Effect of Equipment Parameters on Amount and Dispersion of Wetting Liquid in Planetary Centrifugal Granulation

Abstract

In planetary centrifugal wet granulation, the binder is often mixed into the formulation as a powder, followed by the addition of a wetting liquid, in a single step. Therefore, the amount and dispersion of the wetting liquid are important factors that determining granulation success and granules characteristics. In this study, granulation experiments, according to the Box–Behnken design, were performed. Further, the effects of equipment parameters, namely, processing speed, processing time, and vessel size, on the minimum amount of wetting liquid required to enable granulation and dispersion state in the vessel were statistically analyzed. Placebo granules were formulated with lactose hydrate and corn starch (7 : 3), using sodium carmellose as a binder. Results showed that the amount of wetting liquid decreased with increase in processing speed, processing time, and vessel size; however, the dispersion state of the wetting liquid was not significantly affected. Analysis of the effects of the equipment parameters on granule characteristics showed that a larger vessel size was proportional to a larger median diameter and smaller particle-size distribution width (span), and a faster processing speed was proportional to a smaller span. Furthermore, granules with the target properties could be prepared according to the parameters estimated from the model. In conclusion, the equipment parameters for controlling the amount of wetting liquid, which affected the granule properties, were clarified.

Introduction

Pharmaceutical tablets or capsules are reformulated into powders for children or elderly with undeveloped or impaired swallowing functions, respectively.¹⁾ However, these powders show poor dispensing and dosing as they are easy to scatter and adhesive, and have low flowability.²⁾ In contrast, granules are a dose-adjustable form that solve associated with powders.³⁾ Therefore, granulation is needed as one of the pharmaceutical compounding methods. We developed a small-scale granulation method called planetary centrifugal granulation, for the easier and more efficient preparation of granules in community and hospital pharmacies.⁴⁾ Practical examples include granule formulations of dantrolene sodium and rifampicin reformulated from capsules.^5,6) Although studies on the application of planetary centrifugal granulation are underway, basic studies are also needed to establish this method as one of the standard extemporaneous compounding procedures.

Planetary centrifugal granulation is a batch-type granulation method that uses a planetary centrifugal mixer (revolution–rotation speed ratio, 1 : 1) with a small intersection angle of approximately 15° between the revolution and rotation axes and a sealed cylindrical vessel, such as an ointment container, where granulation is performed by internal convection caused by the revolving and rotating motions.⁷⁾ The mixer settings are processing speed and time, and the device settings are container shape and size. The effects of operating parameters, such as processing speed, processing time, vessel size, and vessel loading rate, on granule characteristics have been investigated separately in previous studies.^4,6,8) However, factors determining the amount of binding liquid, which is the most important factor in wet granulation, and their effect on the granule properties have not been understood yet.

In the planetary centrifugal granulation method, the binder is often mixed in the formulation as a powder and a wetting liquid is added in a single step. The basic amount of the wetting liquid required is determined by the summation of the plastic limit (PL) values of the ingredients in the formulation.⁴⁾ However, this basic amount is present in a wide range, depending on the processing conditions. Therefore, trial-and-error studies need to be conducted, considering questions like: Which parameters should be adjusted to control the amount of wetting liquid? How will changing the amount of wetting liquid affect the granule characteristics of the resulting granules?

Therefore, this study aimed at investigating the effect of equipment parameters on the amount of wetting liquid required for the granulation and the dispersion state in planetary centrifugal wet granulation. The experiment was based on a three-factor, three-level Box–Behnken design. The independent variables were processing speed, processing time, and vessel size, selected based on previous studies.^4,6,8) The dependent variables were the minimum amount of wetting liquid required for the granulation and dispersion state of the wetting liquid in the vessel, expressed as a percentage of the PL value of the formulation (PL%), and the coefficient of variation of the wetting liquid content when sampled from six points in the vessel (CV%). In this study, we used placebo granules without an active pharmaceutical ingredient (API) because API affects PL% and CV%.⁹⁾ The effects of the independent variables on the PL% and CV% were analyzed using multiple regression and response surface methods. Simultaneously, the yield, median particle size, particle size distribution width, sphericity, and granule strength of the obtained granules were measured to evaluate the effect of equipment parameters on granule properties. Furthermore, the relationships between the amount and dispersion state of the wetting liquid and the granule characteristics were examined. Finally, the predictability of the generated mathematical models was evaluated by preparing the targeted granules with the desired characteristics.

Experimental

Materials

Powdered lactose monohydrate (median diameter: 37 µm) and corn starch (median diameter: 37 µm) were purchased from Pfizer Co., Ltd. (Tokyo, Japan) and used as the diluents. Sodium carboxymethyl cellulose (CMC) was obtained from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan) and was used as a binder. Ultrapure water was produced using an Arium Mini Plus (Sartorius Japan Co., Ltd., Tokyo, Japan) and used as the wetting liquid.

