- Email: [email protected]
Fluidised bed agglomeration of particles with different glass transition temperatures
PTEC-10114; No of Pages 8 Powder Technology xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Powder Technology journal homepage: www.elsevier.com/locate/powtec
Fluidised bed agglomeration of particles with different glass transition temperatures Carlos Avilés-Avilés, Elisabeth Dumoulin, Christelle Turchiuli ⁎ AgroParisTech, UMR 1145 Ingénierie Procédés Aliments, 1 Avenue des Olympiades, F-91300, Massy, France INRA, UMR1145 Ingénierie Procédés Aliments, F-91300, Massy, France CNAM, UMR1145 Ingénierie Procédés Aliments, F-91300, Massy, France
a r t i c l e
i n f o
Available online xxxx Keywords: Agglomeration Fluidised bed Growth mechanisms Stickiness
a b s t r a c t Fluidised bed agglomeration of particles consists in spraying liquid drops (water or binder solution) on the surface of particles, fluidised by hot air, in order to create sticky regions to allow formation of agglomerates when the sticky particles collide. Many parameters influence agglomerate growth, especially those controlling the particle circulation, and the water and temperature conditions within the bed that determine drying and particle stickiness linked to the glass transition of amorphous components and to the viscosity of moist zones at the particle surface. Maltodextrin DE12 and DE21 particles with different glass transition temperature domains were agglomerated in a batch bench scale fluidised bed, under constant mechanical constraints, changing the sprayed water feed rate and the fluidisation air temperature in order to investigate the influence of particle stickiness on agglomerate growth kinetics and mechanism. The two powders showed a different sensitivity to the water and temperature constraints applied. Whilst for maltodextrin DE12, the size and growth rate increased significantly with the sprayed water flow rate, only a small variation was observed for maltodextrin DE21. In both cases, agglomeration occurred in two stages: firstly, the association of initial particles and secondly the agglomeration of intermediate agglomerates into larger and more porous structures. The change from one growth mechanism to the other depended on the conditions, and influenced the size distribution and structure of agglomerates. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Agglomeration is widely used in the food and pharmaceutical industry where a great amount of the raw materials, intermediate or final products are in powder form. Agglomerates consist in the assembly of fine particles (~ 200 μm) linked by solid bridges to create large porous structures (~800 μm). By modifying the physical properties of the powder particles (size, shape, porosity, density), agglomeration allows modifying the powder end-use properties (flowability, wettability, instantaneity). Fluidised bed agglomeration consists of spraying a liquid (water or binder aqueous solution) on a bed of solid particles fluidised by hot air. Sprayed liquid droplets collide with the individualised and moving particles, allowing to create sticky zones on their surfaces, either due to the local wetting of the particle surface (water soluble particles) or
⁎ Corresponding author at: AgroParisTech, UMR 1145 Ingénierie Procédés Aliments, 1 Avenue des Olympiades, 91300 Massy, France. Tel.: +33 1 69 93 50 71; fax: +33 1 69 93 50 05. E-mail addresses: [email protected] (C. Avilés-Avilés), [email protected] (E. Dumoulin), [email protected] (C. Turchiuli).
to the deposit of binder solution on the particle surface (non soluble particles). Agglomerate formation takes place when the sticky moving particles collide with others (sticky or not) and form liquid or viscous bridges that are dried and consolidated by the fluidisation hot air. The repetition of these steps (spraying, wetting, collision, adhesion and drying) allows progressive agglomerate growth. Fluidised bed agglomeration is therefore a complex process resulting in different phenomena. Agglomerate growth, structure and properties depend on many parameters linked to the equipment, to the operating conditions, and to the particle and sprayed liquid properties (Table 1) [1]. These parameters determine the mechanical, temperature and water constraints applied inside the particle fluidised bed. Mechanical constraints are mainly due to the agitation of particles by the fluidising air allowing both contacts between particles by friction or collision, and the distribution of the sprayed liquid within the bed. For a given device and geometry, mechanical constraints mainly depend on the fluidising air flow rate; on the particle load in the chamber; and on the particle size, size distribution and density influencing fluidisation. Temperature constraints are linked to the temperature and flow rate of both the hot fluidising air and sprayed liquid. They determine the temperature and the drying conditions within the particle bed and therefore also influence the water constraints. Controlled agglomerate growth occurs
http://dx.doi.org/10.1016/j.powtec.2014.03.026 0032-5910/© 2014 Elsevier B.V. All rights reserved.
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
2
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
Table 1 Main parameters in fluidised bed agglomeration and their contribution to the mechanical, thermal and water constraints. Geometry of granulator
Operating parameters
Product parameters
Chamber * Size * Shape Air distributor Nozzle * Type * Position
Particle load Fluidising air * Flow rate * Temperature Spraying * Liquid feed rate * Liquid drop size
Particles * Composition * Size/size distribution, density * Surface Sprayed liquid * Composition/concentration * Temperature
when the conditions allow bridges to create between colliding particles without over wetting and collapse of the bed. This requires controlling at the same time agitation, wetting and drying of the particle bed (i.e. intensity and duration of the constraints applied). The wetting of the particle surface by the sprayed liquid allows the formation of liquid or viscous bridges between colliding particles thanks to the viscous dissipation of the collision kinetic energy. Actually, a collision between two particles is effective only if the Stokes number St is below a certain critical value St* [2,3]. Above St*, particle rebound is observed and the collision does not allow agglomerate growth: St ¼ 8 ρp r p uc =ð9 μl Þ and St ¼ ð1 þ 1=cÞ ln ðe=ru Þ Where ρp is the particle density (kg m−3), rp is the mean particle radius (m), uc is the relative collision velocity (m s −1), μl is the viscosity of the viscous film (Pa s), c is the collision restitution coefficient (-), e is the viscous film thickness (m), and ru is the particle surface rugosity (-). However, the formed liquid or viscous bridges must also be able to give rise to solid bridges by the evaporation of the solvent (water) during hot-air drying. For the agglomeration of water insoluble particles the spraying of an aqueous solution containing a binder (e.g. amorphous component) is needed to form a viscous adhering film on the particle surface and render it “sticky”. When particles are soluble in water and contain amorphous components, they can be agglomerated by spraying pure water. In this case, water can cause both some partial dissolution and plastification, decreasing the glass transition temperature of the amorphous components on the particle surface [4]. In the specific
temperature conditions of the fluidised bed, these plasticized amorphous components undergo glass transition causing a decrease of their viscosity and the appearance of stickiness when they reach the rubbery state. This corresponds to a critical viscosity range between 108 and 1012 Pa s [5,6]. For a given water content, temperatures for sticky conditions range between the glass transition temperature (Tg), and the “sticky temperature” (Ts); from 20 to 30 °C above Tg for the main food carbohydrates (maltodextrin, lactose, fructose, sucrose, etc.) [7,8]. Tg and Ts depend on the composition of the product considered. During fluidised bed agglomeration, the particle surface water content and temperature permanently vary along the bed and with time due to simultaneous wetting and drying of the particles (1–4, Fig. 1) and to the agitation causing their circulation through the thermal zones within the bed [9,10]. The presence of sticky particles able to agglomerate when colliding therefore depends on their position, on temperature and water conditions within the bed and on the particle surface composition [4,8,11]. Different studies were performed on the effect of parameters such as fluidising air temperature and flow rate; nature, concentration and feed rate of the sprayed solution on agglomerates growth [12–17]. Most of them were performed with non soluble particles and only few investigated the behaviour of particles undergoing glass transition. The aim of this work was to study the agglomerate growth during fluidised bed agglomeration of particles with different glass transition temperature evolution with the water content. Two model powders e.g. maltodextrins with different dextrose equivalent (DE12 and DE21), were agglomerated spraying water. Similar mechanical constraints (same air flow rate, particle load and initial size distribution) were applied, but temperature and water constraints were changed by modifying the sprayed water feed rate and the fluidising air temperature. Agglomerate growth was studied following the evolution of both the median diameter and particle size distribution in order to consider the influence of the studied parameters on both the growth kinetics and growth mechanisms. 2. Materials and methods 2.1. Powders Maltodextrin DE12 and DE21 (Glucidex, Roquette, Fr) were used for agglomeration trials (Table 2). Prior to each trial, powders were sieved (Analysette 3 Spartan, Fritsch, Ge—80 g, 12 min, amplitude of vibration 2.5 mm) to obtain a narrow size distribution. The fraction between 100 and 315 μm was kept in order to have a median diameter d50 of about
Fig. 1. Glass transition and sticky temperatures (Tg and Ts) as a function of water content X and water activity aw for maltodextrin DE12 and evolution of temperature and water content of moist zones on the particle surface during the different steps of fluidized bed agglomeration (1. initial particle; 2. wetting; 3. drying/collision; 4. solid bridge/dry agglomerate).
