- Email: [email protected]
An analysis of the mini-tablet fluidized bed coating process
Accepted Manuscript Title: AN ANALYSIS OF THE MINI-TABLET FLUIDIZED BED COATING PROCESS ˇ Authors: Rok Sibanc, Magdalena Turk, Rok Dreu PII: DOI: Reference:
S0263-8762(18)30134-5 https://doi.org/10.1016/j.cherd.2018.03.020 CHERD 3090
To appear in: Received date: Revised date: Accepted date:
1-9-2017 7-3-2018 14-3-2018
ˇ Please cite this article as: Sibanc, Rok, Turk, Magdalena, Dreu, Rok, AN ANALYSIS OF THE MINI-TABLET FLUIDIZED BED COATING PROCESS.Chemical Engineering Research and Design https://doi.org/10.1016/j.cherd.2018.03.020 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
AN ANALYSIS OF THE MINI-TABLET FLUIDIZED BED COATING PROCESS
Rok Šibanc1,3,*, Magdalena Turk2,3 & Rok Dreu3
A
N
2, 2.5 and 3 mm diameter mini-tablets were coated in two fluidized bed coaters Cycle times of mini-tablets were measured with a photoluminescent system Effect of cycle time variability on the coating variability was between 5 and 28 % Particle volume fraction was determined using transmittance measurements
M
U
HIGHLIGHTS
SC R
*Corresponding author: [email protected], +49 211 81 14225 Other authors: [email protected], [email protected]
IP T
1 Institute of Pharmaceutics and Biopharmaceutics, Heinrich-Heine-University, Universitätsstr. 1, 40225 Düsseldorf, Germany 2 Department of Pharmaceutical Technology, Faculty of Pharmacy, Medical University of Gdansk, Hallera 107, 80-416 Gdansk, Poland 3 Department of Pharmaceutical Technology, Faculty of Pharmacy, University of Ljubljana, Aškerčeva cesta 7, 1000 Ljubljana
TE D
ABSTRACT
Mini-tablets with diameters of 2.0, 2.5, and 3.0 mm are coated in two different lab-scale fluidized bed coaters equipped with a Wurster draft tube. The main focus of the research is to evaluate the
EP
inter-particle coating variability, and to assess the contribution of cycle time variation. Cycle times are measured using a photoluminescent tracer with a detector mounted on the top of the draft tube.
CC
The number of passes variability is represented from 5 to 28% of the total coating variability. Additionally, transmittance measurements at the top of the Wurster draft tube are performed in order to assess the inter-particle sheltering effects. Transmittance results are correlated to the amount of
A
coating deposited per single pass of the spray zone and are converted to solids volume fractions. The dynamics of the transmittance signal further reveal the persistence of a particle arrangement within the draft tube of the two different coaters. The gathered results give insight into the different performance of two fluidized bed coaters in terms of inter-particle coating variability.
KEYWORDS Mini-tablets; fluidized bed coating; coating variability; cycle time; Wurster coater
1
INTRODUCTION Mini-tablets are tablets with a maximum diameter of 3 mm and are becoming an increasingly
popular oral dosage form, especially for pediatric and geriatric populations (Tissen et al., 2011;
IP T
Klingmann et al., 2013, 2015; Kluk et al., 2015; Aleksovski et al., 2015; Gaber et al., 2015; Pinto et al., 2016; Thabet et al., 2018). Tablets are traditionally coated in drum coaters, however, mini-tablets are coated using either fluidized bed coaters (Bodea et al., 2010; Mohamed et al., 2015) or drum
SC R
coaters (Mondelli et al., 2010). In order for the applied coating to perform its function it has to have the required thickness. So, the mean coating thickness of all tablets not only has to be equal to the required thickness, but also the coating thickness of each individual tablet has to be as close as
U
possible to the required thickness or in some of applications at least not less than the required value.
N
The (non)uniformity of coating thickness between tablets is commonly expressed using a coefficient
A
of variation (CV).
Prediction and troubleshooting of the coating process may be performed by modeling. Models are
M
categorized as macro- and micro-levels, where macrolevels describe processes applied to the whole batch, while microlevels consider phoneme of film creation in a single core scale. Macrolevels
TE D
include: renewal theory, compartment models, Monte Carlo methods and population balance models (Turton, 2008, 2010). Renewal theory is a simple model which describes two contributions of the coating process variability. It considers the coating process as a repeating process, where the total
EP
coating amount on a particle m is a sum of its N passes through the coating zone, where a small but variable amount of coating x is received at each pass (Cheng and Turton, 2000a; Mann, 1983; Turton,
CC
2008).
𝑚 = ∑𝑁 𝑖=1 𝑥𝑖
Eq (1)
A
Both N and x are distributions described by mean and variance. The resulting coefficient of variability of coating CV(m) can be considered as a sum of variances of both factors: 𝐶𝑉(𝑚) =
𝑆𝑇𝐷(𝑚) 𝐸(𝑚)
𝑉𝑎𝑟(𝑥)
𝑉𝑎𝑟(𝑁)
= √[𝐸(𝑥)]2 [𝐸(𝑁)] + [𝐸(𝑁)]2
Eq (2)
The E represents the expected value (or mean) of the variable, and if Var/E2 is expressed as CV2 the equation has the following simplified form:
𝐶𝑉(𝑚) = √
[𝐶𝑉(𝑥)]2 𝐸(𝑁)
+ [𝐶𝑉(𝑁)]2
Eq (3)
The number of passes N can be also expressed with the time between passes through the spraying zone - t and the total coating time T (i.e. E(N)=T/E(t)). It was shown (Mann, 1983) that CV(N) can be expressed with the circulation time distribution and Eq(3) expressed as: 𝐶𝑉(𝑚) = √
𝑉𝑎𝑟(𝑥)𝐸(𝑡) [𝐸(𝑥)]2 𝑇
+
𝑉𝑎𝑟(𝑡)
Eq (4)
TE(t)
IP T
Eq (4) is valid when the number of circulations is larger than 30, according to Mann (1983). Magnetic particle tracking was successfully used for describing and visualization of particles movement in fluid bed systems (Sánchez Quintanilla et al., 2014; Yang et al., 2017) including rotor
SC R
fluid bed system (Neuwirth et al., 2013) or spouted bed (Mohs et al., 2009). Cheng and Turton (2000b, 2000a) performed a series of fluidized bed experiments using a magnetic tracer particle to study the cycle times of pellets, and to evaluate the contribution of variation of cycle times to the coating mass variation. Shelukar et al. (2000) performed similar experiments using tablets and performed
U
modification to the distribution plate in order to reduce the variability of tablet cycle times. Their
N
studies establish that the contribution due to variation in circulation time to tablets is between 14 and
A
24%. If the reported data from Cheng and Turton is used and calculated according to Shelukar et al., then the contribution of circulation times for variability of coating on pellets is only up to 3.4%.
M
Positron emission particle tracking (PEPT) is a superior technique with regard to the study of particle cycle times, since it yields a mm resolution position of a radioactive labelled particle over time. As a
TE D
result, it can also detect particle recirculation events within a specific zone and elucidate residence times in different zones of the process chamber. The advantage of the PEPT method is that particle tracking can also be accomplished for process chambers at elevated temperatures and pressures. The
EP
disadvantages of this method are cost of the equipment, size limitation and thickness of process chamber walls, and the need of radioactive tracer. The half-life of a radioactively labeled particle may be a limitation for the required total tracking time in terms of the required spatial resolution of
CC
trajectory. PEPT was used in a few studies to analyze a Wurster fluidized bed (Chang et al., 2013; Li et al., 2015a; Palmer et al., 2007) or a top spray fluidized bed (Depypere et al., 2009). Tablet cycle
A
and residence times in a pan coater can be determined using video imaging, where the orientation of the tablets can also be tracked (Kandela et al., 2010). Cycle and residence times, as well as coating uniformity, can also be studied by means of simulations such as a discrete element method (DEM) (Kumar et al., 2015) or in the case of a fluidized bed with a combination of DEM with computational fluid dynamics (CFD) (Fries et al., 2013; Jiang et al., 2017; Li et al., 2015b).
Another important parameter in the analysis of fluidized bed processes is the amount of material in the spraying zone and is normally expressed as a volume fraction. Volume fraction of solids directly influences the per-pass variability due to sheltering effects (Cheng and Turton, 2000a). Local solids volume fraction measurements can be performed using a reflective optical fiber (Johnsson and Johnsson, 2001) or using transmittance measurements (Šibanc et al., 2016). For determination of volume fraction distribution capacitive or X-ray tomography methods were used (Rautenbach et al., 2013; Hensler et al., 2016; Weber et al., 2018).
IP T
The main motivation for this work is to understand the coating variability of coating of mini-tablets in a Wurster coater. Mini-tablets of three different sizes (diameter of 2.0, 2.5 and 3.0 mm) are used
SC R
in the study. The influence of coater design is also of interest, since significant differences in the coating of pellets were observed in previous research. Therefore, we compare a classical and swirl Wurster coater (Glatt GPCG-1 and Brinox CGD). As a main coating result, we investigate inter-tablet coating variability using dye content analysis. Besides the effect of the inlet air geometry, the effect
U
of fluidizing air flow is investigated at two different air flow rates (130 and 156 m/h3). In this study,
N
the mini-tablet cycle time measurements are performed in order to explain the contribution of number of passes variability to the total coating variability. Additionally, light attenuation measurements
A
(Šibanc et al., 2016) are conducted in order to evaluate the number density of mini-tablets at the top
M
of the Wurster draft tube, and to assess the dynamics of the system. The significance of this work is threefold. The raw data of coating variability of mini-tablets in fluidized beds is of importance,
TE D
especially due to increasing interested in pediatrics and individualized medicine. Secondly, the process understanding is increased and coating variability sources are elucidated using the inline process analytical tools. Lastly, the raw experimental data can be of use for the development of
2
EP
numerical simulations.
MATERIALS AND METHODS Mini-tablets preparation
CC
2.1
Biconvex mini-tablets with a diameter of 2.0 mm (MT2.0), 2.5 mm (MT2.5), and 3.0 mm (MT3.0) are prepared from a tableting mixture containing: lactose FlowLac® 100 (74.07% (w/w)),
A
microcrystalline cellulose Vivapur® 102 (24.68 % (w/w)), colloidal silicon dioxide Aerosil® 200 (0.25 % (w/w)), and sodium stearyl fumarate PRUV® (1.00 % (w/w)). Mini-tablets are pressed using a rotary tablet press (Erweka RTP-D8, Germany) using compacting pressures between 300 and 400 MPa.
2.2
Coating procedure
Mini-tablets are coated with a 10% (w/w) aqueous coating dispersion consisting of: HPMC (8.00% (w/w)), PEG 6000 (1.00% (w/w)), and tartrazine (1.00% (w/w)) as a colorant (Fig. 1). In each coating experiment 1000 grams of mini-tablets are coated with a target theoretical coating thickness of 20 µm. We spray a coating dispersion of 737 g (MT2.0), 587 g (MT2.5), or 524 g (MT3.0) of coating dispersion under the assumption of 70% coating efficiency. Film coatings are performed in a bottom spray fluid bed systems, using either classical (CW) or swirl Wurster (SW) chamber (Glatt GPCG-1,
IP T
Germany and Brinox CGD, Slovenia; Fig. 2). Air flow rates of 130 and 156 m3/h are used when utilizing the CW chamber, and we use an air flow rate of 156 m3/h in the case of an SW chamber. A constant spray rate of 5 g/min, an atomizing air pressure of 2 bar, inlet air temperature of 55°C, and
SC R
a gap of 25 mm are used in all cases. Since the same spray rate is used for all mini-tablet sizes with different specific surfaces and the aim is to apply the same coating thickness regardless of mini-tablet type, the coating times varied between tablet sizes. The coating time is 8735 s for MT2.0, 7040 s for
Analysis of coating uniformity
N
2.3
U
MT2.5, and 6282 s for MT3.0, respectively.
