Epstein N., Grace J.R. Spouted and Spout-Fluid Beds: Fundamentals and Applications

Подождите немного. Документ загружается.

Granulation and particle coating 237

59. J. M. Link, W. Godlieb, N. G. Deen, and J. A. M. Kuipers. Discrete element study of granulation

in spouted-ﬂuidized bed. Chem. Eng. Sci., 62 (2007), 195–207.

60. C. R. Duarte, V. V. Murata, and M. A. S. Barrozo. The use of a population balance model in

the study of inoculation of soybean seeds in a spouted bed. Can. J. Chem. Eng., 82 (2004),

116–121.

61. M. Paulo Filho, S. C. S. Rocha, and A. C. L. Lisboa. Modeling and experimental analysis of

polydisperse particles coating in spouted bed. Chem. Eng. Proc., 45 (2006), 965–972.

62. J. Pi

na, V. Bucal

a, N. S. Schbib, P. Ege, and H. I. de Lasa. Modeling a silicon CVD spouted

bed pilot plant reactor. Intern. J. Chem. Reactor Eng., 4 (2006), 1–19.

63. V. Pont, K. Saleh, D. Steinmetz, and M. Hemati. Inﬂuence of the physicochemical properties

on the growth of solids particles by granulation in ﬂuidized bed. Powder Technol., 120 (2001),

97–104.

64. B. Guignon, E. Regalado, A. Duquenoy, and E. Dumoulin. Helping to choose operating

parameters for a coating ﬂuid bed process. Powder Technol., 130 (2003), 193–198.

65. K. Saleh, D. Steinmetz, and M. Hemati. Experimental study and modeling of ﬂuidized bed

coating and agglomeration. Powder Technol., 130 (2003), 116–123.

66. M. G. A. Vieira, M. W. Donida, and S. C. S. Rocha. Adhesion of an aqueous polymeric

suspension to inert particles in a spouted bed. Drying Technol., 22 (2004), 1069–1085.

67. M. W. Donida, S. C. S. Rocha, and F. Bartholomeu. Inﬂuence of polymeric suspension

characteristics on the particle coating in a spouted bed. Drying Technol., 23 (2005), 1–13.

68. S. C. S. Rocha, M. W. Donida, and A. M. M. Marques. Liquid-particle surface properties on

spouted bed coating and drying performance. Can. J. Chem. Eng., 87 (2009), 695–703.

69. B. Bhandari and T. Howes. Relating the stickiness property of foods undergoing drying and

dried products to their surface energetics. Drying Technol., 23 (2005), 781–797.

70. K. C. Link and E. U. Schl

under. Fluidized bed spray granulation and ﬁlm coating. A new

method for the investigation of the coating process on a single sphere. Drying Technol., 15

(1997), 1827–1843.

71. S. M. Iveson, J. D. Litster, K. Hapgood, and B. J. Ennis. Nucleation, growth and breakage

phenomena in agitated wet granulation process: a review. Powder Technol., 117 (2001), 3–39.

72. A. M. M. Marques. Inﬂuence of particle-suspension adhesion during spouted bed coating with

bottom atomization. M.Sc. dissertation. Univ. of Campinas, Brazil (2007) (in Portuguese).

73. P. I. Spitzner Neto and J. T. Freire. Study of pastes drying in spouted bed: inﬂuence of the

paste presence on the process. In Proceedings of the XXV Brazilian Congress on Particulate

Systems, ed. J. T. Freire (S

ao Carlos, Brazil: Federal Univ. of S

ao Carlos Press, 1997), 2,

pp. 520–525 (in Portuguese).

74. M. F. D. Medeiros, S. C. S. Rocha, O. L. S. Alsina, C. E. M. Jer

onimo, U. K. L. Medeiros,

and A. L. M. L. da M ata. Drying of pulps of tropical fruits in spouted bed: effect of the

composition on dryer performance. Drying Technol., 20 (2002), 855–881.

