You are on page 1of 9

Chemical Engineering Research and Design 1 4 6 ( 2 0 1 9 ) 78–86

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www.elsevier.com/locate/cherd

Drying of Maltodextrin solution in a vacuum spray


dryer

Fernanda de Melo Ramos a,∗ , Job Ubbink b , Vivaldo Silveira Júnior a ,


Ana Silvia Prata a
a Department of Food Engineering, Faculty of Food Engineering, State University of Campinas, P.O. Box 6121,
13083-862 Campinas, SP, Brazil
b Food Science and Nutrition Department, California Polytechnic State University, San Luis Obispo, CA 93442, USA

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

Article history: New drying strategies that use low temperatures can have a significant impact on the
Received 25 September 2018 improvement of food quality, in particular regarding the retention of flavor compounds,
Received in revised form 11 March bioactives and other thermosensitive components. The vacuum spray dryer (VSD) is a spray
2019 dryer that operates with a low-pressure drying chamber, which consequently reduces the
Accepted 24 March 2019 increases the thermodynamic driving force for water removal and allows drying at sig-
Available online 1 April 2019 nificantly reduced temperatures. In order to understand the process behavior and define
operational strategies, a mathematical model that encompasses mass and energy balances
Keywords: was validated with experimental measurements of pressure and temperature during drying
Drying modelling of large chained maltodextrin (dextrose equivalent = 10). Results from experiments carried
Energy balance out in a pilot VSD present a good fit with the proposed model and confirmed its underlying
Maltodextrin assumptions. In addition, comparative analyses were performed regarding physical aspects
Glass transition of particles produced by VSD and by conventional spray dryer (SD) in the same equipment,
Spray drying but without vacuum. Under the tested conditions, VSD particles presented a higher moisture
Vacuum content (8%) and smaller time of wettability than SD particles. The morphological changes
were caused by the vacuum and can be interesting for technological applications.
© 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction tional spray drying, the materials are exposed to high levels of oxygen,
owing to the large amounts of air used in the drying process and the
Spray drying is a process that transforms a liquid raw material into turbulent flow characteristics in the drying chamber.
dried particles by atomizing the solution into a hot drying medium In this context, new drying strategies that use low temperatures
(Keshani et al., 2015). The atomization produces very small droplets and avoid the use of air can have a positive impact on the retention
that are able to efficiently evaporate the water in the very short res- of bioactive compounds. An electrostatic spray-dryer, for example, by
idence time in the drying chamber. However, as a result of the high polarity differences is claimed to promote the migration of polar com-
temperatures that are typically above 130 ◦ C, degradation reactions can pounds, such as the solvent and the carrier, to the outer surface of the
be induced in thermo-sensitive compounds such as vitamins, flavors droplet, improving the active compound retention. The external avail-
or lipids, resulting in decreased quality of the product even after a few ability of water favors the evaporation process requiring lower inlet
seconds of exposure (Anandharamakrishnan et al., 2007; Belingheri temperatures. The patent claims that inlet temperature ranges from
et al., 2015; Fu and Etzel, 1995; Ixtaina et al., 2015; Jafari et al., 2008; 25 ◦ C to 110 ◦ C (Sobel et al., 2016). Similar technology was launched by
®
Menshutina et al., 2010; Mestry et al., 2011; Tonon et al., 2009; Tsotsas, Fluid Air, known as PolarDry , and tests have been done to potential
2012; Wijlhuizen et al., 1979; Zare et al., 2012). In addition, in conven- products.


Corresponding author.
E-mail addresses: fernandalp29@hotmail.com (F. de Melo Ramos), jubbink@calpoly.edu (J. Ubbink), vivaldo@unicamp.br (V. Silveira
Júnior), asprata@unicamp.br (A.S. Prata).
https://doi.org/10.1016/j.cherd.2019.03.036
0263-8762/© 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Chemical Engineering Research and Design 1 4 6 ( 2 0 1 9 ) 78–86 79