An analysis of mini-tablets coating uniformity is based on spectrophotometric measurements of
A
tartrazine concentration. 100 mini-tablets are analyzed using the UV-VIS method. Each of 100 mini-
M
tablets is placed in a single vial and diluted with 4.5 ml of dihydrogen phosphate buffer (6.5 pH). After the dissolving of water soluble ingredients, vials are centrifuged (15 min, 3500 rpm) to separate
TE D
undissolved particles of disintegrated mini-tablet. Samples are analyzed for absorbance (UV spectrophotometer HP8453, Hewlett-Packard, USA) at 425 nm wavelength. Further, based on tartrazine concentration, coating mixture composition and dry film density (1.27 g/cm3) as well as surface of single mini-tablet, film thickness of each mini-tablet and coefficient of variation of per-
Cycle time measurements
CC
2.4
EP
particle coating mass (CV(m)) is calculated.
Cycle time measurements of mini-tablets are performed using a photoluminescent detection system, primarily developed for pellet cycle time measurements. A single tracer mini-tablet is coated
A
with photoluminescent coating (ethanol dispersion of PVP K25 and CuS:EU LED phosphor in 6:1 ratio), whereas all the other mini-tablets are coated with a carbon black coating (8% HPMC, 1% PEG 6000, and 1.7 % active carbon). The detection system is placed on the top of the Wurster draft tube and faces outwards. It consists of an excitation part composed of blue LED (XP-E Royal Blue, Cree Inc., USA) and a detection part comprising of photodiode sensitive to red light (EPD-660-1-0.9, Epigap Optronic GmbH, Germany). In order to improve the signal-to-noise ratio, the excitation part
is equipped with a blue filter transparent for wavelengths from 400 – 500 nm (363 Special Medium Blue, Lee Filters, USA) and the detector with a red color filter transparent for wavelengths above 600 nm (CL182 Cool LED Light Red, Lee Filters USA). A transimpedance amplifier which amplifies and converts photoelectric current to voltage is a part of the custom designed circuit board. The output voltages are recorded using an external analog to digital converter. The measurements are performed for 1.5 h in each case. To simulate coating conditions purified water is sprayed during the measurements. Conversion to number of cycles
IP T
2.4.1
The cycle time to number of passes conversion representative for population of mini-tablets is
SC R
done using a numerical simulation, where random tracer mini-tablet cycle times are drawn from the recorded cycle time population and summed for each of simulated particle until the target simulated coating time is achieved. Target simulated coating time is 7200 s in the first series of simulation, while in the second series the time is set as in the actual coating experiment, depending on the mini-
U
tablet size – specific surface. 50,000 repetitions of the described numerical experiments are performed
N
for each case. The parameters of the number of passes - N distributions do not change significantly
Transmittance measurements
M
2.5
A
above the selected number of repetitions.
Light attenuation measurements are performed in order to evaluate the inter-particle shielding of mini-tablets in the Wurster draft tube, and to assess the dynamics of the system. The measurement
TE D
system consists of a laser source and a 10 x 10 mm2 photodiode detector mounted on a glass Wurster tube. Measurements are performed at the top of the Wurster glass tube (Fig. 4). Transmittance measurements are performed for all mini-tablets sizes in both coater types at both air flow rates. For
EP
each measurement, the signal is recorded for 5 minutes at an acquisition frequency of 50 kHz. More details about the method are available in previous publication where the volume fraction of pellets
CC
was determined using such a system (Šibanc et al., 2016). It was shown that the measurements at the top of the Wurster draft tube correlate with the global volume fractions of solids in the whole Wurster
A
draft tube.
2.5.1
Conversion of transmittance data to volume fractions
The transmittance is dependent on the amount (volume fraction) of mini-tablets in the measurement volume, their shape, orientation, and position. A Monte Carlo simulation like in a previous study (Šibanc et al., 2016) is employed to obtain transmittance distributions for volume
fractions between 0.1 and 10% in 0.1% steps. An exemplary simulated sensor response for five different volume fractions is shown in Fig. 5. The average transmittance follows Beer-Lambert law, which describes exponential decrease of transmittance depending on volume fraction. The reason for distribution of transmittance at a fixed volume fraction is the discrete nature of the relatively low amount of particles in the measurement volume. By varying the position of the same amount of particles, arrangements with varying transmittance are obtained. Recorded transmittance distribution data are then deconvoluted to volume fraction distributions via usage of simulated transmittance
IP T
distribution data database, as described in a previous study. Since the Monte Carlo simulation uses spherical particles for simulation of transmittance, the circle
steps:
SC R
equivalent diameter of average mini-tablet projections are calculated. This was done in the following
1. A 3D model of a mini-tablet based on its average dimensions is constructed in a 3D raytracing
U
program (Blender 2.78, open-source) at location 0,0,0.
N
2. Camera object is set at location 0,-10,0, with a render resolution of 1500 x 1500 px and an
A
orthographic projection which corresponds to parallel laser light rays.
M
3. Mini-tablets are rotated around the x axis from sin(alpha)=0 to sin(alpha)=1 in 50 steps and an image render is created at each step.
TE D
4. The mini-tablet projection size at each angle is determined as the number of pixels belonging to the mini-tablet multiplied by the square of the pixel size.
2.5.2
EP
5. The average projection size is used to calculate the circle equivalent diameter. Autocorrelation
CC
Autocorrelation of the whole transmittance signal (300 s at 50 kHz) is performed for all 12 cases. Autocorrelation of a signal is a correlation of a signal with the same signal shifted by some amount, commonly named as ‘lag.’ Based on preliminary signal studies, autocorrelation with lags up to 20 ms
A
are performed. Autocorrelation is a time domain analysis and provides an insight into particle dynamics in this case. 3
RESULTS AND DISCUSSION
3.1
Coating variability analysis
Coating variability (CV(m)) of mini-tablets based on spectrophotometric measurements of dye concentration (Table 2, Fig. 6) is in a range from 3.08% to 19.08%. The best result is obtained when
coating MT2.5 using SW coater and the worst result using CW coater, while coating MT3.0 at lower fluidizing air flow. By comparing the coater design it can be seen that the mini-tablets coated with SW always have lower coating mass variability. The lower values of the fluidizing air flow, which were tested using CW, always lead to higher coating variability. It is assumed that this is due to higher packing of mini-tablets in the coating zone, as well as due to lower particle circulation rates (with probable recirculation events in the spray zone). Increasing the size of mini-tablets had a negative effect on the coating variability. Presumably, larger mini-tablets had less uniform circulation as gap
IP T
size is maintained at its maximum position – posing greater obstruction for bigger mini-tablets, but also very important the coating time is different for every mini-tablet size and is decreasing with minitablet size in order to achieve the same level of coating thickness. A coating variability of MT2.5 and
SC R
MT3.0 is predicted for the coating time of MT2.0 according to Eq (4), that is that the coating variability decreases proportionally to 1/T0.5. Also, in the case of time extrapolated coating variability, values MT3.0 have the highest coating variability. Cycle time results
U
3.2
N
Mini-tablet cycle time distributions for six experiments (Fig. 7) primarily show the differences between CW and SW coater. The mode values of the cycle time distributions of CW coater are lower
A
than the SW ones, and also the widths of the distributions of CW seem to be lower. 2.0 mm mini-
M
tablets have lower cycle times than the 3.0 mm mini-tablets, which while observing results at the same inlet air flow rate confirms the flow limiting effect of the gap between draft tube and distribution
TE D
plate set to 25 mm. The results of 2.5 mm diameter mini-tablets, which are not shown in the plot, are in all cases between the results of 2.0 and 3.0 mm mini-tablets. The total number of tracer mini-tablet passes N, median, mean, variance, and maximum cycle
EP
times of cycle time measurements are summarized in Table 3. In can be seen that within the measuring time of 5400 s the tracer tablet passes the Wurster partition on average more times in the CW coater.
CC
The mean circulation times in all conducted experiments are between 3.42 and 6.71 s, whereas the median cycle times are between 2.16 and 5.56 s. This indicates an asymmetric distribution, which is confirmed by the skewness parameter, as well as a visual observation of distributions (Fig. 7). All
A
skewness values are positive, meaning the cycle time distributions have a number of long cycle times. The skewness values, as well as the maximum cycle times, are much higher in the case of a CW coater when compared to a SW coater. A very important parameter for uniform coating is the variability of cycle times, which can be described by its variance. The range of variance was 12.88 to 102.51 s2 for the CW coater, and from 9.09 to 11.15 s2 for the SW coater.
Simulation-derived distributions for the number of particle passes simulating equal coating time of 7200 s (Fig. 8) show the swirl chamber has narrower distributions of number of passes, but also generally lower mean values in comparison to the CW chamber. Derived distributions also show that the average number of passes through the spraying zone is not affected by mini-tablet size in the case of a SW coater. This could indicate that the SW operates at a lower particle circulation rate than the CW chamber, and is therefore not limited by the gap size as the mini-tablet size changes. The opposite was observed in the case of a CW. It is evident that the number of passes of MT3.0 in the CW chamber
IP T
was lower than for the other two mini-tablets sizes. The number of passes was also dependent on the air flow rate, and as expected the lower air flow rate led to a lower number of circulations.
SC R
Distributions for the number of particle passes were also calculated for the actual coating times of mini-tablets (Fig. 9). Since the coating time was dependent on the mini-tablet size, the differences in the number of passes for different sized mini-tablets are more pronounced. The largest effect of a mini-tablet size on the mean value of number of particle passes can be seen in the case of CW coater
U
using fluidizing air flow of 156 m3/h (i.e. more than 1000 passes difference between MT2.0 and
Coating variability analysis
A
3.3
N
MT3.0).
M
Coating variability CV(m) is analyzed in terms of the two contributions: variability of the number of passes CV(N) and the amount of deposited coating per pass CV(x) (Table 4). CV(x) is calculated
TE D
using Eq (3) and the values of CV(m), CV(N), and E(N). The term CV(N)2/CV(m)2 is the relative contribution of number of passes variability on total coating mass variability. As discussed above the lower variability of coating thickness is seen in the case of a swirl coater. According to the calculation, both CV(N) and CV(x) attribute to this result. CV(N) is in the case of SW, between 1.47 and 1.63%,
EP
and in the case of a CW coater between 2.29 and 5.42%. This indicates that the SW coater has better mixing or an absence of dead zones in the annulus bed region, and the most probable reason for that
CC
is the curved design of the distribution plate. Calculated CV(x) values range for the SW coater is between 88.8% and 148.6%, and in the case of a CW coater between 279 and 591%. A possible explanation of this phenomenon, besides the lower particle number density, could be that the swirling
A
air in SW coater pushes the particles away from the spraying nozzle thus making mini-tablets more equally distant from the spraying nozzle, which therefore results in a more even deposition per pass (Luštrik et al., 2013). Higher fluidizing air flow rate in all cases increases the number of passes E(N), reduces coating thickness variation CV(m), and reduces per pass variation CV(x). A possible explanation for reduced CV(x) at higher air flows is that with a higher air flow rate particles in the Wurster draft tube
accelerate faster, while horizontal mass flow particles are limited by the gap height and so a reduction of the particle volume fraction occurs, which in turn reduces the sheltering effect of particles from the spray. The relative contribution of number of passes variability is calculated to be between 10.3% and 28.0% for the SW coater, and in range from 4.9% to 19.2% for the CW coater. These results are comparable to the results of Shelukar et al., where they coated 4.5 kg of 7.9 mm tablets in a Niro
Transmittance
3.4.1
SC R
3.4
IP T
Wurster coater (Model MP1).
Measurements
Transmittance measurements are performed to evaluate the effect of process conditions and coater type on the amount (number density) of mini-tablets in the Wurster draft tube. The measurements are
U
performed at the exit of the draft tube (Table 5), which presumably correlates to the conditions in the
N
spray zone. The transmittance values are a measurement of light sheltering, and can be an indicator
A
of inter-particle sheltering from the spray. The transmittance is inversely proportional to the volume fraction of particles in the measuring volume, and is also a function of particle size. Therefore, the
M
transmittance values of differently sized mini-tablets cannot be compared directly.
