European Journal of Pharmaceutical Sciences 20 (2003) 273283

A new process control strategy for aqueous lm coating of pellets in uidised bed
Crilles C. Larsen a,b, , Jrn M. Sonnergaard b , Poul Bertelsen c , Per Holm d
b a Process Technology Development, Ferring A/S, Indertoften 10, DK-2720 Vanlse, Denmark Department of Pharmaceutics, The Danish University of Pharmaceutical Sciences, Universitetsparken 2, DK-2100 Copenhagen, Denmark c Pharmaceutical Development, Nycomed Denmark A/S, Langebjerg 1, DK-4000 Roskilde, Denmark d LifeCycle Pharma A/S, Carl Jacobsens Vej 22, DK-2500 Valby, Denmark

Received 19 March 2003; received in revised form 8 July 2003; accepted 14 July 2003

Abstract The parameters with effect on maximum spray rate and maximum relative outlet air humidity when coating pellets in a uidised bed were investigated. The tested variables include type of water based modied release lm coating (Eudragit NE 30D, Eudragit RS 30D, Aquacoat ECD ) coating principle (top spray, bottom spray), inlet air humidity and type of pellets (sugar spheres, microcrystalline cellulose pellets). The maximum spray rate was not inuenced by the coating principles. The highest spray rate was obtained for the lm polymer with the lowest tackiness which is assumed to be the controlling factor. The type of pellets affected the maximum spray rate. A thermodynamic model for the coating process is employed throughout the process and not just during steady state. The thermodynamic model is incorporated into a new process control strategy. The process control strategy is based on in-process calculation of degree of utilisation of the potential evaporation energy (DUE) of the outlet air and the relative outlet air humidity (RH). The spray rate is maximised using set points of DUE and RH as control parameters. The product temperature is controlled simultaneously by regulating the inlet air temperature. 2003 Elsevier B.V. All rights reserved.
Keywords: Fluidised bed; Aqueous coating; Maximum spray rate; Maximum relative outlet air humidity; Thermodynamic model; Process control

1. Introduction Traditionally, the control of a coating process has been based on set points for the most critical process variables combined with the practical experience of the operator. The critical process variables include the spray rate, the process airow, the atomising airow, the inlet air temperature and the inlet air humidity. By employing the thermodynamics of the coating process the process variables can be analytical processed and more sophisticated process control methods can be developed. Until now thermodynamics have been used to describe and control the mass and energy balance of the coating process under steady-state conditions. Ebey (1987) presented a thermodynamic model which described an aqueous lm coating process. In the presented

Corresponding author. Tel.: +45-38765400; fax: +45-38765411. E-mail address: crilles.casper.larsen@ferring.com (C.C. Larsen).

model the outlet air humidity and outlet air temperature was calculated under steady-state conditions. This model did not account for effects of the atomising air nor energy loss to the surroundings. Furthermore, Ebey (1987) presented an environmental equivalency (EE) factor to give a relative measure of the rate at which water evaporates from the surface of a tablet bed. As long as EE is kept constant any critical process parameter can be modied without inuencing the lm coating quality. Reiland et al. (1983) dened a vaporisation efciency (Es ), which was dened as the actual spray rate divided by the theoretical maximum spray rate calculated under steady-state conditions. The theoretical maximum spray rate was calculated from the saturation humidity of the exhaust air. This model did not account for effects of the atomising air but did employ an energy loss to the surroundings of 5%. Reiland et al. showed that the tackiness of the lm formulation had a signicant effect on the magnitude of the Es . Kage et al. (1996) dened an index R calculated under steady-state conditions similar to the vaporisation

0928-0987/$ see front matter 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0928-0987(03)00206-9

274

C.C. Larsen et al. / European Journal of Pharmaceutical Sciences 20 (2003) 273283

efciency (Es ) introduced by Reiland et al. Kage et al. employed the humidity instead of the spray rate to estimate the vaporisation efciency. This model did not account for effects of the atomising air nor energy loss to the surroundings. Dewettinck et al. (1999) presented a computer model for modelling the steady-state thermodynamics of an aqueous lm coating in uidised bed. The model included the effects of the atomising air and energy loss to the surroundings. Dewettinck et al. estimated a correction factor of 0.918 between their model and the ideal theoretical adiabatic conditions. To assume steady-state requires that all critical process variables are kept constant which is not the case during the critical phases in a coating process like the start up and when the spray rate is altered. In this investigation the maximum spray rate and maximum relative outlet air humidity are identied for the following critical variables: lm coating formulation, coating principle, inlet air humidity and type of substrate. Furthermore, an optimal process control is applied during the described investigation. This process control involves the use of an alternative thermodynamic factor based on enthalpies of the actual and adiabatic outlet air conditions during coating. The factor is applied throughout the process and not just during steady state. Finally, a process control strategy is presented for the control of the aqueous lm coating process in general.