Experimental Design

A three-factor, three-level Box–Behnken design was used to analyze the relationship between the equipment parameters of planetary centrifugal granulation, such as the processing speed, processing time, and vessel size, and the minimum amount of water (wetting liquid) required for granulation and the dispersion state of water in the vessel. Table 1 listed the independent variables and their levels. A design of 15 experiments was formulated for 12 designated points, with three replicates at the center points (Table 2). The same design was used to evaluate the relationship between the equipment parameters and resulting granule properties. To evaluate the interaction effects of the independent variables, we employed a quadratic model (Eq. 1),

  

(1)

Table 1. Independent Variables and Their Levels for Box–Behnken Design

Independent variable	Level
	−1	0	1
X1: Processing speed (rpm)	800	1000	1200
X2: Processing time (s)	40	120	200
X3: Vessel size (mL)	12	35	58

Table 2. Box–Behnken Design, and Amount of Water and Dispersion State of Water

Code	X1	X2	X3	Amount of water		Dispersion state of water (CV %)
				(PL %)	(mL/g)*
1	−1	−1	0	70.8	0.282	7.6
2	1	−1	0	59.5	0.237	5.0
3	−1	1	0	55.3	0.220	6.4
4	1	1	0	33.1	0.177	11.3
5	−1	0	−1	67.1	0.267	13.1
6	1	0	−1	39.5	0.157	10.4
7	−1	0	1	55.3	0.220	4.4
8	1	0	1	33.4	0.133	8.2
9	0	−1	−1	73.0	0.290	13.5
10	0	1	−1	58.7	0.233	6.0
11	0	−1	1	60.8	0.242	6.5
12	0	1	1	42.0	0.167	8.2
13	0	0	0	52.4	0.208	2.9
14	0	0	0	52.4	0.208	2.1
15	0	0	0	52.0	0.207	2.6

*Amount of water per powder weight is presented for ease of understanding.

where Y is the dependent variable, X_i and X_j are the independent variables, X_iX_j are the interaction effects, X_i² are the square effects, b₀ is the intercept term, b_i are the linear terms, b_ij are the interaction terms, and b_ii are the square terms. The data obtained from the experiments were analyzed by ANOVA using Design–Expert software (version10, Stat-Ease, Inc., Minneapolis, MN, U.S.A.). The suitability of the model was evaluated using the coefficient of determination (R²), and p < 0.05 was considered statistically significant. Design–Expert software was also used to generate contour plots and response surface plots, and to predict the conditions for preparing the targeted granules as well as the values of the granule properties and their 95% confidence intervals.

Granulation Experiments

The formulations were composed of a 97% lactose-cornstarch mixture (7 : 3, w/w) and 3% CMC containing no API. For all the experiments, the filling rate of the vessel was fixed at approximately 25%. A predetermined amount of each ingredient was weighed into an ointment container (UG12-mL, 35-mL, or 58-mL, Umano Kagaku Youki Co., Ltd., Osaka, Japan), which had a similar shape, and blended at a speed of 1200 rpm for 45 s in a multi-purpose dispensing mixer (MW-N300DS-1, Beat Sensing Co., Ltd., Sunto-gun Shimizu-cho, Shizuoka-ken, Japan). Then, an appropriate amount of water was added using a metered spray bottle (Mini-trigger spray, Tsumamoto Kogyo Co., Ltd., Tokyo, Japan), and the mixture was granulated using at multi-purpose dispensing mixer according to the experimental sign described in Table 2. The minimum amount of water required for granulation was determined in an exploratory manner as mentioned later (“Determination of the Minimum Amount of Wetting Liquid That Can Be Granulated”). The granules were then dried in an oven (ONW-600S; AS ONE Co., Ltd., Osaka, Japan) at 50 °C until they reached a constant weight.

Evaluation of Dispersion State of Water

After the granulation procedure, the samples were collected from six points in the vessel, as shown in Fig. 1. The amounts of each sample were 6–9% of the loading materials. The samples were weighed before and after drying in an oven (ONW-600S, AS ONE Co., Ltd.) at 50 °C for one day. The water content of the sample and the coefficient of variation of water content were calculated using Eqs. 2 and 3, respectively.

  

(2)

  

(3)

Fig. 1. Upper (a) and Side (b) Views of Sampling Points in the Vessel

The areas surrounded by dotted line indicate sampling points.

Evaluation of Granulation Efficiency and Granule Characteristics

Yield of Granules

Dry granules were passed through a 2800-µm sieve followed by a 150-µm sieve, to remove any oversized agglomerates and fines. The remaining spherical granules were weighed and the yield (%) was calculated using Eq. 4.