TE D
For the same mini-tablet size and air flow the transmittance values obtained using the SW coater are always higher than for the CW coater (Fig. 10). This could at least partially explain why CV(x) values are smaller in the case of the SW coater. Since the measurements are performed across the Wurster draft tube, the resulting transmittance is a result of average volume fraction across the
EP
measuring volume and does not provide any information about the location of the mini-tablets. Increasing air flow rate increases the transmittance when MT2.0 and MT2.5 are used in the CW,
CC
however, the opposite is true for MT3.0. In all cases of the SW coater, air flow rate actually reduces the transmittance, which can indicate a change of particle flow pattern.
A
The idea for measuring the transmittance is that it correlates with the amount of particles present
in the Wurster draft tube, where the coating occurs. This can be used as a measure of inter-particle shielding from the spraying nozzle, which is one of the reasons for per pass variability of deposited coating amount CV(x). In the case of a higher number density of particles in the coating zone, the shielding effect is more pronounced. The correlation between the observed average transmittances and per pass variability CV(x) from Table 4 is in this case valid per mini-tablet size, and with an
outlier at 3.0 mm coated in the CW coater at 156 m3/h (Fig. 11). The reason why the correlation is valid per mini-tablet size and also for the outlier can be that the particle shielding is only one of the phenomena contributing to the per pass variability of the deposited coating amount. Another reason is the residence time or the particle velocity through the coating zone, which is for fluidized beds particle size-dependent. In addition, the particle trajectory through the draft tube also has an importance. Considering a hypothetical case where the same amount of particles are travelling through the draft tube, the particles travelling nearer to the nozzle receive more coating than the
IP T
particles further from the nozzle, purely based on the fact that the spray droplets concentration decreases with the distance from the nozzle. Of course, particle shielding further enhances the importance of trajectory. In the case of a SW coater, the swirling air pushes the particles more towards
SC R
the draft tube wall and unifies the particle trajectories (Luštrik et al., 2013). Increasing the transmittance further, for example with a decreasing gap width would mean having fewer particles in the spray zone. This could then potentially lead to some adverse effects such as spray drying, deposition of the coating material on the Wurster draft tube wall, and particle agglomerationVolume
N
U
fraction
In order to compare the hydrodynamics of the process the transmittance distributions are converted
A
to volume fraction distributions (Table 5). The average volume fractions are fairly constant across
M
mini-tablet size, except for the 3.0 mm tablet at CW156 (Fig. 12). The volume fraction of mini-tablets at the top of the draft tube is in all cases higher for the CW coater. The higher air flow rate decreases
TE D
the volume fraction in the CW coater and increases it in the SW coater. The air flow rate has a couple of effects on the fluidized bed. The higher air flow rate means higher air velocities in the draft tube, and a faster and more dilute flow under the assumption of a constant mass flow of particles. Secondly, higher air velocity also results in a higher suction force towards the inside of the draft tube in the
EP
region of the distribution plate. Lastly, a higher air flow rate reduces resistance to the flow of material in the annular region of the coater and enables a higher mass flow rate of particles, and consequently
CC
higher volume fractions. As a result, it is hard to attribute the results to a single phenomenon, but it is clear that their contributions are different for both coater types. The relative standard deviations
A
(RSD) of volume fractions are quite comparable for the same mini-tablet size, except for SW130, where the RSD is always the highest (Table 5). This is due to much lower average values of volume fractions. In terms of absolute volume fraction deviations, it is the lowest in such a combination of coater type and air flow rate, i.e. for all mini-tablets sizes.
3.4.2
Dynamics
Autocorrelation of a transmittance signal time series show how quickly the correlation vanishes and can be related to the particle dynamics. A faster decreasing correlation may indicate a faster change of the particle arrangement in the measurement volume. It can be postulated that higher autocorrelation values indicate movement of particles in clusters with little local mixing, which in terms facilitates (amplifies) the influence of the inter-particle sheltering at the same transmittance value. Fig. 13 shows autocorrelation signals of all 12 transmittance measurements. The thin lines
IP T
represent 2.0 mm mini-tablets, the dashed for 2.5 mm, and the 3.0 mm are represented with thick lines, whereas the darker color represents a lower air flow rate. The upper part of Fig. 13 shows the autocorrelations of measurements in the CW coater. Differences between all six cases can be seen,
SC R
meaning the method and analysis is sensitive at least for relatively big changes. An effect of the air flow rate can be observed for the 2.0 and 2.5 mm mini-tablets, as the autocorrelation decreases quicker for higher air flow rates. This is expected since the higher air flow rate normally results in higher particle velocity. The effect of air flow rate can be seen in the SW as well, although to a lesser extent
U
then in the CW. In case of the SW coater, mini-tablet size plays a much smaller role than in the CW
N
coater. This lower particle size sensitivity is also seen in the number of circulation results, and shows
A
that the SW coater can be considered more suitable when coating a distribution of particles and an equal coating thickness is applicable to all particles. Lower autocorrelation values of the SW
M
compared to the CW supports the particle cluster movement hypothesis, as also the total coating variability is on average smaller than in the case of the CW, and analysis has shown the particle per
TE D
pass effect has a predominant part in the total coating amount variability. The transmittance – CV(x) correlation outlier (CW156 - MT3.0, Fig. 11) can be explained by the autocorrelation curve, which is higher in this instance than expected for such an air flow rate. As higher autocorrelation values mean
4
CC
EP
less local particle mixing, this situation leads to higher CV(x) even at intermediate transmittance valu
CONCLUSIONS
Mini-tablets are successfully coated in two different fluidized bed coaters yielding inter-particle
A
coating variability in the range from 3.1% to 19.1%. We observe that the swirl coater performs better in all three sizes of mini-tablets, and has the maximum coating variability of 4.8% – using 3.0 mm mini-tablets. Coating of smaller mini-tablets always results in smaller coating variability, independent of coater type. Cycle time results show that the swirl coater has a lower number of passes variability, although the number of passes was lower than in the classical Wurster coater with the same coating parameters.
Based on the cycle time and coating variability results, the contribution of the number of passes variability and the particle per pass variability of the deposited coating amount is calculated. The contribution of the number of passes variability is in a range from 5 to 28%. The per pass variability ranges from 89 to 591%, and is always lower in cases of the SW coater. It is also shown that the transmittance measurements partially explain the particle per pass coating variability. Higher transmittance measured at the top of the Wurster draft tube correlates with lower per pass variability for a specific mini-tablet size. Transmittance data is also converted to volume
IP T
fraction data, and is also used to perform additional dynamics evaluation. Volume fractions are in a range from 2.3 to 5.7% across all cases, and as such can be useful for validation of simulations. Also,
SC R
autocorrelations of transmittance signal primarily show the difference between both coater types – that is, the SW coater shows less dependence on mini-tablet size for many measured parameters: coating variability, number of circulations, and volume fractions.
U
Future work will be focused on a broader experimental plan of cycle times. Cycle times will be measured for a different design of distributor plates, and the effect of the gap between the Wurster
N
draft tube and the distributor plate will be also evaluated. Additionally, studies of intra-tablet coating
M
ACKNOWLEDGEMENTS
A
variability are also planned.
TE D
The authors would like to thank Domen Kitak for the help with the development of the cycle time measuring system, and Grega Ratek for the help with the cycle time measurements. Funding: This work was supported by The Applied Research Program of The National Centre for
Foundation.
CC
APPENDIX
EP
Research and Development (grant number PBS1/A7/3/2012), and the Alexander von Humboldt
A
List of abbreviationsm mass of coating t cycle time x amount of coating per pass CFD computational fluid dynamics CV coefficient of variation
CW classical Wurster coater DEM discrete element method PEPT positron emission particle tracking RSD relative standard deviation
IP T
MT mini-tablet N number of passes
SC R
SW swirl Wurster coater
A
CC
EP
TE D
M
A
N
U
T total coating time
REFERENCES
A
CC
EP
TE D
M
A
N
U
SC R
IP T
Aleksovski, A., Dreu, R., Gašperlin, M., Planinšek, O., 2015. Mini-tablets: a contemporary system for oral drug delivery in targeted patient groups. Expert Opin. Drug Deliv. 12, 65–84. https://doi.org/10.1517/17425247.2014.951633 Bodea, M., Tomuţă, I., Leucuta, S., 2010. Film coating preparation of metoprolol tartrate mini-tablets and in vitro drug release studies. Clujul Med. 83, 457–464. Chang, Y.-F., Adamsen, T.C.H., Pisarev, G.I., Hoffmann, A.C., 2013. PEPT: An invaluable tool for 3-D particle tracking and CFD simulation verification in hydrocyclone studies. EPJ Web Conf. 50, 05001. https://doi.org/10.1051/epjconf/20135005001 Cheng, X.X., Turton, R., 2000a. The prediction of variability occurring in fluidized bed coating equipment. II. The role of nonuniform particle coverage as particles pass through the spray zone. Pharm. Dev. Technol. 5, 323–332. Cheng, X.X., Turton, R., 2000b. The prediction of variability occurring in fluidized bed coating equipment. I. The measurement of particle circulation rates in a bottom-spray fluidized bed coater. Pharm. Dev. Technol. 5, 311–322. Depypere, F., Pieters, J.G., Dewettinck, K., 2009. PEPT visualisation of particle motion in a tapered fluidised bed coater. J. Food Eng. 93, 324–336. https://doi.org/10.1016/j.jfoodeng.2009.01.042 Fries, L., Antonyuk, S., Heinrich, S., Dopfer, D., Palzer, S., 2013. Collision dynamics in fluidised bed granulators: A DEM-CFD study. Chem. Eng. Sci. 86, 108–123. https://doi.org/10.1016/j.ces.2012.06.026 Gaber, D.M., Nafee, N., Abdallah, O.Y., 2015. Mini-tablets versus pellets as promising multiparticulate modified release delivery systems for highly soluble drugs. Int. J. Pharm. 488, 86–94. https://doi.org/10.1016/j.ijpharm.2015.04.021 Hensler, T., Firsching, M., Bonilla, J.S.G., Wörlein, T., Uhlmann, N., Wirth, K.-E., 2016. Noninvasive investigation of the cross-sectional solids distribution in CFB risers by X-ray computed tomography. Powder Technol. 297, 247–258. https://doi.org/10.1016/j.powtec.2016.04.030 Jiang, Z., Bück, A., Tsotsas, E., 2017. CFD-DEM study of residence time, droplet deposition and collision velocity for a binary particle mixture in a wurster fluidized bed coater. Dry. Technol. https://doi.org/10.1080/07373937.2017.1319852 Johnsson, H., Johnsson, F., 2001. Measurements of local solids volume-fraction in fluidized bed boilers. Powder Technol. 115, 13–26. https://doi.org/10.1016/S0032-5910(00)00270-9 Kandela, B., Sheorey, U., Banerjee, A., Bellare, J., 2010. Study of tablet-coating parameters for a pan coater through video imaging and Monte Carlo simulation. Powder Technol. 204, 103–112. https://doi.org/10.1016/j.powtec.2010.07.024 Klingmann, V., Seitz, A., Meissner, T., Breitkreutz, J., Moeltner, A., Bosse, H.M., 2015. Acceptability of Uncoated Mini-Tablets in Neonates—A Randomized Controlled Trial. J. Pediatr. 167, 893–896.e2. https://doi.org/10.1016/j.jpeds.2015.07.010 Klingmann, V., Spomer, N., Lerch, C., Stoltenberg, I., Frömke, C., Bosse, H.M., Breitkreutz, J., Meissner, T., 2013. Favorable Acceptance of Mini-Tablets Compared with Syrup: A Randomized Controlled Trial in Infants and Preschool Children. J. Pediatr. 163, 1728– 1732.e1. https://doi.org/10.1016/j.jpeds.2013.07.014 Kluk, A., Sznitowska, M., Brandt, A., Sznurkowska, K., Plata-Nazar, K., Mysliwiec, M., Kaminska, B., Kotlowska, H., 2015. Can preschool-aged children swallow several minitablets at a time? Results from a clinical pilot study. Int. J. Pharm. 485, 1–6. https://doi.org/10.1016/j.ijpharm.2015.02.068 Kumar, R., Freireich, B., Wassgren, C., 2015. DEM–compartment–population balance model for particle coating in a horizontal rotating drum. Chem. Eng. Sci., Pharmaceutical Particles and Processing 125, 144–157. https://doi.org/10.1016/j.ces.2014.06.046