In a coating process true steady state cannot be achieved as dry matter is applied throughout the process with subsequent increase in the mass of the product. Furthermore, as a uidised bed is not a totally isolated system heat is transferred by convection and conduction to the surroundings and the specic enthalpy is therefore lower than what is calculated from adiabatic conditions. The enthalpy balance of a coating process can be expressed as hin m g + hl m l + hsi m s + hmix m l d ms loss + d(hs ms ) ; m = hout m g+Q s = dt dt and implies g + hl m l + hsi m s + hmix m l hin m loss + hs m = hout m g+Q s

(1a)

(1b)

2. Thermodynamic calculations During coating in a uidised bed a water based lm coating is sprayed onto the pellets bed creating a wet surface on the pellets. The rate of evaporation from this wet surface to the surrounding air is determined by the difference between the water vapour pressure at the liquid surface and the water vapour pressure in the surrounding air. In an enclosed space, evaporation continues until the two vapour pressures are equal, i.e. until the air is saturated with water at the same temperature as the surface. However, if unsaturated air is constantly supplied like in the uidised bed, the wet surface will reach an equilibrium temperature at which the cooling effect due to the evaporation is equal to the heat transfer to the liquid by conduction and convection from the air. In the ideal evaporation process (adiabatic process) the heat is transferred to the liquid and the specic enthalpy of the air is nearly constant as no energy loss to the surroundings occurs. The deviation from constant enthalpy is due to difference in temperature of the added water and of the product. If more liquid is added than can be evaporated by the process air over wetting and subsequently agglomeration of the pellets can occur. This is caused by build-up of surplus water which cannot be evaporated due to insufcient energy supply.

l the mass liquid ow, m where m g is the mass airow, m s the mass solid ow, hin the enthalpy of the inlet air, hl the enthalpy of the water, hsi the enthalpy of the solid in the spray, hmix m l the mixing enthalpy of water and solid, hout loss the energy loss to the the enthalpy of the outlet air, Q surroundings, d(hs ms )/dt the increase in enthalpy of the solid in the uidised bed and hs is the enthalpy of the solid loss is positive in a process where in the uidised bed. Q the surroundings have a lower temperature than the product temperature of the process. If the surroundings have a higher temperature than the product temperature of the process an inux of energy to the process is seen instead of an efux loss becomes negative. from the process and Q In this investigation, it is presupposed that contribution from the mixing enthalpy of water and solid along with the enthalpy of the solid in the spray and in the uidised bed are negligible, thereby Eq. (1b) can be rewritten as Eq. (2). loss hin m g + hl m l = hout m g+Q (2)

A more simplied model using the Mollier diagram (Fig. 1) can be used to explain the difference between an adiabatic and a non-adiabatic evaporation process. In Fig. 1, the adiabatic process is marked by the numbers 13 showing an ideal process where all the energy is used to evaporate water. The non-adiabatic process is marked by the numbers 1, 2 and 4 showing the actual process where not all of the energy is used for evaporation due to heat loss to the surroundings. For both processes the inlet air is heated from room temperature to a specied higher temperature corresponding to the line 1 to 2. During heating the enthalpy of the air is increased and as humidity is constant the line is vertical. The heated air is passed over the wet surface (pellets). In the adiabatic process the specic enthalpy of the air is close to constant as the increase in humidity is compensated by a decrease in temperature corresponding to the line 2 to 3. In the non-adiabatic process, the specic enthalpy of the

C.C. Larsen et al. / European Journal of Pharmaceutical Sciences 20 (2003) 273283

275

Fig. 1. Schematic Mollier diagram of the adiabatic (13) and non-adiabatic (1, 2 and 4) enthalpy conditions in a vaporisation process.