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
3 Valve
Table 2 Properties of maltodextrin powders used (data Roquette, Fr).
DE Glucose (%w/w) Maltose (%w/w) Oligo and polysaccharides (%w/w) Tg dry powder (°C)
Maltodextrin DE12
Maltodextrin DE21
11–14 1 2 97 180
20–23 3 7 90 140
Bag filter
T
200 μm. Sorption isotherms and evolution of glass transition temperature with water activity (aw) are given for the two powders in Fig. 2. 2.2. Equipment and operating conditions
Air
Bi-fluid nozzle Particle fluidised bed
Liquid
17 cm
Sampling tubes
Tb
8 cm
Agglomeration trials were performed in a pilot scale batch fluidised bed granulator (UniGlatt, Glatt, Ge) (Fig. 3). Water (20 °C) was top sprayed using a bi-fluid nozzle (type 970, Schlick, Fr) with a relative air pressure of 1 bar. For all the trials, the initial particle load was 350 g. It was fluidised using a constant fluidisation air flow rate of 80 kg h −1. Before entering the chamber, ambient air was heated up to a set-point temperature Tc depending on the trial (Table 3). Its temperature was measured during the trial 1 cm under the grid (Tinlet), 1 cm above the grid (Tbottom) and in the particle bed, 6 cm above the grid (Tb). Tb was considered as the bed average temperature since, as reported by Jimenez et al. [16], it represents a good indicator of the drying conditions in the fluidised bed and can be used as a control process parameter. Prior to water spraying (t = 0 min) the powder was pre-heated until Tb was
(a)
Tbottom Tinlet
Grid Fluidisation air
Tc Electrical heater
Fig. 3. Batch fluidised bed granulator (UniGlatt).
stable. Then water was sprayed with a constant feed rate QL set to a value depending on the trial (Table 3). During the spraying stage, powder samples (2–3 g) were taken within the particle bed after 4, 8 and 12 min, using a perforated tube
100.00
X(g water/100 g dry solids)
90.00 80.00 70.00 60.00
MD21
50.00 40.00 30.00
MD12
20.00 10.00 0.00 0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
aw
Tg (oC)
(b) 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0 0.00
X sorption isotherm
MD12
MD 21 MD 12 aw
MD21
0.10
X = f(aw) from sorption isotherms at 23°C
0.20
0.30
0.40
MD 12
5.31
6.22
7.35
4.45
5.86
7.39
0.50
aw
8.89
0.60
0.70
0.80
0.90
11.17
14.95
22.48
X
12.07
16.52
25.27
MD 21
9.33
1.00
Fig. 2. Sorption isotherms at 23 °C (a) and evolution of glass transition temperature Tg as a function of water activity and water content (b) for maltodextrin DE12 and DE21 [18].
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
4
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
Table 3 Experimental conditions of agglomeration trials (initial particle load, 350 g; Initial particle median diameter, ~200 μm; fluidisation air flow rate, 80 kg h −1; atomisation air relative pressure, 1 bar). Trial
Maltodextrin DE
QL (ml min−1)
Tc (°C)
Tinlet (°C)
Tbottom (°C)
Tb (°C)
Sprayed water mass (g)
A.1 A.2 A.3 A.4 A.5 B.1 B.2 B.3 B.4 B.5
12
3.3 1.7 3.2 4.0 4.9 3.2 3.4 3.7 3.8 4.9
60 70 70 70 70 60 70 70 70 70
60 66 69 66 68 60 68 69 69 67
44 52 56 49 46 46 55 52 53 49
45 53 54 50 49 46 53 52 52 49
55 30 50 65 77 55 63 60 61 77
21
inserted at 8 or 17 cm above the grid depending on the particle bed height. For both maltodextrins, at these sampling times (inferior to 15 min) and in the studied conditions, similar median mass diameter d50 and size distribution were obtained for the two sampling positions (Fig. 4). It can therefore be assumed that there is no size segregation of the particles between these two positions and samples taken at any of them during the different agglomeration trials can be compared.
2.3. Particle and liquid droplet characterization Particle size distribution of the initial powder and of the samples was measured by manual sieving (1–2 g, 2 min) ensuring a gentle shaking, using a series of 15 sieves (diameter 5 cm) with a mesh between 100 and 2500 μm according to a 21/3 geometric progression. Water content of the initial powder and samples was measured by oven drying (103 °C, 24 h). For observation under optical microscope (BX51, Olympus, Fr), particles were immersed in silicon oil (47v20, Rhodorsil, Fr). The size of the sprayed water droplets was measured by laser diffraction (Spraytec, Malvern, Fr) placing the nozzle outside the chamber. The Sauter diameter d3,2 of the sprayed water droplets increased from 23 to 27 μm when QL was increased from 1.7 to 4.9 ml min −1.
3. Results and discussion Maltodextrin DE12 and DE21 particles were agglomerated in similar conditions (initial median diameter, ~ 200 μm; particle load, 350 g; fluidisation air flow rate, 80 kg h −1; atomisation air relative pressure, 1 bar) changing either the sprayed water flow rate QL or the inlet air set-point temperature Tc in order to change temperature and water constraints in the particle bed while keeping constant mechanical constraints.
3.1. Air temperature in the fluidised bed In all cases, the inlet air temperature Tinlet, measured just below the grid, was equal or less than 4 °C inferior to Tc allowing to consider that, inlet air temperature was similar for trials A.2 to A.5 and B.2 to B.5 (e.g. 67.5 °C ± 1.5 °C since between 66 and 69 °C), and for trials A.1 and B.1 (e.g. 60 °C) (Table 3). The observed small differences were due to heat losses between the heater and the grid that depend on the hot air and ambient air temperature: the difference was bigger when ambient air temperature was lower (e.g. 16 °C for trials A.2 and A.4 instead of ~20 °C) or when hot air temperature was higher. Due to intense mixing in the fluidised bed, Tbottom, the air temperature measured just above the grid, was inferior to Tinlet and corresponded to the bed temperature Tb (± 2 °C). Air temperature in the particle bed decreased due to the spraying of water (at 20 °C), to water evaporation (drying), and to heat losses through the walls of the chamber. For the same inlet air temperature, Tb and Tbottom decreased when the sprayed liquid feed rate was increased. The evolution of Tb with QL was similar for the two maltodextrins (Fig. 5) with Tb decreasing from about 53 °C for QL = 1.7 ml min −1 down to 49 °C for QL = 4.9 ml min −1 for Tinlet ≈ 67.5 °C. A higher quantity of cold water (20 °C) arriving into the agglomeration chamber increases the water concentration differential between the fluidisation air and the solid particles, so drying also increases and the temperature within the fluidised bed decreases.
3.2. Agglomerate growth rate In order to follow agglomerate growth during the atomisation step, samples were regularly taken in the isothermal zone of the particle bed, where Tb was measured. Dried, non sticky particles/agglomerates were collected and their size distribution was measured by manual
Fig. 4. Comparison of median mass diameter d50 of samples taken at 8 and 17 cm above the grid during maltodextrin DE12 (a) and DE21 (b) agglomeration trials.