A
CC
EP
TE D
M
A
N
U
SC R
IP T
Li, L., Rasmuson, A., Ingram, A., Johansson, M., Remmelgas, J., von Corswant, C., Folestad, S., 2015a. PEPT study of particle cycle and residence time distributions in a Wurster fluid bed. AIChE J. 61, 756–768. https://doi.org/10.1002/aic.14692 Li, L., Remmelgas, J., van Wachem, B.G.M., von Corswant, C., Johansson, M., Folestad, S., Rasmuson, A., 2015b. Residence time distributions of different size particles in the spray zone of a Wurster fluid bed studied using DEM-CFD. Powder Technol. 280, 124–134. https://doi.org/10.1016/j.powtec.2015.04.031 Luštrik, M., Šibanc, R., Srčič, S., Perpar, M., Žun, I., Dreu, R., 2013. Characteristics of pellet flow in a Wurster coater draft tube utilizing piezoelectric probe. Powder Technol. 235, 640–651. https://doi.org/10.1016/j.powtec.2012.11.034 Mann, U., 1983. Analysis of spouted-bed coating and granulation. 1. Batch operation. Ind. Eng. Chem. Process Des. Dev. 22, 288–292. https://doi.org/10.1021/i200021a019 Mohamed, F.A.A., Roberts, M., Seton, L., Ford, J.L., Levina, M., Rajabi-Siahboomi, A.R., 2015. Film-coated matrix mini-tablets for the extended release of a water-soluble drug. Drug Dev. Ind. Pharm. 41, 623–630. https://doi.org/10.3109/03639045.2014.891128 Mohs, 7G., Gryczka, O., Heinrich, S., Mörl, L., 2009. Magnetic monitoring of a single particle in a prismatic spouted bed. Chem. Eng. Sci. 64, 4811–4825. https://doi.org/10.1016/j.ces.2009.08.025 Mondelli, G., Albertini, B., Funaro, C., Passerini, N., 2010. Minitablets Coated in a Solid-Wall Pan for Theophylline Sustained-Release Capsules. PharmTech 2010. Neuwirth, J., Antonyuk, S., Heinrich, S., 2013. Particle dynamics in the fluidized bed: Magnetic particle tracking and discrete particle modelling. AIP Conf. Proc. 1542, 1098–1101. https://doi.org/10.1063/1.4812127 Palmer, S., Ingram, A., Fan, X., Fitzpatrick, S., Seville, J., 2007. Investigation of the Sources of Variability in the Wurster Coater: Analysis of Particle Cycle Times using PEPT. 12th Int. Conf. Fluid. - New Horiz. Fluid. Eng. Pinto, J.T., Brachkova, M.I., Fernandes, A.I., Pinto, J.F., 2016. Evaluation of the ability of powdered milk to produce minitablets containing paracetamol for the paediatric population. Chem. Eng. Res. Des., 7th International Granulation Workshop, 2015 110, 171–182. https://doi.org/10.1016/j.cherd.2016.04.014 Rautenbach, C., Mudde, R.F., Yang, X., Melaaen, M.C., Halvorsen, B.M., 2013. A comparative study between electrical capacitance tomography and time-resolved X-ray tomography. Flow Meas. Instrum. 30, 34–44. https://doi.org/10.1016/j.flowmeasinst.2012.11.005 Sánchez Quintanilla, M.Á., Espin, M.J., Valverde, J.M., 2014. Effect of magnetic field orientation on fluidized beds of magnetic particles: Theory and experiment. Particuology, Special issue on conveying and handling of particulate solids – Challenges of discrete element simulation, application and calibration 12, 54–63. https://doi.org/10.1016/j.partic.2013.03.002 Shelukar, S., Ho, J., Zega, J., Roland, E., Yeh, N., Quiram, D., Nole, A., Katdare, A., Reynolds, S., 2000. Identification and characterization of factors controlling tablet coating uniformity in a Wurster coating process. Powder Technol. 110, 29–36. https://doi.org/10.1016/S00325910(99)00265-X Šibanc, R., Žun, I., Dreu, R., 2016. Measurement of particle concentration in a Wurster coater draft tube using light attenuation. Chem. Eng. Res. Des., 7th International Granulation Workshop, 2015 110, 20–31. https://doi.org/10.1016/j.cherd.2016.01.027 Thabet, Y., Walsh, J., Breitkreutz, J., 2018. Flexible and precise dosing of enalapril maleate for all paediatric age groups utilizing orodispersible minitablets. Int. J. Pharm. 541, 136–142. https://doi.org/10.1016/j.ijpharm.2018.02.037 Tissen, C., Woertz, K., Breitkreutz, J., Kleinebudde, P., 2011. Development of mini-tablets with 1mm and 2mm diameter. Int. J. Pharm. 416, 164–170. https://doi.org/10.1016/j.ijpharm.2011.06.027
A
CC
EP
TE D
M
A
N
U
SC R
IP T
Turton, R., 2010. The application of modeling techniques to film-coating processes. Drug Dev. Ind. Pharm. 36, 143–151. https://doi.org/10.3109/03639040903452278 Turton, R., 2008. Challenges in the modeling and prediction of coating of pharmaceutical dosage forms. Powder Technol. 181, 186–194. https://doi.org/10.1016/j.powtec.2006.12.006 Weber, J.M., Bobek, M.M., Breault, R.W., Mei, J.S., Shadle, L.J., 2018. Investigation of core-annular flow in an industrial scale circulating fluidized bed riser with electrical capacitance volume tomography (ECVT). Powder Technol. 327, 524–535. https://doi.org/10.1016/j.powtec.2017.12.094 Yang, L., Padding, J.T., Buist, K.A., Kuipers, J.A.M., 2017. Three-dimensional fluidized beds with rough spheres: Validation of a Two Fluid Model by Magnetic Particle Tracking and discrete particle simulations. Chem. Eng. Sci. 174, 238–258. https://doi.org/10.1016/j.ces.2017.09.014
IP T SC R U N A M TE D
A
CC
EP
Fig. 1. 2.5 mm mini-tablets coated using CW coater at air flow of 156 m3/h
IP T SC R
A
CC
EP
TE D
M
A
N
U
Fig. 2. Schematic of classical Wurster (CW) and swirl Wurster (SW) coater, arrows indicate the inlet air velocity vectors as derived by use of corresponding air distribution plates
Fig. 3. Schematic of cycle time measurement setup (left: side view, right: top view).
IP T SC R U N
A
CC
EP
TE D
M
A
Fig. 4. Schematic transmittance measurement setup (left: side view, right: top view).
Fig. 5. Simulated transmittance distributions for varying volume fraction of 2.0 mm mini-tablets
IP T SC R U N
A
CC
EP
TE D
M
A
Fig. 6. Coating variability of different mini-tablets coated using different coater and coating conditions
M
A
N
U
SC R
IP T
Fig. 7. Cycle times distributions for 2.0 and 3.0 mm mini-tablets, showing only cycle times limited to 13 s
A
CC
EP
TE D
Fig. 8. Simulation derived distributions for number of particle passes within 7200 s coating time
IP T SC R U N
A
CC
EP
TE D
M
A
Fig. 9. Number of passes distributions for coating times per mini-tablet size. 8735 s for MT2.0, 7040 s for MT2.5, and 6282 s for MT3.0.
U
SC R
IP T
Fig. 10. Average transmittance at the top of the Wurster draft tube
N
.
A
CC
EP
TE D
M
A
Fig. 11. Coater independent correlation between per pass variability C(X) and average transmittance at the top of the Wurster draft tube
IP T SC R U N
A
CC
EP
TE D
M
A
Fig. 12. Average volume fraction of mini-tablet at the top of the Wurster draft tube
IP T SC R U N A M TE D EP CC
A
Fig. 13. Autocorrelation of transmittance signal for mini-tablets of different size in both coaters (top – CW coater, bottom - SW coater) at two different fluidization air flow rates
Table 1. Mini-tablets characteristics
3
Crushing force/N
Cap radius/mm Height/mm
Mass/mg
Density/g/cm
2.0
3.00
2.02
7.8
1.30
13.3
2.5
3.75
1.82
11.7
1.34
22.1
3.0
4.50
2.21
20.0
1.36
28.3
IP T
Diameter/mm
Experiment
Average coating thickness/µm
CV(m)/%
2.0_CW130
18.7
14.51
=
2.0_CW156
19.9
6.15
=
2.0_SW156
18.4
3.19
2.5_CW130
18.9
16.95
2.5_CW156
21.9
10.28
9.23
2.5_SW156
17.9
M
Table 2. Coating variability results and time extrapolated values of coating variability for coating time of MT2.0
3.08
2.77
3.0_CW130
18.8
19.08
16.18
3.0_CW156
18.2
12.45
10.55
3.0_SW156
18.3
4.84
4.11
SC R
U N
A
TE D
EP CC A
CV(m)T=2.0/%
=
15.22
Table 3. Mini-tablet cycle time measurements results within 5400 s of measuring time
Mean/s
Var/s2
Max/s
Skewness
2.0_CW130 1111
2.79
4.85
102.51
156.78
8.70
2.0_CW156 1568
2.16
3.42
21.73
66.26
7.00
2.0_SW156
938
4.81
5.74
10.90
36.09
2.90
2.5_CW130 987
3.59
5.46
64.15
143.70
8.54
2.5_CW156 1259
2.57
3.50
12.88
50.80
7.27
2.5_SW156
883
5.23
6.11
11.45
30.51
2.29
3.0_CW130 760
4.35
6.71
99.17
3.0_CW156 1089
3.04
4.94
91.26
3.0_SW156
5.56
5.98
9.09
900
IP T
Median/s
SC R
N
118.43
6.41
181.16
10.87
22.76
1.04
N
U
Experiment
CV(x)/%
CV(N)2/CV(m)2/%
4.92
579
11.5
2551 6.15
2.70
279
19.2
1521 3.19
1.47
110
21.4
1290 16.95
4.09
591
5.8
2.0_CW130
1799 14.51
2.0_CW156
EP
2.5_CW130
CV(N)/%
M
E(N) CV(m)/%
TE D
Experiment
2.0_SW156
A
Table 4. Mini-tablet coating variability source analysis
2014 10.28
2.29
450
4.9
2.5_SW156
1153 3.08
1.63
89
28.0
3.0_CW130
937
19.08
4.85
565
6.5
3.0_CW156
1273 12.45
5.42
400
19.0
1050 4.84
1.56
149
10.3
A
CC
2.5_CW156
3.0_SW156
Table 5. Transmittance data (T) and volume fractions (VF) of mini-tablets at the top of Wurster draft tube
T SD/%
T RSD/%
VF/%
VF SD/%
VF RSD/%
2.0_CW130 19.2
8.2
42.9
4.94
1.11
22.4
2.0_CW156 24.3
9.1
37.5
4.24
0.98
23.1
2.0_SW130 44.6
11.1
24.9
2.42
0.68
28.0
2.0_SW156 26.0
9.3
35.8
4.05
0.93
23.1
2.5_CW130 24.2
10.3
42.6
4.99
1.30
2.5_CW156 29.5
10.9
36.8
4.29
1.13
2.5_SW130 51.4
12.4
24.0
2.33
2.5_SW156 35.8
11.3
31.7
3.61
3.0_CW130 35.5
13.6
38.3
4.64
3.0_CW156 27.8
12.7
45.7
5.74
3.0_SW130 58.2
14.2
24.3
3.0_SW156 43.6
13.9
32.0
26.1
33.0
0.97
27.0
1.52
32.7
1.82
31.7
2.39
0.99
41.3
3.71
1.27
34.3
U
SC R
26.4
0.77
N A
M TE D EP CC A
IP T
Experiment T/%
S0263-8762(18)30134-5 https://doi.org/10.1016/j.cherd.2018.03.020 CHERD 3090
To appear in: Received date: Revised date: Accepted date:
1-9-2017 7-3-2018 14-3-2018
ˇ Please cite this article as: Sibanc, Rok, Turk, Magdalena, Dreu, Rok, AN ANALYSIS OF THE MINI-TABLET FLUIDIZED BED COATING PROCESS.Chemical Engineering Research and Design https://doi.org/10.1016/j.cherd.2018.03.020 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
AN ANALYSIS OF THE MINI-TABLET FLUIDIZED BED COATING PROCESS
Rok Šibanc1,3,*, Magdalena Turk2,3 & Rok Dreu3
A
N
2, 2.5 and 3 mm diameter mini-tablets were coated in two fluidized bed coaters Cycle times of mini-tablets were measured with a photoluminescent system Effect of cycle time variability on the coating variability was between 5 and 28 % Particle volume fraction was determined using transmittance measurements
M
U
HIGHLIGHTS
SC R
*Corresponding author: [email protected], +49 211 81 14225 Other authors: [email protected], [email protected]
IP T
1 Institute of Pharmaceutics and Biopharmaceutics, Heinrich-Heine-University, Universitätsstr. 1, 40225 Düsseldorf, Germany 2 Department of Pharmaceutical Technology, Faculty of Pharmacy, Medical University of Gdansk, Hallera 107, 80-416 Gdansk, Poland 3 Department of Pharmaceutical Technology, Faculty of Pharmacy, University of Ljubljana, Aškerčeva cesta 7, 1000 Ljubljana
TE D
ABSTRACT
Mini-tablets with diameters of 2.0, 2.5, and 3.0 mm are coated in two different lab-scale fluidized bed coaters equipped with a Wurster draft tube. The main focus of the research is to evaluate the
EP
inter-particle coating variability, and to assess the contribution of cycle time variation. Cycle times are measured using a photoluminescent tracer with a detector mounted on the top of the draft tube.