air is not constant, as the increase in humidity is compensated by a decrease in temperature plus heat loss to the surroundings. This is visualised by the dotted line 2 to 4. In a coating process the ambient air temperature corresponds to position 1, the inlet air temperature to position 2 and the outlet air temperature to position 4. Position 3 (adiabatic outlet air temperature) along with the enthalpies have to be calculated. The specic enthalpy h can be calculated according to Eq. (3). h = (cg + Xcv )(T T0 ) + 0 X (3) where T0 is the reference temperature, here 0 C. cg is the mean specic heat for air containing no moisture (dry air) at a given temperature range T0 to T, and cv is the mean specic heat for water vapour at a given temperature range T0 to T. In the temperature range 0100 C, cg and cv can be considered as constants with cg = 1.01 kJ/(kg K) and cv = 1.87 kJ/(kg K). 0 is the latent heat of vaporisation for water at T0 (0 = 2501 kJ/kg). X is the absolute humidity in kg water per kg dry air.

To calculate the specic enthalpy of the outlet air for the non-adiabatic conditions, employing Eq. (3), the outlet air temperature and absolute outlet air humidity are known. The outlet air temperature is measured and the absolute outlet air humidity is calculated by adding the measured absolute inlet air humidity to the measured water addition determined from the spray rate. It is a key factor to use the spray rate as basis for the calculation of the outlet air humidity since a direct measurement of outlet humidity would not respond to overwetting of product caused by a too high spray rate since the surplus of water would not evaporate. Furthermore, a more accurate value of the outlet air humidity is obtained using the precise measurement of the inlet air humidity as basis. This is due to delayed response of the outlet air humidity sensor when changing spray rate compared to the constant humidity level of the inlet air. To calculate the specic enthalpy of the adiabatic conditions the absolute outlet air humidity is known while the outlet air temperature is unknown. Combining the following mass balance for water (Eq. (4)) and enthalpy balance (Eq. (5)) the specic enthalpy of the adiabatic conditions

276

C.C. Larsen et al. / European Journal of Pharmaceutical Sciences 20 (2003) 273283

can be calculated according to Eq. (4). m g Xin + m l =m g Xout Xout Xin = m l m g (4)

where Xin is the absolute inlet air humidity and Xout is the absolute outlet air humidity. hin m g + hl m l = hout-adia m g hout-adia = hin + m l m g hl (5) where hout-adia is the enthalpy of the outlet air under adiabatic conditions. The enthalpy of the water can be calculated by multiplying the mean specic heat for water cl at a given temperature range T0 to T1 with the difference (T1 T0 ) hout-adia = hin + (Xout Xin ) cl (Tl T0 ) (6)

The lm coating formulations were applied to either sugar spheres (Nu-pareil PG sugar spheres NF; mesh 2025, DMV, The Netherlands) or microcrystalline cellulose pellets (Celphere CP-507 JPE; mesh 2535, Asahi Kasei Corp., Japan). The pellets were characterised by particle size distribution expressed as geometric-weight mean diameter (dgw ) and geometric standard deviation (sg ), granular density (Ph.Eur.) and specic surface area according to the methods described in a previous article (Larsen et al., 2003). The data are summarised in Table 1. 3.2. Equipment The coating process was performed in a uidised bed (MP1 ; Aeromatic-Fielder AG, Switzerland). A top-spray coating insert and a bottom spray coating insert (Precision coater ) (Walter, 1997, 1998) were used. A two-uid spray nozzle was used in all coating inserts. The set-up of the uidised bed with the sorption dehumidier and the individual measuring and control points is shown in Fig. 2. The following process parameters were recorded to a digital chart recorder with intervals of 5 s. Absolute inlet air humidity (g water/kg dry air). Inlet air temperature ( C). Product temperature ( C). Outlet air temperature ( C). Mass airow (kg/h; In-Flow F106; Bronkhorst High-Tech B.V., The Netherlands). Mass liquid ow (g/min; Liqui-Flow L30; Bronkhorst High-Tech B.V., The Netherlands). The atomising airow was monitored but not recorded. A sorption dehumidier (ML270; Munters A/S, Denmark) was connected to the inlet side of the uidised bed to control the inlet air humidity. The lower capacity for the sorption dehumidier was 2 g water/kg dry air at 490 kg/h airow corresponding to a dew point temperature of 8.4 C. The dehumidied air was passed through a cooling surface to lower the temperature of the air before entering the uidised bed. The temperature of the dehumidied air after cooling was monitored but not recorded to the digital chart recorder. The uidised bed was equipped with a feedback system between product and inlet air temperature. The product temperature was set to a given temperature and the PLC then
Table 1 Physical characteristics for the employed pellets Type of pellets Nu-pareil Particle size distribution, dgw (m) sg Granular density (Ph.Eur.) (g/cm3 ) Specic surface area (cm2 /cm3 ) 804 1.14 1.457 74.2 Celphere 620 1.16 1.523 98.3