Nomenclature

Acronyms
C.V. Control volume
dm/dt Mass accumulate in the system over the time
Ṁinlet Mass flow rate of the inlet streams [kg/s]
Ṁoutlet Mass flow rate of the outlet streams [kg/s]
F Feed solution stream [kg/s]
Ai Atomization air stream (inlet) [kg/s]
Ao Air stream carrying the water evaporated (out-
let) [kg/s]
P Product stream [kg/s]
Gi Mass flow rate of atomization dry air in the inlet
of dryer [kg dry air/s] Fig. 1 – Global scheme of streams in the VSD.
Go Mass flow rate of dry air in the outlet of dryer
[kg dry air/s]
XF Mass fraction of water present in the feed solu-
In the same way, the vacuum spray dryer introduced in this study
tion [kg water/ kg feed solution]
is a drying technology under development (Ramos et al., 2016a, 2016b)
XP Mass fraction of water present in the product
that combines the advantages of the atomization, by increasing the sur-
[kg water/ kg product]
face area to improve the heat transfer, with the same strategy used for
Yi Absolute humidity of the atomizing air at the freeze-drying process, i.e., the reduction of the pressure inside the dry-
inlet [kg water/ kg dry air] ing chamber for decreasing the evaporation temperature of the water.
Yo Absolute humidity of the air at the outlet [kg of This combination of factors enables the processing of sensitive prod-
water/ kg dry air] ucts in a reduced time comparatively to freeze drying processes and
HF Enthalpy of the feed solution [J/s] under conditions that minimize product degradation.
Hi Enthalpy of the atomization air (inlet) [J/s] The operation principle of a VSD can be subdivided into three steps
HP Enthalpy of the product [J/s] that take place inside the drying chamber under vacuum, as shown in
Ho Enthalpy of the outlet air [J/s] the control volume (C.V.) around the nozzle (Fig. 1).
The first step, described as contact, occurs when the feed solution
Ha Enthalpy of the air stream (dry air (g) + water
enters in contact with the atomization air, in co-current flow, still inside
vapor (v)) [J/s]
the atomizing nozzle. The two stream entering in the system are those
Hstream Enthalpy of the stream containing solids [J/s]
from the feed solution and from the atomization air. Contrary to con-
Q Heat flow from the environment to the system ventional spray drying processes, the VSD’s chamber is not fed with
[J/s] a hot air stream and, as the only air that flows into this dryer is the
Cpg Specific heat of the air at the inlet and outlet compressed air used in the atomization, the low pressure inside the
streams [kJ/kg ◦ C] chamber can be maintained.
Cpv Specific heat of vaporization [kJ/kg ◦ C] Secondly, in the expansion step, the droplets of the feed solution
Cps Specific heat of the solids [kJ/kg ◦ C] are exposed to the low pressure inside the drying chamber. If its vapor
Cpw Specific heat of the liquid water [kJ/kg ◦ C] pressure of the water in the droplets is higher than the water vapor
pressure inside the chamber, part of the water molecules in the liquid
 Latent heat of vaporization [kJ/kg]
phase will evaporate driven by the mass gradient, since the vapor is not
Ta Air temperature [◦ C]
saturated. Subsequently, for continuing the evaporation of the liquid,
Tref Reference temperature (Tref = 0 ◦ C)
the latent heat of vaporization must be supplied by the internal energy
Ts Temperature of the product [◦ C] of the liquid. As the consequence, the evaporating droplet will reach
Tw Liquid water temperature [◦ C] the wet bulb temperature. With the established temperature gradient
Pw Partial pressure water vapor [kPa] (Tvapor –Twet bulb ), the evaporation step achieves a steady state that will
Pt Total pressure of the system [kPa] continue while there is excess of the water phase in the droplet. During
Ps Saturation vapor pressure [kPa] the evaporation step, a water content gradient in the drying droplet
ϕ Relative humidity [%] will be formed. Under certain conditions, a crust may be formed, with
Tg Glass transition temperature of the system [◦ C] kinetics depending on the glass transition of the material and the water
Tg,m Glass transition temperature of the dry matrix content dependence of the diffusion coefficient of water (Parker and
Ring, 1995). This crust, when formed, helps to avoid the sticking of the
[◦ C]
material to the wall of the dryer (Adhikari et al., 2007a) but also hinders
Tg,w Glass transition temperature of the water [◦ C]
the migration of the water phase to the surface because of the low rates
Qs Weight fraction solids in system [kg solids/ kg
of diffusion of water at low water contents (Räderer et al., 2002). All
total] these steps occur simultaneously inside of the drying chamber.
Qw Weight fraction water in system [kg water/ kg For spray drying, the basic phenomena include numerous aspects
total] of fluid mechanics, heat and mass-transfer, particle technology, and
kGT Gordon-Taylor coefficient materials science (Chen, 2004) and a number of models are available
in the literature whether as looking for the droplets drying kinetics
(Huang et al., 2003; Mezhericher et al., 2010; Woo et al., 2008) or flow
fields of the drying agent/particles (Kieviet, 1997; Langrish et al., 2004;
Zbiciński, 1995). Both aspects can be solved using discrete approxima-
tions Eulerian-Lagrangian by computational fluid dynamics (CFD) due
in part to the complexity of the gas and particle flow patterns and in
part to the overall complexity of the spray drying process (Kuriakose
and Anandharamakrishnan, 2010; Langrish, 2009).
80 Chemical Engineering Research and Design 1 4 6 ( 2 0 1 9 ) 78–86

Fig. 2 – Scheme of process and instrumentation of the vacuum spray dryer (VSD).

An alternative approach is coarse-scale modeling that can deter- 1.2 m3 /h at a pressure of 94.24 kPa and a temperature of 26.9 ◦ C.
mine in a relatively simple manner the processes conditions to be used The pressure inside the chamber was kept around at 27 kPa.
for short-chamber spray dryers based on a limited number of well- After that, a flow rate of 0.008 g/s was used, for 3 repetitions.
chosen assumptions (Langrish, 2009). The heat and mass transfer rates
Data of these trials are reported in this work. During the exper-
from the droplets to the gas phase are estimated, assuming well-mixed
iments, temperature and relative humidity data were acquired
conditions inside the chamber as well as equilibrium between the gas
every 30 s using an Arduino-Radiuíno platform and software
and particles in the outlet.
Drying in a vacuum spray dryer is an expensive technology. A Scada BR for the data acquisition by a computer.
detailed and quantitative understanding of the process is therefore The same solution was dried in conventional spray dryer
needed to optimize process parameters and reduce the costs of a poten- (MSD 1.0 model, Labmaq, Ribeirão Preto, Brazil) in order to
tial commercial technology. For this reason, we perform experiments compare particles produced by the two techniques. The flow
in the laboratory on pilot scale, which we corroborate with a modeling rate of the drying air was 81 m3 /h, the air pressure was 4 bar,
approach. In the present work, a coarse-scale approach was used to the compressed air flow rate was 2.4 m3 /h, and the feed rate
develop a mathematical model based on overall mass and energy bal- was 0.0007 m3 /h. The temperature of the drying air was set at
ances in the spray dryer. This model was validated through the drying
190 ◦ C.
of a maltodextrin solution by the vacuum spray drying process.