75. R. L. R. Lefevre and M. S. T. Price, Coated nuclear fuel particles: the coating process and its

model. Nuclear Technol., 35 (1977), 227–237,

76. E. H. Voice. Coatings of pyrocarbon and silicon carbide by chemical vapour deposition. Chem.

Engr. (Dec. 1974), 785–792.

77. D. W. Marshall, Idaho National Laboratory. Private communication to J. R. Grace (2009).

14 The Wurster coater

Sarah Palmer, Andrew Ingram, and Jonathan Seville

14.1 Introduction: particle coating techniques

The Wurster coater is a special variant of a ﬂuidized or spouted bed (Figure 14.1) that

resembles a spouted bed in being of conical-cylindrical construction, but normally has

a gas distributor at its base, through which an upfacing spray nozzle projects. This

nozzle is usually of the two-ﬂuid type and is used to introduce a coating material –

usually a solution, but sometimes a suspension or melt. It is customary to install a draft

tube into the column, so particles are constrained to circulate upward inside the draft

tube and downward in the annulus surr ounding the draft tube. In this respect, Wurster

coaters have similarities to spout-ﬂuid beds containing draft tubes. Their main use is for

coating pharmaceutical tablets and pellets, for which this is one of the most common

methods.

Within the pharmaceutical industry, particulate materials are coated for a number of

reasons

1,2

: to produce controlled release dosage forms, to mask unpleasant tastes, to

protect ingredients within the tablet/pellet from the environment (light, moisture, air, or

undesirable pH), for aesthetic appeal, and to protect the consumer from toxic compounds

that may cause gastric distress or nausea.

In industrial practice, particle coating is accomplished using several types of equip-

ment, most notably rotating drum (or perforated pan) and ﬂuidized bed coaters.

(Fluidized beds are also used to agglomerate particles, but agglomeration is unde-

sirable in coaters, which are therefore designed to minimize it.) Rotating drums are

used primarily to coat tablets (usually having a particle diameter  6.0 mm). Although

rotating drums provide a low-mechanical-stress environment for tablet coating, their use

can result in high tablet-to-tablet coating variability and/or long batch processing times

because of poor heat and mass transfer compared with ﬂuidized beds.

3,4

To minimize

variability in coating thickness, as required for precision coating of the active ingredient

onto tablets, bottom-spray (Wurster) type beds are advantageous.

4–6

Wurster coaters are

sometimes considered to be ﬂuidized beds, rather than spouted beds, but they meet the

spouting conditions outlined in Chapter 1.

The Wurster coater is used extensively in the pharmaceutical industry for ﬁlm coating

of pellets, tablets, and other particulate dosage forms; in the food industry for coating

Spouted and Spout-Fluid Beds: Fundamentals and Applications, eds. Norman Epstein and John R. Grace.

Published by Cambridge University Press.



Cambridge University Press 2011.

The Wurster coater 239

Chamber

Fountain region

(deceleration region)

Draft tube

(upbed region)

Annulus

(downbed region)

Partition gap (horizontal transport region)

Dished air

distribution plate

Fluidizing air

Two-fluid

spray nozzle

Coating solution and

atomization air feeds

(a)

(b)

550

500

450

400

350

300

250

200

150

100

−150 −100 −50 0 50 100 150

(c)

Figure 14.1. Wurster coater: (a) schematic

; (b) photograph

; (c) time-average s olids motion

(black arrows) and a single particle trajectory (line).

240 Sarah Palmer, Andrew Ingram, and Jonathan Seville

food ingredients; and in the agricultural industry f or production of controlled-release

fertilizers.

14.2 Features of the Wurster coater

The Wurster coater was invented and developed by Dale E. Wurster in the early 1950s at

the University of Wisconsin.

In 1954, Abbott Laboratories produced the ﬁrst commer-

cially available ﬁlm-coated tablet using this device.

Although originally designed to

coat tablets, the process is now widely used to coat smaller particles, such as the pellets

used to ﬁll pharmaceutical capsules.