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
Fig. 5. Bed temperature Tb as a function of the sprayed water feed rate QL during maltodextrin DE12 and DE21 agglomeration trials with Tc = 70 °C.
sieving. For the two maltodextrin powders, in the studied conditions, the median diameter d50 of the taken samples increased with time (Fig. 6). For maltodextrin DE12, when the sprayed water feed rate QL was higher, particles grew into bigger agglomerates (Fig. 6a): the median diameter d50 of agglomerates obtained after 12 min of spraying was about 260 μm for a QL of 1.7 ml min −1 and 720 μm for a QL of 4.9 ml min −1. This corresponded to an increase of the particle growth rate d(d50)/dt from 6.9 to 39 μm min −1 when QL was increased from 1.7 to 4.9 ml min −1 (Fig. 7). In similar conditions, the growth rate of maltodextrin DE21 particles was less affected by QL. No significant difference in the evolution of the median diameter d50 of agglomerates with time was observed when increasing the water feed rate from 3.4 to 4.9 ml min −1 (Fig. 6b), and only a small increase of d(d50)/dt from 24.4 to 25.2 μm min −1 was observed (Fig. 7). In any case, for the same water feed rate QL, the size increase and growth rate were lower for maltodextrin DE21 compared to maltodextrin DE12. A higher water feed rate corresponds to a higher number of slightly larger water droplets sprayed with therefore more water dispersed on the particle surface into more numerous and larger moist zones available for bridges formation. This is expected to lead to the formation of more liquid bridges between the particles and to allow agglomerates to grow to larger sizes within a shorter period of time as it was observed for non soluble particles agglomerated with a binder solution [17]. However, for soluble maltodextrin particles, able to undergo glass transition, the particle surface stickiness and viscosity also take part to the bridge formation. They depend on the maltodextrin DE and they are modified when the water content and temperature are modified:
(a)
Fig. 7. Evolution of the agglomerate growth rate d(d50)/dt with the sprayed water feed rate QL for maltodextrin DE12 and DE21 (Tc = 70 °C).
maltodextrin solutions are less viscous as their DE increases and the viscosity decrease with the water content increase is smaller [19,20]; for the same water activity, maltodextrins with higher DE undergo glass transition at a lower temperature (Fig. 2). Therefore, when the water content is increased, the viscosity in the moist zones decreases causing an increase of the Stokes number that may lead to no agglomeration when μl is too low (St N St*: viscous dissipation of the collision kinetic energy not sufficient to avoid rebound when particles collide). For maltodextrin DE12, the viscosity is significantly higher than for maltodextrin DE21 whatever the water content and, in the conditions studied, the decrease may be not significant enough to reach St*. Therefore, when increasing QL, agglomerate growth rate still increases due to the more numerous and larger moist zones at the particle surface and to the decrease of the glass transition temperature with the water content that allows particle to still be sticky despite the lower temperature observed in the bed. For maltodextrin DE21 particles, the different behaviour may be due to the lower viscosity values causing not sufficient viscous dissipation of energy (St N St*) and inefficient collisions. In this case, no increase of the growth rate was observed when QL was increased. Whatever the tested conditions, the water content of the collected samples was inferior to 7 g water/100 g dry solids and air temperature Tb in the particle bed was inferior to 54 °C. This corresponded, for the two maltodextrins, to non sticky particles dried below the glass transition temperature Tg (Fig. 2). Particle stickiness allowing the formation of viscous bridges therefore arises when particles are still in the wetting zone, below the nozzle, where temperatures are inferior to Tb and
(b)
800
4.9 ml/min
Maltodextrine DE 12
800
Maltodextrin DE 21 4.9 ml·min-
4.0 ml/min 600
3.8 ml·min -1
600
3.2 ml/min 400
d50 (µm)
d50 (µm)
5
3.7 ml·min -1 3.4 ml·min -1
400
1.7 ml/min 200
200
0 0
5
10
time (min)
15
0 0
5
10
15
time (min)
Fig. 6. Evolution of the median diameter d50 with time for different sprayed water feed rate QL during maltodextrin DE12 (a) and maltodextrin DE21 (b) agglomeration trials with Tc = 70 °C.
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
6
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
humidity is high. According to the curves on Fig. 2, in the temperature range of the fluidized bed, maltodextrin DE21 with lower Tg values than maltodextrin DE12, should be sticky for a wider range of water content before drying below Tg, and therefore, should show a faster and more important agglomeration than maltodextrin DE12 particles. However, both maltodextrins have different drying kinetics with probably a faster drying for the less viscous maltodextrin DE21. In order to evaluate the influence of the fluidisation air temperature on the agglomeration kinetics, trials were performed with the same water feed rate (e.g. ~ 3.3 ml min −1) at Tc = 60 °C (A.1 and B.1) and Tc = 70 °C (A.3 and B.2) with both maltodextrin powders. A lower fluidisation air temperature corresponded to a lower bed temperature, with Tb decreasing from about 54 °C, for Tc = 70 °C, to about 46 °C, for Tc = 60 °C. When comparing the size increase with time for the two maltodextrins, different effects of the air temperature were obtained. Whilst for maltodextrin DE12 (Fig. 8a), the size increase was lower when the air temperature was decreased, for maltodextrin DE21 (Fig. 8b) it was higher. A lower air temperature corresponds to a lower particle temperature and to a smaller drying capacity of air with probably a higher humidity in the particle bed. Under these conditions, the water content of the moist zones on the surface of the particles is probably higher causing a decrease of the particle surface viscosity. For maltodextrin DE12 particles, that may lead to Stokes number values above the critical value with a moist layer at the particle surface not viscous enough to allow bridge formation. But, at the same time, when Tc is decreased, the drying rate is lower and particles remain wet longer with a progressive increase of the surface viscosity as drying proceeds. This may explain the slight positive effect of the decrease of Tc on the size increase for maltodextrin DE21 particles that may also be the result of lower glass transition temperatures allowing particles to remain sticky for a longer time period. These results confirm the importance of drying in wet fluidized bed agglomeration [17]. The different behaviours observed for maltodextrin DE12 and DE21 particles are the result of different competitive phenomena linked to both operating parameters and particle properties occurring simultaneously with different time scale and intensity. This may lead to different growth mechanisms and to agglomerates with different properties. 3.3. Agglomerate growth mechanisms The association of initial particles into bigger agglomerates can occur either by the progressive incorporation of initial particles into larger and larger agglomerates (particle-cluster agglomeration) or by first the formation of initial clusters of initial particles and then the association of
0.5 0.25 0
Final particles t = 12 min
0.25 0
t = 8 min xm 0.25 0
t = 4 min 0.25 0
t = 0 min 0.25 0 100
1000
di (µ µm) Fig. 9. Typical evolution of the particle size distribution during agglomeration trials (example: Trial A.3—maltodextrin DE12; Tc = 70 °C; QL = 3.2 ml min −1).
these clusters into larger agglomerates (cluster-cluster agglomeration). Depending on the agglomeration mechanism at play, the obtained agglomerates will have different structures, with probably more compact agglomerates obtained in the first case. In order to determine which of these two mechanisms occurred during agglomerates growth, the evolution of the particle size distribution during spraying was followed. Fig. 9 shows the typical evolution of the particle size distribution during agglomeration for the different trials performed. As soon as water was sprayed as small droplets, it started wetting the particle surface inducing stickiness and as a result, smaller initial particles (b200 μm) started to agglomerate. The first structures to be formed were small agglomerates of a few particles stuck together. As a consequence, the first peak decreased, giving rise to a second peak at about 200 μm (t = 4 min). This second peak then started to decrease and a third one appeared corresponding to even bigger agglomerates (N350 μm) formed by the progressive association of initial agglomerates (t N 4 min). This third peak progressively increased and moved to larger and larger diameter while agglomerates grew. This progressive growth with formation of an intermediate population in the first minutes of agglomeration has already been reported for fluidised bed agglomeration of milk powder [21]. Depending on the conditions, the evolution of the different peaks was more or less rapid, but it was similar.
Fig. 8. Size increase as a function of time for different fluidisation air temperature Tc during maltodextrin DE12 (a) and DE21 (b) agglomeration trials (QL = 3.3 ml min−1).
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
(a)
7
(b)
0,40
1,0
Initial particles
<315 µm
Intermediate population
0,30
315 - 630 µm
0,8
Agglomerates
>630 µm
xm
xm
0,6 0,20
0,4 0,10
0,2
0,00
0,0 0
3
6
9
12
15
0
3
6
time (min)
9
12
15
time (min)
Fig. 10. Evolution of the mass fraction Xm of particles in each size class (a) and for initial particles, intermediate population and agglomerates (b) during agglomeration trial A3 (maltodextrin DE12; Tc = 70 °C; QL = 3.2 ml min−1).