CC
The number of passes variability is represented from 5 to 28% of the total coating variability. Additionally, transmittance measurements at the top of the Wurster draft tube are performed in order to assess the inter-particle sheltering effects. Transmittance results are correlated to the amount of
A
coating deposited per single pass of the spray zone and are converted to solids volume fractions. The dynamics of the transmittance signal further reveal the persistence of a particle arrangement within the draft tube of the two different coaters. The gathered results give insight into the different performance of two fluidized bed coaters in terms of inter-particle coating variability.
KEYWORDS Mini-tablets; fluidized bed coating; coating variability; cycle time; Wurster coater
1
INTRODUCTION Mini-tablets are tablets with a maximum diameter of 3 mm and are becoming an increasingly
popular oral dosage form, especially for pediatric and geriatric populations (Tissen et al., 2011;
IP T
Klingmann et al., 2013, 2015; Kluk et al., 2015; Aleksovski et al., 2015; Gaber et al., 2015; Pinto et al., 2016; Thabet et al., 2018). Tablets are traditionally coated in drum coaters, however, mini-tablets are coated using either fluidized bed coaters (Bodea et al., 2010; Mohamed et al., 2015) or drum
SC R
coaters (Mondelli et al., 2010). In order for the applied coating to perform its function it has to have the required thickness. So, the mean coating thickness of all tablets not only has to be equal to the required thickness, but also the coating thickness of each individual tablet has to be as close as
U
possible to the required thickness or in some of applications at least not less than the required value.
N
The (non)uniformity of coating thickness between tablets is commonly expressed using a coefficient
A
of variation (CV).
Prediction and troubleshooting of the coating process may be performed by modeling. Models are
M
categorized as macro- and micro-levels, where macrolevels describe processes applied to the whole batch, while microlevels consider phoneme of film creation in a single core scale. Macrolevels
TE D
include: renewal theory, compartment models, Monte Carlo methods and population balance models (Turton, 2008, 2010). Renewal theory is a simple model which describes two contributions of the coating process variability. It considers the coating process as a repeating process, where the total
EP
coating amount on a particle m is a sum of its N passes through the coating zone, where a small but variable amount of coating x is received at each pass (Cheng and Turton, 2000a; Mann, 1983; Turton,
CC
2008).
𝑚 = ∑𝑁 𝑖=1 𝑥𝑖
Eq (1)
A
Both N and x are distributions described by mean and variance. The resulting coefficient of variability of coating CV(m) can be considered as a sum of variances of both factors: 𝐶𝑉(𝑚) =
𝑆𝑇𝐷(𝑚) 𝐸(𝑚)
𝑉𝑎𝑟(𝑥)
𝑉𝑎𝑟(𝑁)
= √[𝐸(𝑥)]2 [𝐸(𝑁)] + [𝐸(𝑁)]2
Eq (2)
The E represents the expected value (or mean) of the variable, and if Var/E2 is expressed as CV2 the equation has the following simplified form:
𝐶𝑉(𝑚) = √
[𝐶𝑉(𝑥)]2 𝐸(𝑁)
+ [𝐶𝑉(𝑁)]2
Eq (3)
The number of passes N can be also expressed with the time between passes through the spraying zone - t and the total coating time T (i.e. E(N)=T/E(t)). It was shown (Mann, 1983) that CV(N) can be expressed with the circulation time distribution and Eq(3) expressed as: 𝐶𝑉(𝑚) = √
𝑉𝑎𝑟(𝑥)𝐸(𝑡) [𝐸(𝑥)]2 𝑇
+
𝑉𝑎𝑟(𝑡)
Eq (4)
TE(t)
IP T
Eq (4) is valid when the number of circulations is larger than 30, according to Mann (1983). Magnetic particle tracking was successfully used for describing and visualization of particles movement in fluid bed systems (Sánchez Quintanilla et al., 2014; Yang et al., 2017) including rotor
SC R
fluid bed system (Neuwirth et al., 2013) or spouted bed (Mohs et al., 2009). Cheng and Turton (2000b, 2000a) performed a series of fluidized bed experiments using a magnetic tracer particle to study the cycle times of pellets, and to evaluate the contribution of variation of cycle times to the coating mass variation. Shelukar et al. (2000) performed similar experiments using tablets and performed
U
modification to the distribution plate in order to reduce the variability of tablet cycle times. Their
N
studies establish that the contribution due to variation in circulation time to tablets is between 14 and
A
24%. If the reported data from Cheng and Turton is used and calculated according to Shelukar et al., then the contribution of circulation times for variability of coating on pellets is only up to 3.4%.
M
Positron emission particle tracking (PEPT) is a superior technique with regard to the study of particle cycle times, since it yields a mm resolution position of a radioactive labelled particle over time. As a
TE D
result, it can also detect particle recirculation events within a specific zone and elucidate residence times in different zones of the process chamber. The advantage of the PEPT method is that particle tracking can also be accomplished for process chambers at elevated temperatures and pressures. The
EP
disadvantages of this method are cost of the equipment, size limitation and thickness of process chamber walls, and the need of radioactive tracer. The half-life of a radioactively labeled particle may be a limitation for the required total tracking time in terms of the required spatial resolution of
CC
trajectory. PEPT was used in a few studies to analyze a Wurster fluidized bed (Chang et al., 2013; Li et al., 2015a; Palmer et al., 2007) or a top spray fluidized bed (Depypere et al., 2009). Tablet cycle
A
and residence times in a pan coater can be determined using video imaging, where the orientation of the tablets can also be tracked (Kandela et al., 2010). Cycle and residence times, as well as coating uniformity, can also be studied by means of simulations such as a discrete element method (DEM) (Kumar et al., 2015) or in the case of a fluidized bed with a combination of DEM with computational fluid dynamics (CFD) (Fries et al., 2013; Jiang et al., 2017; Li et al., 2015b).
Another important parameter in the analysis of fluidized bed processes is the amount of material in the spraying zone and is normally expressed as a volume fraction. Volume fraction of solids directly influences the per-pass variability due to sheltering effects (Cheng and Turton, 2000a). Local solids volume fraction measurements can be performed using a reflective optical fiber (Johnsson and Johnsson, 2001) or using transmittance measurements (Šibanc et al., 2016). For determination of volume fraction distribution capacitive or X-ray tomography methods were used (Rautenbach et al., 2013; Hensler et al., 2016; Weber et al., 2018).
IP T
The main motivation for this work is to understand the coating variability of coating of mini-tablets in a Wurster coater. Mini-tablets of three different sizes (diameter of 2.0, 2.5 and 3.0 mm) are used
SC R
in the study. The influence of coater design is also of interest, since significant differences in the coating of pellets were observed in previous research. Therefore, we compare a classical and swirl Wurster coater (Glatt GPCG-1 and Brinox CGD). As a main coating result, we investigate inter-tablet coating variability using dye content analysis. Besides the effect of the inlet air geometry, the effect
U
of fluidizing air flow is investigated at two different air flow rates (130 and 156 m/h3). In this study,
N
the mini-tablet cycle time measurements are performed in order to explain the contribution of number of passes variability to the total coating variability. Additionally, light attenuation measurements
A
(Šibanc et al., 2016) are conducted in order to evaluate the number density of mini-tablets at the top
M
of the Wurster draft tube, and to assess the dynamics of the system. The significance of this work is threefold. The raw data of coating variability of mini-tablets in fluidized beds is of importance,
TE D
especially due to increasing interested in pediatrics and individualized medicine. Secondly, the process understanding is increased and coating variability sources are elucidated using the inline process analytical tools. Lastly, the raw experimental data can be of use for the development of
2
EP
numerical simulations.
MATERIALS AND METHODS Mini-tablets preparation
CC
2.1
Biconvex mini-tablets with a diameter of 2.0 mm (MT2.0), 2.5 mm (MT2.5), and 3.0 mm (MT3.0) are prepared from a tableting mixture containing: lactose FlowLac® 100 (74.07% (w/w)),
A
microcrystalline cellulose Vivapur® 102 (24.68 % (w/w)), colloidal silicon dioxide Aerosil® 200 (0.25 % (w/w)), and sodium stearyl fumarate PRUV® (1.00 % (w/w)). Mini-tablets are pressed using a rotary tablet press (Erweka RTP-D8, Germany) using compacting pressures between 300 and 400 MPa.
2.2
Coating procedure
Mini-tablets are coated with a 10% (w/w) aqueous coating dispersion consisting of: HPMC (8.00% (w/w)), PEG 6000 (1.00% (w/w)), and tartrazine (1.00% (w/w)) as a colorant (Fig. 1). In each coating experiment 1000 grams of mini-tablets are coated with a target theoretical coating thickness of 20 µm. We spray a coating dispersion of 737 g (MT2.0), 587 g (MT2.5), or 524 g (MT3.0) of coating dispersion under the assumption of 70% coating efficiency. Film coatings are performed in a bottom spray fluid bed systems, using either classical (CW) or swirl Wurster (SW) chamber (Glatt GPCG-1,
IP T
Germany and Brinox CGD, Slovenia; Fig. 2). Air flow rates of 130 and 156 m3/h are used when utilizing the CW chamber, and we use an air flow rate of 156 m3/h in the case of an SW chamber. A constant spray rate of 5 g/min, an atomizing air pressure of 2 bar, inlet air temperature of 55°C, and
SC R
a gap of 25 mm are used in all cases. Since the same spray rate is used for all mini-tablet sizes with different specific surfaces and the aim is to apply the same coating thickness regardless of mini-tablet type, the coating times varied between tablet sizes. The coating time is 8735 s for MT2.0, 7040 s for
Analysis of coating uniformity
N
2.3
U
MT2.5, and 6282 s for MT3.0, respectively.
