application of the fluidized bed process for formulation

Pow der Technology, 79 (1994) 135-146 135

Application of the fluidized bed process for formulation of WG-type pesticide granules

I.A. Al AminInstitu te o f Organic Industry, W arsaw (Poland)

A.K. Biń*F aculty o f C hem ica l a n d Process Engineering, W arsaw University o f Technology, W arsaw (Poland)

(R ece iv ed Ju ly 2, 1993; in rev ised fo rm D e c e m b e r 22, 1993)

Abstract

E x p e r im e n ts w e re c a r r ie d o u t in a la b o ra to ry - s c a le a p p a r a tu s to o p tim iz e c o n d it io n s o f th e f lu id iz e d b e d p ro c e s s a n d ty p e s o f a d d it iv e s in o r d e r to o b ta in p e s t ic id e g r a n u le s o f th e p r o p e r s iz e , re a d ily d is p e rs ib le in w a te r a n d fo rm in g s ta b le w a te r su s p e n s io n s , a n d o f su ff ic ie n t m e c h a n ic a l s t r e n g th . T h e m o d e l o f H a r a d a e t al. h a s b e e n fo u n d a d e q u a te to d e s c r ib e th e k in e t ic s o f g r a n u le g ro w th f o r th e m a te r ia ls s tu d ie d , a n d th e p a r a m e te r s o f th is m o d e l h a v e b e e n e s t im a te d . T h e a g g lo m e ra t io n c o e f f ic ie n t in th is m o d e l is a n in c re a s in g fu n c t io n o f th e g ra n u la t in g s o lu tio n d o s a g e ra te s . P o s s ib le c o m p a r is o n s w ith r e f e r e n c e d a ta a r e p r e s e n te d a n d d isc u sse d .

Introduction

O ne of the principle objectives of pesticide form ulation is to obtain products which deliver the necessary biological perform ance and have a lengthy shelf life. The final decision as far as optim al pesticide form ulations are concerned, apart from their biological performance, should take into account their toxicity, environm ental im pact and production costs. The most common pesticide form ulations include soluble liquids (SL), emulsifiable concentrates (EC ), w ettable powders (W P) and suspension concentrates (SC) [1].

C urrent trends in pesticide form ulation are to m anufacture and offer new types of products such as SC, capsule suspensions (CS), w ater-dispersible granules (W G) and emulsions in w ater (EW ). Am ong them, the W G form ulations seem to be promising: currently over fifty active substances are commercially offered by leading pesticide m anufacturers as W G products. In all work on W G-type products, special atten tion m ust be given to dispersibility, suspendibility, particle hardness and dust properties.

The aim o f the present work was to check the applicability of the fluidized bed granulation process for selected pesticides and auxiliary m aterials and to determ ine the process conditions enabling w ater-dispersible granules to be obtained which would be m e

* A u th o r to w hom co rre sp o n d e n c e sh o u ld b e ad d re sse d .

chanically resistant and which would form stable water suspensions.

Powder granulation in a fluidized bed

BackgroundA ulton and Banks [2] sum m arized the effects of

different factors which may affect product quality obtained in a fluidized bed granulation process. These can be divided into equipm ent, process and m aterial factors. Among the equipm ent param eters, the effects of the distributor, shape of the granulator, location of the spray nozzle, m ethod of operation (high or low pressure), and the equipm ent scale can be m entioned. The process param eters may involve the initial m aterial load, fluidizing agent flowrate, its inlet tem perature and humidity, and factors which determ ine the spray nozzle perform ance. The m aterial param eters concern type and quantity of binder, p roperties of the starting m aterial (powder form, hydrophobicity) and its ability to be fluidized.

M ost of the published work on fluidized bed granulation has concentrated on the effects o f process param eters. In general, fluidized bed granulation seems to be very sensitive to relatively small changes in process conditions. Suitable selection of fluidising agent flowrate should provide an appropriate bed expansion and re moval of solvent by drying, thus avoiding local over

0032-5910/94/S07.00 © 1994 E lsev ie r S eq u o ia . A ll rig h ts re se rv ed S S D I 0 0 3 2 -5 9 1 0Pobrano z http://repo.pw.edu.pl / Downloaded from Repository of Warsaw University of Technology 2022-04-10

136

wetting and clumping of solid particles. The energy balance requirem ents m ust then be adjusted to the proper course of-granulation (agglom eration), which dem ands sufficient liquid to be presen t at the particle surface during the process [3, 5, 8-10]. This will be reflected in the effect of the flowrate of binder solution on the size distribution of granules, their friability and porosity.

T here is experim ental evidence [11, 12] which dem onstrates that the size o f granules obtained in the process of agglomeration is directly dependent upon the size of the drops of b inder solution. The size distribution of binder solution drops will be a function of the physical properties of the solution and the perform ance of the spraying nozzle.

A ttem pts to in terpret phenom ena and mechanisms of the granulation processes, including those carried out in a fluidized bed, in term s of physical background are presented by Ennis et al. [14, 15]. These authors considered the physical m echanisms o f wet granulation processes of powders using binder solutions, which occur at the micro- and macrolevels. The form er concerns physical phenom ena which take place a t the particle surface (e.g., involving dynamic pendular liquid bridges). The necessary conditions for two colliding granules to coalesce have been form ulated by m eans of a critical viscous Stokes’ num ber. If the values of this num ber exceed the critical value, noninertial granulation occurs (i.e. each collision of particles results in their coalescence). The process ra te is independent of the kinetic energy of the particles and of the viscosity of the binder solution; however, it critically hinges on the presence and distribution of binder. For this regime of granulation such operating param eters as spray drop size and spray distribution will control the granulation o f fine powders. If the value of the Stokes’ num ber is close to the critical value, inertial granulation will occur, for which both the kinetic energy of the granules as well as the binder viscosity become im portant. The ra te of granule coalescence will increase with increasing viscosity of binder solution and with decreasing kinetic energy of the granules. O n further increase in granule size, the local values of the Stokes’ num ber can be g reater than the critical value and a transition to the coating regime will occur, w here fu rther granule growth is lim ited due to coalescence.