When representing the evolution of the mass fraction Xm of maltodextrin DE12 particles in each size class over time, different curve shapes were observed corresponding to the different populations identified (Fig. 10a). The first one (black dashed lines) corresponded to the initial particles in the smaller size classes (b315 μm). Their mass fraction decreased rapidly during the first minutes of water spraying. This corresponds to the association of initial particles to form intermediate structures corresponding to the second peak (between 315 and 630 μm). For this second group (plain grey lines), the mass fraction started increasing as soon as water was sprayed and slightly decreased or remained constant after several minutes due to the association of intermediates structures into bigger agglomerates (N630 μm) corresponding to the third group (light grey dashed lines). At the beginning of the trial, no particles of this population were present. They appeared after several minutes of water spraying and their mass fraction started to increase. When grouping the different curves obtained for each population, the three typical evolutions can be observed (Fig. 10b): initial particles (b 315 μm), intermediate population (315–630 μm) and agglomerates (N630 μm). Similar curves were obtained for both maltodextrin powders and whatever the sprayed water feed rate QL. The only difference observed between maltodextrin DE12 and DE21 was the width of intermediate population that extended up to 800 μm for maltodextrin DE21. Formation of intermediate particles by the association of small initial particles was the main growth mechanism in this case, and only few larger agglomerates resulting of the association of these intermediate particles appeared since bridges were not sufficient in this case to overcome mechanical constraints.
For maltodextrin DE12, the modification of the water feed rate did not change the general shape of the three curves (Fig. 11). The only difference concerned the position of the change in the slope of the different curves corresponding to the change of the main growth mechanisms (particlecluster or cluster-cluster). When using a lower QL of 1.7 ml min −1, initial particles disappeared more slowly than with a higher QL, and at the end of the trial there was still an important concentration of these particles (about 75% of the whole mass compared to less than 20% for a QL of 4.9 ml min −1). The change in the slope appeared at the same time for the three curves, but it was delayed when the liquid feed rate was decreased (after about 12 min when using a QL of 1.7 ml min −1 instead of about 4 min for a QL of 4.9 ml min −1). For both maltodextrin particles, with a lower water feed rate, a limited size increase was observed since only a small part of the initial particles agglomerated into intermediate agglomerates and only few bigger agglomerates were formed. Increasing the water feed rate therefore not only allowed a faster growth, but also allowed obtaining an agglomerated powder with a more homogeneous particle size distribution and no more initial particles. Optical microscope observation of particles from the three size classes (Fig. 12) confirmed a different structure for intermediate population and larger agglomerates. Intermediate population consisted in compact associations of a few initial particles linked by solid bridges. Final agglomerate population consisted of the intermediate agglomerates linked by solid bridges and forming a larger and more open structure within which intermediate particles could be recognized. Optical microscope observations did not show any difference in the structure of the
Initial particles
(a)
Initial particles
Intermediate population
1,0
(b)
Intermediate population 1,0
Agglomerates
0,8
0,6
0,6
Xm
Xm
0,8
Agglomerates
0,4
0,4
0,2
0,2
1.7 ml·min-1
0,0 0
3
6
9
time (min)
12
15
0,0 0
3
6
9
12
15
time (min)
Fig. 11. Evolution of the mass fraction Xm of the three populations for different water feed rates QL during maltodextrin DE12 (a) and DE21 (b) agglomeration trials (Tc = 70 °C).
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
8
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
Maltodextrin DE12
Maltodextrin DE21
Fig. 12. Optical microscope images of initial particles (a, d), intermediate particles (b, e) and agglomerates (c, f) from maltodextrin DE12 (a, b, c) and DE21 (d, e, f) agglomeration trials (Tc = 70 °C, QL = 4.9 ml min −1).
two populations when changing the maltodextrin powder or the agglomeration conditions. 4. Conclusion Maltodextrin DE12 and DE21 showed a different sensitivity to water and temperature constraints during fluidised bed agglomeration. Especially whilst for maltodextrin DE12, the size increase and growth rate increased significantly when the sprayed water feed rate was increased, almost no variation was observed for maltodextrin DE21. This was attributed to a faster drying of maltodextrin DE21, which also corresponded to lower glass transition temperatures and to less viscous solutions, less efficient for the formation of viscous bridges between particles. Whatever the conditions tested, agglomeration occurred in two stages with, first, the association of initial particles into intermediate agglomerates and second, the association of the intermediate structures into larger and more porous agglomerates. The change from the first growth mechanism to the second one was delayed when decreasing the sprayed water feed rate with only few large agglomerates formed in this case. This may also influence the agglomerate structure and the powder properties. Further to allow increasing the agglomeration growth rate in the case of maltodextrin DE12 agglomeration, the increase of the sprayed water feed rate also allowed obtaining agglomerated powders with a more homogeneous particle size distribution. Acknowledgments The authors would like to acknowledge the financial support of CONACYT (Mexico) (310189) and the French Government (734297G) for the PhD scholarships provided to Carlos Avilés-Avilés. References [1] M.E. Aulton, M. Banks, Fluidised bed granulation—factors influencing the quality of the product, Int. J. Pharm. Technol. Prod. Manuf. 4 (1981) 24–29. [2] G.I. Tardos, M.I. Khan, P.R. Mort, Critical parameters and limiting conditions in binder granulation of fine powders, Powder Technol. 94 (1997) 245–258.
[3] B.J. Ennis, G.T. Tardos, R. Pfeffer, A microlevel-based characterization of granulation phenomena, Powder Technol. 65 (1991) 257–272. [4] S. Palzer, Contrôle de l'agglomération de polymères de dextrose par l'application du concept de la transition vitreuse, Ind. Aliment. Agric. 122 (2005) 20–27. [5] S. Palzer, Agglomeration of pharmaceutical, detergent, chemical and food powders— similarities and differences of materials and processes, Powder Technol. 206 (2011) 2–17. [6] B. Bhandari, T. Howes, Relating the stickiness of food undergoing drying and dried products to their surface energetics, Dry. Technol. 23 (2005) 781–797. [7] C. Hennigs, T.K. Kockel, T.A.G. Langrish, New measurements of the sticky behavior of skim milk powder, Dry. Technol. 19 (2001) 471–484. [8] L. Ozmen, T.A.G. Langrish, Comparison of glass transition temperature and sticky point temperature for skim milk powder, Dry. Technol. 20 (2006) 37–41. [9] C. Turchiuli, T. Jimenez, E. Dumoulin, Identification of thermal zones and population balance modelling of fluidized bed spray granulation, Powder Technol. 208 (2011) 542–552. [10] S.J. Maronga, P. Wnukowski, Establishing temperature and humidity profiles in fluidized bed particulate coating, Powder Technol. 94 (1997) 181–185. [11] Y. Roos, M. Karel, Applying state diagrams to food processing and development, Food Technol. 45 (1991) 66–68. [12] F. Ronsse, J. Depelchin, J.G. Pieters, Particle surface moisture content estimation using population balance modelling in fluidised bed agglomeration, J. Food Eng. 109 (2012) 347–357. [13] K. Terrazas-Velarde, M. Peglow, E. Tsotsas, Kinetics of fluidized bed spray agglomeration for compact and porous particles, Chem. Eng. Sci. 66 (2011) 1866–1878. [14] A. Seo, P. Holm, T. Schaefer, Effects of droplet size and type of binder on the agglomerate growth mechanisms by melt agglomeration in a fluidised bed, Eur. J. Pharm. Sci. 16 (2002) 95–105. [15] T.A.M. Hirata, G.C. Dacanal, F.C. Menegalli, Effect of operational conditions on the properties of pectin powder agglomerated in pulsed fluid bed, Powder Technol. 245 (2013) 174–181. [16] T. Jimenez, C. Turchiuli, E. Dumoulin, Particles agglomeration in a conical fluidized bed in relation with air temperature profiles, Chem. Eng. Sci. 61 (2006) 5954–5961. [17] K. Terrazas-Velarde, M. Peglow, E. Tsotsas, Investigation of the kinetics of fluidized bed spray agglomeration based on stochastic methods, AIChE J 57 (11) (2011) 3012–3026. [18] A. Gianfrancesco, Spray drying engineering: particle stickiness in relation with agglomeration, (PhD thesis) AgroParisTech, France, 2009. (192 pp.). [19] P. Dokic, J. Jakovljevic, L. Dokic-Baucal, Molecular characteristics of maltodextrins and rheological behaviour of diluted and concentrated solutions, Colloids Surf., A 141 (1998) 435–440. [20] F. Avaltroni, P.E. Bouquerand, V. Normand, Maltodextrin molecular weight distribution influence on the glass transition temperature and viscosity in aqueous solutions, Carbohydr. Polym. 58 (2004) 323–334. [21] C. Turchiuli, R. Smail, E. Dumoulin, Fluidized bed agglomeration of skim milk powder: analysis of sampling for the follow-up of agglomerate growth, Powder Technol. 238 (2012) 161–168.