2.1. Characterization of the powders


2. Experimental design
2.1.1. Moisture content and water activity
A VSD model equipment was developed based on previ- The moisture content of the dried powders was determined
ous studies of Ramos et al. (2016a, 2016b) and consists in a gravimetrically at 105 ◦ C ± 5 ◦ C until constant weight following
sequence of vacuum pumps (0.3–0.9 kW), with rotary vanes, the official AOAC method (AOAC, 2006). Each powder replicate
adapted to a cylindrical-cone chamber to promote low pres- was tested in triplicate.
sure condition, as shown in Fig. 2. A double fluid atomizer Water activity (aw ) of the particles was measured at 25 ◦ C
(Labmaq, Brazil) is located at the top of the drying chamber in triplicate by a water activity analyzer (AquaLab Series 3TE,
and a small amount of air is supplied from a compressor MSV Decagon, Pullman, USA).
6 (Schulz, Brazil) to allow the atomization. The stainless steel
drying chamber is 200 mm in diameter and 680 mm tall. The 2.1.2. Scanning electron microscopy (SEM)
cyclone has a total height of 540 mm and its largest diameter is The samples were coated under vacuum with gold in a
95 mm. The system instrumentation used three temperature Sputter-Coater EMITECH unit (K450, Kent, UK). The thickness
sensors (resistance thermometers) to measure the air temper- of gold layer is estimated to be 200 Å.
ature at several spots in the system (TT 101 to 103 in Fig. 2). The micrographs were obtained in a scanning electron
Inside the vacuum line, a pressure sensor (PCG-750, Agilent microscope with Energy Dispersive X-ray Detection (SEM: Leo
Technologies) was placed to monitor the vacuum level (PT101 440i, EDS: 6070, SEM/EDS: LEO Electron Microscopy/ Oxford,
in Fig. 2). In this setup a relative humidity and temperature Cambridge, England). The analyses were performed with an
sensor (RHT-WM, Novus, Brazil) is located (MT101 and TT 103 accelerating voltage of 5 kV and a 50 pA beam current to obtain
respectively in Fig. 2). the micrographs.
A solution of 40% w/w of maltodextrin DE 10 (MOR-REX
DE 10, Ingredion, Brazil) was prepared by dissolving the mate- 2.1.3. Particle size distribution and mean diameter
rial in deionized water at 25 ◦ C for 24 h. The atomization of Size distribution and mean diameter of powders were deter-
the maltodextrin solution was performed using a peristaltic mined by light scattering technique using laser diffraction
pump (Masterflex, United States) with atomization airflow of (Mastersizer Hydro 2000 MU, Malvern Instruments Ltd.,
Chemical Engineering Research and Design 1 4 6 ( 2 0 1 9 ) 78–86 81

Malvern, UK). The mean diameter was determined based on 3.2. Energy balance
the mean diameter of a sphere of the same volume, the
Brouckere diameter (D43 ) (Eq. (1)). The samples were analyzed The general energy balance in the VSD can be obtained by an
by wet method, with dispersion in 99.5% ethanol. enthalpy balance in each stream in the C.V.. In this way, the
 4 energy balance becomes:
ni .di
D[4,3] =  (1)
3
ni .di (FHF + Gi Hi ) = (PHP + Go Ho ) (6)

where, di is the average particle diameter and ni is the number However, this process can be not completely adiabatic,
of particles. as some heat exchange between the system and its neigh-
borhood occurs. A term of the losses or gain in enthalpy
2.1.4. Wettability is therefore inserted in the balance equation. The gain can
Samples of each powder (∼1 g) were spread on the surface of occurs if the temperature inside the chamber is very low,
a beaker containing 100 mL of distilled water at 25 ◦ C without resulting in a heat flow from the environment to the chamber,
stirring in order to determine the wettability of the powders and thus:
(Fuchs et al., 2006). The time spent for immersing or wetting
the last particle of powder was used as wettability response of (FHF + Gi Hi + Q) = (PHP + Go Ho ) (7)
the samples.
whereby Q is the energy supplied to the drying chamber. For
2.1.5. Tapped density the air stream (dry air (g) + water vapor (v)), the enthalpy may
Tapped density in g. cm−3 was determined by measuring the generically be stated by the equation:
volume occupied by 2 g of powder sample in a 50 ml gradu-     
ated cylindrical glass tube (24 mm diameter, 198 mm height) at Ha = Cpg Ta − Tref + Y  + Cpv Ta − Tref (8)
room temperature (Goula and Adamopoulos, 2004) after tap-
ping the bottom of cylinder for ten times on a hard surface where, the specific heat of the air was calculated at the cor-
with interval of two seconds between each beat. The tapped respondent temperature of the stream (Jumah et al., 1996)
density was calculated as ratio between mass to volume. This and for Cpg the value of 1.04 kJ kg−1 C−1 at the inlet and
analysis was performed in triplicate. outlet streams was used. The specific heat of vaporization,
Cpv , was estimated as 1.8 kJ kg−1 C−1 , and the latent heat of
3. Model development vaporization, ␭, was taken at wet bulb temperature as 2500 kJ
kg−1 (Zhang et al., 2011) Ta is the air temperature (in ◦ C) and
3.1. Mass balance Tref = 0 ◦ C is the reference temperature.
The air absolute humidity (Y) was estimated using the
˙ psychometric correlations for the environmental air, entering
The general mass balance, in mass flow rate (M), for the VSD
with the temperature of the air in the following equation (Keey,
can be written as:
1978; Strumillo and Kudra, 1986):
dm  
= Ṁinlet − Ṁoutlet (2)
dt Y = 0.622Pw /(Pt − Pw ) (9)

The model is solved using overall mass and energy bal- where, Pw is partial pressure water vapor and Pt is the total
ances, and the C.V. chosen for the study of the process was pressure of the system. In the inlet stream the pressure was
the drying chamber (Fig. 1). Thus, one can find in the inlet, the 94.24 kPa, whereas in the outlet stream was 29.33 kPa.
feed solution stream (F) and atomizing air stream (Ai ), and in At the inlet of the dryer, the air temperature (Ta ) was
the outlet, the air stream carrying the water evaporated (Ao ) 26.94 ◦ C and the relative humidity was 44.68%. Firstly, the max-
and the product stream (P). No chemical reactions occur in the imum vapor pressure at the inlet air temperature, i.e., the
material and the mass of product accumulated outside of the saturation vapor pressure (Ps ), needed to be calculated. One
control volume (Figs. 1 and 2) was considered to be negligible way of calculating it was through the Antoine equation, one
(dm/dt = 0). Thus, the general mass balance in the VSD can be version of which is (Langrish, 2009):
described as:
 3816.44

Ps (Pa) = 133.3 exp 18.3036 − (10)
(F + Ai ) = (P + Ao ) [kg/s] (3) T (◦ C) + 229.02