The authors consider also the process of granule consolidation. This directly concerns granule voidage, which may be of im portance for granulated m aterials such as pesticides which are expected to disperse in a fluid agent.

Ennis et al. [15] assum ed the presence of coalescence and consolidation mechanisms only, and neglected other phenom ena such as attrition of granules, wetting of particles and the rate of the spraying zone penetration

throughout the bed. In general, the au thors’ conclusions have been confirmed by experim ental evidence. However, since the variables defining the Stokes’ num ber are difficult to determ ine quantitatively, so far this theory has a mainly qualitative character.

Rowe [16] discussed theoretically the problem s involving cohesion and adhesion betw een solid particle and binder, together with the phenom enon of its spreading over the particle surface. The surface energy of the binder and the polarity of the granulated m aterial are of significance here. This enabled the author to determ ine the optim al choice of binder. The binding strength betw een the particles and liquid binder and the resulting properties of granules and solid-liquid dispersions are also discussed (see e.g. [14]).

Granulation process kineticsThe particular form of a relationship describing

growth ra te during granulation in a fluidized bed will depend on the operating m ethod, the fractional distribution of particles, the o rder o f growth kinetics, and the occurrence of side processes such as particle entrainm ent by gas stream or granule attrition in the bed. T he simplest kinetic models o f granulation in a fluidized bed describe growth due to coating o f m onodisperse m aterials assuming uniform growth of particles over their entire surface, neglecting fines entrainm ent and attrition. M ore complex m odels of granulation take into account polydisperse particulate beds, entrainm ent and breakdow n of granules, recirculation of seed and effects of the residence tim e of particles in the bed. Brief sum m aries of models of the granulation processes in a fluidized bed have been p resen ted by Biń [17, 18]. For the presen t situation (batch process of fine particle agglom eration), the m odel o f a pure agglomeration process proposed by H arada et al. [3, 19] seems to be the most relevant:

d p = d p0 exp(cć) ( 1 )

It was derived by assuming tha t two particles coalesce to form a granule. E quation (1) contains an agglomeration coefficient, c, which should be dependent upon the operating param eters. H arada [3] suggested an em pirical correlation for c:

c = 52.9 OTł^ ~ * L* -1 .2 9 (2 )m 0

In this correlation m L should be expressed in kg h - 1 and с is obtained in h _1. The range of applicability of this equation is relatively narrow: The authors varied the values of the term m L( 1 —х ъ)/т 0 from 0.035 to 0.07 hr1.

F or a granulation process in which the growth rate of granules is proportional to their total surface area, a linear m odel of growth can be obtained:

Pobrano z http://repo.pw.edu.pl / Downloaded from Repository of Warsaw University of Technology 2022-04-10

137

which dem onstrates that the granule growth kinetics is of the zeroth order:

dp = d p0 + at (3)

i ddp t A= — = const.dt (4)

It is interesting to note tha t at c / < l , eqn. (1) would simplify to eqn. (3) and a ~ d p0c.

H ere, the way to define a representative particle (granule) size may be of in terest (see discussion by Bin et al. [18]). In general, the following expression can be used for polydisperse m aterials:

d„ (5)

w here m and n are the natural num ber exponents. For m = 3 and n = 2, the area-weighted m ean particle size, more frequently known as the Sauter diam eter, d 32, is obtained. For m = 4 and n = 3, the mass-weighted m ean particle size can be calculated (d43). Since sieve analyses are usually available, simple relations can be derived for d32 and d43:

1

2 (* ,4 > ;), d ,i3 "jiijixidpj)

Experimental

ApparatusThe fluidized bed granulator was m anufactured by

A erom atic (type STRA ), having a capacity of 2 kg of pow der (Fig. 1). The main part of this apparatus was a granulator m ade of a tapered glass section of diam eters 100/250 mm and height 320 mm, and a cylindrical

section o f 250 mm diam eter and 160 mm height. The upper p a rt of the granulator was equipped with four bag filters to capture the entrained fines and return them to the bed. The filters were automatically blown with a stream o f com pressed air. The granulator was equipped with o ther devices, such as m anom eters to control pressure drop over the grid and over the filters, therm om eters to control the inlet and outlet air tem peratures, control valves for the air stream through the granulator and for com pressed air used to spray binder solution, and a pum p to feed binder solution to the spray nozzle. A two-phase spray nozzle was used in the experim ents (hole d iam eter for liquid was 0.9 mm with air annulus diam eters 1.2/2.5 mm) located 275 mm from the grid plate. The grid was m ade of three parts: a perforated bottom plate with perforations 2.8 mm diam eter, a middle portion of 60 /ллп diam eter, and an upper section of 246 p.m diam eter.

Maximum air flowrate available in the apparatus was 120 m 3 h - 1 and maximum air tem perature was 80 °C.

Materials and experimental conditionsT hree powders were used in the granulation exper

iments: kaolin, cupric hydroxychloride (HCM ) and sulfur. T heir basic properties are listed in Table 1. Mean initial particle sizes were determ ined using two methods: sedim entation balance and a C oulter apparatus. The differences betw een the results obtained from these m ethods for kaolin and sulfur can most likely be attribu ted to different electrolytes applied in the m easurem ents, which could affect possible agglomeration o r breakdow n o f particles during size determ ination. This conclusion was confirmed in separate experiments carried ou t for suspensions obtained from the same but granulated m aterials. Clearly smaller mean sizes w ere recorded, which can be explained by the presence

Outlet air

t

Fig. 1. E x p e rim e n ta l se tup .