Having calculated hout-adia the adiabatic outlet air temperature can be calculated by rearranging Eqs. (3)(7). Tout-adia = hout-adia 0 Xout T0 cg + Xout cv (7)

2.1. Degree of utilisation of the potential evaporation energy (DUE) of the outlet air The maximum utilisation of the evaporation energy of the outlet air is lower than what is calculated for the adiabatic evaporation process due to heat loss to the surroundings as described above. The degree of utilisation of the potential evaporation energy (DUE) of the outlet air is calculated according to Eq. (8). DUE (%) = hout href 100; hout-adia href href = 0 kJ/kg (8)

href is the enthalpy at the reference state, here dry air on gas form and liquid water at 1013 mbar and 0 C. If DUE exceeds 100% the coating process gets critical resulting in a potential risk of over wetting of the coated pellets.

3. Materials and methods 3.1. Materials The lm polymers used as aqueous dispersions were Aquacoat ECD (FMC, USA), Eudragit NE 30 and Eudragit RS 30D (Rhm Pharma, Germany). Triethyl citrate (Citroex-2 ; Morex Inc., USA) and glycerol monostearate (Imwitor 900; Sasol Germany GmbH, Germany) were used in the lm coating formulations as plasticiser and glidant, respectively. Polyoxyethylenesorbitan monooleate 80 (Tween 80; SigmaAldrich, USA) was used to enhance the dispersability of glycerol monostearate in water. Di-methyl siloxan (Struktol SB2071; Schill & Seilacher GmbH, Germany) was used as an anti-foam agent in the lm coating formulations.

C.C. Larsen et al. / European Journal of Pharmaceutical Sciences 20 (2003) 273283

277

Fig. 2. PI diagram for the set-up of the uidised bed MP1 with the individual measuring and control points. TI: temperature censor; MI: moisture censor; FI: ow censor; PI: pressure censor.

regulated the inlet air temperature to maintain the preset product temperature. 3.3. Methods 3.3.1. Film coating formulations The compositions of the lm coating formulations shown in Table 2 were chosen in accordance with the recommendation given by the manufactures. The lm coating formulations based on Eudragit NE 30D and RS 30D were made in two steps. A mixture of water, Imwitor 900, Tween 80, Struktol SB2071 and Citroex-2
Table 2 Compositions of the employed lm coating formulations Film coating formulation (% (w/w)) Eudragit NE 30D Eudragit NE 30D Eudragit RS 30D Aquacoat ECD Citroex-2 Imwitor 900 Tween 80 Struktol SB2071 Water Dry matter total (%) Film polymer (%) 50.00 50.00 3.00 0.75 0.30 0.15 45.80 19.10 15.00 50.00 3.00 Eudragit RS 30D Aquacoat ECD

for the Eudragit RS 30D was heated to 75 C. The suspension was subsequently homogenised on a high-speed homogeniser (Ultra-Turrax T25; Jahnke and Kunkel, Germany) at 20,000 rpm for 20 min and cooled to room temperature. The Eudragit dispersion was gently mixed with the cooled homogenised suspension. The Eudragit RS 30D lm coating formulation was plasticised by gently stirring for 1 h with a paddle mixer before use (Petereit et al., 1995). The lm coating formulation based on Aquacoat ECD was made by plasticising the Aquacoat ECD suspension with Citroex-2 for 1 h by gently stirring with a paddle mixer. Water was added until the desired polymer concentration of 15.0% was reached. 3.3.2. Process conditions The load of uncoated pellets was 3000 g in all experiments. The inlet mass airow of the uidised bed was in the range of 7090 kg dry air/h. The atomising air pressure was set to 2 bar (4.5 kg dry air/h). The product temperature was preset to 21, 25 and 35 C for the experiments utilising the Eudragit NE 30D, Eudragit RS 30D and Aquacoat ECD lm coating formulations, respectively, as recommended by the manufactures. The inlet air temperature was set with no maximum except for the upper limit of the heater in the uidised bed at 120 C. The spray rate was in the rst hour of coating xed to 4 g/min in the rst 30 min and 6 g/min in the next 30 min. This was done, to avoid excessive water uptake in the core for the microcrystalline cellulose pellets and to avoid