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
Contents lists available at ScienceDirect
Powder Technology journal homepage: www.elsevier.com/locate/powtec
Fluidised bed agglomeration of particles with different glass transition temperatures Carlos Avilés-Avilés, Elisabeth Dumoulin, Christelle Turchiuli ⁎ AgroParisTech, UMR 1145 Ingénierie Procédés Aliments, 1 Avenue des Olympiades, F-91300, Massy, France INRA, UMR1145 Ingénierie Procédés Aliments, F-91300, Massy, France CNAM, UMR1145 Ingénierie Procédés Aliments, F-91300, Massy, France
a r t i c l e
i n f o
Available online xxxx Keywords: Agglomeration Fluidised bed Growth mechanisms Stickiness
a b s t r a c t Fluidised bed agglomeration of particles consists in spraying liquid drops (water or binder solution) on the surface of particles, fluidised by hot air, in order to create sticky regions to allow formation of agglomerates when the sticky particles collide. Many parameters influence agglomerate growth, especially those controlling the particle circulation, and the water and temperature conditions within the bed that determine drying and particle stickiness linked to the glass transition of amorphous components and to the viscosity of moist zones at the particle surface. Maltodextrin DE12 and DE21 particles with different glass transition temperature domains were agglomerated in a batch bench scale fluidised bed, under constant mechanical constraints, changing the sprayed water feed rate and the fluidisation air temperature in order to investigate the influence of particle stickiness on agglomerate growth kinetics and mechanism. The two powders showed a different sensitivity to the water and temperature constraints applied. Whilst for maltodextrin DE12, the size and growth rate increased significantly with the sprayed water flow rate, only a small variation was observed for maltodextrin DE21. In both cases, agglomeration occurred in two stages: firstly, the association of initial particles and secondly the agglomeration of intermediate agglomerates into larger and more porous structures. The change from one growth mechanism to the other depended on the conditions, and influenced the size distribution and structure of agglomerates. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Agglomeration is widely used in the food and pharmaceutical industry where a great amount of the raw materials, intermediate or final products are in powder form. Agglomerates consist in the assembly of fine particles (~ 200 μm) linked by solid bridges to create large porous structures (~800 μm). By modifying the physical properties of the powder particles (size, shape, porosity, density), agglomeration allows modifying the powder end-use properties (flowability, wettability, instantaneity). Fluidised bed agglomeration consists of spraying a liquid (water or binder aqueous solution) on a bed of solid particles fluidised by hot air. Sprayed liquid droplets collide with the individualised and moving particles, allowing to create sticky zones on their surfaces, either due to the local wetting of the particle surface (water soluble particles) or
⁎ Corresponding author at: AgroParisTech, UMR 1145 Ingénierie Procédés Aliments, 1 Avenue des Olympiades, 91300 Massy, France. Tel.: +33 1 69 93 50 71; fax: +33 1 69 93 50 05. E-mail addresses: [email protected] (C. Avilés-Avilés), [email protected] (E. Dumoulin), [email protected] (C. Turchiuli).
to the deposit of binder solution on the particle surface (non soluble particles). Agglomerate formation takes place when the sticky moving particles collide with others (sticky or not) and form liquid or viscous bridges that are dried and consolidated by the fluidisation hot air. The repetition of these steps (spraying, wetting, collision, adhesion and drying) allows progressive agglomerate growth. Fluidised bed agglomeration is therefore a complex process resulting in different phenomena. Agglomerate growth, structure and properties depend on many parameters linked to the equipment, to the operating conditions, and to the particle and sprayed liquid properties (Table 1) [1]. These parameters determine the mechanical, temperature and water constraints applied inside the particle fluidised bed. Mechanical constraints are mainly due to the agitation of particles by the fluidising air allowing both contacts between particles by friction or collision, and the distribution of the sprayed liquid within the bed. For a given device and geometry, mechanical constraints mainly depend on the fluidising air flow rate; on the particle load in the chamber; and on the particle size, size distribution and density influencing fluidisation. Temperature constraints are linked to the temperature and flow rate of both the hot fluidising air and sprayed liquid. They determine the temperature and the drying conditions within the particle bed and therefore also influence the water constraints. Controlled agglomerate growth occurs
http://dx.doi.org/10.1016/j.powtec.2014.03.026 0032-5910/© 2014 Elsevier B.V. All rights reserved.
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
2
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
Table 1 Main parameters in fluidised bed agglomeration and their contribution to the mechanical, thermal and water constraints. Geometry of granulator
Operating parameters
Product parameters
Chamber * Size * Shape Air distributor Nozzle * Type * Position
Particle load Fluidising air * Flow rate * Temperature Spraying * Liquid feed rate * Liquid drop size
Particles * Composition * Size/size distribution, density * Surface Sprayed liquid * Composition/concentration * Temperature
when the conditions allow bridges to create between colliding particles without over wetting and collapse of the bed. This requires controlling at the same time agitation, wetting and drying of the particle bed (i.e. intensity and duration of the constraints applied). The wetting of the particle surface by the sprayed liquid allows the formation of liquid or viscous bridges between colliding particles thanks to the viscous dissipation of the collision kinetic energy. Actually, a collision between two particles is effective only if the Stokes number St is below a certain critical value St* [2,3]. Above St*, particle rebound is observed and the collision does not allow agglomerate growth: St ¼ 8 ρp r p uc =ð9 μl Þ and St ¼ ð1 þ 1=cÞ ln ðe=ru Þ Where ρp is the particle density (kg m−3), rp is the mean particle radius (m), uc is the relative collision velocity (m s −1), μl is the viscosity of the viscous film (Pa s), c is the collision restitution coefficient (-), e is the viscous film thickness (m), and ru is the particle surface rugosity (-). However, the formed liquid or viscous bridges must also be able to give rise to solid bridges by the evaporation of the solvent (water) during hot-air drying. For the agglomeration of water insoluble particles the spraying of an aqueous solution containing a binder (e.g. amorphous component) is needed to form a viscous adhering film on the particle surface and render it “sticky”. When particles are soluble in water and contain amorphous components, they can be agglomerated by spraying pure water. In this case, water can cause both some partial dissolution and plastification, decreasing the glass transition temperature of the amorphous components on the particle surface [4]. In the specific
temperature conditions of the fluidised bed, these plasticized amorphous components undergo glass transition causing a decrease of their viscosity and the appearance of stickiness when they reach the rubbery state. This corresponds to a critical viscosity range between 108 and 1012 Pa s [5,6]. For a given water content, temperatures for sticky conditions range between the glass transition temperature (Tg), and the “sticky temperature” (Ts); from 20 to 30 °C above Tg for the main food carbohydrates (maltodextrin, lactose, fructose, sucrose, etc.) [7,8]. Tg and Ts depend on the composition of the product considered. During fluidised bed agglomeration, the particle surface water content and temperature permanently vary along the bed and with time due to simultaneous wetting and drying of the particles (1–4, Fig. 1) and to the agitation causing their circulation through the thermal zones within the bed [9,10]. The presence of sticky particles able to agglomerate when colliding therefore depends on their position, on temperature and water conditions within the bed and on the particle surface composition [4,8,11]. Different studies were performed on the effect of parameters such as fluidising air temperature and flow rate; nature, concentration and feed rate of the sprayed solution on agglomerates growth [12–17]. Most of them were performed with non soluble particles and only few investigated the behaviour of particles undergoing glass transition. The aim of this work was to study the agglomerate growth during fluidised bed agglomeration of particles with different glass transition temperature evolution with the water content. Two model powders e.g. maltodextrins with different dextrose equivalent (DE12 and DE21), were agglomerated spraying water. Similar mechanical constraints (same air flow rate, particle load and initial size distribution) were applied, but temperature and water constraints were changed by modifying the sprayed water feed rate and the fluidising air temperature. Agglomerate growth was studied following the evolution of both the median diameter and particle size distribution in order to consider the influence of the studied parameters on both the growth kinetics and growth mechanisms. 2. Materials and methods 2.1. Powders Maltodextrin DE12 and DE21 (Glucidex, Roquette, Fr) were used for agglomeration trials (Table 2). Prior to each trial, powders were sieved (Analysette 3 Spartan, Fritsch, Ge—80 g, 12 min, amplitude of vibration 2.5 mm) to obtain a narrow size distribution. The fraction between 100 and 315 μm was kept in order to have a median diameter d50 of about
Fig. 1. Glass transition and sticky temperatures (Tg and Ts) as a function of water content X and water activity aw for maltodextrin DE12 and evolution of temperature and water content of moist zones on the particle surface during the different steps of fluidized bed agglomeration (1. initial particle; 2. wetting; 3. drying/collision; 4. solid bridge/dry agglomerate).