Inside the chamber, the medium consists of three phases: The partial pressure of water vapor divided by the sat-
solid (s), liquid (l) and gas (g). The stream Ai is a moist air uration vapor pressure is the relative humidity(ϕ), which is
stream, and it is constituted by water vapor (v) and dry air (G). commonly expressed in %. Therefore, using the Eqs. (9)–(11),
The G streams, in the inlet (i) and in the outlet (o) of the dryer, we have that the saturation pressure and the partial pressure
are exactly the same. Then, the mass balance for the water water vapor was calculated as 3.53 and 1.57 kPa respectively.
component, in wet basis, can be described as:
Pw
ϕ= . 100% (11)
Ps
(FXF + Gi Yi ) = (PXP + Go Yo ) (4)
In the outlet stream, the air temperature (Ta ) was 26.86 ◦ C
For the solid components, we have: and the relative humidity was 14.8%. Consequently, using Eqs.
(9)–(11) we have that the saturation pressure and the partial
F (1 − XF ) = P (1 − XP ) (5) pressure water of vapor was calculated as 3.52 and 0.52 kPa
82 Chemical Engineering Research and Design 1 4 6 ( 2 0 1 9 ) 78–86

Table 1 – Characterization of the streams.


Parameters Inlet Outlet

Feed solution Air inlet Product Air outlet

ṁ [g/s] 0.008 0.359 0.003 0.359


Xi [kg water/ kg total] 0.600 – 0.079 –
Y [kg water/kg dry air] – 0.0106 – 0.0112

where ṁ =mass flow of the stream.

respectively. Thus, the absolute humidity values in the inlet Some discussions in the literature consider the energy sav-
and outlet conditions was calculated as 0.0106 and 0.0112 kg ing of vacuum systems as much inferior to convective dryers
water vapor/kg dry air respectively. (Motevali et al., 2014, 2011). The main reason for that, as for
The enthalpy of the stream containing solids (F and P) is freeze-dryers, is the thickness of the product layer that lim-
given by sum of the enthalpy of water and solids and can be its the diffusion of water through the material, increasing
described, in a generic way, by the equation: the processing time. However, in the VSD, the exposed area
is higher than in conventional vacuum drying due to atom-
ˆ     ization of the feed and the time to water evaporation is thus
Hstream = Cps Ts − Tref (1 − Xi ) + Cpw Tw − Tref (Xi ) (12) significantly reduced. On the other hand, conventional spray
dryer has a blower and a heater, which, for the MSD 1.0 model
(Labmaq, Ribeirão Preto, Brazil) that we employed together
For the specific heat of solids, Cps , a value of 1.5 kJ kg−1
add up to 4000 W of power. The vacuum system used in this
C−1 was used (Frías et al., 2001) and the specific heat of
work has three vacuum pumps which consume less than half
liquid water, Cpw is 4.18 kJ kg−1 C−1 (Baik et al., 2001). The
(1500 W) of this energy.
enthalpy was calculated in the correspondent temperature of
While we recognize that this prototype has some limita-
the stream (i.e., 24 ◦ C (F stream) and 26.9 ◦ C (P stream)). Ts is
tions, it turns out that these are mainly related to the design
the temperature of the product, ◦ C, considered in equilibrium
parameters of the dryer, and not to the intrinsic physics of the
with Tw , the liquid water temperature, ◦ C.
process. The drying process could be conducted even without
external energy source. As part of furthering the innovation
4. Results and discussion of this equipment, the prototype developed in this work is
being improved taking into account the results and analysis
Drying process normally involves high temperatures which reported in this article. Our current results however already
causes sensorial and nutritional changes in food products show that for the mass flow as applied here (0.008 g/s), the
and convective air to accelerate the heat and mass transfer VSD uses about 25% less energy (102.6 J/ kg) in comparison
and which consequently increments the oxygen supply. Both with a conventional spray dryer (424.2 J/ kg). The total energy
parameters are harmful for thermo-labile and oxidable prod- involved by a SD is higher than that used for VSD due to the
ucts as vitamins, colorants and oils. heating of the flow air. Of course, a higher thermal input would
Also, at these conditions, a high energy consumption is allow greater transfer of energy to the system, providing better
required due to the latent heat supply for water evaporation, conditions for water evaporation and favoring an increase of
the sensible heat for heating the air and the energy require- the flow rate of the solution. Thus, even with additional heat
ments for blowers. For example, in a conventional spray dryer, supplied to the process to improve the flow rate, the VSD can
about 3900 W of energy-input is required to process 1 kg/s of not be considered more energy expensive than conventional
a 40% w/w solution using 15 kg/s of dry airflow (0.01 kg water SD.
vapor/kg dry air). From this amount, 85% of the energy is used All values of incoming and outgoing stream in the drying
to heat the air stream (m.Cp .dT) up to the inlet temperature chamber required by the mass balance were experimentally
of the spray dryer (200 ◦ C). High temperatures are required determined and these values are presented in Table 1.
by considering the small residence time of the particles in The energy balance developed considers that one part
the path of the spray dryer, which is around 4 s (Birchal and of the heat required to evaporate the water is supplied by
Passos, 2005). However, when observing the water withdrawal the environment. The parameters used and results from this
potential of a stream of dry air, which is ideally the adiabatic energy balance developed can be seen in Table 2.
saturation limit, it is seen that air exits with about 60% of its In the literature, a limited number of studies (Aoyama et al.,
water saturation capacity, indicating that this energy portion 2009; Islam et al., 2017, 2016; Kitamura et al., 2009; Ramos et al.,
can be optimized by reducing the temperature or the air flow 2016a, 2016b; Semyonov et al., 2011) are published with dry-
employed. ers based on the vacuum principle. Several groups employed
In the VSD developed in this work, auxiliary air is not infrared heaters installed near the atomizer (Aoyama et al.,
employed for that the drying occurs and due to significant 2009; Kitamura et al., 2009) to supply energy through electro-
reduction of the amount of oxygen in the chamber, the magnetic waves as auxiliary source of heat for evaporation. In
quality of products prone to oxidation can be improved. Dry- another study (Islam et al., 2017, 2016) superheated steam at
ing can account for 60–70% of the total energy consuming 200 ◦ C was used. All the groups circulated hot water at 50 ◦ C
of a process (Bahu, 1991) and therefore the evaluation of inside the chamber jacket to increase the temperature of the
drying systems regarding sustainability beyond the require- inner wall of the chamber and prevent the sticking of materi-
ments of product quality have a strong interest to allow its als to the wall. Our study is different in this respect, as there
implementation.
Chemical Engineering Research and Design 1 4 6 ( 2 0 1 9 ) 78–86 83

Table 2 – Energy parameters of inlet and outlet stream in VSD.