0.75 0.30 0.15 48.80 16.10 15.00

47.00 18.00 15.00

278

C.C. Larsen et al. / European Journal of Pharmaceutical Sciences 20 (2003) 273283

dissolving the sugar contained in the sugar spheres. After the rst hour the spray rate was increased every 1530 min until the maximum spray rate was reached (cf. Section 3.3.4). 3.3.3. Factorial design investigation The experimental design was based on a full 3 2 2 factorial design with lm coating formulation on three levels and coating insert along with inlet humidity on two levels. A total of 14 experiments were performed in the factorial design investigation including replication of two separate experiments. To substantiate a hypothesis concerning the inuence of the sugar spheres on the maximum spray rate two additional experiments were completed with microcrystalline cellulose pellets instead of sugar spheres. Maximum spray rate and maximum relative outlet air humidity were used as response variables in the factorial design investigation. The statistical analyses were performed in Statgraphics Version 5 (Manugistics Inc., USA) for Windows. P < 0.05 was set as the statistical signicance level. 3.3.4. Estimation of the maximum spray rate and relative outlet air humidity The maximum spray rate was dened as the highest possible spray rate at which a continuous product ow could still be maintained, as pauses in the product ow increases the risk for agglomeration of coated pellets. To ease the comparison of effects, the maximum spray rate was normalised relative to the process mass airow. The spray rate was expressed in grams of water per kilogram of dry air rather than in grams of lm coating formulation per kilogram of dry air. Since the spray rate was expressed as the water spray rate the difference in polymer content between formulations did not affect the analysis and mass/enthalpy balance calculations.
Table 3 Results from the individual experiments Coating insert Precision coater Precision coater Precision coater Precision coater Precision coater Precision coater Top spray Top spray Top spray Top spray Top spray Top spray Precision coater Precision coater Precision coater Precision coater Film coating formulation Aquacoat ECD Aquacoat ECD Eudragit RS 30D Eudragit RS 30D Eudragit NE 30D Eudragit NE 30D Aquacoat ECD Aquacoat ECD Eudragit RS 30D Eudragit NE 30D Eudragit NE 30D Eudragit RS 30D Eudragit RS 30D Eudragit NE 30D Eudragit NE 30D Eudragit NE 30D Inlet air humidity (g water/kg dry air) 2 6 2 6 2 6 2 6 6 2 6 2 2 6 6 2

The maximum relative outlet air humidity (RHmax) was calculated on basis of the maximum spray rate as the highest possible relative outlet air humidity which did not pause product ow.

4. Results and discussion In Table 3, the identied maximum spray rates and the maximum relative outlet air humidities from the individual experiments are summarised. In all experiments, the coating efciency expressed as amount of coating applied (yield of coating) and amount of agglomerated pellets larger than 1000 m were quantied. The coating efciency (average S.D.) was 89.5% 9.8% and the amount of agglomerated pellets was 1.9% 2.1% and the processes were therefore assessed to have been completed satisfactorily. A visual inspection of the coating insert for any residue of product sticking to the container was performed at the end of each experiment. The product was free owing in all experiments and no residues were seen. 4.1. Maximum spray rate The ANOVA showed that lm coating formulation, inlet air humidity and type of pellets had a statistically signicant effect on the maximum spray rate. The lm coating formulation had a statistically signicant effect (P < 0.001). A multiple range test at the 99.9% condence level using the StudentNewmanKeuls method demonstrated that the three different lm coating formulations had statistically different maximum spray rates. The maximum spray rate for the Aquacoat ECD based lm coating formulation, with an average of 22.9 g water/kg dry air, was signicantly higher than both of the Eudragit based lm coating formulations. The Eudragit RS 30D had an

