,0 0,5 1,0

F1 : 75,71%

Active
Additional
B
3

PR35I
2
R35I

1
F2: 11,99%

PR53I R35D
0 PR35N PR35T
PR53N
R35N R35T
PR53D PR35D
-1
PR53T

-2
-6 -4 -2 0 2 4
F1: 75,71% Active

Fig. 4. PCA projection of critical relative humidity RHc values, monolayer coverage values (wm and awm), Tg values, water content (w) and water activity (aw) of the “as is” sample
(active variables), and carbohydrate osmolality (Osmol*), type of Carbohydrate (C*), type of Fat (F*), Carbohydrate-to-fat-material ratio (C:F ratio*) (*additional variables) (A) and
PCA projection of the analysed powders on principal components surface (B).
342 E. Domian et al. / Food Hydrocolloids 73 (2017) 335e343
the information. Component 1 corresponding to the highest
eigenvalue (8.3) explained 75.71%, whereas Component 2
explained 11.99% of the total variance. Component 1 was strongly
positively correlated with carbohydrate osmolality (r ¼ 0.66) and
water activity of the “as is” powder (r ¼ 0.94), whereas negative
contributions (from r ¼ 0.82 to r ¼ 0.94) of this component were
found for: relative humidity RHc, monolayer coverage values wm
and awm for the GAB model, as well as the MDSC Tg values.
Considering the above dependencies, the first principal component
may be interpreted as a measure of traits characterizing powders
stability, whereas the second one as a measure of the moisture of
spray-dried powders. The degree to which each variable is repre-
sented by the current principal components was presented in
Fig. 4A. The further away is the variable from the center of the
circle, the better it is represented by the current system of
coordinates.
The points representing spray-dried fat-filled powders distrib-
uted across the space of the principal components form three
clusters (Fig. 4B). The first cluster is constituted by the powders
containing wheat dextrin soluble fiber Nutriose, the second one e
by the powders containing inulin, and the third one e by powders
formulated with polydextrose and trehalose. This first group is
located to the left from the center (i.e. has negative values of co-
ordinates for the first factor). Hence, these powders e formulated
with the carbohydrate having the lowest osmolality, are charac-
terised by the lowest water activity after spray drying and by the
highest stability, i.e. by the highest values of: critical relative hu-
midity RHc, monolayer coverage, as well as the MDSC Tg. Location
to the right from the center of the third group of powders formu-
lated with carbohydrates having the highest osmolality is indicative
of their different characteristics, the highest water activity after
spray drying and the lowest stability. The powders from the second
group (with inulin) display greater stability compared to the third
group, but do not differ extremely, which is indicated by their close
location.
This study demonstrated that spray-dried fat-filled pea protein
based powders represented a complex system as there an overall
glass transition is influenced by carbohydrate components. In this
solid complex system carbohydrate components showed immisci-
bility with other components, like other powders of this type, for
example milk powders (Jouppila & Roos, 1994). The carbohydrate
(including Nutriose, inulin, trehalose and polydexstrose) and pea
proteins were not compatible and the presence of fat did not in-
fluence the glass transition. Likewise, Kelly, O’Mahony, Kelly, and
O’Callaghan (2014) demonstrated that oil type had no significant
effect on the principal physical characteristics of fat-filled dairy-
based powders, i.e. moisture and aw, glass transition or water
sorption characteristics.
4. Conclusions
This study demonstrated that the critical relative humidity
(RHc) values, obtained using automatic water vapor sorption in-
struments, can be used to detect the glassy to rubbery transition in
multicomponent complex amorphous food systems. This water
sorption method shows promise as a practical tool to control
storage stability. Study results demonstrated that changes in the
RHc of the spray-dried fat powders were significantly affect by the
type of carbohydrate. Neither the type of fat nor its content caused
significant differences in the critical relative humidity RHc of the
powders. The powders with Nutriose showed the highest RHc at
25
C (70e80%), followed by the powders with inulin (RHc 30e40%)
and these with trehalose and polydextrose (RHc 30e35%). A strong