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
3 Valve
Table 2 Properties of maltodextrin powders used (data Roquette, Fr).
DE Glucose (%w/w) Maltose (%w/w) Oligo and polysaccharides (%w/w) Tg dry powder (°C)
Maltodextrin DE12
Maltodextrin DE21
11–14 1 2 97 180
20–23 3 7 90 140
Bag filter
T
200 μm. Sorption isotherms and evolution of glass transition temperature with water activity (aw) are given for the two powders in Fig. 2. 2.2. Equipment and operating conditions
Air
Bi-fluid nozzle Particle fluidised bed
Liquid
17 cm
Sampling tubes
Tb
8 cm
Agglomeration trials were performed in a pilot scale batch fluidised bed granulator (UniGlatt, Glatt, Ge) (Fig. 3). Water (20 °C) was top sprayed using a bi-fluid nozzle (type 970, Schlick, Fr) with a relative air pressure of 1 bar. For all the trials, the initial particle load was 350 g. It was fluidised using a constant fluidisation air flow rate of 80 kg h −1. Before entering the chamber, ambient air was heated up to a set-point temperature Tc depending on the trial (Table 3). Its temperature was measured during the trial 1 cm under the grid (Tinlet), 1 cm above the grid (Tbottom) and in the particle bed, 6 cm above the grid (Tb). Tb was considered as the bed average temperature since, as reported by Jimenez et al. [16], it represents a good indicator of the drying conditions in the fluidised bed and can be used as a control process parameter. Prior to water spraying (t = 0 min) the powder was pre-heated until Tb was
(a)
Tbottom Tinlet
Grid Fluidisation air
Tc Electrical heater
Fig. 3. Batch fluidised bed granulator (UniGlatt).
stable. Then water was sprayed with a constant feed rate QL set to a value depending on the trial (Table 3). During the spraying stage, powder samples (2–3 g) were taken within the particle bed after 4, 8 and 12 min, using a perforated tube
100.00
X(g water/100 g dry solids)
90.00 80.00 70.00 60.00
MD21
50.00 40.00 30.00
MD12
20.00 10.00 0.00 0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
aw
Tg (oC)
(b) 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0 0.00
X sorption isotherm
MD12
MD 21 MD 12 aw
MD21
0.10
X = f(aw) from sorption isotherms at 23°C
0.20
0.30
0.40
MD 12
5.31
6.22
7.35
4.45
5.86
7.39
0.50
aw
8.89
0.60
0.70
0.80
0.90
11.17
14.95
22.48
X
12.07
16.52
25.27
MD 21
9.33
1.00
Fig. 2. Sorption isotherms at 23 °C (a) and evolution of glass transition temperature Tg as a function of water activity and water content (b) for maltodextrin DE12 and DE21 [18].
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
4
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
Table 3 Experimental conditions of agglomeration trials (initial particle load, 350 g; Initial particle median diameter, ~200 μm; fluidisation air flow rate, 80 kg h −1; atomisation air relative pressure, 1 bar). Trial
Maltodextrin DE
QL (ml min−1)
Tc (°C)
Tinlet (°C)
Tbottom (°C)
Tb (°C)
Sprayed water mass (g)
A.1 A.2 A.3 A.4 A.5 B.1 B.2 B.3 B.4 B.5
12
3.3 1.7 3.2 4.0 4.9 3.2 3.4 3.7 3.8 4.9
60 70 70 70 70 60 70 70 70 70
60 66 69 66 68 60 68 69 69 67
44 52 56 49 46 46 55 52 53 49
45 53 54 50 49 46 53 52 52 49
55 30 50 65 77 55 63 60 61 77
21
inserted at 8 or 17 cm above the grid depending on the particle bed height. For both maltodextrins, at these sampling times (inferior to 15 min) and in the studied conditions, similar median mass diameter d50 and size distribution were obtained for the two sampling positions (Fig. 4). It can therefore be assumed that there is no size segregation of the particles between these two positions and samples taken at any of them during the different agglomeration trials can be compared.
2.3. Particle and liquid droplet characterization Particle size distribution of the initial powder and of the samples was measured by manual sieving (1–2 g, 2 min) ensuring a gentle shaking, using a series of 15 sieves (diameter 5 cm) with a mesh between 100 and 2500 μm according to a 21/3 geometric progression. Water content of the initial powder and samples was measured by oven drying (103 °C, 24 h). For observation under optical microscope (BX51, Olympus, Fr), particles were immersed in silicon oil (47v20, Rhodorsil, Fr). The size of the sprayed water droplets was measured by laser diffraction (Spraytec, Malvern, Fr) placing the nozzle outside the chamber. The Sauter diameter d3,2 of the sprayed water droplets increased from 23 to 27 μm when QL was increased from 1.7 to 4.9 ml min −1.
3. Results and discussion Maltodextrin DE12 and DE21 particles were agglomerated in similar conditions (initial median diameter, ~ 200 μm; particle load, 350 g; fluidisation air flow rate, 80 kg h −1; atomisation air relative pressure, 1 bar) changing either the sprayed water flow rate QL or the inlet air set-point temperature Tc in order to change temperature and water constraints in the particle bed while keeping constant mechanical constraints.
3.1. Air temperature in the fluidised bed In all cases, the inlet air temperature Tinlet, measured just below the grid, was equal or less than 4 °C inferior to Tc allowing to consider that, inlet air temperature was similar for trials A.2 to A.5 and B.2 to B.5 (e.g. 67.5 °C ± 1.5 °C since between 66 and 69 °C), and for trials A.1 and B.1 (e.g. 60 °C) (Table 3). The observed small differences were due to heat losses between the heater and the grid that depend on the hot air and ambient air temperature: the difference was bigger when ambient air temperature was lower (e.g. 16 °C for trials A.2 and A.4 instead of ~20 °C) or when hot air temperature was higher. Due to intense mixing in the fluidised bed, Tbottom, the air temperature measured just above the grid, was inferior to Tinlet and corresponded to the bed temperature Tb (± 2 °C). Air temperature in the particle bed decreased due to the spraying of water (at 20 °C), to water evaporation (drying), and to heat losses through the walls of the chamber. For the same inlet air temperature, Tb and Tbottom decreased when the sprayed liquid feed rate was increased. The evolution of Tb with QL was similar for the two maltodextrins (Fig. 5) with Tb decreasing from about 53 °C for QL = 1.7 ml min −1 down to 49 °C for QL = 4.9 ml min −1 for Tinlet ≈ 67.5 °C. A higher quantity of cold water (20 °C) arriving into the agglomeration chamber increases the water concentration differential between the fluidisation air and the solid particles, so drying also increases and the temperature within the fluidised bed decreases.
3.2. Agglomerate growth rate In order to follow agglomerate growth during the atomisation step, samples were regularly taken in the isothermal zone of the particle bed, where Tb was measured. Dried, non sticky particles/agglomerates were collected and their size distribution was measured by manual
Fig. 4. Comparison of median mass diameter d50 of samples taken at 8 and 17 cm above the grid during maltodextrin DE12 (a) and DE21 (b) agglomeration trials.