6,8,9

As indicated earlier, the key feature of the Wurster-type coater is the draft tube (or

“partition”), which is separated by a gap from the gas distributor plate. Compared with

conventional spouted beds, the more complex gas distribution arrangement yields some

of the properties of both spouted and ﬂuidized beds, enabling the operating range to be

extended downward in particle size from Geldart Group D particles,

which are easily

spouted, to Geldart Groups A and B.

The ﬂuidization airﬂow serves several functions; aerating the annulus, enabling pneu-

matic transport of particles up through the draft tube, and providing drying capacity.

Typically, the distributor plate possesses larger holes in the central region (beneath the

draft tube) and smaller holes in the outer region (beneath the annulus), favoring gas ﬂow

in the former.

Christensen and Bertelsen

identiﬁed four different regions in the Wurster coater

according to particle motion: upbed (draft tube), deceleration (fountain), downbed (annu-

lus), and horizontal transport regions. These are characterized as follows:

Upbed region. Particles pass through the gap between the distributor and the draft

tube and are pneumatically transported upward while receiving coating from the spray

nozzle located at the center of the distribution plate. Spray droplets strike the particle

surfaces and may then spread, coalesce, and dry to form a ﬁlm. The air and particle

velocities are not uniform across the draft tube; velocities at the center signiﬁcantly

exceed those at the wall. Particles close to the wall rise more slowly or may even move

downward along the inside of the draft tube. In unfavorable circumstances, internal

circulation may occur within the draft tube.

Deceleration region. After leaving the draft tube, the particles enter the fountain

region, where they decelerate, continue drying, and fall downward onto the top of the

downbed (annulus) region.

Downbed region. Particles within the annulus are in close contact with one another

and move slowly downward as a moving packed bed. It is in this region that sticking

and agglomeration are most likely to occur if the particles are not sufﬁciently dry;

this is undesirable in terms of coating quality. Air volume in the downbed depends on

the size and number of holes in the outer region of the distributor plate. According to

Jones,

the airﬂow in t he annulus should be at or close to that necessary for incipient

ﬂuidization, to ensure continuous downward movement toward the partition gap.

The Wurster coater 241

Horizontal transport region. This name is given to the region below the draft tube, in

which particles pass horizontally through the partition gap. From this region, particles

reenter the upbed region. The partition gap controls the ﬂow of particles into the upbed

region, and thus the solids circulation rate.

This cyclic pattern of solids motion is repeated until the desired thickness of coating is

deposited on the particles.

In addition to better coating performance than with the top-spray methods,

advantages

of the Wurster coating process include excellent heat and mass transfer within the bed,

leading to short batch processing times, ability to handle particles as small as 100 μm,

and mechanical simplicity, with no moving parts. A disadvantage is that the process is

mechanically harsh (compared with that of rotating drums); particle attrition can become

an issue with increasing particle size because of increased momentum of collisions at

the higher air velocities needed for circulation.

Spray nozzle blockage is also an issue

because, should it occur, the problem cannot be easily resolved, as nozzle removal

during operation is impossible.

6,13

For tablet coating, dead zones can exist within the

annulus near the inner chamber wall; this is undesirable because dead zones contribute

signiﬁcantly to tablet-to-tablet coating variability. A further limitation of the process

is that ﬁne (<100 μm) particles tend to agglomerate rather than coat (unless special

enhancements are used) and may adhere to chamber walls.

In industrial practice, optimum operating conditions are based largely on past experi-

ence and experimentation via trial and error; the results from such methods are therefore

system-speciﬁc. Despite the widespread industrial use of Wurster coaters, the funda-

mental mechanisms controlling the process are not well understood.

14.3 Uniformity and quality of coating

Depending on the application, the performance of coated particles in their end use may

depend crucially on the unifor mity and quality of the coating. Uniformity can vary from

particle to par ticle and also from point to point on the surface of each par ticle. Most

studies have concentrated on the former, although both can be important. Methods for

measuring both are reviewed by Palmer.