138

T A B L E 1. P hysical p ro p e r t ie s o f g ra n u la te d pow ders

P o w d er R e a i d en sity (kg m -3 )

In itia l av erage g ra n u le d ia m e te r u m ‘(m m s ” 1)

и b Mmb(m m s _1)

S ed im . b a lan ce C o u lte r c o u n te r

^43(yu,m)

^32(/xm)

d n

O m )<̂ 32( д т )

K ao lin 2600 4.5 3.34 16.4 13.5 0.20 1.6H C M 3000 5.8 3.53 4.2 3.8 0.02 0.42S ulfur 2070 13.5 7.03 14.2 11.9 0.13 1.4

“V a lu es u sing R e mf = A r/1250; bv a lu es u sin g Hmb = 100 d i3.

of additives used during granulation of these m aterials which positively influenced the dispersibility o f the agglomerates.

Typically an air flowrate of 40 m 3 h ” 1 (or 50 m 3 h _1 during form ation of larger granules) a t 20 °C was applied, corresponding to superficial air velocities in the apparatus from 0.23 to 1.46 m s - 1 (or 0.29 to 1.82 m s_1), depending on the vertical distance m easured from the grid plate.

During an experim ental run, granulation solutions containing different additives (dispersants, wetting and binding agents) w ere sprayed by m eans of a two-phase nozzle into the fluidized bed of particles. The main properties of these additives are given in Table 2. The applied feed rates of the granulating solutions, containing from 1 1 0 to 180 cm3 o f w ater, were from 0.06 to 0.23 g s _ 1 (with respect to the solvent, w ater), which corresponded to the pendular regime of granulation. The spraying air param eters were: flowrate 0.15 m3 h - 1 and 0.12 M Pa pressure.

A fter term inating the granulation stage (dosing of granulating solution), which lasted from 6 to 45 min, depending on the volume and feed ra te of the binding agent solution, the granulated product was dried with a stream of air at 50-80 °C for about 15 min.

T A B L E 2. P ro p e rtie s o f g ra n u la tin g so lu tio n s3

T ype o f d ispe rs in g ag en t

S o lu tio n d en sity (kg m -3 )

S o lu tio n v iscosity (m P a s)

S u rface ten s io n (m N m _ I)

p H

W a te r + w e ttin gag en t only 1.27 31.4

T am o l N N 8609 1053 1.80 44.0 7.5V a n isp e rse CB 1053 2.60 35.8 9.0P olyfon I I 1056 1.97 37.8 9.0Syrew eal 1058 1.95 36.7 6.9K lu tan 1054 1.80 36.0 4.7

“C o m p o sitio n o f so lu tions: 6 w t.% o f d isp e rs in g agen t; 0.5 w t.% o f w ettin g ag e n t (w t.% w ith re sp e c t to d ry g ran u les).

ProcedureThe principle objective of the experim ental inves

tigations was to determ ine the optim al conditions of the fluidized bed process needed to obtain granules of the p roper size, readily dispersible in w ater, forming stable w ater suspensions, and having sufficient m echanical strength. For this purpose, the applicability of selected additives in the form of aqueous solutions to be used in the granulation process, the effects of their concentration and feed rates have been examined.

In the final stage of experim ent, various attem pts to determ ine the kinetics of the granulation process in a fluidized bed were undertaken for the selected conditions of this process, i.e. those at which the most advantageous product p roperties were obtained.

The kinetic m easurem ents were carried out for a batch process of granulation. These were m ade using a weighed am ount of about 0.4 kg of powder, prepared earlier in a je t mill, and a predeterm ined am ount of granulating solution containing 110-180 cm3 o f wate r + max. 10 wt.% of additives (with respect to dry mass of granulated product). In the initial stage of experim ent with kaolin, with constant process param eters (feed rate of granulating solution, fluidizing agent flowrate and tem perature, and spraying conditions of the solution), the granulation process was carried out in an in terrupted m anner, i.e. the process was in terrup ted after a certain time interval, samples of granules w ere taken and analysed, they were then returned to the bed and the run was continued. This procedure was la ter considerably changed, since it was established tha t the granules were not sufficiently dry, some of the m oisture had been lost during these m anipulations, and the bed param eters (tem perature and m oisture) could vary as well, affecting reproducibility of the results.

In the modified procedure, the granulation run was com pleted (together with the drying stage), and the experim ents were carefully planned for a predeterm ined time interval. The experim ental series was designed to obtain at least four data points for a given set of the process param eters.


139

The following product properties were m easured:a) granulometric composition, by m eans o f sieve analysis using an autom atic vacuum sieving apparatus (Alpine Co.), taking a sample of 10 g and sieving it for 5 min;b) suspendibility (ability to form stable suspensions in w ater) was determ ined by pouring a portion of granules (0.75 to 1.0 g) into standard hard w ater contained in a cylinder of the sedim entation balance (Sartorius Co.); the percentage of solids m aintained in suspension after 30 min of pouring the granules which dispersed in the w ater was taken from an analysis of the suspension curve as a m easure of suspendibility;

c) friability of granules was determ ined using a device similar to tha t described by N iessen et al. [20] and called a Roche Friabilator. A sample of granules (50 g), free of dusty fractions, was subjected to attrition for 15 min by rotating and mixing with a blade at a rotation frequency of 20 m in -1 . A fter com pleting this test, the sample was separated on a sieve of 71 fxm. d iam eter. The mass fraction below 71 i±m was taken as the final m easure of product friability.