positive correlation was demonstrated between the RHc and the
monolayer coverage values obtained via DVS water vapor isotherm,
as well as the MDSC Tg values. Also, a strong negative correlation
was found between the RHc and the specific water activity “as is” of
the powders after spray-drying. In contrast, no correlation was
detected between the moisture content of powders after their
spray-drying and RHc, the monolayer coverage values, as well as
the MDSC Tg. Hence, the powders formulated with the carbohy-
drate of the lowest osmolality e Nutriose, were characterised by
the lowest water activity after spray drying and by the lowest
values of the critical relative humidity RHc, monolayer coverage, as
well as the MDSC Tg. The powders formulated with the carbohy-
drates having the lowest osmolality e polydextrose and trehalose,
were characterised by the highest water activity after spray drying
and by the least stability. The powders with inulin were more stable
compared to these containing polydextrose and trehalose, but the
differences were not extreme.
References
Auerbach, M., & Dedman, A. K. (2012). Bulking agentsemulti-functional ingredients.
In K. O’Donnell, & M. Kearsley (Eds.), Sweeteners and sugar alternatives in food
technology (pp. 433e470). John Wiley & Sons.
Bajaj, P. R., Tang, J., & Sablani, S. S. (2015). Pea protein isolates: Novel wall materials
for microencapsulating flaxseed oil. Food and Bioprocess Technology, 8(12),
2418e2428.
Bakry, A. M., Abbas, S., Ali, B., Majeed, H., Abouelwafa, M. Y., Mousa, A., et al. (2016).
Microencapsulation of oils: A comprehensive review of benefits, techniques,
and applications. Comprehensive Reviews in Food Science and Food Safety, 15(1),
143e182.
Bhandari, B. R., & Hartel, R. W. (2005). Phase transitions during food powder pro-
duction and powder stability. In C. Onwulata (Ed.), Encapsulated, powdered,
foods (pp. 261e291). New York: Taylor and Francis.
Bhandari, B., & Roos, Y. H. (2016). Non-equilibrium states and glass transitions in
Foods: Processing effects and product-specific implications. Woodhead Publishing.
Boquet, R., Chirife, J., & Iglesias, H. A. (1978). Equations for fitting water sorption
isotherms of foods. International Journal of Food Science & Technology, 13(4),
319e327.
Burnett, D. J., Thielmann, F., & Booth, J. (2004). Determining the critical relative
humidity for moisture-induced phase transitions. International Journal of
Pharmaceutics, 287(1), 123e133.
Carter, B. P., & Schmidt, S. J. (2012). Developments in glass transition determination
in foods using moisture sorption isotherms. Food Chemistry, 132(4), 1693e1698.
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), 43e46.
Domian, E., Brynda-Kopytowska, A., & Marzec, A. (2017). Functional properties and
oxidative stability of flaxseed oil microencapsulated by pray drying using
legume proteins in combination with soluble fiber or trehalose. Food and Bio-
process Technology. http://dx.doi.org/10.1007/s11947-017-1908-1.
Domian, E., Sułek, A., Cenkier, J., & Kerschke, A. (2014). Influence of agglomeration
on physical characteristics and oxidative stability of spray-dried oil powder
with milk protein and trehalose wall material. Journal of Food Engineering, 125,
34e43.
Gharsallaoui, A., Saurel, R., Chambin, O., & Voilley, A. (2012). Pea (Pisum sativum, L.)
protein isolate stabilized emulsions: A novel system for microencapsulation of
lipophilic ingredients by spray drying. Food and Bioprocess Technology, 5(6),
2211e2221.
Ho, T. M., Truong, T., & Bhandari, B. (2016). Spray-drying and non-equilibrium
states/glass transition. In B. Bhandari, & Y. H. Roos (Eds.), Non-Equilibrium
states and glass transitions in Foods: Processing effects and product-specific im-
plications (pp. 111e136). Woodhead Publishing.
Iglesias, H. A., & Chirife, J. (1995). An alternative to the Guggenheim, Anderson and
De Boer model for the mathematical description of moisture sorption isotherms
of foods. Food Research International, 28(3), 317e321.
Jakubczyk, E., Ostrowska-Ligeza, E., & Gondek, E. (2010). Moisture sorption char-
acteristics and glass transition temperature of apple puree powder. Interna-
tional Journal of Food Science & Technology, 45(12), 2515e2523.
Jayasundera, M., Adhikari, B., Aldred, P., & Ghandi, A. (2009). Surface modification of
spray dried food and emulsion powders with surface-active proteins: A review.