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
Fig. 5. Bed temperature Tb as a function of the sprayed water feed rate QL during maltodextrin DE12 and DE21 agglomeration trials with Tc = 70 °C.
sieving. For the two maltodextrin powders, in the studied conditions, the median diameter d50 of the taken samples increased with time (Fig. 6). For maltodextrin DE12, when the sprayed water feed rate QL was higher, particles grew into bigger agglomerates (Fig. 6a): the median diameter d50 of agglomerates obtained after 12 min of spraying was about 260 μm for a QL of 1.7 ml min −1 and 720 μm for a QL of 4.9 ml min −1. This corresponded to an increase of the particle growth rate d(d50)/dt from 6.9 to 39 μm min −1 when QL was increased from 1.7 to 4.9 ml min −1 (Fig. 7). In similar conditions, the growth rate of maltodextrin DE21 particles was less affected by QL. No significant difference in the evolution of the median diameter d50 of agglomerates with time was observed when increasing the water feed rate from 3.4 to 4.9 ml min −1 (Fig. 6b), and only a small increase of d(d50)/dt from 24.4 to 25.2 μm min −1 was observed (Fig. 7). In any case, for the same water feed rate QL, the size increase and growth rate were lower for maltodextrin DE21 compared to maltodextrin DE12. A higher water feed rate corresponds to a higher number of slightly larger water droplets sprayed with therefore more water dispersed on the particle surface into more numerous and larger moist zones available for bridges formation. This is expected to lead to the formation of more liquid bridges between the particles and to allow agglomerates to grow to larger sizes within a shorter period of time as it was observed for non soluble particles agglomerated with a binder solution [17]. However, for soluble maltodextrin particles, able to undergo glass transition, the particle surface stickiness and viscosity also take part to the bridge formation. They depend on the maltodextrin DE and they are modified when the water content and temperature are modified:
(a)
Fig. 7. Evolution of the agglomerate growth rate d(d50)/dt with the sprayed water feed rate QL for maltodextrin DE12 and DE21 (Tc = 70 °C).
maltodextrin solutions are less viscous as their DE increases and the viscosity decrease with the water content increase is smaller [19,20]; for the same water activity, maltodextrins with higher DE undergo glass transition at a lower temperature (Fig. 2). Therefore, when the water content is increased, the viscosity in the moist zones decreases causing an increase of the Stokes number that may lead to no agglomeration when μl is too low (St N St*: viscous dissipation of the collision kinetic energy not sufficient to avoid rebound when particles collide). For maltodextrin DE12, the viscosity is significantly higher than for maltodextrin DE21 whatever the water content and, in the conditions studied, the decrease may be not significant enough to reach St*. Therefore, when increasing QL, agglomerate growth rate still increases due to the more numerous and larger moist zones at the particle surface and to the decrease of the glass transition temperature with the water content that allows particle to still be sticky despite the lower temperature observed in the bed. For maltodextrin DE21 particles, the different behaviour may be due to the lower viscosity values causing not sufficient viscous dissipation of energy (St N St*) and inefficient collisions. In this case, no increase of the growth rate was observed when QL was increased. Whatever the tested conditions, the water content of the collected samples was inferior to 7 g water/100 g dry solids and air temperature Tb in the particle bed was inferior to 54 °C. This corresponded, for the two maltodextrins, to non sticky particles dried below the glass transition temperature Tg (Fig. 2). Particle stickiness allowing the formation of viscous bridges therefore arises when particles are still in the wetting zone, below the nozzle, where temperatures are inferior to Tb and
(b)
800
4.9 ml/min
Maltodextrine DE 12
800
Maltodextrin DE 21 4.9 ml·min-
4.0 ml/min 600
3.8 ml·min -1
600
3.2 ml/min 400
d50 (µm)
d50 (µm)
5
3.7 ml·min -1 3.4 ml·min -1
400
1.7 ml/min 200
200
0 0
5
10
time (min)
15
0 0
5
10
15
time (min)
Fig. 6. Evolution of the median diameter d50 with time for different sprayed water feed rate QL during maltodextrin DE12 (a) and maltodextrin DE21 (b) agglomeration trials with Tc = 70 °C.
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
6
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
humidity is high. According to the curves on Fig. 2, in the temperature range of the fluidized bed, maltodextrin DE21 with lower Tg values than maltodextrin DE12, should be sticky for a wider range of water content before drying below Tg, and therefore, should show a faster and more important agglomeration than maltodextrin DE12 particles. However, both maltodextrins have different drying kinetics with probably a faster drying for the less viscous maltodextrin DE21. In order to evaluate the influence of the fluidisation air temperature on the agglomeration kinetics, trials were performed with the same water feed rate (e.g. ~ 3.3 ml min −1) at Tc = 60 °C (A.1 and B.1) and Tc = 70 °C (A.3 and B.2) with both maltodextrin powders. A lower fluidisation air temperature corresponded to a lower bed temperature, with Tb decreasing from about 54 °C, for Tc = 70 °C, to about 46 °C, for Tc = 60 °C. When comparing the size increase with time for the two maltodextrins, different effects of the air temperature were obtained. Whilst for maltodextrin DE12 (Fig. 8a), the size increase was lower when the air temperature was decreased, for maltodextrin DE21 (Fig. 8b) it was higher. A lower air temperature corresponds to a lower particle temperature and to a smaller drying capacity of air with probably a higher humidity in the particle bed. Under these conditions, the water content of the moist zones on the surface of the particles is probably higher causing a decrease of the particle surface viscosity. For maltodextrin DE12 particles, that may lead to Stokes number values above the critical value with a moist layer at the particle surface not viscous enough to allow bridge formation. But, at the same time, when Tc is decreased, the drying rate is lower and particles remain wet longer with a progressive increase of the surface viscosity as drying proceeds. This may explain the slight positive effect of the decrease of Tc on the size increase for maltodextrin DE21 particles that may also be the result of lower glass transition temperatures allowing particles to remain sticky for a longer time period. These results confirm the importance of drying in wet fluidized bed agglomeration [17]. The different behaviours observed for maltodextrin DE12 and DE21 particles are the result of different competitive phenomena linked to both operating parameters and particle properties occurring simultaneously with different time scale and intensity. This may lead to different growth mechanisms and to agglomerates with different properties. 3.3. Agglomerate growth mechanisms The association of initial particles into bigger agglomerates can occur either by the progressive incorporation of initial particles into larger and larger agglomerates (particle-cluster agglomeration) or by first the formation of initial clusters of initial particles and then the association of
0.5 0.25 0
Final particles t = 12 min
0.25 0
t = 8 min xm 0.25 0
t = 4 min 0.25 0
t = 0 min 0.25 0 100
1000
di (µ µm) Fig. 9. Typical evolution of the particle size distribution during agglomeration trials (example: Trial A.3—maltodextrin DE12; Tc = 70 °C; QL = 3.2 ml min −1).
these clusters into larger agglomerates (cluster-cluster agglomeration). Depending on the agglomeration mechanism at play, the obtained agglomerates will have different structures, with probably more compact agglomerates obtained in the first case. In order to determine which of these two mechanisms occurred during agglomerates growth, the evolution of the particle size distribution during spraying was followed. Fig. 9 shows the typical evolution of the particle size distribution during agglomeration for the different trials performed. As soon as water was sprayed as small droplets, it started wetting the particle surface inducing stickiness and as a result, smaller initial particles (b200 μm) started to agglomerate. The first structures to be formed were small agglomerates of a few particles stuck together. As a consequence, the first peak decreased, giving rise to a second peak at about 200 μm (t = 4 min). This second peak then started to decrease and a third one appeared corresponding to even bigger agglomerates (N350 μm) formed by the progressive association of initial agglomerates (t N 4 min). This third peak progressively increased and moved to larger and larger diameter while agglomerates grew. This progressive growth with formation of an intermediate population in the first minutes of agglomeration has already been reported for fluidised bed agglomeration of milk powder [21]. Depending on the conditions, the evolution of the different peaks was more or less rapid, but it was similar.
Fig. 8. Size increase as a function of time for different fluidisation air temperature Tc during maltodextrin DE12 (a) and DE21 (b) agglomeration trials (QL = 3.3 ml min−1).
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
(a)
7
(b)
0,40
1,0
Initial particles
<315 µm
Intermediate population
0,30
315 - 630 µm
0,8
Agglomerates
>630 µm
xm
xm
0,6 0,20
0,4 0,10
0,2
0,00
0,0 0
3
6
9
12
15
0
3
6
time (min)
9
12
15
time (min)
Fig. 10. Evolution of the mass fraction Xm of particles in each size class (a) and for initial particles, intermediate population and agglomerates (b) during agglomeration trial A3 (maltodextrin DE12; Tc = 70 °C; QL = 3.2 ml min−1).
When representing the evolution of the mass fraction Xm of maltodextrin DE12 particles in each size class over time, different curve shapes were observed corresponding to the different populations identified (Fig. 10a). The first one (black dashed lines) corresponded to the initial particles in the smaller size classes (b315 μm). Their mass fraction decreased rapidly during the first minutes of water spraying. This corresponds to the association of initial particles to form intermediate structures corresponding to the second peak (between 315 and 630 μm). For this second group (plain grey lines), the mass fraction started increasing as soon as water was sprayed and slightly decreased or remained constant after several minutes due to the association of intermediates structures into bigger agglomerates (N630 μm) corresponding to the third group (light grey dashed lines). At the beginning of the trial, no particles of this population were present. They appeared after several minutes of water spraying and their mass fraction started to increase. When grouping the different curves obtained for each population, the three typical evolutions can be observed (Fig. 10b): initial particles (b 315 μm), intermediate population (315–630 μm) and agglomerates (N630 μm). Similar curves were obtained for both maltodextrin powders and whatever the sprayed water feed rate QL. The only difference observed between maltodextrin DE12 and DE21 was the width of intermediate population that extended up to 800 μm for maltodextrin DE21. Formation of intermediate particles by the association of small initial particles was the main growth mechanism in this case, and only few larger agglomerates resulting of the association of these intermediate particles appeared since bridges were not sufficient in this case to overcome mechanical constraints.