Type of pellets Nu-pareil Nu-pareil Nu-pareil Nu-pareil Nu-pareil Nu-pareil Nu-pareil Nu-pareil Nu-pareil Nu-pareil Nu-pareil Nu-pareil Nu-pareil Nu-pareil Celphere Celphere

Maximum spray rate (g/kg dry air) 29.2 20.2 8.5 8.9 5.1 2.4 21.1 21.2 9.7 5.2 2.4 12.3 12.1 1.8 8.3 10.4

Maximum relative outlet air humidity (%) 85.2 73.1 54.9 71.1 44.0 53.3 71.0 84.0 75.1 44.1 54.6 71.2 68.2 46.3 83.2 73.0

Critical water spray rate normalised in proportion to the process mass airow.

C.C. Larsen et al. / European Journal of Pharmaceutical Sciences 20 (2003) 273283

279

average maximum spray rate of 10.0 g water/kg dry air and the Eudragit NE 30D had an average maximum spray rate of 3.4 g water/kg dry air. Wesseling et al. (1999) have set-up an ascending order of tackiness for lm coating formulations based on the following aqueous polymer dispersions; Aquacoat ECD < Eudragit RL 30D < Eudragit RS 30D < Eudragit NE 30D. This sequence supports the ndings in this investigation that a lower tackiness is reected in a higher maximum spray rate. Inlet air humidity had a statistically signicant effect (P < 0.05) on maximum spray rates. When the inlet air humidity was set at 2 g water/kg dry air (actual readout 2.1) the average maximum spray rate was 13.7 g water/kg dry air whereas when the inlet air humidity was set at 6 g water/kg dry air (actual readout 5.7) the average maximum spray rate was 10.7 g water/kg dry air. The difference in maximum spray rate between the to levels of inlet air humidity was 3.0 g water/kg dry air which was close to the actual difference in inlet humidity of 3.6 g water/kg dry air at the two levels. The maximum spray rate for the microcrystalline cellulose pellets coated with Eudragit NE 30D was 9.4 g water/kg dry air in average, and considerably higher than the maximum spray rate for the coating of sugar

spheres, with an average of 3.4 g water/kg dry air. The observed difference in maximum spray rate for the two types of pellets is most likely due to the water soluble sugar contained in the sugar spheres which increases the tackiness of the lm coating and thereby decreases the maximum spray rate. 4.2. Maximum relative outlet air humidity The ANOVA showed that lm coating formulation had a statistically signicant effect on the RHmax (P < 0.001). A multiple range test at the 99.9% condence level using the StudentNewmanKeuls method demonstrated that the Eudragit NE 30D (48% RH) lm coating formulation differed signicantly from the Aquacoat ECD (78% RH) and Eudragit RS 30D (69% RH) lm coating formulations. Although the Aquacoat ECD lm coating formulation could be applied by a signicantly higher spray rate than the Eudragit RS 30D lm coating formulation no signicant difference in RHmax could be found between the two lm coating formulations. The reason for this was the different product temperatures and hence outlet temperatures employed in the two situations. The Aquacoat ECD lm coating formulation was applied at a product temperature of 35 C whereas the Eudragit RS 30D lm coating formulation was applied at 25 C.

Fig. 3. Recorded process data and calculated utilisation of the potential evaporation energy for an Aquacoat ECD coating in precision coater insert at 2 g/kg dry air inlet humidity on Nu-pareil pellets.

280

C.C. Larsen et al. / European Journal of Pharmaceutical Sciences 20 (2003) 273283

The RHmax for the coating of microcrystalline cellulose pellets with Eudragit NE 30D, with an average of 78%, was signicantly higher than the RHmax for the coating of sugar spheres, with an average of 48%. This is in accordance with the result on maximum spray rate explained previously. The two different levels of inlet air humidity did not affect the maximum relative outlet air humidity. As shown above, a decrease in inlet air humidity results in an increase in spray rate and thereby maintaining a constant RHmax for the two tested levels of inlet air humidity. A maximum relative outlet air humidity threshold can be set as an upper limit for the maximum spray rate for any given coating system. The upper limit is mainly depending on the tackiness of the coating system. Changes in inlet air humidity do not signicantly affect this threshold but do affect the maximum spray rate. 4.3. Computation on the degree of utilisation of the potential evaporation energy (DUE) of the outlet air The logged data was processed in a spreadsheet (Excel ; Microsoft Corp., USA) and adiabatic and non-adiabatic process conditions were calculated and evaluated. DUE was computed using equation 8. Fig. 3 shows the logged and calculated process data for one Aquacoat ECD based coating. The set points for the product temperature were maintained