Journal of Food Engineering, 93(3), 266e277.
Jouppila, K., & Roos, Y. H. (1994). Glass transitions and crystallization in milk
powders. Journal of Dairy Science, 77(10), 2907e2915.
Kelly, G. M., O’Mahony, J. A., Kelly, A. L., & O’Callaghan, D. J. (2014). Physical char-
acteristics of spray-dried dairy powders containing different vegetable oils.
Journal of Food Engineering, 122, 122e129.
Lewicki, P. P. (2004). Water as the determinant of food engineering properties. A
review. Journal of Food Engineering, 61(4), 483e495.
Li, Q. E., & Schmidt, S. J. (2011). Use of ramping and equilibrium water vapor
sorption methods to determine the critical relative humidity at which the
glassy to rubbery transition occurs in polydextrose. Journal of Food Science,
76(1), 149e157.
 E. Domian et al. / Food Hydrocolloids 73 (2017) 335e343
Liu, S., Elmer, C., Low, N. H., & Nickerson, M. T. (2010). Effect of pH on the functional
behaviour of pea protein isolateegum Arabic complexes. Food Research Inter-
national, 43(2), 489e495.
Lomauro, C. J., Bakshi, A. S., & Labuza, T. P. (1985). Evaluation of food moisture
sorption isotherm equations. Part I: Fruit, vegetable and meat products. Leb-
ensmittel-Wissenschaft und Technologie, 18(2), 111e117.
Nesterenko, A., Alric, I., Silvestre, F., & Durrieu, V. (2013). Vegetable proteins in
microencapsulation: A review of recent interventions and their effectiveness.
Industrial Crops and Products, 42, 469e479.
Nesterenko, A., Alric, I., Violleau, F., Silvestre, F., & Durrieu, V. (2014). The effect of
vegetable protein modifications on the microencapsulation process. Food Hy-
drocolloids, 41, 95e102.
Pierucci, A. P. T., Andrade, L. R., Farina, M., Pedrosa, C., & Rocha-Lea ~o, M. H. M.
(2007). Comparison of a-tocopherol microparticles produced with different
wall materials: Pea protein a new interesting alternative. Journal of Microen-
capsulation, 24(3), 201e213.
Rahman, M. S., Al-Marhubi, I. M., & Al-Mahrouqi, A. (2007). Measurement of glass
transition temperature by mechanical (DMTA), thermal (DSC and MDSC), water
diffusion and density methods: A comparison study. Chemical Physics Letters,
440(4), 372e377.
Rodríguez, J., Martín, M. J., Ruiz, M. A., & Clares, B. (2016). Current encapsulation
strategies for bioactive oils: From alimentary to pharmaceutical perspectives.
Food Research International, 83, 41e59.
Rong, Y., Sillick, M., & Gregson, C. M. (2009). Determination of dextrose equivalent
value and number average molecular weight of maltodextrin by osmometry.
Journal of Food Science, 74(1), C33eC40.
Roos, Y. H. (1995). Glass transition-related physicochemical changes in foods:
Chemical and rheological changes during phase transition in food. Food
343
Technology, 49(10), 97e102.
Roos, Y. H. (2002). Importance of glass transition and water activity to spray drying
and stability of dairy powders. Le Lait, 82(4), 475e484.
Sablani, S. S., Kasapis, S., & Rahman, M. S. (2007). Evaluating water activity and glass
transition concepts for food stability. Journal of Food Engineering, 78(1),
266e271.
Sheu, T. Y., & Rosenberg, M. (1995). Microencapsulation by spray drying ethyl
caprylate in whey protein and carbohydrate wall systems. Journal of Food Sci-
ence, 60(1), 98e103.
Sillick, M., & Gregson, C. M. (2010). Critical water activity of disaccharide/malto-
dextrin blends. Carbohydrate Polymers, 79(4), 1028e1033.
Tamm, F., Herbst, S., Brodkorb, A., & Drusch, S. (2016). Functional properties of pea
protein hydrolysates in emulsions and spray-dried microcapsules. Food Hy-
drocolloids, 58, 204e214.
Van den Berg, C. (1985). Development of BET-like models for sorption of water on
foods, theory and relevance. In Properties of Water in Foods (pp. 119e131).
Netherlands: Springer.
Vega, C., & Roos, Y. H. (2006). Invited review: Spray-dried dairy and dairy-like
emulsionsdcompositional considerations. Journal of Dairy Science, 89(2),
383e401.
Wolf, W., Spiess, W. E. L., Jung, G., Weisser, H., Bizot, H., & Duckworth, R. B. (1984).
The water-vapour sorption isotherms of microcrystalline cellulose (MCC) and of
purified potato starch. Results of a collaborative study. Journal of Food Engi-
neering, 3(1), 51e73.
Yuan, X., Carter, B. P., & Schmidt, S. J. (2011). Determining the critical relative hu-
midity at which the glassy to rubbery transition occurs in polydextrose using an
automatic water vapor sorption instrument. Journal of Food Science, 76(1),
78e89.