Results and discussion

KaolinT he experim ents carried out using kaolin were treated

as a prelim inary stage, to check the experim ental p rocedure, though this m aterial is applied in pesticide technologies as a filler. Thus for kaolin the effects of the type of additive, the feeding rate and the volume o f granulating solution on the granulation kinetics and the product properties were studied. The size distributions of granules obtained in these experim ents can be approxim ated by a log-normal distribution, although in some cases they were distinctly of bim odal character. T he m ean size of granule depends on the concentration of the additive (Fig. 2). A rapid variation in granule

growth tendency above 0.75 wt.% Tam ol is seen. This can be explained by a possible change in the growth mechanism. The initial point (at zero Tam ol content) corresponds to granulation of kaolin using w ater and2 wt.% of the wetting agent Rokafenol N 8 . A t amounts of Tam ol above the threshold value of 0.75 wt.% , further granule growth with increasing Tam ol content is approximately linear, and the lines for d 32 and d43 are practically parallel. In general, the results obtained in these experim ents are in agreem ent with the literature data [4, 11, 21-26]. The feed ra te and the volume of granulating solution also affect the granule size (Fig. 3). As seen from this figure, the granule size, expressed as d43, increases with the dose rate, m L, m ore rapidly with higher am ounts of solvent in the solution. These results also show good correspondence with the conclusions p resen ted earlier.

The suspendibility and friability of the kaolin granules are plotted against Tam ol content in Fig. 4 (at 2 wt.% of the wetting agent, Rokafenol N 8). Increasing the

Fig. 3. D e p e n d e n c e o f g ra n u le m e a n size (d32) on th e fe e d ra te o f g ra n u la tin g so lu tio n a n d th e w a te r vo lum e.

I I I I I0 1 2 3 4 5 6

% wt. T a m o l

Fig. 2. D e p e n d e n c e o f g ra n u le m ean size on T am o l N N 8609 c o n te n t.

-204

% wt. Tamol

Fig. 4. D e p e n d e n c e o f su sp en d ib ility in h a rd w a te r an d friability o f k ao lin g ra n u le s on T am o l 8609 c o n te n t.

wt. Rokafenol N8 = 0,142 g / s

150 m l w ater

• fri. о susp .

>4■P *H I—I •H-HT3с(UaСЯ p40 со

K aolin


140

am ount of Tam ol to about 2 wt.% results in a significant increase in suspendibility of kaolin particles obtained after dispersion of granules of this m aterial, then the curve levels off at a value of about 65-67% . However, a further increase in Tamol content in the kaolin granules results in greater m echanical strength, expressed in term s of friability. F u rther im provement in the suspendibility of granulated kaolin can be achieved by adjusting the am ount of wetting agent. This is illustrated in Fig. 5, which dem onstrates that the optimum am ount of this substance is 0.5 wt.% , corresponding to about 73% of the suspended solids.

The choice o f dispersing agent may influence the granulom etric composition of the product. Figures 6 and 7 show that increasing the content of PVP K90 from 0.1 to 0.4 w t.% results in larger granules, and a similar effect is attained by introducing the appropriate am ount of K lutan (6 w t.% ) instead of Polyfon at constant PVP K90 content. The best solid suspendibility has

Fig. 5. D e p e n d e n c e o f su sp en d ib ility in h a rd w a te r a n d friab ility o f k ao lin g ran u le s on R o k a fe n o l N 8 (w e ttin g a g e n t) c o n te n t.

3s 50ок 20

105

1

- - - - - - - - - - - - - - - - - 1- - - - - - - - - - - 1-- - - - - - 1- - - - - - 1- - - - - 1--- 1— r T I- - - - - - - - - - - - - - - - - - - 1- - - - - - - - - - - 1-- - - - - - 1- - - - - - 1— |—

K aolin6 % w t. P o ly fon H + 2 % w t. R o k a fe n o l N8 + 150 m l w a te rrhL = 0 ,1 4 2 g / s *

% w t. PVP K90 0,1 0,4

—1---1---1—Г П -----9 i 7 ■ В10 2

—I---1—I—1111—4 9 I 7 > •Ф (д m ) 1°

10

Fig. 6. C u m u la tiv e d is tr ib u tio n o f g ra n u la te d kao lin .

99

50

К 20

10

5

-i—i i 'i i iKaolin

0.4 % wt. PVP K90 + 2 % wt. R okafeno l N8 + 150 m l w a te r m L = 0,142 g / s

% wt. о 6 Po lyfon H о 4 Polyfon H о 6 K lu tan

e •

* •

di (/mi)

Fig. 7. C u m u la tiv e d is tr ib u tio n o f g ra n u la te d kao lin .

Fig. 8. D e p e n d e n c e o f g ra n u le m e a n size (d 43) on g ra n u la tio n tim e: 1 - eq n . (1 ), A 43 = 26.74 /im ; c43 = 0.1237 m in _ I; 2 - eqn . (1 ), Л 43 = 34.93 /u n ; c43 = 0.1554 m in -1 .

been found for granulated kaolin containing Polyfon О as a dispersing agent.

The kinetics of kaolin granule growth with addition of 6 wt.% of K lutan (dispersing agent) is plotted in Figs. 8 and 9 for two dose rates of the granulation solution. D espite the ra ther large scatter of the data points, caused mainly by the experim ental procedure applied at this stage of the investigation, the H arada et al. model of agglom eration, eqn. ( 1 ), can be considered to approxim ate these data. The obtained values of the agglomeration coefficients cmn and the constants A mn are listed in Table 3. The constant A mn should formally correspond to the initial particle size d p0 (i.e. at / = 0). This correspondence seems to be satisfactory if the Sauter granule size is used; however, for d43 this correspondence fails. This discrepancy can be attributed most likely to ‘dry’ particle agglomeration due to cohesion before even starting the actual granulation process, so that the initial particle size may be quite different


141

Fig. 9. D e p e n d e n c e o f g ra n u le m e a n size (d 32) o n g ra n u la tio n tim e: 1 - e q n . ( 1 ), A 32= 13.72 ju.m; c32 = 0.1166 m in ~ !; 2 - eqn . (1 ), A 32 = 1 3 .8 3 fim ; c32 = 0.1744 m in “ '.

from tha t given in Table 1. The values of cmn given in Table 3 for kaolin are the average of 17 or 18 experim ents which were jointly trea ted for a given dose ra te of granulation solution. It should be em phasised that a statistically much be tte r fit to eqn. ( 1 ) has been obtained for the individual runs (in term s of the correlation coefficient) than for the jo in t set of the data. However, due to the achievable reproducibility and accuracy of these experim ents, it was decided to trea t them jointly. W ith this sort of approach one can conclude that there is no real difference betw een the values of c32 and c43, though there is an effect o f the dose ra te on these coefficients — higher values are evident at higher dose rates. This conclusion will become m ore evident for the o ther m aterials studied.