For maltodextrin DE12, the modification of the water feed rate did not change the general shape of the three curves (Fig. 11). The only difference concerned the position of the change in the slope of the different curves corresponding to the change of the main growth mechanisms (particlecluster or cluster-cluster). When using a lower QL of 1.7 ml min −1, initial particles disappeared more slowly than with a higher QL, and at the end of the trial there was still an important concentration of these particles (about 75% of the whole mass compared to less than 20% for a QL of 4.9 ml min −1). The change in the slope appeared at the same time for the three curves, but it was delayed when the liquid feed rate was decreased (after about 12 min when using a QL of 1.7 ml min −1 instead of about 4 min for a QL of 4.9 ml min −1). For both maltodextrin particles, with a lower water feed rate, a limited size increase was observed since only a small part of the initial particles agglomerated into intermediate agglomerates and only few bigger agglomerates were formed. Increasing the water feed rate therefore not only allowed a faster growth, but also allowed obtaining an agglomerated powder with a more homogeneous particle size distribution and no more initial particles. Optical microscope observation of particles from the three size classes (Fig. 12) confirmed a different structure for intermediate population and larger agglomerates. Intermediate population consisted in compact associations of a few initial particles linked by solid bridges. Final agglomerate population consisted of the intermediate agglomerates linked by solid bridges and forming a larger and more open structure within which intermediate particles could be recognized. Optical microscope observations did not show any difference in the structure of the
Initial particles
(a)
Initial particles
Intermediate population
1,0
(b)
Intermediate population 1,0
Agglomerates
0,8
0,6
0,6
Xm
Xm
0,8
Agglomerates
0,4
0,4
0,2
0,2
1.7 ml·min-1
0,0 0
3
6
9
time (min)
12
15
0,0 0
3
6
9
12
15
time (min)
Fig. 11. Evolution of the mass fraction Xm of the three populations for different water feed rates QL during maltodextrin DE12 (a) and DE21 (b) agglomeration trials (Tc = 70 °C).
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026
8
C. Avilés-Avilés et al. / Powder Technology xxx (2014) xxx–xxx
Maltodextrin DE12
Maltodextrin DE21
Fig. 12. Optical microscope images of initial particles (a, d), intermediate particles (b, e) and agglomerates (c, f) from maltodextrin DE12 (a, b, c) and DE21 (d, e, f) agglomeration trials (Tc = 70 °C, QL = 4.9 ml min −1).
two populations when changing the maltodextrin powder or the agglomeration conditions. 4. Conclusion Maltodextrin DE12 and DE21 showed a different sensitivity to water and temperature constraints during fluidised bed agglomeration. Especially whilst for maltodextrin DE12, the size increase and growth rate increased significantly when the sprayed water feed rate was increased, almost no variation was observed for maltodextrin DE21. This was attributed to a faster drying of maltodextrin DE21, which also corresponded to lower glass transition temperatures and to less viscous solutions, less efficient for the formation of viscous bridges between particles. Whatever the conditions tested, agglomeration occurred in two stages with, first, the association of initial particles into intermediate agglomerates and second, the association of the intermediate structures into larger and more porous agglomerates. The change from the first growth mechanism to the second one was delayed when decreasing the sprayed water feed rate with only few large agglomerates formed in this case. This may also influence the agglomerate structure and the powder properties. Further to allow increasing the agglomeration growth rate in the case of maltodextrin DE12 agglomeration, the increase of the sprayed water feed rate also allowed obtaining agglomerated powders with a more homogeneous particle size distribution. Acknowledgments The authors would like to acknowledge the financial support of CONACYT (Mexico) (310189) and the French Government (734297G) for the PhD scholarships provided to Carlos Avilés-Avilés. References [1] M.E. Aulton, M. Banks, Fluidised bed granulation—factors influencing the quality of the product, Int. J. Pharm. Technol. Prod. Manuf. 4 (1981) 24–29. [2] G.I. Tardos, M.I. Khan, P.R. Mort, Critical parameters and limiting conditions in binder granulation of fine powders, Powder Technol. 94 (1997) 245–258.
[3] B.J. Ennis, G.T. Tardos, R. Pfeffer, A microlevel-based characterization of granulation phenomena, Powder Technol. 65 (1991) 257–272. [4] S. Palzer, Contrôle de l'agglomération de polymères de dextrose par l'application du concept de la transition vitreuse, Ind. Aliment. Agric. 122 (2005) 20–27. [5] S. Palzer, Agglomeration of pharmaceutical, detergent, chemical and food powders— similarities and differences of materials and processes, Powder Technol. 206 (2011) 2–17. [6] B. Bhandari, T. Howes, Relating the stickiness of food undergoing drying and dried products to their surface energetics, Dry. Technol. 23 (2005) 781–797. [7] C. Hennigs, T.K. Kockel, T.A.G. Langrish, New measurements of the sticky behavior of skim milk powder, Dry. Technol. 19 (2001) 471–484. [8] L. Ozmen, T.A.G. Langrish, Comparison of glass transition temperature and sticky point temperature for skim milk powder, Dry. Technol. 20 (2006) 37–41. [9] C. Turchiuli, T. Jimenez, E. Dumoulin, Identification of thermal zones and population balance modelling of fluidized bed spray granulation, Powder Technol. 208 (2011) 542–552. [10] S.J. Maronga, P. Wnukowski, Establishing temperature and humidity profiles in fluidized bed particulate coating, Powder Technol. 94 (1997) 181–185. [11] Y. Roos, M. Karel, Applying state diagrams to food processing and development, Food Technol. 45 (1991) 66–68. [12] F. Ronsse, J. Depelchin, J.G. Pieters, Particle surface moisture content estimation using population balance modelling in fluidised bed agglomeration, J. Food Eng. 109 (2012) 347–357. [13] K. Terrazas-Velarde, M. Peglow, E. Tsotsas, Kinetics of fluidized bed spray agglomeration for compact and porous particles, Chem. Eng. Sci. 66 (2011) 1866–1878. [14] A. Seo, P. Holm, T. Schaefer, Effects of droplet size and type of binder on the agglomerate growth mechanisms by melt agglomeration in a fluidised bed, Eur. J. Pharm. Sci. 16 (2002) 95–105. [15] T.A.M. Hirata, G.C. Dacanal, F.C. Menegalli, Effect of operational conditions on the properties of pectin powder agglomerated in pulsed fluid bed, Powder Technol. 245 (2013) 174–181. [16] T. Jimenez, C. Turchiuli, E. Dumoulin, Particles agglomeration in a conical fluidized bed in relation with air temperature profiles, Chem. Eng. Sci. 61 (2006) 5954–5961. [17] K. Terrazas-Velarde, M. Peglow, E. Tsotsas, Investigation of the kinetics of fluidized bed spray agglomeration based on stochastic methods, AIChE J 57 (11) (2011) 3012–3026. [18] A. Gianfrancesco, Spray drying engineering: particle stickiness in relation with agglomeration, (PhD thesis) AgroParisTech, France, 2009. (192 pp.). [19] P. Dokic, J. Jakovljevic, L. Dokic-Baucal, Molecular characteristics of maltodextrins and rheological behaviour of diluted and concentrated solutions, Colloids Surf., A 141 (1998) 435–440. [20] F. Avaltroni, P.E. Bouquerand, V. Normand, Maltodextrin molecular weight distribution influence on the glass transition temperature and viscosity in aqueous solutions, Carbohydr. Polym. 58 (2004) 323–334. [21] C. Turchiuli, R. Smail, E. Dumoulin, Fluidized bed agglomeration of skim milk powder: analysis of sampling for the follow-up of agglomerate growth, Powder Technol. 238 (2012) 161–168.
Please cite this article as: C. Avilés-Avilés, et al., Fluidised bed agglomeration of particles with different glass transition temperatures, Powder Technol. (2014), http://dx.doi.org/10.1016/j.powtec.2014.03.026