throughout the experiments and no signicant differences between product temperature and outlet temperature were seen. Therefore, only the outlet temperature is shown in the following gures. Each time the spray rate was increased the mass and energy balance of the process was disturbed as the evaporation energy was insufcient to compensate the increase in spray rate. This was indicated by an utilisation above 100% for the DUE. The increased spray rate causes a decrease in product temperature if unattended. If the spray rate is increased too much the dew point of the outlet air could be reached leading to over wetting with subsequent agglomeration of the product. As shown in Fig. 3, an increase in spray rate was followed by an almost simultaneous increase in inlet air temperature in order to keep the product temperature constant. This regulation was controlled by the feedback system between product temperature set point and inlet air temperature. The data shown in Fig. 3 indicates that the process was in control. When the spray rate was increased during coating the DUE approached the 100% threshold. The 100% threshold was not exceeded for longer periods at any time. The energy loss to the surroundings was close to being constant throughout the experiment, however, as the inlet air temperature rises so does the available energy. Consequently, the difference between non-adiabatic and adiabatic conditions was decreased throughout the process.

Fig. 4. Recorded process data and calculated utilisation of the potential evaporation energy for a Eudragit RS 30D coating in top-spray insert at 2 g/kg dry air inlet humidity on Nu-pareil pellets.

C.C. Larsen et al. / European Journal of Pharmaceutical Sciences 20 (2003) 273283

281

Fig. 5. Recorded process data and calculated utilisation of the potential evaporation energy for a Eudragit NE 30D coating in top-spray insert at 2 g/kg dry air inlet humidity on Nu-pareil pellets.

In Fig. 4, logged and calculated process data from one Eudragit RS 30D based coating is shown. The coating process shown in Fig. 4 was in control similar to the coating process shown in Fig. 3. The DUE was only above the 100% threshold for short periods of time. However, the DUE was closer to 100% threshold throughout the experiment compared to the coating process shown in Fig. 3. The experiments utilising the Aquacoat as lm polymer operates with a product temperature of 35 C whereas the experiments utilising the Eudragit RS as lm polymer operates with a product temperature of 25 C. The small difference in temperature between equipment and surroundings in the latter results in less energy lost to the surroundings and therefore the non-adiabatic conditions is closer to the adiabatic. In Fig. 5, logged and calculated process data from one Eudragit NE 30D based coating is shown. Throughout the experiment illustrated in Fig. 5, the DUE was close to 100%. Unlike the other experiments based on Aquacoat and Eudragit RS the DUE in the Eudragit NE based experiments were above the 100% threshold. The reason for this was probably the fact that the room temperature was close to or above the product temperature with a difference of a few degrees centigrade, possibly also the overall accuracy of the measurement of inlet air humidity, inlet air ow and spray rates affects the calculation of DUE.

4.4. Process control strategy The maximum spray rate when coating with aqueous based polymer dispersions was demonstrated to be determined by the maximum relative outlet air humidity (RHmax) for the specic lm coating formulation presuming the degree of utilisation of the potential evaporation energy of the outlet air did not exceed the 100% threshold and that the product temperature Tp was within the required limits for the specic lm coating formulation. These three sets of process control parameters are basis for a new process control strategy as shown schematically in Fig. 6. The strategy is based on two separate control loops including an inner control loop controlling the product temperature by the inlet air temperature. An outer control loop controlling the relative outlet air humidity (RH) and the degree of utilisation of the potential evaporation energy by the spray rate. If the product temperature Tp changes as a result of a change in spray rate the inlet air temperature Ti is altered to counteract this change. The outer control loop is based on set points, for RHmax(s) and DUEmax(s). As shown in Fig. 6 the spray rate is increased as long as the actual relative outlet air humidity is below RHmax(s) under the prerequisite that the

282

C.C. Larsen et al. / European Journal of Pharmaceutical Sciences 20 (2003) 273283