Cheng and Turton

list the variables that affect coating uniformity and quality, as

summarized in Table 14.1. Coating occurs only within the spray region. It is then

reasonable to assume, as suggested by Mann et al.

and Cheng and Tur t on,

that the

particle-to-particle variability in coating arises from a combination of (1) variation in

the number of passes made by particles through the spray region and (2) variation in the

amount of coating per pass (Figure 14.2).

The total mass of coating, M

total

, on a given particle is given by

total



i=1

, (14.1)

242 Sarah Palmer, Andrew Ingram, and Jonathan Seville

Table 14.1. Variables affecting coating uniformity and quality.

The factors in bold are most signiﬁcant.

Design variable Process variable Product variable

Type of equipment

Type of nozzle

Type of distributor plate

Diameter of draft tube

Spacing of draft tube to

base (partition gap)

Length of draft tube

Temperature of ﬂuidizing air

Humidity of ﬂuidizing air

Spray rate (liquid and air)

Atomization pressure

Batch size

Velocity of air (inner and

outer regions)

Solvent (water or organic)

Coating material

Coating solution composition

Particle properties

Spray

zone

Circulation

path of

particle

Figure 14.2. Circulation of a particle through the spray zone (after Cheng and Turton

where N is the number of passes through the spray zone and x

is the amount of coating

applied during the ith pass. Mann et al.

showed that the coefﬁcient of variation (CV )

for coating mass per particle can be expressed as:

CV =

total











, (14.2)

where the two terms on the right-hand side refer to the variation in mass deposited per

pass and the variation in the number of passes, respectively. Mann

showed that CV

could be represented more usefully in terms of the circulation time distribution:

CV =

total







coat





coat

, (14.3)

where t

coat

is the total coating time. The coefﬁcient of variation is therefore proportional

√

1/t

coat

, implying that any desired CV can be obtained by extending the run time

appropriately.

Cheng and Turton

and Shelukar et al.

demonstrated the usefulness of this approach

experimentally and have shown that in most cases it is the variation in coating per pass

through the spray zone that dominates the overall particle-to-particle coating variability.

This has important consequences for the design of Wurster coaters, as discussed later.

The Wurster coater 243

14.4 Particle motion studies

Mann and Crosby

investigated particle cycle times in a spouted/ﬂuidized bed with

a draft tube in which the air supplies to the spout and annulus were independently

controlled. They used a magnetically labeled particle and a detecting coil around the top

of the draft tube to record each pass of the particle. With increasing air supply to the

annulus, the cycle time decreased and became less variable. It was concluded that the

increased airﬂow slightly expanded the bed, reducing the friction between particles and

the adjacent walls and allowing higher and more uniform particle ﬂow toward the gap.

Wesdyk et al.

investigated the effect of bead size in the range 850 to 1400 μmon

ﬁlm thickness in a Wurster coating unit. Results from dissolution testing and scanning

electron microscopy (SEM) analysis indicated that larger beads received a thicker coating

than smaller ones. This effect was said to be caused by differences in particle trajectories:

smaller beads projected further into the expansion chamber of the coater and therefore

cycled fewer times through the coating zone. It was also argued that lighter beads may

accelerate more rapidly through the coating zone, therefore receiving less ﬁlm coating.

Cheng and Turton

carried out extensive batch coating on 1-mm spherical particles

in which the cycle time of a magnetically labeled particle and the ﬁnal spread in coat-

ing within the batch were measured. They investigated the effect of partition gap, gas

ﬂowrate, partition diameter, and partition height on particle motion and coating unifor-

mity. For the range of parameters studied, the partition gap exerted the strongest inﬂuence

on the rate of particle circulation. Cheng and Turton also found that the mean and stan-

dard deviation of the particle cycle time distribution were approximately proportional

over a wide range of operating conditions. By assuming that the time-averaged behavior

of the labeled particle was representative of the ensemble behavior of the bed, they

were able to correlate coating uniformity with operating conditions, using the approach

described in Section 14.3. The authors argued that nonuniformity in coating per pass

derives from variation in trajectories through the draft tube: particles passing close to

the spray source pick up more coating and shelter those farthest away. This important

observation will be considered further in relation to modeling.