Parameters Inlet Outlet

Gi F Q Go P

h [kJ/kg] 55.1 74.6 – 56.6 46.0


H [J/s] 19.78 0.59 0.09a 20.31 0.14
T [◦ C] 26.9 24 – 26.9 26.9

a
Calculated value from energy balance.

The dependence of the glass transition temperature on the


water content can be modeled by the Gordon-Taylor equation
(Gordon and Taylor, 1952):

T g,m .Q s + kGT .Q w .T g,w


Tg = (13)
T g,m + kGT .T g,w

where, Tg , Tg,m , Tg,w : glass transition temperature of system,


dry matrix and water, Qs , Qw : weight fraction solids and water
in system, kGT : Gordon-Taylor coefficient, and for maltodex-
trin corresponds to 7.3.
Equation (13) is conventionally used to fit the water con-
tent dependence of the glass transition temperature of food
materials to determine the end point for drying. If the glass
temperature of one component is increased, the glass transi-
Fig. 3 – State diagram of maltodextrin DE-12. The state tion temperature of the system is increased as well, implying
diagram is composed of the ice melting line Tm , the boiling in that the amount of water that needs to be removed is lower
line Tb , the glass transition line Tg and the glass transition to reach the glassy state. In addition, the drying is aided by
temperature of the maximally cryo-concentrated state Tg’ . the high solids content of the feed, which further limits the
Tb is plotted atmospheric pressure (101.35 kPa; amount of water that needs to be removed during the drying
conventional spray drying) and a pressure of 3.5 kPa operation. The current material is therefore not very sensitive
(vacuum spray drying). As maltodextrins do not show a to sticking that often occurs in spraying towers and that is
clear crystallization limit, the solubility line, commonly due to the high product’s outlet temperature relative to its Tg .
plotted for simple carbohydrate-water systems, is omitted. Such more sensitive materials remain in the rubbery state at
The approximate process routes for conventional and the end of the drying (Adhikari et al., 2007a, 2007b, 2004).
vacuum spray drying are indicated by the dashed arrows, The critical condition for any spray drying process is the
which represent the sample trajectory from the initial water content below which the powder is in the glassy state at
solution at 0.4 w/w maltodextrin content to the final either the outlet temperature of the dryer or the storage or the
powder with a water content of 0.08 w/w. The state storage temperature of the powder. For a storage temperature
diagram is compiled using data from Kilburn et al. (2005). of 25 ◦ C, the powder is at the glass transition for a weight frac-
tion water of 0.103 (wet basis). For any water content below this
value, the maltodextrin powder will thus be in the glassy state.
As the powder after drying in the VSD attained at water con-
was no external way of supplying vaporization heat to the sys- tent of 0.079 kg water/kg product, the powder is in the glassy
tem, not even in the feed solution that was processed at room state after drying. We conclude that the current vacuum spray
temperature. process is able to dry solutions and dispersions to a sufficient
In a schematic state diagram (Fig. 3), the pathway for the extent to allow the production of physically-stable powders,
process vacuum operated as well as the pathway that repre- notwithstanding that its water content is higher than that
sent the evaporation of the water in the conventional process which is normally obtained using a conventional spray dryer
are indicated by dotted lines. (water weight fraction in the range of 0.04) (Fang and Bhandari,
The dried sample is a model product, consisting of pure 2012), can be considered stable. While the water content of the
maltodextrin with a dextrose equivalent (DE) of 10 and a glass current powder prepared by VSD is low enough for the powder
transition temperature in the dry state of 164 ◦ C (Kilburn et al., to be in the glassy state, it is still higher than what is usually
2005). These characteristics enable a more efficient drying by considered to be optimal in case oxidation sensitive bioactives
reducing the sticking of the material to the wall of the spray are encapsulated in the maltodextrin matrix. A further lower-
dryer when compared to carbohydrates of lower molecular ing of the water content should is definitely be possible using
weight. Sticking occurs when the viscosity of the material our process, the goal of our research is however to demon-
is sufficiently low, i.e., in the early stages of the drying, and strate that we could obtain a powder in the glassy state using
the empirical parameter to quantify the propensity to stick- the VSD, not to fully optimize powder stability in the light of
ing, the so-called, sticky point, was shown to be related to the stability of a potential encapsulant.
the glass transition temperature (Tg ) of the material to be The powder obtained in VSD was apparently more porous
dried (Roudaut et al., 2002). The Tg , in turn, is influenced (Fig. 4b) when compared to the equivalent powder obtained in
by both the molecular weight of the material and its water conventional spray dryer using the same nozzle and solution
content. of maltodextrin in the same concentration.
84 Chemical Engineering Research and Design 1 4 6 ( 2 0 1 9 ) 78–86

Fig. 5 – Particle size distributions for the particles produced


in VSD and conventional spray dryer.

