To cite this article: Mary Tanya am Ende & Alfred Berchielli (2005) A Thermodynamic Model for
Organic and Aqueous Tablet Film Coating, Pharmaceutical Development and Technology, 10:1,
47-58, DOI: 10.1081/PDT-35915
Download by: [Copyright Clearance Center] Date: 03 May 2017, At: 00:11
Pharmaceutical Development and Technology, 1:47–58, 2005
Copyright D 2005 Taylor & Francis Inc.
ISSN: 1083-7450 print / 1097-9867 online
DOI: 10.1081/PDT-200035915
47
Order reprints of this article at www.copyright.rightslink.com
48 M. T. am Ende and A. Berchielli
Table 1
Overview of published film coating models
Reference Mass balance Energy balance Heat transfer Aqueous Organic Coating pan
p p
Reiland, Seitz, Accela-Cota 24
Yeager,
Brusenback 1983[1]
p p
Stetsko, Banker, Accela-Cota 48
Peck 1983[2]a
p p p p
Ebey 1987[3]b Accela-Cota
p p p
Liu, Litster 1993[4] Fluid bed coating
p p p
Rodriguez et al. 1996[5] Pelligrini
p p p p
am Ende, Berchielli, Vector LDCS-20,
Chidlaw 2001[6] HCT-30,HCT-60
and HC-130L;
Compulab 36;
Accela-Cota 48
a
Ph.D. research funded by Thomas Engineering.
b
Thermodynamic Analysis of Aqueous Coating, or TAAC, computer model by Thomas Engineering.
energy balances are performed. The inlet streams consist are depicted in Figure 1 to distinguish between the flow
of the drying air, the film-coating solution sprayed from through the perforated pan for the Thomas Engineering
the nozzle, and the compressed air stream used to atomize coaters (Compulab-36 and Accela-Cota-48) and the flow
the coating solution. Two different inlet air stream options from the front or back region across the spray nozzle for
Figure 1. Schematic representation of the film-coating process flow diagram for side-vented coating pans.
A Thermodynamic Model for Tablet Film Coating 49
In order to make this model generally applicable to and incorporates the adjustment in molar volume of air due
different side-vented coating pan designs, a HLF was to the elevated temperature of Tair,in at the inlet flow meter.
experimentally determined and included in the model
calculations. The HLF is defined as the heat transfer
coefficient (hloss) for the coating pan multiplied by the pan Energy Balance
surface area (A). The HLF was determined empirically
for six different coating pans including 1 kg (HCT-30, The total energy balance for this system can be
LDCS-20), 12 kg (HCT-60), 60 kg (Compulab-36), 120 kg expressed by the First Law of Thermodynamics as
(Accela-Cota-48), and 220 kg (HC-130L) scales. follows:[8]
This equation converts the volumetric flow rate of the The energy balance equation is rearranged to
inlet air stream at the inlet temperature to a mass flow rate solve for the unknown outlet temperature, Tair,out: (see
A Thermodynamic Model for Tablet Film Coating 51
^ ^
fmair;in Cp;air Tair;in þ xw mcoat Cp;w Tcoat xw mcoat DH v;w þ xorg mcoat Cp;org Tcoat xorg mcoat DH v;org þ HLF TRT g
Tair;out ¼
½mair;in Cp;air þ xw mcoat Cp;w þ xorg mcoat Cp;org þ HLF
½13
Eq. 13 above). The thermodynamic model detailed above manufactured by Thomas Engineering (Hoffman Estates,
provides a direct relationship between the temperature, Illinois). This equipment covers a range from laboratory
humidity and flow rate of the inlet air stream and the (1 kg) to production scale (120 – 220 kg). Laboratory scale
coating solution spray rate to the temperature and units that are designed to be explosion proof were used for
humidity of the exhaust air stream. Once the HLF is both aqueous- and organic solvent-based processes.