An analysis of mini-tablets coating uniformity is based on spectrophotometric measurements of
A
tartrazine concentration. 100 mini-tablets are analyzed using the UV-VIS method. Each of 100 mini-
M
tablets is placed in a single vial and diluted with 4.5 ml of dihydrogen phosphate buffer (6.5 pH). After the dissolving of water soluble ingredients, vials are centrifuged (15 min, 3500 rpm) to separate
TE D
undissolved particles of disintegrated mini-tablet. Samples are analyzed for absorbance (UV spectrophotometer HP8453, Hewlett-Packard, USA) at 425 nm wavelength. Further, based on tartrazine concentration, coating mixture composition and dry film density (1.27 g/cm3) as well as surface of single mini-tablet, film thickness of each mini-tablet and coefficient of variation of per-
Cycle time measurements
CC
2.4
EP
particle coating mass (CV(m)) is calculated.
Cycle time measurements of mini-tablets are performed using a photoluminescent detection system, primarily developed for pellet cycle time measurements. A single tracer mini-tablet is coated
A
with photoluminescent coating (ethanol dispersion of PVP K25 and CuS:EU LED phosphor in 6:1 ratio), whereas all the other mini-tablets are coated with a carbon black coating (8% HPMC, 1% PEG 6000, and 1.7 % active carbon). The detection system is placed on the top of the Wurster draft tube and faces outwards. It consists of an excitation part composed of blue LED (XP-E Royal Blue, Cree Inc., USA) and a detection part comprising of photodiode sensitive to red light (EPD-660-1-0.9, Epigap Optronic GmbH, Germany). In order to improve the signal-to-noise ratio, the excitation part
is equipped with a blue filter transparent for wavelengths from 400 – 500 nm (363 Special Medium Blue, Lee Filters, USA) and the detector with a red color filter transparent for wavelengths above 600 nm (CL182 Cool LED Light Red, Lee Filters USA). A transimpedance amplifier which amplifies and converts photoelectric current to voltage is a part of the custom designed circuit board. The output voltages are recorded using an external analog to digital converter. The measurements are performed for 1.5 h in each case. To simulate coating conditions purified water is sprayed during the measurements. Conversion to number of cycles
IP T
2.4.1
The cycle time to number of passes conversion representative for population of mini-tablets is
SC R
done using a numerical simulation, where random tracer mini-tablet cycle times are drawn from the recorded cycle time population and summed for each of simulated particle until the target simulated coating time is achieved. Target simulated coating time is 7200 s in the first series of simulation, while in the second series the time is set as in the actual coating experiment, depending on the mini-
U
tablet size – specific surface. 50,000 repetitions of the described numerical experiments are performed
N
for each case. The parameters of the number of passes - N distributions do not change significantly
Transmittance measurements
M
2.5
A
above the selected number of repetitions.
Light attenuation measurements are performed in order to evaluate the inter-particle shielding of mini-tablets in the Wurster draft tube, and to assess the dynamics of the system. The measurement
TE D
system consists of a laser source and a 10 x 10 mm2 photodiode detector mounted on a glass Wurster tube. Measurements are performed at the top of the Wurster glass tube (Fig. 4). Transmittance measurements are performed for all mini-tablets sizes in both coater types at both air flow rates. For
EP
each measurement, the signal is recorded for 5 minutes at an acquisition frequency of 50 kHz. More details about the method are available in previous publication where the volume fraction of pellets
CC
was determined using such a system (Šibanc et al., 2016). It was shown that the measurements at the top of the Wurster draft tube correlate with the global volume fractions of solids in the whole Wurster
A
draft tube.
2.5.1
Conversion of transmittance data to volume fractions
The transmittance is dependent on the amount (volume fraction) of mini-tablets in the measurement volume, their shape, orientation, and position. A Monte Carlo simulation like in a previous study (Šibanc et al., 2016) is employed to obtain transmittance distributions for volume
fractions between 0.1 and 10% in 0.1% steps. An exemplary simulated sensor response for five different volume fractions is shown in Fig. 5. The average transmittance follows Beer-Lambert law, which describes exponential decrease of transmittance depending on volume fraction. The reason for distribution of transmittance at a fixed volume fraction is the discrete nature of the relatively low amount of particles in the measurement volume. By varying the position of the same amount of particles, arrangements with varying transmittance are obtained. Recorded transmittance distribution data are then deconvoluted to volume fraction distributions via usage of simulated transmittance
IP T
distribution data database, as described in a previous study. Since the Monte Carlo simulation uses spherical particles for simulation of transmittance, the circle
steps:
SC R
equivalent diameter of average mini-tablet projections are calculated. This was done in the following
1. A 3D model of a mini-tablet based on its average dimensions is constructed in a 3D raytracing
U
program (Blender 2.78, open-source) at location 0,0,0.
N
2. Camera object is set at location 0,-10,0, with a render resolution of 1500 x 1500 px and an
A
orthographic projection which corresponds to parallel laser light rays.
M
3. Mini-tablets are rotated around the x axis from sin(alpha)=0 to sin(alpha)=1 in 50 steps and an image render is created at each step.
TE D
4. The mini-tablet projection size at each angle is determined as the number of pixels belonging to the mini-tablet multiplied by the square of the pixel size.
2.5.2
EP
5. The average projection size is used to calculate the circle equivalent diameter. Autocorrelation
CC
Autocorrelation of the whole transmittance signal (300 s at 50 kHz) is performed for all 12 cases. Autocorrelation of a signal is a correlation of a signal with the same signal shifted by some amount, commonly named as ‘lag.’ Based on preliminary signal studies, autocorrelation with lags up to 20 ms
A
are performed. Autocorrelation is a time domain analysis and provides an insight into particle dynamics in this case. 3
RESULTS AND DISCUSSION
3.1
Coating variability analysis
Coating variability (CV(m)) of mini-tablets based on spectrophotometric measurements of dye concentration (Table 2, Fig. 6) is in a range from 3.08% to 19.08%. The best result is obtained when
coating MT2.5 using SW coater and the worst result using CW coater, while coating MT3.0 at lower fluidizing air flow. By comparing the coater design it can be seen that the mini-tablets coated with SW always have lower coating mass variability. The lower values of the fluidizing air flow, which were tested using CW, always lead to higher coating variability. It is assumed that this is due to higher packing of mini-tablets in the coating zone, as well as due to lower particle circulation rates (with probable recirculation events in the spray zone). Increasing the size of mini-tablets had a negative effect on the coating variability. Presumably, larger mini-tablets had less uniform circulation as gap
IP T
size is maintained at its maximum position – posing greater obstruction for bigger mini-tablets, but also very important the coating time is different for every mini-tablet size and is decreasing with minitablet size in order to achieve the same level of coating thickness. A coating variability of MT2.5 and
SC R
MT3.0 is predicted for the coating time of MT2.0 according to Eq (4), that is that the coating variability decreases proportionally to 1/T0.5. Also, in the case of time extrapolated coating variability, values MT3.0 have the highest coating variability. Cycle time results
U
3.2
N
Mini-tablet cycle time distributions for six experiments (Fig. 7) primarily show the differences between CW and SW coater. The mode values of the cycle time distributions of CW coater are lower
A
than the SW ones, and also the widths of the distributions of CW seem to be lower. 2.0 mm mini-
M
tablets have lower cycle times than the 3.0 mm mini-tablets, which while observing results at the same inlet air flow rate confirms the flow limiting effect of the gap between draft tube and distribution
TE D
plate set to 25 mm. The results of 2.5 mm diameter mini-tablets, which are not shown in the plot, are in all cases between the results of 2.0 and 3.0 mm mini-tablets. The total number of tracer mini-tablet passes N, median, mean, variance, and maximum cycle
EP
times of cycle time measurements are summarized in Table 3. In can be seen that within the measuring time of 5400 s the tracer tablet passes the Wurster partition on average more times in the CW coater.
CC
The mean circulation times in all conducted experiments are between 3.42 and 6.71 s, whereas the median cycle times are between 2.16 and 5.56 s. This indicates an asymmetric distribution, which is confirmed by the skewness parameter, as well as a visual observation of distributions (Fig. 7). All
A
skewness values are positive, meaning the cycle time distributions have a number of long cycle times. The skewness values, as well as the maximum cycle times, are much higher in the case of a CW coater when compared to a SW coater. A very important parameter for uniform coating is the variability of cycle times, which can be described by its variance. The range of variance was 12.88 to 102.51 s2 for the CW coater, and from 9.09 to 11.15 s2 for the SW coater.
Simulation-derived distributions for the number of particle passes simulating equal coating time of 7200 s (Fig. 8) show the swirl chamber has narrower distributions of number of passes, but also generally lower mean values in comparison to the CW chamber. Derived distributions also show that the average number of passes through the spraying zone is not affected by mini-tablet size in the case of a SW coater. This could indicate that the SW operates at a lower particle circulation rate than the CW chamber, and is therefore not limited by the gap size as the mini-tablet size changes. The opposite was observed in the case of a CW. It is evident that the number of passes of MT3.0 in the CW chamber
IP T
was lower than for the other two mini-tablets sizes. The number of passes was also dependent on the air flow rate, and as expected the lower air flow rate led to a lower number of circulations.
SC R
Distributions for the number of particle passes were also calculated for the actual coating times of mini-tablets (Fig. 9). Since the coating time was dependent on the mini-tablet size, the differences in the number of passes for different sized mini-tablets are more pronounced. The largest effect of a mini-tablet size on the mean value of number of particle passes can be seen in the case of CW coater
U
using fluidizing air flow of 156 m3/h (i.e. more than 1000 passes difference between MT2.0 and
Coating variability analysis
A
3.3
N
MT3.0).
M
Coating variability CV(m) is analyzed in terms of the two contributions: variability of the number of passes CV(N) and the amount of deposited coating per pass CV(x) (Table 4). CV(x) is calculated
TE D
using Eq (3) and the values of CV(m), CV(N), and E(N). The term CV(N)2/CV(m)2 is the relative contribution of number of passes variability on total coating mass variability. As discussed above the lower variability of coating thickness is seen in the case of a swirl coater. According to the calculation, both CV(N) and CV(x) attribute to this result. CV(N) is in the case of SW, between 1.47 and 1.63%,
EP
and in the case of a CW coater between 2.29 and 5.42%. This indicates that the SW coater has better mixing or an absence of dead zones in the annulus bed region, and the most probable reason for that
CC
is the curved design of the distribution plate. Calculated CV(x) values range for the SW coater is between 88.8% and 148.6%, and in the case of a CW coater between 279 and 591%. A possible explanation of this phenomenon, besides the lower particle number density, could be that the swirling
A
air in SW coater pushes the particles away from the spraying nozzle thus making mini-tablets more equally distant from the spraying nozzle, which therefore results in a more even deposition per pass (Luštrik et al., 2013). Higher fluidizing air flow rate in all cases increases the number of passes E(N), reduces coating thickness variation CV(m), and reduces per pass variation CV(x). A possible explanation for reduced CV(x) at higher air flows is that with a higher air flow rate particles in the Wurster draft tube
accelerate faster, while horizontal mass flow particles are limited by the gap height and so a reduction of the particle volume fraction occurs, which in turn reduces the sheltering effect of particles from the spray. The relative contribution of number of passes variability is calculated to be between 10.3% and 28.0% for the SW coater, and in range from 4.9% to 19.2% for the CW coater. These results are comparable to the results of Shelukar et al., where they coated 4.5 kg of 7.9 mm tablets in a Niro
Transmittance
3.4.1
SC R
3.4
IP T
Wurster coater (Model MP1).
Measurements
Transmittance measurements are performed to evaluate the effect of process conditions and coater type on the amount (number density) of mini-tablets in the Wurster draft tube. The measurements are
U
performed at the exit of the draft tube (Table 5), which presumably correlates to the conditions in the
N
spray zone. The transmittance values are a measurement of light sheltering, and can be an indicator
A
of inter-particle sheltering from the spray. The transmittance is inversely proportional to the volume fraction of particles in the measuring volume, and is also a function of particle size. Therefore, the
M
transmittance values of differently sized mini-tablets cannot be compared directly.