Cupric hydroxychloride (HCM)For this m aterial, similar log-normal granules size

distributions were obtained, though again in some cases bim odal distributions were evident. The variations in m ean granule size with the additive content (Vanisperse

or PVP K30) are shown in Figs. 10 and 11. In the first case, the granule size will increase if the Vanisperse content exceeds 4 wt.%; however, m ore pronounced granule growth can be achieved if the la tter substance (PVP K30) is applied. It is advantageous to use more than 0.4 wt.% of this binding agent, since an almost linear increase in m ean granule size can be observed, much as for kaolin on the addition of Tamol.

Suspendibility and friability of the granulated HCM depend on the Vanisperse content and the possible addition of PVP K30 (Figs. 12 and 13). It is interesting to note that stabilization of product suspendibility occurs at a V anisperse content above about 4 wt.% , accompanied by very good m echanical p roperties of the granulated product, in term s of < 0 .1 % friability within this range of Vanisperse addition. Further addition of PVP K30 as a binding agent betw een 0.1 to 1.6 wt.% does not affect product suspendibility. However it does improve the m echanical strength of the granules since their friability drops to about 0 .0 2 % at the optimum level of 0.8 w t.% PVP K30.

The kinetics of granule growth for HCM was examined using a set of additives: 6 w t.% Vanisperse, 1.6 wt.% PVP K30, 0.5 wt.% Rokafenol N 8 and 180 cm3 of w ater at three different dose rates of granulation solution:0.060, 0.100 and 0.142 g s -1 , respectively. A n example of the growth kinetics for this m aterial is shown in Fig. 14. These data have also been approxim ated by the H arada et al. model, eqn. (1). The values of cmn and A mn are listed in Table 3. The agglomeration coefficient, cmn, is dependent upon the dosing rate, m L, and a power-law approxim ation for this dependence can be suggested:

,= b mnrńj? (6)T he following values of bmn and the exponent have been obtained: b43 = 0.677 m in - 1 and /7 = 0.951 (r = 0.9861); b 32 = 0.623 m in - 1 a n d p = 0.853 (r = 0.9965); ^5o% = 0-507 m in - 1 and /> = 0.838 (r=0.9981). These values are applicable if m L is expressed in g s -1 . It

T A B L E 3. K in e tic p a ra m e te rs o f flu id izing b e d g ra n u la tio n

T yp e o f m a te ria l

(g s -1 )A 43 ( H

CA3(m in -1 )

r ^32M

C32(m in * 1)

r Л 50%((*m)

c 50%( m i n 1)

r

K aolin 0.125 26.74 0.1237 0.9088 13.72 0.1166 0.9360( + 6% K lu tan ) 0.155 34.93 0.1554 0.9190 13.83 0.1744 0.9514

0.060 83.72 0.0481 0.9812 36.74 0.0572 0.9530 47.03 0.0485 0.9775H C M “ 0.100 122.3 0.0699 0.9797 54.39 0.0844 0.9948 56.55 0.0718 0.9822

0.142 95.24 0.1108 0.9799 60.80 0.1213 0.9610 58.24 0.1003 0.9702

0.100 125.6 0.0288 0.9764 62.34 0.0402 0.8990 79.23 0.0309 0.9858S u lfu r6 0.167 148.2 0.0526 0.9110 91.04 0.0678 0.9954 105.40 0.0378 0.8265

0.231 96.56 0.1042 0.9826 39.40 0.1647 0.9671 57.36 0.1180 0.9805

a6 w t.% V a n isp e rse CB + 1.6 w t.% P V P K30; b8 w t.% V a n isp e rse CB + 1.6 w t.% P V P K30.


142

Fig. 10. D e p e n d e n c e o f H C M g ra n u le m e a n size o n V a n isp e rse c o n te n t.

F ig. 11. D e p e n d e n c e o f H C M g ra n u le m e a n size o n P V P K30 c o n te n t.

% w t. V a n is p e rs e CB

Fig. 12. D e p e n d e n c e o f su sp en d ib ility in h a rd w a te r an d friab ility o f H C M g ra n u le s o n V a n isp e rse c o n ten t.

F ig. 13. D e p e n d e n c e o f su sp en d ib ility in h a rd w a te r a n d friab ility o f H C M g ra n u le s on P V P K 30 c o n te n t.

F ig. 14. D e p e n d e n c e o f H C M g ra n u le m e a n size on g ran u la tio n tim e fo r m L = 0.100 g s _ I .

can be concluded tha t for this m aterial some effect of the definition of the representative granule size on the agglom eration coefficient cmn is observed, and the ratio C32/C43 varies betw een 1 . 1 0 and 1 .2 1 , whereas the ratio c50%/c43 changes from 0.90 to 1.03. The values of A mn are m uch higher than the initial particle size of HCM (see Table 1) and here the same reasons as those for kaolin may be offered.

SulfurFor this m aterial, log-normal distributions of the

granule size were obtained using Vanisperse CB and0.5 w t.% Rokafenol N 8 + 150 cm3 of w ater at m L =0.142 g s -1 . The m ean size of sulfur granules varies with increasing am ounts of Vanisperse, however, the trend is different (see Fig. 15). A ddition of 1.8 wt.% PVP K30 to a mixture containing 8 wt.% Vanisperse in 136 cm3 o f w ater at the same dose rate increases the m ean size of the granules by 30%. Effects of other additives (8 wt.% Syreweal + 0.5 wt.% Rokafenol N 8;


143

F ig . 15. D e p e n d e n c e o f su lfu r g ra n u le m e a n size o n V a n isp e rse c o n te n t.