determined for a coating pan, this model can be used to Larger production-scale equipment was dedicated for
predict the effect of changing spray rate, or other inlet either aqueous or solvent use. Tablet-coating experiments
conditions on the exhaust temperature. were performed using both Vector and Thomas Engi-
neering coating pans. Figure 1 shows a simplified diagram
of a coating pan. Typically, the inlet air enters the pan in
Percent RH Determination the front or back for the Vector design and from the side
in the Thomas Engineering design. Both manufacturers’
The percent relative humidity (RH) of the exhaust air pans are side vented with different perforation designs.
stream can be calculated from a ratio of the partial All Vector coating pans used in this work were partially
pressure of water vapor in the exhaust air (calculated from perforated (contain a solid nonperforated area between
the measured dew point of the exhaust air) to the vapor exhaust manifolds), and all Thomas Engineering pans
pressure of water at the outlet temperature, Tair,out, as were fully perforated.
shown in Eq. 14.
The theoretical HLF was calculated based on the heat measure the tablet bed temperature during a typical
transfer coefficient for the pan materials and the pan coating run in the HCT-60. As depicted in Figure 2, the
surface area, as depicted in Eq. 1. The heat transfer temperature of the tablets approximately equaled the
coefficient for the material of construction used in these exhaust air temperature after 10 minutes of coating (i.e.,
pans was determined according to Kern (1950) as during the steady state coating phase). The inlet air
hloss =KDT0.25.[9] In this equation, K was estimated to be temperature was adjusted throughout the run to maintain a
0.32 based on assuming the dryer surface was four vertical constant exhaust temperature.
plates and one horizontal plate. The temperature differ-
ence, is DT between the coating pan surface in °F
(assumed equal to Tair,out) and ambient temperature. Film-Coating Pan Heat Loss Factors
The heat transfer coefficient units of BTU/hr ft2 °F
were converted to cal/min m2 K for data reported in HLF Determination from the Thermodynamic Model
this article.
As an example of the HLF calculation, as described
previously for an HCT-60 coating pan, 10 film-coating
RESULTS AND DISCUSSIONS trials were performed using aqueous or mixed organic
formulations. The aqueous film coatings were comprised of
Check of Assumption T air,out T tablet 94% (w/w) water and 6% solids with process variables
ranging from 40 to 60 g/min for spray rate, 64 to 80°C for
This thermodynamic film-coating model was devel- inlet air temperature, and 20 to 5°C for inlet dew point,
oped from a model originally developed for tablet coating while the inlet volumetric flow rate and ambient temper-
in a fluid bed. In that model, the sensible heat temperature ature were maintained constant at 300 cfm and 22°C, re-
gradients are calculated assuming the inlet drying air spectively. The mixed aqueous/organic film-coating trials
temperature equals the tablet bed temperature (i.e., consisted of 69% acetone, 19% water, and 12% solids. As
Tair,in Ttablet). For a fluid-bed coating system, this with the previous aqueous coating process conditions, the
assumption is reasonable since the distance between the inlet volumetric flow rate and ambient temperature were
spray nozzle and tablets is less than an inch. However, for maintained constant at 300 cfm and 22°C, respectively. The
the more commonly used side-vented pan, the spray process conditions were varied from 185 to 205 g/min
nozzle is several inches from the cascading tablet bed. As for spray rate, and 50 to 60°C for inlet air temperature.
a consequence, we assume in our model that the tablet bed Using the experimental results for the actual Tair,out
temperature equals the outlet air temperature (i.e., from the 10 HCT-60 studies listed earlier, the HLF value
Ttablet Tair,out). was solved iteratively and then fine tuned from 350, 360,
To verify this tablet bed temperature assumption, an and 370 (Table 3). The resulting difference between the
infrared gun was used in one experiment to periodically predicted and actual Tair,out was summed into a group
average number of 0.19, 0.01, and 0.21, respectively.
Therefore, the HLF in this case was chosen to be the
minimum value of 360.
A summary of calculated HLF values for six different
types and scales of coating pans is listed in Table 4. The
HLF generally increases with coating pan scale from 1 to
120 kg, covering those pans with once-through drying
airflow patterns. This trend is expected based on the
increased heat transfer surface area, as indicated by Eq. 1
where HLF = hloss*A.