processes, conventional spray dryer and VSD, is an important


factor to be understood for explaining the differences between
the structures formed.
Fig. 4 – SEM micrographs of particles obtained by spray Results of moisture content, water activity, wettability,
dryer and vacuum spray dryer. apparent density and mean diameter obtained for the par-
ticles produced in the SD and VSD are shown in Table 3.
The conventional spray drying process implies an inten- Particles produced by VSD were significantly different from
sive moisture evaporation at the surface of droplets, and the those obtained with SD. Wettability is the time required for
heat of evaporation of water is responsible to keep the droplets water to imbibe to the particles until they sediment. The wet-
at the wet bulb temperature until the evaporation of water is tability of VSD particles was approximately 7 times lower than
significantly restricted by low diffusion rates at the low mois- the particles from SD (Table 3). This is most likely related to
ture crusty formed at the particle surface. The driving force the porous structure formed by these particles in VSD, which
for water removal in this process is much higher than in VSD can be represented by their higher apparent density. The mean
and the evaporation process impacts on the development of diameter of particles can contribute with this effect, because
the particle morphology. smaller particles tends to be together and hinder the water
In the case of spray drying, the high pressures gener- hydration.
ate inside of particles can produce particles that inflate In principle, the mean diameter of particles produced by
(Fig. 4a) significantly producing smooth and spherical sur- atomization depends on the type of atomizer (Filkova and
faces (Oakley, 1997). Under vacuum, the particles will also Mujumdar, 1995). However, even though the same atomizer
puff and porous channels are formed, which is a well-known was used, particles produced by VSD were larger (159 ␮m) than
phenomenon occurring in freeze–dried materials and can be SD particles (88 ␮m) – Table 3. However, as can be seen in Fig. 5,
attributed to the expansion of air and water vapor at sub- the size distribution occurs in two peaks. The first one, in about
ambient pressures (Motevali et al., 2014). But we hypothesize 20 ␮m, represents the size of VSD particles that, as expected,
that two more reasons are responsible for the observed behav- presented some morphological alterations as compared to SD
ior. In VSD, the water vapor is more evenly released due the particles (50 ␮m). The peak at 500 ␮m refers to the recovered
longer time took to form the surface crust in contact with the powder from the wall of the VSD. The particles recovered from
drying environment. This can reduce the resistance to mass the wall of the drying chamber in this process were larger.
transfer close to the particle surface as the water diffusion For SD, the size measurement was done only on particles col-
coefficient is very low in the crust. This effect is reinforced lected in the cyclone. These observations are consistent with
by the migration rate of water from the matrix being more the morphology of the particles shown in Fig. 4. The value of
homogeneous due to the lower driving force for heat and mass the polydispersity index (PDI) indicates the degree of unifor-
transfer, further reducing the internal pressure. Another fac- mity of the distribution. For VSD particles, the value obtained
tor that could have an influence is that the water removal was 17.2 while for powders obtained by SD it was 3.46, which
becomes more difficult over time due to an increased vis- confirms the less uniform size distribution for VSD particles.
cosity linked to the low temperatures, which is additionally We discard the possibility of agglomeration, because at the
be responsible to restrain the formed structure from inflating final moisture content (7%) reached at the end of vacuum dry-
(Fig. 4b). The difference between the driven force between the ing process, the water activity of 10 DE maltodextrin is 0.65.

Table 3 – Moisture content, water activity, wettability, apparent density and mean diameter of the powders produced in
conventional SD and VSD.
Equipment Moisture content (%) Water activity (aw ) Wettability (s) Apparent density (g/cm3 ) Mean diameter (␮m)

Spray dryer 1.13 ± 0.22a 0.043 ± 0.006a 1341.00 ± 28.00b 0.286 ± 0.016a 87.68 ± 13.68a
Vacuum spray dryer 7.88 ± 0.09b 0.381 ± 0.003b 192.00 ± 20.07a 0.412 ± 0.004b 159.51 ± 4.78b

Different letters indicate statistically significant differences between the samples produced in the different types of equipment (p ≤ 0.05).
Chemical Engineering Research and Design 1 4 6 ( 2 0 1 9 ) 78–86 85