TE D
For the same mini-tablet size and air flow the transmittance values obtained using the SW coater are always higher than for the CW coater (Fig. 10). This could at least partially explain why CV(x) values are smaller in the case of the SW coater. Since the measurements are performed across the Wurster draft tube, the resulting transmittance is a result of average volume fraction across the
EP
measuring volume and does not provide any information about the location of the mini-tablets. Increasing air flow rate increases the transmittance when MT2.0 and MT2.5 are used in the CW,
CC
however, the opposite is true for MT3.0. In all cases of the SW coater, air flow rate actually reduces the transmittance, which can indicate a change of particle flow pattern.
A
The idea for measuring the transmittance is that it correlates with the amount of particles present
in the Wurster draft tube, where the coating occurs. This can be used as a measure of inter-particle shielding from the spraying nozzle, which is one of the reasons for per pass variability of deposited coating amount CV(x). In the case of a higher number density of particles in the coating zone, the shielding effect is more pronounced. The correlation between the observed average transmittances and per pass variability CV(x) from Table 4 is in this case valid per mini-tablet size, and with an
outlier at 3.0 mm coated in the CW coater at 156 m3/h (Fig. 11). The reason why the correlation is valid per mini-tablet size and also for the outlier can be that the particle shielding is only one of the phenomena contributing to the per pass variability of the deposited coating amount. Another reason is the residence time or the particle velocity through the coating zone, which is for fluidized beds particle size-dependent. In addition, the particle trajectory through the draft tube also has an importance. Considering a hypothetical case where the same amount of particles are travelling through the draft tube, the particles travelling nearer to the nozzle receive more coating than the
IP T
particles further from the nozzle, purely based on the fact that the spray droplets concentration decreases with the distance from the nozzle. Of course, particle shielding further enhances the importance of trajectory. In the case of a SW coater, the swirling air pushes the particles more towards
SC R
the draft tube wall and unifies the particle trajectories (Luštrik et al., 2013). Increasing the transmittance further, for example with a decreasing gap width would mean having fewer particles in the spray zone. This could then potentially lead to some adverse effects such as spray drying, deposition of the coating material on the Wurster draft tube wall, and particle agglomerationVolume
N
U
fraction
In order to compare the hydrodynamics of the process the transmittance distributions are converted
A
to volume fraction distributions (Table 5). The average volume fractions are fairly constant across
M
mini-tablet size, except for the 3.0 mm tablet at CW156 (Fig. 12). The volume fraction of mini-tablets at the top of the draft tube is in all cases higher for the CW coater. The higher air flow rate decreases
TE D
the volume fraction in the CW coater and increases it in the SW coater. The air flow rate has a couple of effects on the fluidized bed. The higher air flow rate means higher air velocities in the draft tube, and a faster and more dilute flow under the assumption of a constant mass flow of particles. Secondly, higher air velocity also results in a higher suction force towards the inside of the draft tube in the
EP
region of the distribution plate. Lastly, a higher air flow rate reduces resistance to the flow of material in the annular region of the coater and enables a higher mass flow rate of particles, and consequently
CC
higher volume fractions. As a result, it is hard to attribute the results to a single phenomenon, but it is clear that their contributions are different for both coater types. The relative standard deviations
A
(RSD) of volume fractions are quite comparable for the same mini-tablet size, except for SW130, where the RSD is always the highest (Table 5). This is due to much lower average values of volume fractions. In terms of absolute volume fraction deviations, it is the lowest in such a combination of coater type and air flow rate, i.e. for all mini-tablets sizes.
3.4.2
Dynamics
Autocorrelation of a transmittance signal time series show how quickly the correlation vanishes and can be related to the particle dynamics. A faster decreasing correlation may indicate a faster change of the particle arrangement in the measurement volume. It can be postulated that higher autocorrelation values indicate movement of particles in clusters with little local mixing, which in terms facilitates (amplifies) the influence of the inter-particle sheltering at the same transmittance value. Fig. 13 shows autocorrelation signals of all 12 transmittance measurements. The thin lines
IP T
represent 2.0 mm mini-tablets, the dashed for 2.5 mm, and the 3.0 mm are represented with thick lines, whereas the darker color represents a lower air flow rate. The upper part of Fig. 13 shows the autocorrelations of measurements in the CW coater. Differences between all six cases can be seen,
SC R
meaning the method and analysis is sensitive at least for relatively big changes. An effect of the air flow rate can be observed for the 2.0 and 2.5 mm mini-tablets, as the autocorrelation decreases quicker for higher air flow rates. This is expected since the higher air flow rate normally results in higher particle velocity. The effect of air flow rate can be seen in the SW as well, although to a lesser extent
U
then in the CW. In case of the SW coater, mini-tablet size plays a much smaller role than in the CW
N
coater. This lower particle size sensitivity is also seen in the number of circulation results, and shows
A
that the SW coater can be considered more suitable when coating a distribution of particles and an equal coating thickness is applicable to all particles. Lower autocorrelation values of the SW
M
compared to the CW supports the particle cluster movement hypothesis, as also the total coating variability is on average smaller than in the case of the CW, and analysis has shown the particle per
TE D
pass effect has a predominant part in the total coating amount variability. The transmittance – CV(x) correlation outlier (CW156 - MT3.0, Fig. 11) can be explained by the autocorrelation curve, which is higher in this instance than expected for such an air flow rate. As higher autocorrelation values mean
4
CC
EP
less local particle mixing, this situation leads to higher CV(x) even at intermediate transmittance valu
CONCLUSIONS
Mini-tablets are successfully coated in two different fluidized bed coaters yielding inter-particle
A
coating variability in the range from 3.1% to 19.1%. We observe that the swirl coater performs better in all three sizes of mini-tablets, and has the maximum coating variability of 4.8% – using 3.0 mm mini-tablets. Coating of smaller mini-tablets always results in smaller coating variability, independent of coater type. Cycle time results show that the swirl coater has a lower number of passes variability, although the number of passes was lower than in the classical Wurster coater with the same coating parameters.
Based on the cycle time and coating variability results, the contribution of the number of passes variability and the particle per pass variability of the deposited coating amount is calculated. The contribution of the number of passes variability is in a range from 5 to 28%. The per pass variability ranges from 89 to 591%, and is always lower in cases of the SW coater. It is also shown that the transmittance measurements partially explain the particle per pass coating variability. Higher transmittance measured at the top of the Wurster draft tube correlates with lower per pass variability for a specific mini-tablet size. Transmittance data is also converted to volume
IP T
fraction data, and is also used to perform additional dynamics evaluation. Volume fractions are in a range from 2.3 to 5.7% across all cases, and as such can be useful for validation of simulations. Also,
SC R
autocorrelations of transmittance signal primarily show the difference between both coater types – that is, the SW coater shows less dependence on mini-tablet size for many measured parameters: coating variability, number of circulations, and volume fractions.
U
Future work will be focused on a broader experimental plan of cycle times. Cycle times will be measured for a different design of distributor plates, and the effect of the gap between the Wurster
N
draft tube and the distributor plate will be also evaluated. Additionally, studies of intra-tablet coating
M
ACKNOWLEDGEMENTS
A
variability are also planned.
TE D
The authors would like to thank Domen Kitak for the help with the development of the cycle time measuring system, and Grega Ratek for the help with the cycle time measurements. Funding: This work was supported by The Applied Research Program of The National Centre for
Foundation.
CC
APPENDIX
EP
Research and Development (grant number PBS1/A7/3/2012), and the Alexander von Humboldt
A
List of abbreviationsm mass of coating t cycle time x amount of coating per pass CFD computational fluid dynamics CV coefficient of variation
CW classical Wurster coater DEM discrete element method PEPT positron emission particle tracking RSD relative standard deviation
IP T
MT mini-tablet N number of passes
SC R
SW swirl Wurster coater
A
CC
EP
TE D
M
A
N
U
T total coating time
REFERENCES
A
CC
EP
TE D
M
A
N
U
SC R
IP T