A possible explanation for the lowest HLF being the
HC-130L coater in spite of its large size is that this
particular system operates at faster organic coating spray
rates relative to the inlet drying airflow rates. This rapid
spray results in more evaporative cooling, which lowers
Tair, out to below ambient conditions, thus eliminating the
Figure 2. Comparison of tablet bed temperature, Ttablet (^), to thermal driving force for heat loss. In addition, this coater
the inlet drying air temperature, Tair,in (.), and the exhaust air is designed as a more efficient process because the unit
temperature, Tair,out (&), during a coating run in the HCT-60. contains a recirculation loop for solvent recovery. This
A Thermodynamic Model for Tablet Film Coating 53
Table 3
The heat loss factor, HLF, for each coating pan was determined from an initial series of experiments in which the second unknown
variable in Eq. 13 was modified iteratively until the predicted Tair,out matched the actual Tair,out
recovery process lowers the moisture remaining in the 6.67 m2 for the HC-130L. The theoretical HLF deter-
inlet air stream that is recirculated back into the pan and, mined for the Vector Hi-Coater pans ranged from 0 to
therefore, has a greater capacity to pick up water. The inlet 3000 cal/min°C over the inlet air temperature ranges
air stream dew points for the HC-130L and the HCT-60 investigated (Table 4). The HC-130L operates with an
were measured as minus 30°C and 10°C, respectively. outlet temperature below the room conditions due to the
These data provide supporting evidence that the larger solvent recovery on the recirculation loop of the drying air
coating pan has less heat loss due to the high organic spray stream, and therefore, no heat loss is expected for this
rate and recirculation loop compared to the other pans, system. The experimentally determined HLF based on the
which are all once-through systems. model is 0, as is the theoretical value. The explanation for
this may reside in the fact that the volumetric flow meter
location is not at Tair,in, and therefore, the model is
Theoretical HLF underpredicting Tair,out. The minimum theoretical HLF
value of 64 is consistent with the experimentally
The pan surface areas for the Hi-Coater series determined HLF of 62 for the HCT-30. However, the
provided by Vector Corporation were 0.25 m2 for the theoretical HLF values for the LDCS-20 and HCT-60 are
LDCS-20 and HCT-30, 1.12 m2 for the HCT-60, and twice the experimental findings at 47 and 520 vs. 24 and
Table 4
Effect of coating pan scale on the experimental HLF and comparison to calculated HLF for Vector Hi-Coater pans
360, respectively. All of these experimentally determined spray rate were 8°C below the outlet temperature
HLF were of the same magnitude as the theoretical values measured from the empty pan/no spray studies. When
and support the physical interpretation of the heat loss the empty pan/no spray data were modified to include
factor used in this model. these coating conditions, then the original HLF value
from the model predicts the correct Tair,out within 1°C for
both HCT-30 and LDCS-20 pans. These findings suggest
HLF Comparison to Empty Pan with No Spray that the evaporative cooling of the coating is the major
draw on the inlet air enthalpy that lowers Tair,out, and that
The HLF determined from the empty pan/no spray the sensible heating of tablets to the outlet temperature is
studies for the laboratory scale coaters are compared to not a significant factor. These results support the
the results obtained during tablet film-coating trials in assumption made in developing this model to neglect
Table 4. The inlet and outlet temperature profiles for the the sensible heat changes in the tablets and coatings.
LDCS-20 and HCT-30 empty pan/no spray coating
studies are shown in Figure 3. These data show that the
LCDS-20 pan reaches its set point more rapidly than does Comparison of Model Predictions to Actual
the HCT-30. In addition, the temperature drop across the Results in the Same Pan
pan at steady state was 13°C for the LDCS-20
(HLFEP =65) vs. 20°C for the HCT-30 (HLFEP = 104). As a check of the versatility of the model represent-
This corroborates the HLF determinations from the model ed by Eq. 13, 20 experiments were carried out to vary
that indicate more heat is lost in the HCT-30 pan a number of conditions while keeping the HLF constant.