Agglomeration would only be important if relative humidity Baik, O.D., Marcotte, M., Sablani, S.S., Castaigne, F., 2001. Thermal
of drying environment (45%) were higher than 65%. and physical properties of bakery products. Crit. Rev. Food Sci.
Nutr. 41, 321–352, http://dx.doi.org/10.1080/20014091091832.
Belingheri, C., Giussani, B., Rodriguez-Estrada, M.T., Ferrillo, A.,
5. Conclusion Vittadini, E., 2015. Oxidative stability of high-oleic sunflower
oil in a porous starch carrier. Food Chem. 166, 346–351,
A conventional spray dryer was adapted to be operated under http://dx.doi.org/10.1016/j.foodchem.2014.06.029.
vacuum and was evaluated by the drying of atomized droplets Birchal, V.S., Passos, M.L., 2005. Modeling and simulation of milk
of a DE 10 maltodextrin solution. We adopted a coarse-grained emulsion drying in spray dryers. Brazilian J. Chem. Eng. 22,
mathematical model to account for the heat and mass bal- 293–302, http://dx.doi.org/10.1590/S0104-66322005000200018.
Chen, X.D., 2004. Heat-mass transfer and structure formation
ances in this system, and we validated it with data of pressure
during drying of single food droplets. Dry. Technol. 22, 1–2,
and temperature acquired on line. The results show that the
http://dx.doi.org/10.1081/DRT-120028226.
proposed mathematical model represents the process well. Fang, Z., Bhandari, B., 2012. Spray drying, freeze drying and
As the vacuum drying process takes place without external related processes for food ingredient and nutraceutical
source of heat, the saves VSD energy when compared with encapsulation. Encapsulation Technol. Deliv. Syst. Food
freeze drying and conventional spray drying processes. Also, Ingredients Nutraceuticals, 73–109,
as the process does not involve an air stream, the oxygen http://dx.doi.org/10.1016/B978-0-85709-124-6.50004-4.
Filkova, I., Mujumdar, A.S., 1995. Industrial spray drying Systems.
availability inside the chamber is reduced. This should bring
Handbook of Industrial Drying, vol. 1., 2nd ed. Inc., New York,
significant benefits in the drying of oxygen sensitive bioactive pp. 263–308.
compounds. Because the lack of an external heat source, only Frías, J.M., Oliveira, J.C., Schittkowski, K., 2001. Modeling and
a very low amount of solution could be processed. Notwith- parameter identification of a maltodextrin DE 12 drying
standing, the drying kinetics in the VSD are sufficiently fast to process in a convection oven. Appl. Math. Model. 25, 449–462,
enable the conversion of the maltodextrin DE 10 solution into http://dx.doi.org/10.1016/S0307-904X(00)00060-3.
Fu, W.-Y., Etzel, M.R., 1995. Spray drying of Lactococcus lactis ssp.
an amorphous powder in the glassy state at the end of the
lactis c2 and cellular injury. J. Food Sci. 60, 195–200.
drying process. The powder characteristics of both processes
Fuchs, M., Turchiuli, C., Bohin, M., Cuvelier, M.E., Ordonnaud, C.,
were significantly different, and the VSD particles showed a Peyrat-Maillard, M.N., Dumoulin, E., 2006. Encapsulation of oil
porous aspect contributing with their wettability. However, in powder using spray drying and fluidised bed
the effect of internal pressure of the chamber on the particles agglomeration. J. Food Eng. 75, 27–35,
characteristics warrants further exploration. The evaluation http://dx.doi.org/10.1016/j.jfoodeng.2005.03.047.
of the moisture content and water activity of atomized mate- Gordon, M., Taylor, J.S., 1952. Ideal copolymers and the
second-order transitions of synthetic rubbers. i.
rial during the process may help to design a new dryer that
non-crystalline copolymers. J. Appl. Chem. 2, 493–500,
allows to optimize direct correlating with the product quality.
http://dx.doi.org/10.1002/jctb.5010020901.
Goula, A.M., Adamopoulos, K.G., 2004. Spray drying of tomato
Acknowledgement pulp: effect of feed concentration. Dry. Technol. 22, 2309–2330,
http://dx.doi.org/10.1081/DRT-200040007.
This study was financed in part by the Coordenação de Huang, L., Kumar, K., Mujumdar, A.S., 2003. A parametric study of
the gas flow patterns and drying performance of co-current
Aperfeiçoamento de Pessoal de Nível Superior – Brasil (CAPES)
spray dryer: results of a computational fluid dynamics study.
– Finance Code 001.
Dry. Technol. 21, 957–978,
http://dx.doi.org/10.1081/DRT-120021850.
References Islam, M.Z., Kitamura, Y., Yamano, Y., Kitamura, M., 2016. Effect
of vacuum spray drying on the physicochemical properties,
Adhikari, B., Howes, T., Bhandari, B.R., Troung, V., 2004. Effect of water sorption and glass transition phenomenon of orange
addition of maltodextrin on drying kinetics and stickiness of juice powder. J. Food Eng. 169, 131–140,
sugar and acid-rich foods during convective drying: http://dx.doi.org/10.1016/j.jfoodeng.2015.08.024.
experiments and modelling. J. Food Eng. 62, 53–68, Islam, M.Z., Kitamura, Y., Kokawa, M., Monalisa, K., Tsai, F.H.,
http://dx.doi.org/10.1016/S0260-8774(03)00171-7. Miyamura, S., 2017. Effects of micro wet milling and vacuum
Adhikari, B., Howes, T., Shrestha, A., Bhandari, B.R., 2007a. Effect spray drying on the physicochemical and antioxidant
of surface tension and viscosity on the surface stickiness of properties of orange (Citrus unshiu) juice with pulp powder.
carbohydrate and protein solutions. J. Food Eng. 79, 1136–1143, Food Bioprod. Process. 101, 132–144,
http://dx.doi.org/10.1016/j.jfoodeng.2006.04.002. http://dx.doi.org/10.1016/j.fbp.2016.11.002.
Adhikari, B., Howes, T., Shrestha, A.K., Bhandari, B.R., 2007b. Ixtaina, V.Y., Julio, L.M., Wagner, J.R., Nolasco, S.M., Tomás, M.C.,
Development of stickiness of whey protein isolate and lactose 2015. Physicochemical characterization and stability of chia
droplets during convective drying. Chem. Eng. Process. 46, oil microencapsulated with sodium caseinate and lactose by
420–428, http://dx.doi.org/10.1016/j.cep.2006.07.014. spray-drying. Powder Technol. 271, 26–34,
Anandharamakrishnan, C., Rielly, C.D., Stapley, A.G.F., 2007. http://dx.doi.org/10.1016/j.powtec.2014.11.006.
Effects of process variables on the denaturation of whey Jafari, S.M., Assadpoor, E., He, Y., Bhandari, B., 2008.
proteins during spray drying. Dry. Technol. 25, 799–807, Encapsulation efficiency of food flavours and oils during spray
http://dx.doi.org/10.1080/07373930701370175. drying. Dry. Technol. 26, 816–835,
AOAC, 2006. Official Methods of Analysis of the Association of http://dx.doi.org/10.1080/07373930802135972.
Official Analytical Chemists. Association of Official Analytical Jumah, R.Y., Mujumdar, A.S., Raghavan, G.S.V.A., 1996. A
Chemists, Gaitherburg, MD. mathematical model for constant and intermittent batch
Aoyama, R., Kitamura, Y., Yamazaki, K., 2009. Experimental drying of grains in a novel rotating jet spouted bed. Dry.
analysis of spraying and drying characteristics in vacuum Technol. 14, 765–802,
spray dryer. Japan J. Food Eng. 1 (0), 127–133, http://dx.doi.org/10.1080/07373939608917124.
http://dx.doi.org/10.11301/jsfe.10.127. Keey, R.B., 1978. Introduction to Industrial Drying Operations.
Bahu, R.E., 1991. Energy consumption in dryer design. In: Pergamon Press, Oxfor, UK.
Mujumdar, A.S., Filkova, I. (Eds.), Drying’91. Elsevier, Keshani, S., Daud, W.R.W., Nourouzi, M.M., Namvar, F., Ghasemi,
Amsterdam, pp. 553–557. M., 2015. Spray drying: an overview on wall deposition,
86 Chemical Engineering Research and Design 1 4 6 ( 2 0 1 9 ) 78–86