By using both a magnetic tracer particle and dye pulse injection in a bed with

a draft tube, Shelukar et al.

were able to measure both the cycle time variability

and the variation in coating per pass and to relate these to coating distribution as in

Eq. (14.3).

The effects of partition gap and auxiliary ﬂuidization airﬂow rate on solids circulation

rate, airﬂow distribution between annulus and draft t ube, and pressure drop were studied

in Wurster coaters of different sizes by Choi,

with solids circulation rate estimated

visually through a sight glass at the wall. This work again identiﬁed the partition gap as

being critical in determining the overall circulation rate of solids.

Fitzpatrick et al.

examined the effects of operating conditions (airﬂow rate and

particle loading) and coater conﬁguration (partition gap) on solids motion using positron

emission particle tracking (PEPT) to follow the trajectory of a single tracer particle in

the absence of coating liquid spray. Mean particle cycle time was found to increase with

244 Sarah Palmer, Andrew Ingram, and Jonathan Seville

Hopper flow

Fluidized/hydrostatic

1/2

Flow rate, W ( D – d

)

5/2

W independent of height, H

Figure 14.3. Two alternative analyses for solids discharge through partition gap.

increasing batch and tablet size and with decreasing airﬂow rate and partition gap. The

variability (relative standard deviation) in particle cycle time increased with decreasing

airﬂow rate and batch size. This was attributed primarily to the formation of dead zones

in the lower portion of the annulus, associated with ordered packing at the wall.

Palmer

studied the coating of 710–1000-μm spherical microcrystalline cellulose

pellets, using PEPT to correlate particle motion with coating uniformity over a range

of operating conditions (partition height, batch size, and gas ﬂowrate). Coating was

most variable at the high and low extremes of batch size and partition gap. Variability

tended to correlate with irregularity in the ﬂow and trajectories of the particles. In

particular, certain conditions gave rise to recirculation of particles within the draft tube,

with particles then passing several times through the spray zone before exiting.

An important question in understanding Wurster coater operation is: What controls

the ﬂow through the partition gap? There are two possibilities, according to whether or

not the annulus is ﬂuidized (Figure 14.3). If the annulus is not ﬂuidized, solids ﬂow from

the annulus is hopperlike,

and obeys a Beverloo-type equation:

W = Cρg

0.5

− kd

)

5/2

, (14.4)

where W is the mass discharge rate through the oriﬁce, D

is the hydraulic diameter of

the discharge oriﬁce (equal to 4 × (cross-sectional area)/(outlet perimeter)); d

is the

particle diameter; C and k are empirical discharge and shape coefﬁcients, respectively;

ρ is the ﬂowing bulk density; and g is the acceleration from gravity. Given that, as in

hopper ﬂow, the weight of the solids is supported mostly by friction at the walls, the

solids circulation rate is little affected by the head of solids in the annulus and might

therefore be relatively independent of batch size. On the other hand, if the annulus is

ﬂuidized, the ﬂuidized solids behave as a liquid and their ﬂow through the partition gap

is then governed by a Bernoulli expression (taking pressure as constant):

1/2ρv

+ ρgH = constant, (14.5)

where ρ is the solids bulk density, v the solids discharge velocity, and H the height of

material in the annulus, so that W ∝ H

1/2

– that is, the mass ﬂow rate is proportional

to the square root of the solids head. The complicating factor is that the division of gas

ﬂow between the draft tube and the annulus is itself affected by the head of solids in

The Wurster coater 245

Particles far

from spray

receive little

coating

Particles

close to spray

receive more

coating and

shield others

No particles pass

through spray in

this re

ion

Draft tube radius R

Hollow

cone spray

region

Figure 14.4. Effect of sheltering in the spray region (Cheng and Turton

). Note that the spray is of

the hollow cone type.

the annulus, and therefore by the batch size, and by the solids entrainment rate into the

draft tube. Experimental results from the literature suggest that different combinations

of design and operating variables have resulted in operation in both these modes. No

comprehensive hydrodynamic model yet exists.