(HLF =62.2) compared to the LDCS-20 (HLF = 24). In these experiments, coating compositions ranged from
The discrepancy between the HLF determined by 100% aqueous to 78% organic/22% aqueous, inlet temper-
empty pan/no spray conditions and film-coating studies atures from 50 to 80°C, and spray rates of 40 to 200 g/min.
amounted to 41 cal/min°C for both pans. The two major From this wide range of inlet variables, the model
causes for the higher HLF values for the empty pan/no predicts Tair,out for the HCT-60 to within 3.0°C (see
spray results are 1) lack of an evaporative cooling effect Table 5).
from the coating process and 2) lack of tablets in the pan
to absorb heat.
The Tair,out determined from the actual film-coating Scale-Up Predictions Vs. Experimental Results
trials with 7 g/min aqueous spray rate or 21 g/min organic
The additional experimental studies that were not
used to calculate the pan HLF can also be used to assess
the predictive capabilities of this model for scaling-up the
coating process. For aqueous film coatings, the predicted
Tair,out was within 3°C of the actual reading (Table 6).
One approach to scaling-up coating systems is to maintain
the same inlet and outlet temperatures and target new
conditions for the drying airflow rate and coating spray
rate. For this approach, the data in Table 6 indicate that
Vair,in and aqueous coating spray rates should be scaled by
10-fold and 20-fold to scale from the HCT-30 (1 kg) to
the HCT-60 and CL 36 (60 kg) coaters, respectively.
Alternatively, the model can be used to predict the target
Tair,out even when the main input variables of Tair,in and
spray rate are changed. This case is demonstrated by the
results of predicting Tair,out from the HCT-30 (1 kg) using
Tair,in of 65°C to 80°C in the HCT-60 (12 kg) and AC 48
(120 kg) within 1.2°C of the target 59°C.
For organic film coatings, the predicted Tair,out was
Figure 3. Heating profiles for the LDCS-20 (Tair,in = .; within 5.7°C of the actual reading (Table 7). The best
Tair,out = 6) and HCT-30 (Tair,in = &; Tair,out = 5) coating pans prediction is obtained when scaling from the LDCS-20
used to determine the HLFs for empty pans with no spray to the HCT-60. Scaling from the HCT-30 to HC-130L,
coating when the inlet air temperature controller was set at 70°C. the Vair,in was increased 40-fold while the spray rate
A Thermodynamic Model for Tablet Film Coating 55
Table 5
Comparison of model prediction versus the actual Tair,out for 20 additional coating trials in the HCT-60 using a HLF of 360
Table 6
Scale-up predictions of Tair,out from HCT-30 to HCT-60 and AC48 for aqueous film-coatings (94% water and 6% solids) compared
to actual results
Table 7
Scale-up predictions of Tair,out from HCT-30 to HCT-60 and HC-130L for organic film-coatings
Predicted Difference
% % % Tair,in Vair,in Spray rate Tair,out Actual [predicted-actual]
Coating pan Acetone Water Solids (°C) (cfm) (g/min) (°C) Tair,out (°C) (°C)
LDCS-20 Acetone 50 43 7
Comparison of Enthalpy Terms for Aqueous HCT-30 Acetone 50 29 21
Vs. Organic Coatings HCT-60 Acetone 49 39 12
HC-130L Acetone 47 53 0
The individual enthalpy terms in the energy balance HCT-60 Water 50 30 20
CL-36 Water 47 31 22
equation, Eq. 6, were analyzed in terms of their relative
AC-48 Water 50 37 13
contribution to the overall balance. The results of the
A Thermodynamic Model for Tablet Film Coating 57
rate during evaporation, and, therefore, results in higher 220 kg. These results indicate that equivalent environ-
exhaust temperatures. ments (temperature and humidity) can be established from
For the final simulation, the effect of coating spray small scale (1 to 12 kg) to large scale (60 to 220 kg) for
rate on Tair,out was evaluated when inlet variables were both aqueous and organic coatings.
held constant for Tair,in at 65°C and drying airflow rate at
280 cfm (Figure 7). An increase in spray rate results in a
higher energy demand to evaporate the solvent. This ACKNOWLEDGMENTS
enthalpy consumed for latent heat of vaporization leaves
less heat in the air exhaust stream, and, therefore, Tair,out The authors would like to thank P.D. Daugherity,
decreases. The rate of decline in Tair,out with increased M.B. Fergione, and L.A. Miller for sharing process data
spray rate is greater for the aqueous coating since water for inclusion in this paper. We would also like to thank
requires four times more energy to evaporate per gram D.J. am Ende, K. Waterman, A. Thombre, and S. Herbig
than acetone. for their technical advice. We would also like to thank M.