F ig . 16. D e p e n d e n c e o f su sp en d ib ility in h a rd w a te r an d friab ility o f su lfu r g ra n u le s o n V a n isp e rse c o n te n t.

8 w t.% Tam ol + 0.5 wt.% Nekal BX Dry and Lutensit A-BO ) were also examined. The largest granules were obtained using a granulating m ixture containing Tamol,and the smallest for that containing Syreweal.

Suspendibility and friability of the granulated product containing 8 w t.% V anisperse are shown in Fig. 16 asa function of the am ount o f this additive. A t 8 wt.% Vanisperse, suspendibility reaches a satisfactory value of about 80%. A significant im provem ent in friabilityis also observed in this case. F u rther addition of Tamolresulted in a very substantial im provem ent in the friability of granulated sulfur.

The kinetics of granule growth for sulfur was investigated using a com bination of 8 wt.% Vanisperse,16 wt.% PVP K30 and 0.5 w t.% Rokafenol N 8 in 180 cm3 o f w ater at th ree dose rates: 0.100, 0.167 and 0.231 g s " 1. The results (see for example Fig. 17) are approxim ated by the H arada et al. model, eqn. (1), and the num erical values are listed in Table 3. It can be concluded that this m odel is also applicable to sulfur

Fig . 17. D e p e n d e n c e o f su lfu r g ra n u le m e a n size on g ra n u la tio n tim e fo r m L = 0.100 g s -1 .

granulation, although a poorer fit to the m odel is evident (r> 0 .9 0 ). Again, the values of the agglomeration coefficient are dependent on the dose rate. However, because of the w ider deviations of the data points in this case, no attem pts to find a possible empirical relationship w ere made. Also, the values of A mn are much higher than the initial particle sizes of sulfur (cf. Table 1) for the same reasons given earlier.

Comparison with other worksThe present data can be com pared with the results

o f o ther authors, the m ost relevant data of both qualitative and quantitative character for a batch granulation process in a fluidized bed having been published in refs. 3, 6 , 7, 9, 11-13, 19, 21-23 and 27-34. M ost of these authors concluded that the granulated product obtained has a log-normal or close to it size distribution. Orm ós et al. [9, 23, 28, 29, 32-34], in a series of papers on batch granulation in a fluidized bed, w here they used different solid m aterials and different granulating agents, studied the effects of the am ount, concentration and dose rates of these solutions, as well as o ther param eters (fluidizing air superficial velocity, initial size of solid particles, bed height, application of mechanical mixing) on the size distributions and some mechanical properties of granules. These authors concluded that the m ean granulated product size is linearly dependent upon the time of the process (cf. eqn. (3)), o r alternatively, on the volum etric ratio o f the granulating liquid to the solid particles. The constants in this type of expression were found to be dependent on the type of binding agent or on its viscosity, and therefore on its tem perature and concentration.

As was pointed out earlier, eqn. (3) may be treated as a special case of eqn. (1) if ct< 1. Thus, linear time relationships were also sought for the present data points. For most of the cases, a statistically equivalent


144

fit, as in the H arada et al. model, eqn. (1), was obtained. However, the main shortcom ing of the linear model resulted from obtaining negative values for the constant term which should correspond m ore or less to the initial size of the solid m aterial (d p0), and clearly this has no physical meaning. T herefore, it can be concluded that the linear tim e-dependent m odel will not adequately describe the batch granulation process of the m aterials used in this work, and extrapolation, especially to the initial phases of the process, is unacceptable.

The kinetic data quoted by Ennis et al. [15] for a batch granulation of glass beads can also be approxim ated either by the H arada et al. m odel or by the linear tim e-dependent model, since here the condition ct< 1 is fulfilled.

Increasing am ount and/or concentration of a binding agent results in greater granule size [9, 13, 19, 21, 23, 28-34], and similar effects were observed in the present experim ents (cf. Figs. 2 ,10,11 and 15). The experim ental data of Schaefer and W 0 rts [13] for a m ixture of lactose and maize starch granulated with a gelatine solution indicate a ra ther characteristic change in granule growth trend with certain am ounts of this solution, which exhibits a variation in granulation m echanism similar to that noted in our own experim ents. A n increase in granule size at higher dose rates of the granulating solution has also been reported by Rankell et al. [36] and Schaefer and W 0 rts [7], while only Orm ós et al.[9] suggested the opposite tendency.

The dependence of granule size on size of the drops of granulating solution is of great relevance. Schaefer and W 0 rts [7], and W aldie [12] found an almost linear relationship betw een the granule size and the drop size. Ennis et al. [15] used the experim ental evidence of these authors to confirm their hypothesis that, as a drop falls onto a bed of particles, it is adsorbed onto their surface. Since particle agglom erates will not have sufficient energy to break down or coalesce, they will m aintain the drop structure, and a granule of size corresponding to that of the drop will finally form. W aldie [12] suggested a theoretical model in which it is assum ed that a granulating solution fills only a fraction of the pores in the initially-formed wet granule, thus a power correlation betw een the granule size and the drop size is m ore realistic:

d pa d dn (7)

w here « = 0.80-0.85, and thus sm aller than unity.In the present work no direct m easurem ents of the

drop sizes of the sprayed granulating solutions were undertaken. The spraying conditions applied using the two-phase nozzle corresponded to a ra ther low range of linear outlet air velocity (about 17 m s -1 ). Indirect estim ates give a drop size of betw een 200 and 300 д т diam eter, which is in satisfactory correspondence with

the relationship suggested by W aldie (eqn. (7)), if the size of granules obtained in this work is considered.