14.5 Modeling of coating thickness variation

Turton

summarized the modeling approaches taken by different investigators:

Classical chemical engineering compartment models, typiﬁed by the two-

compartment model of Sherony

and further developed, for example, by Crites

and Tur t on

; this category of model sometimes includes population balances and

size-dependent growth.

Monte Carlo methods, in which particle movement is a random walk, sometimes

related to experimentally observed particle motion. The amount of coating depends

on time and location within the spray zone.

Fully computational methods such as computational ﬂuid dynamics (CFD), including

discrete element methods (DEM).

A conceptual model developed by Cheng and Turton

is interesting from the point

of view of equipment design; it focuses on the coating received per pass. Within the

spray region, the concentration of particles is sufﬁciently high that particles shield

each other from the spray – a “sheltering” effect. The simplest model assumption was

a binary one, in which particles receive either a “standard” coating amount or none.

However, this resulted in predictions that were in poor agreement with experimental

results. In a more sophisticated approach (Figure 14.4), the amount of coating material

received per pass was assumed to be proportional to the particle projected area (∝ d

246 Sarah Palmer, Andrew Ingram, and Jonathan Seville

to decrease with distance from the spray nozzle, and to depend on the void fraction

within the draft tube (spray region). Because of the sheltering effect, models of this kind

predict an increase in the coefﬁcient of variation as the void fraction in the spray region

decreases. Moreover, as the radius of the draft tube increases – for instance, because of

equipment scaleup – increased sheltering again causes an increase in the coefﬁcient of

variation.

Palmer

adopted a similar approach, previously developed by Seiler et al.

for

conventional spouted bed coating, to predict the evolution of coating mass distributions

in a Wurster coater using particle cycle time distributions and time-averaged spray

region void fraction data obtained from PEPT experiments. Probability parameters were

used to deter mine the entry height of particles from the annulus to the spout and

spray-to-par ticle contact at each height interval within the spray zone. To determine the

extent of coating, a statistical collection factor was calculated, which included the spray

density effect. The predicted coating mass distributions compared well with experimental

distributions.

As Turton

noted, none of the existing models offers a priori prediction of coating

behavior, but DEM may make this possible in the near future.

14.6 Developments in design

Several researchers and industrial manufacturers of Wurster-type coaters have inves-

tigated improvements to the basic design, as indicated in Figure 14.5. Identifying the

transition between the cone and distributor as a cause of “dead” zones or regions of

little particle motion, Shelukar et al.

used an additional cone of shallower angle to

prevent this problem. Subramanian et al.

reduced the variation in coating per pass by

adding a proﬁled “tablet deﬂector” around the spray nozzle, the major effect being to

increase the void fraction near the spray zone, hence reducing problematic local wetting

and enhancing coating uniformity.

GEA Pharma Systems, a pharmaceutical equipment manufacturer, introduced a vari-

ant on the bottom-spray coater in which some of the inlet air is introduced with signif-

icant swirl around the spray nozzle to create a high-velocity low-pressure spray zone,

thereby enhancing particle movement into the zone and accelerating subsequent drying

of coated particles. This design demonstrated improved coating relative to the con-

ventional Wurster design, including reduced particle agglomeration and fewer coating

defects.

Acknowledgments

The authors gratefully acknowledge ﬁnancial assistance from the UK Engineering &

Physical Sciences Research Council and Merck Sharp & Dohme Ltd, and loan of

equipment from GEA Pharma Systems.