Chidlaw, D. Millard, and D. Newbold for their work on
the early stages of this model.
Summary of Key Variable Effects on Outlet
Temperature and Humidity
REFERENCES
For the simulation conditions performed in the
‘‘Comparison of Enthalpy Terms for Aqueous Vs. 1. Reiland, T.L.; Seitz, J.A.; Yeager, J.L.; Brusenback,
Organic Coatings’’ section, the variables that most R.A. Aqueous film-coating vaporization efficiency.
impacted the exhaust temperature and/or humidity were Drug Dev. Ind. Pharm. 1983, 9 (6), 945 – 958.
identified. These variables, termed critical variables, 2. Stetsko, G.; Banker, G.S.; Peck, G.E. Mathematical
include the drying airflow rate, mair,in, the inlet air modeling of an aqueous film coating process.
temperature, Tair,in, the coating composition, and the Pharm. Technol. 1983, 7 (11), 50 – 62.
spray rate, mcoat,in. Inlet variables that resulted in minor 3. Ebey, G.C. A thermodynamic model for aqueous
changes in the exhaust conditions include the inlet air film-coating. Pharm. Technol. 1987, 11 (4), 40 –50.
humidity, %RHair,in, and the coating solution tempera- 4. Liu, L.X.; Litster, J.D. Coating mass distribution
ture, Tcoat,in. from a spouted bed seed coater: experimental and
This model can be a resource-conserving approach to modeling studies. Powder Technol. 1993, 74, 259 –
identifying critical variable sensitivities to changes in inlet 270.
conditions, and as a means to predict target conditions on 5. Rodriguez, L.; Grecchi, R.; Cini, M.; Passerini, N.;
scale-up. This may be most important for products where Caputo, O.; Vecchio, C. Variation of operational
the coating processes are humidity sensitive. parameters and process optimization in aqueous film
coating. Pharm. Technol. 1996, 10, 76– 86.
6. am Ende, M.T.; Berchielli, A.; Chidlaw, M.D. A
CONCLUSIONS Thermodynamic Model for Organic and Aqueous
Tablet Film-Coating, AIChE National Meeting,
A general thermodynamic film-coating model for Reno, NV, November 9, 2001.
side-vented pan coating was developed to provide for- 7. am Ende, M.T.; Herbig, S.M.; Korsmeyer, R.W.
mulators with a means of predicting target operating con- Osmotic drug delivery from asymmetric membrane
ditions for optimization, scale-up, and robustness studies. film-coated dosage forms. In Handbook of Pharma-
This model provides a precise relationship between ceutical Controlled Release Technology; Wise, D.L.,
the inlet drying air (temperature, humidity, and flow rate), Ed.; Marcel Dekker, Inc.: New York, 2000; 751 –
coating solution (composition, spray rate, and tempera- 785.
ture) and exhaust air stream (temperature, humidity, and 8. Felder, R.M.; Rousseau, R.W. Elementary Princi-
flow rate) for both aqueous- and organic-based coatings. ples of Chemical Processes; John Wiley & Sons:
This model is broadly applicable to a wide range of New York, 1978.
coating pan sizes, equipment types, and coating compo- 9. Kern, D.Q. Process Heat Transfer; McGraw-Hill:
sitions. The model prediction of Tair,out closely matched New York, 1950; 214– 215.
the actual Tair,out. In fact, the model shows that there is a 10. Perry, R.H.; Green, D. Perry’s Chemical Engi-
linear correlation between the predicted and actual Tair,out neers’ Handbook, 6th Ed.; McGraw-Hill: New
for the six pans, two sites, and scale ranging from 1 to York, 1984.