Fig. 6. Process control strategy ow diagram containing an inner and an outer control loop. The inner loop controls the product temperature (Tp ) by regulating the inlet air temperature (Ti ). The outer loop controls the relative outlet humidity (RH) and the utilisation of the potential evaporation energy (DUE) by regulating the spray rate. Tp (s): set point for the product temperature; RHmax: set point for the maximum relative outlet air humidity; DUEmax: set point for the maximum utilisation of the potential evaporation energy.

degree of utilisation of the potential evaporation energy is below DUEmax(s). The DUEmax(s) would logically be set to 100%. If the actual relative outlet air humidity or the energy utilisation is above their respective set points the spray rate is decreased to avoid uncontrolled coating. By employing the DUE factor for regulating the spray rate a fast response is achieved which is critical to prevent over wetting in coating. Traditionally, employing the inlet air temperature as a regulator when changing spray rate is not optimal due to the time delay before mass and energy balance is re-established.

5. Conclusion The maximum spray rate for Eudragit NE 30D, Eudragit RS 30D and Aquacoat ECD based lm formulations were not inuenced by the coating principles comparing topand bottom-spray precision coater. The maximum spray rate was determined by the maximum relative outlet air humidity (RHmax) obtainable for the specic coating formulation. The levels of RHmax were in the same order as the tackiness of the polymer; Aquacoat ECD < Eudragit RS 30SD < Eudragit NE 30D. The highest maximum spray rate was obtained for the lm polymer with the lowest tackiness, which is assumed to be the controlling factor.

The inlet air humidity did not affect RHmax as a decrease in inlet air humidity allowed an increase in maximum spray rate. The type of pellets affected the maximum spray rate in case of Eudragit NE 30D. Higher maximum coating spray rate was obtained for microcrystalline cellulose pellets compared to sugar spheres. A process control strategy was suggested, maximising the coating spray rate for aqueous modied release lm coating and keeping the process in mass and energy balance. The process control is based on in-process calculation of the degree of utilisation of the potential evaporation energy and out let air humidity (RH). The spray rate is maximised using set points of DUE and RH as control parameters. The inlet air temperature simultaneously controls the product temperature. The set points of the RH and product temperature have to be determined for the specic coating formulation. This new coating control strategy is assumed to be applicable in aqueous coating in general. Furthermore, the assessment of maximum spray rate when upscaling will minimise the number of trials which have to be made in production scale. Acknowledgements H. Lundbeck A/S is acknowledged for the nancial support to the research project.

C.C. Larsen et al. / European Journal of Pharmaceutical Sciences 20 (2003) 273283

283

References
Dewettinck, K., De Visscher, A., Deroo, L., Huyghebaert, A., 1999. Modeling the steady-state thermodynamic operation point of top-spray uidized bed processing. J. Food Eng. 39, 131 143. Ebey, G.C., 1987. A thermodynamic model for aqueous lm-coating. Pharm. Technol. 11 (4), 4050. Kage, H., Takahashi, T., Yoshida, T., Ogura, H., Matsuno, Y., 1996. Coating efciency of seed particles in a uidized bed by atomization of a powder suspension. Powder Technol. 86, 243 250. Larsen, C.C., Sonnergaard, J.M., Bertelsen, P., Holm, P., 2003. Validation of an image analysis method for estimating coating thickness on pellets. Eur. J. Pharm. Sci. 18 (2), 191196.

Petereit, H.U., Assmus, M., Lehmann, K., 1995. Glyceryl monostearate as a glidant in aqueous lm-coating formulations. Eur. J. Pharm. Biopharm. 41 (4), 219228. Reiland, T.L., Seitz, J.A., Yeager, J.L., Brusenback, R.A., 1983. Aqueous lm-coating vaporization efciency. Drug Dev. Ind. Pharm. 9, 945 958. Walter, K., 1997. Coating of particulars freely suspended in a gas stream. In: Bourgin, P. (Ed.), Proceedings of the Euromech 367 on Fluid Mechanics of Coating Processes. Universiti Louis Pasteur, Strasbourg, France, pp. 495504. Walter, K., 1998. Apparatus for coating solid particles. US Patent 5,718,764, Aeromatic-Fielder AG. Wesseling, M., Kuppler, F., Bodmeier, R., 1999. Tackiness of acrylic and cellulosic polymer lms used in the coating of solid dosage forms. Eur. J. Pharm. Biopharm. 47 (1), 7378.