U sing the presen t experim ental data concerning kinetics of the granulation process of kaolin, HCM and sulfur, the applicability o f eqn. (2) was tested. It should be noted th a t the solvent (w ater) flowrate introduced to the bed appears on the right-hand side of this equation, thus, according to H arada et al. [3], the agglom eration coefficient depends on this param eter, and not necessarily on the total am ount of granulating solution. This suggestion had earlier been shared by Davies and G loor [6], however, the H ungarian authors [9, 23, 27-34] claim tha t in the linear tim e-dependent eqn. (3) the kinetic coefficient depends on the mass flowrate of binding agent only (not accounting for the am ount o f solvent introduced into the system). Schaefer and W 0rts [13] p resen ted figures which dem onstrate that the granule size increases with increasing total am ount of the solution.

A dependence of the agglom eration coefficient с on the ratio m j m 0 for the p resen t data and that of Schaefer and W 0 rts [13] is shown in Fig. 18. Equation (2) given by H arada et al. [19] is also plo tted in this figure. It is evident that the range of the ratio m j m 0 is much extended com pared with tha t for which eqn. (2 ) should be valid, and this also applies to the values of the agglom eration coefficient c. The present data points can be approxim ated by a correlation similar to eqn. (2):

C 3 2 = k i ( — ) - * 2 ( 8 )\m 0J

F or kaolin and HCM , к г = 6.62 and A:2 = 1.01 h - 1 ( r = 0.9390), w hereas for sulfur k t = 6.23 and k 2 = 5A9 h - 1 (r = 0.9206), and с is expressed in h ~ !). The validity range of eqn. (8) is fo rm w/mo = 0.6 to 2.3 h ~ \ A similar

Fig. 18. D e p e n d e n c e o f th e ag g lo m era tio n coeffic ien t c32 in the H a ra d a el al. m o d e l on m water/m 0: 1 - eq n . (2 ); 2 - eq n . (8) fo r k ao lin an d H C M ; 3 - eq n . ( 8) fo r su lfu r.


145

picture can be obtained if d 43 is used instead of d32. Such an evident difference betw een the data points for kaolin and HCM and sulfur can m ost likely be attributed to their properties: kaolin and H CM are hydrophilic substances, whereas sulfur is a hydrophobic m aterial.

It is also evident tha t the p resen t data points are not a continuation o f the line displaying eqn. (2 ) of H arada et al. [19]. T herefore, an extrapolation of this line beyond its validity range may lead to large errors.

No data points of the H ungarian authors have been plotted in Fig. 18, since these authors related the kinetic constants o f the granulation process to the am ount of binding agent only, and in view of the presen t discussion this seems to be unjustified.

The m echanical properties of the granulated products obtained in this work in general conform to the conclusions suggested in refs. 5, 9, 13, 21, 23 and 27-35; increasing the am ount of binding agent results in a less friable granular product, however, according to Orm ós et al. [33] there is an optim um relative concentration of binding agent, above which no further im provem ent in m echanical properties of the granules occurs.

A t present, the available range of experim ental data on the kinetics of pow der granulation in a fiuidized bed is limited, and it is still no t possible to provide a sound generalization of such inform ation on all powders and binding m aterials encountered in industrial practice. Taking into account technological and application requirem ents, it can be concluded tha t experim ental assessment o f the optim um conditions indispensable for the granulation process in a fiuidized bed (among them the proper choice and am ount of additives), as well as investigation of the process kinetics, is still necessary.

Conclusions

(1) Based on our experim ents with selected fine powders (kaolin, H CM and sulfur) it can be concluded that the fiuidized bed granulation process of these m aterials is possible and viable. U nder the appropriate process conditions granules of these m aterials having the required properties, such as granulom etric composition, suspendibility and m echanical strength, can be obtained.

(2) The most advantageous types and amounts of additives indispensable to the granulation process for the powders studied and ensuring m anufacture of WG- type products with suspendibility not less than 75% and friability below 1.6% have been determ ined. These are sum m arized in T able 4. The recom m ended amount of solvent (w ater) is from 30 to 50 wt.% with respect to dry solids.

(3) The values of the kinetic param eters describing the granule growth of the studied powders have been estim ated for dose rates of granulating solutions of optim um com position ranging from 0.06 to 0.23 g s " 1. The applicability of the H arada et al. model, eqn. (1), was tested. T he agglom eration coefficients cmn depends to some extent on the assum ed definition of the representative granule size dmn, and increases with increasing dose rates of the granulating solution. The linear tim e-dependent m odel o f the granule growth is not applicable for m ost cases of our data, since it yields physically unsound negative values for the constant term in this model.

(4) A n attem pt to generalize the available experim ental data on the agglom eration coefficient indicated severe lim itations on the eventual extrapolation of the empirical correlation, eqn. (2) of H arada et al., which couples cmn with the term m w/m 0. F u rther refinement should be m ade to explain which factor is more significant in the granulation process: the am ount of binding agent or the am ount of solvent used in the process.

List of symbols

a kinetic coefficient (s_1, m in -1 , h -1)A m„ constant in H arada et al. model (m, pm ) A r A rchim edes num ber =gdp3pG(ps - pg)/mg2bmn constant in eqn. (6) (s -1 , m in-1 , h -1 )с agglom eration coefficient in eqn. ( 1 ) (s-1 ,

m in -1 , h -1 )cmn agglom eration coefficient determ ined by drnn

(s_1, m in -1 , h -1 ) d d drop size (m, /m i)dmn m ean particle size in polydisperse mixture,

eqn. (5) (m, p,m) d p particle size (m, дгп)dpi particle diam eter of :th fraction (m, pm )

T A B L E 4. O p tim u m fo rm u la tio n s

M a te ria l D isp e rs in g a g e n t w t.% B in d e r w t.% W e ttin g w t.%ag en t