process and modeling. J. Food Eng. 1 (46), 152–162, Räderer, M., Besson, A., Sommer, K., Bauer, W., Karel, M., 2002.
http://dx.doi.org/10.1016/j.jfoodeng.2014.09.004. Drying of viscous, shrinking products: modelling and
Kieviet, F.G., 1997. Modelling Quality in Spray Drying., experimental validation. Dry. Technol. 20, 539–541,
http://dx.doi.org/10.2298/TSCI121205224E. http://dx.doi.org/10.1081/DRT-120002556.
Kilburn, D., Claude, J., Schweizer, T., Alam, A., Ubbink, J., 2005. Ramos, F.D.M., Oliveira, C.C.M. de, Soares, A.S.P., Silveira Júnior,
Carbohydrate polymers in amorphous states: an integrated V., 2016a. BR 10 2016 020038 5 300816.
thermodynamic and nanostructural investigation. Ramos, F.D.M., Oliveira, C.C.M., de Soares, A.S.P., Silveira Júnior, V.,
Biomacromolecules 6, 864–879, 2016b. Assessment of differences between products obtained
http://dx.doi.org/10.1021/bm049355r. in conventional and vacuum spray dryer. Food Sci. Technol.
Kitamura, Y., Itoh, H., Echizen, H., Satake, T., 2009. Experimental 36, 724–729, http://dx.doi.org/10.1590/1678-457X.09216.
vacuum spray drying of probiotic foods included with lactic Roudaut, G., Dacremont, C., Vallès Pàmies, B., Colas, B., Le Meste,
acid bacteria. J. Food Process. Preserv. 33, 714–726, M., 2002. Crispness: a critical review on sensory and material
http://dx.doi.org/10.1111/j.1745-4549.2008.00299.x. science approaches. Trends Food Sci. Technol. 13, 217–227,
Kuriakose, R., Anandharamakrishnan, C., 2010. Computational http://dx.doi.org/10.1016/S0924-2244(02)00139-5.
fluid dynamics (CFD) applications in spray drying of food Semyonov, D., Ramon, O., Shimoni, E., 2011. Using ultrasonic
products. Trends Food Sci. Technol. 21, 383–398, vacuum spray dryer to produce highly viable dry probiotics.
http://dx.doi.org/10.1016/j.tifs.2010.04.009. LWT - Food Sci. Technol. 44, 1844–1852,
Langrish, T.A.G., 2009. Multi-scale mathematical modelling of http://dx.doi.org/10.1016/j.lwt.2011.03.021.
spray dryers. J. Food Eng. 93, 218–228, Sobel, R.M., Bun, B., Su, C-P., Gundlach, M., Ackerman, Jr., T.E., St.
http://dx.doi.org/10.1016/j.jfoodeng.2009.01.019. Peter, G.R., 2016. WO 2016/123224 Al 04082016.
Langrish, T.A.G., Williams, J., Fletcher, D.F., 2004. Simulation of Strumillo, C., Kudra, T., 1986. Drying: Principles, Application and
the effects of inlet swirl on gas flow patterns in a pilot-scale Design. Gordon and Breach, New York, NY.
spray dryer. Chem. Eng. Res. Des. 82, 821–833, Tonon, R.V., Brabet, C., Hubinger, M.D., 2009. Influência da
http://dx.doi.org/10.1205/0263876041596661. temperatura do ar de secagem e da concentração de agente
Menshutina, N.V., Gordienko, M.G., Voinovskiy, A.A., Zbicinski, I., carreador sobre as propriedades físico-químicas do suco de
2010. Spray drying of probiotics: process development and açaí em pó. Ciência e Tecnol. Aliment. 29, 444–450,
scale-up. Dry. Technol. 28, 1170–1177, http://dx.doi.org/10.1590/S0101-20612009000200034.
http://dx.doi.org/10.1080/07373937.2010.483043. Tsotsas, E., 2012. Influence of drying kinetics on particle
Mestry, A.P., Mujumdar, A.S., Thorat, B.N., 2011. Optimization of formation: a personal perspective. Dry. Technol. 30,
spray drying of an innovative functional food: fermented 1167–1175, http://dx.doi.org/10.1080/07373937.2012.685139.
mixed juice of carrot and watermelon. Dry. Technol. 29, Wijlhuizen, A.E., Kerkhof, P.J.A.M., Bruin, S., 1979. Theoretical
1121–1131, http://dx.doi.org/10.1080/07373937.2011.566968. study of the inactivation of phosphatase during spray-drying
Mezhericher, M., Levy, A., Borde, I., 2010. Theoretical models of of skim-milk. Chem. Eng. Sci. 34, 651–660,
single droplet drying kinetics: a review. Dry. Technol. 28, http://dx.doi.org/10.1016/0009-2509(79)85110-6.
278–293, http://dx.doi.org/10.1080/07373930903530337. Woo, M.W., Daud, W.R.W., Mujumdar, A.S., Talib, M.Z.M., Hua,
Motevali, A., Minaei, S., Khoshtaghaza, M.H., Amirnejat, H., 2011. W.Z., Tasirin, S.M., 2008. Comparative study of droplet drying
Comparison of energy consumption and specific energy models for CFD modelling. Chem. Eng. Res. Des. (86),
requirements of different methods for drying mushroom 1038–1048, http://dx.doi.org/10.1016/j.cherd.2008.04.003.
slices. Energy 36, 6433–6441, Zare, D., Salehi, A., Niakousari, M., 2012. Determination of
http://dx.doi.org/10.1016/j.energy.2011.09.024. physical properties of sour orange juice powder produced by a
Motevali, A., Minaei, S., Banakar, A., Ghobadian, B., spray dryer. An ASABE Meet. Present., 1–14.
Khoshtaghaza, M.H., 2014. Comparison of energy parameters Zbiciński, I., 1995. Development and experimental verification of
in various dryers. Energy Convers. Manage. 87, 711–725, momentum, heat and mass transfer model in spray drying.
http://dx.doi.org/10.1016/j.enconman.2014.07.012. Chem. Eng. J. 58, 123–133,
Oakley, D.E., 1997. Produce uniform particles by spray drying. http://dx.doi.org/10.1016/0923-0467(94)02943-1.
Chem. Eng. Prog. 93, 48–54. Zhang, W., Yao, Y., He, B., Wang, R., 2011. The energy-saving
Parker, R., Ring, S.G., 1995. Diffusion in maltose-water mixtures at characteristic of silica gel regeneration with high-intensity
temperatures close to the glass transition. Carbohydr. Res. ultrasound. Appl. Energy 88, 2146–2156,
273, 147–155, http://dx.doi.org/10.1016/0008-6215(95)00120-I. http://dx.doi.org/10.1016/j.apenergy.2010.12.023.

You might also like