Aleksovski, A., Dreu, R., Gašperlin, M., Planinšek, O., 2015. Mini-tablets: a contemporary system for oral drug delivery in targeted patient groups. Expert Opin. Drug Deliv. 12, 65–84. https://doi.org/10.1517/17425247.2014.951633 Bodea, M., Tomuţă, I., Leucuta, S., 2010. Film coating preparation of metoprolol tartrate mini-tablets and in vitro drug release studies. Clujul Med. 83, 457–464. Chang, Y.-F., Adamsen, T.C.H., Pisarev, G.I., Hoffmann, A.C., 2013. PEPT: An invaluable tool for 3-D particle tracking and CFD simulation verification in hydrocyclone studies. EPJ Web Conf. 50, 05001. https://doi.org/10.1051/epjconf/20135005001 Cheng, X.X., Turton, R., 2000a. The prediction of variability occurring in fluidized bed coating equipment. II. The role of nonuniform particle coverage as particles pass through the spray zone. Pharm. Dev. Technol. 5, 323–332. Cheng, X.X., Turton, R., 2000b. The prediction of variability occurring in fluidized bed coating equipment. I. The measurement of particle circulation rates in a bottom-spray fluidized bed coater. Pharm. Dev. Technol. 5, 311–322. Depypere, F., Pieters, J.G., Dewettinck, K., 2009. PEPT visualisation of particle motion in a tapered fluidised bed coater. J. Food Eng. 93, 324–336. https://doi.org/10.1016/j.jfoodeng.2009.01.042 Fries, L., Antonyuk, S., Heinrich, S., Dopfer, D., Palzer, S., 2013. Collision dynamics in fluidised bed granulators: A DEM-CFD study. Chem. Eng. Sci. 86, 108–123. https://doi.org/10.1016/j.ces.2012.06.026 Gaber, D.M., Nafee, N., Abdallah, O.Y., 2015. Mini-tablets versus pellets as promising multiparticulate modified release delivery systems for highly soluble drugs. Int. J. Pharm. 488, 86–94. https://doi.org/10.1016/j.ijpharm.2015.04.021 Hensler, T., Firsching, M., Bonilla, J.S.G., Wörlein, T., Uhlmann, N., Wirth, K.-E., 2016. Noninvasive investigation of the cross-sectional solids distribution in CFB risers by X-ray computed tomography. Powder Technol. 297, 247–258. https://doi.org/10.1016/j.powtec.2016.04.030 Jiang, Z., Bück, A., Tsotsas, E., 2017. CFD-DEM study of residence time, droplet deposition and collision velocity for a binary particle mixture in a wurster fluidized bed coater. Dry. Technol. https://doi.org/10.1080/07373937.2017.1319852 Johnsson, H., Johnsson, F., 2001. Measurements of local solids volume-fraction in fluidized bed boilers. Powder Technol. 115, 13–26. https://doi.org/10.1016/S0032-5910(00)00270-9 Kandela, B., Sheorey, U., Banerjee, A., Bellare, J., 2010. Study of tablet-coating parameters for a pan coater through video imaging and Monte Carlo simulation. Powder Technol. 204, 103–112. https://doi.org/10.1016/j.powtec.2010.07.024 Klingmann, V., Seitz, A., Meissner, T., Breitkreutz, J., Moeltner, A., Bosse, H.M., 2015. Acceptability of Uncoated Mini-Tablets in Neonates—A Randomized Controlled Trial. J. Pediatr. 167, 893–896.e2. https://doi.org/10.1016/j.jpeds.2015.07.010 Klingmann, V., Spomer, N., Lerch, C., Stoltenberg, I., Frömke, C., Bosse, H.M., Breitkreutz, J., Meissner, T., 2013. Favorable Acceptance of Mini-Tablets Compared with Syrup: A Randomized Controlled Trial in Infants and Preschool Children. J. Pediatr. 163, 1728– 1732.e1. https://doi.org/10.1016/j.jpeds.2013.07.014 Kluk, A., Sznitowska, M., Brandt, A., Sznurkowska, K., Plata-Nazar, K., Mysliwiec, M., Kaminska, B., Kotlowska, H., 2015. Can preschool-aged children swallow several minitablets at a time? Results from a clinical pilot study. Int. J. Pharm. 485, 1–6. https://doi.org/10.1016/j.ijpharm.2015.02.068 Kumar, R., Freireich, B., Wassgren, C., 2015. DEM–compartment–population balance model for particle coating in a horizontal rotating drum. Chem. Eng. Sci., Pharmaceutical Particles and Processing 125, 144–157. https://doi.org/10.1016/j.ces.2014.06.046
A
CC
EP
TE D
M
A
N
U
SC R
IP T
Li, L., Rasmuson, A., Ingram, A., Johansson, M., Remmelgas, J., von Corswant, C., Folestad, S., 2015a. PEPT study of particle cycle and residence time distributions in a Wurster fluid bed. AIChE J. 61, 756–768. https://doi.org/10.1002/aic.14692 Li, L., Remmelgas, J., van Wachem, B.G.M., von Corswant, C., Johansson, M., Folestad, S., Rasmuson, A., 2015b. Residence time distributions of different size particles in the spray zone of a Wurster fluid bed studied using DEM-CFD. Powder Technol. 280, 124–134. https://doi.org/10.1016/j.powtec.2015.04.031 Luštrik, M., Šibanc, R., Srčič, S., Perpar, M., Žun, I., Dreu, R., 2013. Characteristics of pellet flow in a Wurster coater draft tube utilizing piezoelectric probe. Powder Technol. 235, 640–651. https://doi.org/10.1016/j.powtec.2012.11.034 Mann, U., 1983. Analysis of spouted-bed coating and granulation. 1. Batch operation. Ind. Eng. Chem. Process Des. Dev. 22, 288–292. https://doi.org/10.1021/i200021a019 Mohamed, F.A.A., Roberts, M., Seton, L., Ford, J.L., Levina, M., Rajabi-Siahboomi, A.R., 2015. Film-coated matrix mini-tablets for the extended release of a water-soluble drug. Drug Dev. Ind. Pharm. 41, 623–630. https://doi.org/10.3109/03639045.2014.891128 Mohs, 7G., Gryczka, O., Heinrich, S., Mörl, L., 2009. Magnetic monitoring of a single particle in a prismatic spouted bed. Chem. Eng. Sci. 64, 4811–4825. https://doi.org/10.1016/j.ces.2009.08.025 Mondelli, G., Albertini, B., Funaro, C., Passerini, N., 2010. Minitablets Coated in a Solid-Wall Pan for Theophylline Sustained-Release Capsules. PharmTech 2010. Neuwirth, J., Antonyuk, S., Heinrich, S., 2013. Particle dynamics in the fluidized bed: Magnetic particle tracking and discrete particle modelling. AIP Conf. Proc. 1542, 1098–1101. https://doi.org/10.1063/1.4812127 Palmer, S., Ingram, A., Fan, X., Fitzpatrick, S., Seville, J., 2007. Investigation of the Sources of Variability in the Wurster Coater: Analysis of Particle Cycle Times using PEPT. 12th Int. Conf. Fluid. - New Horiz. Fluid. Eng. Pinto, J.T., Brachkova, M.I., Fernandes, A.I., Pinto, J.F., 2016. Evaluation of the ability of powdered milk to produce minitablets containing paracetamol for the paediatric population. Chem. Eng. Res. Des., 7th International Granulation Workshop, 2015 110, 171–182. https://doi.org/10.1016/j.cherd.2016.04.014 Rautenbach, C., Mudde, R.F., Yang, X., Melaaen, M.C., Halvorsen, B.M., 2013. A comparative study between electrical capacitance tomography and time-resolved X-ray tomography. Flow Meas. Instrum. 30, 34–44. https://doi.org/10.1016/j.flowmeasinst.2012.11.005 Sánchez Quintanilla, M.Á., Espin, M.J., Valverde, J.M., 2014. Effect of magnetic field orientation on fluidized beds of magnetic particles: Theory and experiment. Particuology, Special issue on conveying and handling of particulate solids – Challenges of discrete element simulation, application and calibration 12, 54–63. https://doi.org/10.1016/j.partic.2013.03.002 Shelukar, S., Ho, J., Zega, J., Roland, E., Yeh, N., Quiram, D., Nole, A., Katdare, A., Reynolds, S., 2000. Identification and characterization of factors controlling tablet coating uniformity in a Wurster coating process. Powder Technol. 110, 29–36. https://doi.org/10.1016/S00325910(99)00265-X Šibanc, R., Žun, I., Dreu, R., 2016. Measurement of particle concentration in a Wurster coater draft tube using light attenuation. Chem. Eng. Res. Des., 7th International Granulation Workshop, 2015 110, 20–31. https://doi.org/10.1016/j.cherd.2016.01.027 Thabet, Y., Walsh, J., Breitkreutz, J., 2018. Flexible and precise dosing of enalapril maleate for all paediatric age groups utilizing orodispersible minitablets. Int. J. Pharm. 541, 136–142. https://doi.org/10.1016/j.ijpharm.2018.02.037 Tissen, C., Woertz, K., Breitkreutz, J., Kleinebudde, P., 2011. Development of mini-tablets with 1mm and 2mm diameter. Int. J. Pharm. 416, 164–170. https://doi.org/10.1016/j.ijpharm.2011.06.027
A
CC
EP
TE D
M
A
N
U
SC R
IP T
Turton, R., 2010. The application of modeling techniques to film-coating processes. Drug Dev. Ind. Pharm. 36, 143–151. https://doi.org/10.3109/03639040903452278 Turton, R., 2008. Challenges in the modeling and prediction of coating of pharmaceutical dosage forms. Powder Technol. 181, 186–194. https://doi.org/10.1016/j.powtec.2006.12.006 Weber, J.M., Bobek, M.M., Breault, R.W., Mei, J.S., Shadle, L.J., 2018. Investigation of core-annular flow in an industrial scale circulating fluidized bed riser with electrical capacitance volume tomography (ECVT). Powder Technol. 327, 524–535. https://doi.org/10.1016/j.powtec.2017.12.094 Yang, L., Padding, J.T., Buist, K.A., Kuipers, J.A.M., 2017. Three-dimensional fluidized beds with rough spheres: Validation of a Two Fluid Model by Magnetic Particle Tracking and discrete particle simulations. Chem. Eng. Sci. 174, 238–258. https://doi.org/10.1016/j.ces.2017.09.014
IP T SC R U N A M TE D
A
CC
EP
Fig. 1. 2.5 mm mini-tablets coated using CW coater at air flow of 156 m3/h
IP T SC R
A
CC
EP
TE D
M
A
N
U
Fig. 2. Schematic of classical Wurster (CW) and swirl Wurster (SW) coater, arrows indicate the inlet air velocity vectors as derived by use of corresponding air distribution plates
Fig. 3. Schematic of cycle time measurement setup (left: side view, right: top view).
IP T SC R U N
A
CC
EP
TE D
M
A
Fig. 4. Schematic transmittance measurement setup (left: side view, right: top view).
Fig. 5. Simulated transmittance distributions for varying volume fraction of 2.0 mm mini-tablets
IP T SC R U N
A
CC
EP
TE D
M
A
Fig. 6. Coating variability of different mini-tablets coated using different coater and coating conditions
M
A
N
U
SC R
IP T
Fig. 7. Cycle times distributions for 2.0 and 3.0 mm mini-tablets, showing only cycle times limited to 13 s
A
CC
EP
TE D
Fig. 8. Simulation derived distributions for number of particle passes within 7200 s coating time
IP T SC R U N
A
CC
EP
TE D
M
A
Fig. 9. Number of passes distributions for coating times per mini-tablet size. 8735 s for MT2.0, 7040 s for MT2.5, and 6282 s for MT3.0.
U
SC R
IP T
Fig. 10. Average transmittance at the top of the Wurster draft tube
N
.
A
CC
EP
TE D
M
A
Fig. 11. Coater independent correlation between per pass variability C(X) and average transmittance at the top of the Wurster draft tube
IP T SC R U N
A
CC
EP
TE D
M
A
Fig. 12. Average volume fraction of mini-tablet at the top of the Wurster draft tube
IP T SC R U N A M TE D EP CC
A
Fig. 13. Autocorrelation of transmittance signal for mini-tablets of different size in both coaters (top – CW coater, bottom - SW coater) at two different fluidization air flow rates
Table 1. Mini-tablets characteristics
3
Crushing force/N
Cap radius/mm Height/mm
Mass/mg
Density/g/cm
2.0
3.00
2.02
7.8
1.30
13.3
2.5
3.75
1.82
11.7
1.34
22.1
3.0
4.50
2.21
20.0
1.36
28.3
IP T
Diameter/mm
Experiment
Average coating thickness/µm
CV(m)/%
2.0_CW130
18.7
14.51
=
2.0_CW156
19.9
6.15
=
2.0_SW156
18.4
3.19
2.5_CW130
18.9
16.95
2.5_CW156
21.9
10.28
9.23
2.5_SW156
17.9
M
Table 2. Coating variability results and time extrapolated values of coating variability for coating time of MT2.0
3.08
2.77
3.0_CW130
18.8
19.08
16.18
3.0_CW156
18.2
12.45
10.55
3.0_SW156
18.3
4.84
4.11
SC R
U N
A
TE D
EP CC A
CV(m)T=2.0/%
=
15.22
Table 3. Mini-tablet cycle time measurements results within 5400 s of measuring time
Mean/s
Var/s2
Max/s
Skewness
2.0_CW130 1111
2.79
4.85
102.51
156.78
8.70
2.0_CW156 1568
2.16
3.42
21.73
66.26
7.00
2.0_SW156
938
4.81
5.74
10.90
36.09
2.90
2.5_CW130 987
3.59
5.46
64.15
143.70
8.54
2.5_CW156 1259
2.57
3.50
12.88
50.80
7.27
2.5_SW156
883
5.23
6.11
11.45
30.51
2.29
3.0_CW130 760
4.35
6.71
99.17
3.0_CW156 1089
3.04
4.94
91.26
3.0_SW156
5.56
5.98
9.09
900
IP T
Median/s
SC R
N
118.43
6.41
181.16
10.87
22.76
1.04
N
U
Experiment
CV(x)/%
CV(N)2/CV(m)2/%
4.92
579
11.5
2551 6.15
2.70
279
19.2
1521 3.19
1.47
110
21.4
1290 16.95
4.09
591
5.8
2.0_CW130
1799 14.51
2.0_CW156
EP
2.5_CW130
CV(N)/%
M
E(N) CV(m)/%
TE D
Experiment
2.0_SW156
A
Table 4. Mini-tablet coating variability source analysis
2014 10.28
2.29
450
4.9
2.5_SW156
1153 3.08
1.63
89
28.0
3.0_CW130
937
19.08
4.85
565
6.5
3.0_CW156
1273 12.45
5.42
400
19.0
1050 4.84
1.56
149
10.3
A
CC
2.5_CW156
3.0_SW156
Table 5. Transmittance data (T) and volume fractions (VF) of mini-tablets at the top of Wurster draft tube
T SD/%
T RSD/%
VF/%
VF SD/%
VF RSD/%
2.0_CW130 19.2
8.2
42.9
4.94
1.11
22.4
2.0_CW156 24.3
9.1
37.5
4.24
0.98
23.1
2.0_SW130 44.6
11.1
24.9
2.42
0.68
28.0
2.0_SW156 26.0
9.3
35.8
4.05
0.93
23.1
2.5_CW130 24.2
10.3
42.6
4.99
1.30
2.5_CW156 29.5
10.9
36.8
4.29
1.13
2.5_SW130 51.4
12.4
24.0
2.33
2.5_SW156 35.8
11.3
31.7
3.61
3.0_CW130 35.5
13.6
38.3
4.64
3.0_CW156 27.8
12.7
45.7
5.74
3.0_SW130 58.2
14.2
24.3
3.0_SW156 43.6
13.9
32.0
26.1
33.0
0.97
27.0
1.52
32.7
1.82
31.7
2.39
0.99
41.3
3.71
1.27
34.3
U
SC R
26.4
0.77
N A
M TE D EP CC A
IP T
Experiment T/%