K aolin Syrew eal 4 -8 R o k a fe n o l 0 .5 -2 .0H C M V a n isp e rse C B 4 -8 P V P K 30 0 .2 -2 .0 R o k a fe n o l 0 .5 -2 .0S u lfu r V a n isp e rse C B 4 -8 P V P K 30 0 .2 -2 .0 R o k a fe n o l 0 .5 -2 .0


dpo initial particle size (m, ju,m)^ 5 0 % particle size corresponding to 50% of cu

mulative distribution (m, д т )8 acceleration due to gravity (m s “ 2)

constant in eqn. (8) (h -1 )k2 constant in eqn. (8) (h '1)m exponentmL mass flowrate of solution (or solvent) (g s - *)mw water flowrate (g s ” 1)m 0 initial mass of bed particles (g)n exponentN, num ber of particles in ith fractionP exponentr correlation coefficientR emf = rfpHmfpG//xG, Reynolds num ber a t m ini

mum fluidization velocityt time (s, min, h)^ m b minimum bubbling velocity (m s -1 , mm s ” 1)M mt minimum fluidization velocity (m s -1 , mm

s ' 1)mass fraction of additive

Greek lettersA coefficient of linear granule growthM g gas dynamic viscosity, (Pa s)Pa gas density (kg m -3 )Ps solid density (kg m -3 )

References

1 D . S eam an , Pestic. Sci., 29 (1990) 437.2 M .E . A u lto n an d M . B anks, Int. J. P harm . Tech. Prod. M a n u f,

2 (1981) 24.3 K. H a ra d a , J . F u jita a n d M . Y o sh im u ra , K agaku Kogaku, 33

(1969) 793.4 M .E . A u lto n an d M . B anks, M anuf. C hem . A eroso l News,

(1978) 50.5 U . T h u rn , M isch en , G ra n u lie re n u n d T ro c k n e n p h a rm a z e u -

tisch e r G ru n d sto ffe in h e te ro g e n e n W irb e lsch ich ten , Ph.D . Thesis, E T H Z u rich , 1970.

6 W .L . D avies an d W .T . G lo o r , J r ., J. Pharm . Sci., 60 (1971) 1869.

7 T . S c h a e fe r an d O . W 0rts , A rch . P harm . C hem . Sci. E d ., 6 (1978) 1.

8 B. D e n e s a n d Z . O rm ó s, H ung. J. Ind . C hem ., 6 (1978) 313.9 Z . O rm ó s, К . P a ta k i a n d B. C sukas, H ung. J. Ind. C hem ., I

(1973) 463.10 Z . O rm ó s a n d K. P a ta k i, H ung. J. Ind . Chem ., 7 (1979) 237.11 T . S c h a e fe r a n d O . W 0rts , A rch . P harm . C hem . Sci. E d., 6

(1978) 51.12 B. W ald ie , C hem . Eng. Sci., 46 (1989) 2781.13 T . S c h a e fe r a n d O . W 0rts , A rch . P harm . C hem . Sci. E d., 6

(1978) 14.14 B .J. E n n is , J . L i, G .I . T a d ro s an d R . P fe ffer, C hem . Eng.

Sci., 45 (1990) 3071.15 B .J. E n n is , G .I . T a d ro s an d R . P fe ffer, Pow der Technol., 65

(1991)257.16 R .C . R o w e , Int. J. P harm ., 52 (1989) 149.17 A .K . B in , Inz. A pa r. C hem ., (2 ) (1980) 25 (in P o lish).18 A .K . B iń , J . W arych a n d R . K om orow ski, P ow der Technol.,

41 (1985) 1.19 K. H a ra d a , K agaku K ogaku, 36 (1972) 1237.20 H .J . N iessen , H . H a u s m a n n an d H . N e u m a ie r , Pesticide C h e m

istry, H u m a n W elfare a n d E nvironm ent, V ol. 4, IU P A C , P er- g am o n , O x fo rd , 1983, p. 317.

21 W .L . D av ies a n d W .T . G lo o r , J r ., J. Pharm . Sci., 61 (1972) 618.

22 W .L . D av ies a n d W .T . G lo o r , J r ., J. Pharm . Sci., 62 (1973) 170.

23 Z . O rm ó s, К . P a ta k i a n d B. C sukas, Hung. J. Ind. Chem ., 7 (1973) 307.

24 R . Jo h n so n , J .E . R e e s an d F . S en d a ll, J. Pharm . Pharm acol., 27, S upp l. (1975) 80P.

25 A .R . G u p te , Pharm . Ind ., 35 (1973) 17.26 A . B a n k , D . B ezzegh a n d P. F e k e te , Proc. 2 n d Conf. A ppl.

Phys. C hem ., 2 (1971) 687.27 Z . O rm ó s, H ung. J. Ind . C hem ., 7 (1979) 153.28 Z . O rm ó s, К . P a ta k i an d B. C sukas, Hung. J. Ind. Chem ., 1

(1973) 475.29 Z . O rm ó s, В. C su k as a n d K. P a ta k i, Hung. J. Ind. Chem ., 3

(1975) 193.30 Z . O rm ó s, В. C su k as an d K. P a ta k i, Hung. J. Ind. Chem ., 3

(1975) 631.31 Z . O rm ó s a n d K. P a ta k i, Hung. J. Ind. C hem ., 7 (1979) 89.32 Z . O rm ó s a n d K. P a ta k i, Hung. J. Ind. Chem ., 7 (1979) 105.33 Z . O rm ó s, К . P a ta k i a n d B. S tefko , Hung. J. Ind. C hem ., 7

(1979) 131.34 Z . O rm ó s, К . P a ta k i a n d B. S tefko , Hung. J. Ind. C hem ., 7

(1979) 141.35 G .A . B ell, Pestic. Sci., 29 (1990) 467.36 A .S . R a n k e ll, M .W . S co tt, H .A . L ie b e rm an , F.S. C how an d

J.V . B a ttis ta , J. Pharm . Sci., 53 (1964) 320.


application of the fluidized bed process for formulation

Documents