  

(4)

Determination of Particle Size Distribution

The particle size distribution of the granules was measured using a series of standard sieves (Tokyo Screen Co., Ltd., Tokyo, Japan), including 75, 150, 212, 250, 300, 355, 425, 500, 600, 710, 850, 1000, 1180, 1400, 1700, 2000, 2360, and 2800 µm. After shifting, the individual sieves were weighed to estimate the number of remaining granules. The particle size distribution and median particle size (d50) were obtained from the data. The relative size distribution width, span, was defined using Eq. 5.

  

(5)

where d10, d50, and d90 represent the cumulative 10, 50, and 90% of the particle size diameters, respectively.

Estimation of Sphericity

The sphericity of the granules was estimated using the circularity of the projected image of granules with a diameter of 850–1700 µm. The granules were observed and projected using a microscope (Dino-Lite 3.0 R&D set, AnMo Electronics Co., Taiwan) at a magnification of 30. The images obtained were analyzed using the ImageJ software (version 1.51, National Institutes of Health, U.S.A.) for circularity (Eq. 6).

  

(6)

A circularity value of 1.0 indicated a perfect circle. Data are presented as the mean ± standard deviation (S.D.) of 30 granules.

Measurement of Granule Strength

For the granule strength measurements, 30 granules with diameters of 850–1700 µm were randomly chosen, and crushed using a digital force gauge (ZTA-50N, IMADA Co., Ltd., Toyohashi, Aichi-ken, Japan). The granule strength was calculated using Eq. 7.¹⁰⁾

  

(7)

where, P is the crushing force (N). Data are presented as the mean ± S.D. of 30 granules.

Statistical Analysis

Correlation analysis was performed using EZR software (version 1.54), a modified version of R commander designed to add statistical functions frequently used in biostatistics.¹¹⁾ Statistical significance was set at p < 0.05.

Results and Discussion

Determination of the Minimum Amount of Wetting Liquid That Can Be Granulated

The adequate amount of water in planetary centrifugal wet granulation is considered to be approximately 65% of the PL value of the formulation.⁵⁾ The composition of all formulations used in this study was identical, and the PL value was 0.40 mL/g.⁴⁾ However, the optimum value varied depending on the equipment conditions as well as the operator’s requests, for example, lowering the amount of water required to reduce the drying cost. In this study, we considered the minimum amount of water required for granulation to be the optimal value. For example, the minimum amount of water required at the center point of the experimental design (Code 15 in Table 2) is determined as follows: Fig. 2 shows the particle size distribution of the granules obtained when the amount of water added was increased from 1.10 to 1.30 mL. The fraction below 250 µm was considered to be ungranulated. The ungranulated fraction decreased with an increase in the amount of water and almost disappeared when the amount of water was 1.25 mL. The amount of water added at this time was considered the minimum amount required, and was expressed as a percentage of the PL value (PL%) as well as an amount of water per powder weight (mL/g). The results for all points in the experimental design are listed in Table 2. Granulation was then performed with the minimum amount of water and the dispersion state of water in the vessel was measured immediately after granulation. The results for all points are also shown in Table 2.

Fig. 2. Particle Size Distribution of Granules Prepared from Formulation Code 15 Using Various Amounts of Water

The amounts of water, 1.10, 1.15, 1.20, 1.25, and 1.30 mL represent the amounts of water per powder weight, 0.183, 0.192, 0.200, 0.208, and 0.217 mL/g, respectively.

Effect of Equipment Parameters on Amount and Dispersion State of Water

Using the values listed in Table 2, multiple regression analysis was performed using a quadratic model to examine the effect of the equipment parameters on the amount and dispersion state of water. Results show that a significant model was obtained for PL% with a determination coefficient of 0.9734 (Table 3). A greater coefficient indicates that the independent variable has a stronger effect on the response. Each main parameter was a significant influencing factor (p < 0.05), in the order of processing speed, processing time, followed by vessel size. No significant effects were observed for the interaction terms. Figure 3 shows contour plots of the effects of the processing speed and processing time for each vessel. A faster speed and longer time were sufficient to lower the amount of water, regardless of vessel size. In general, granulation progresses with increasing operating speed and time.¹²⁾ In this study, the processing speed and time also caused granulation to progress with a smaller amount of water.

Table 3. Multiple Regression Analyses of Amount and Dispersion State of Water

Term	PL%		CV%
	Coefficient	p-Value	Coefficient	p-Value
X1	−10.4	0.000	0.425	0.623
X2	−9.38	0.001	−0.088	0.920
X3	−5.85	0.005	−1.96	0.064
X1X2	−2.73	0.167	1.88	0.171
X2X3	1.43	0.437	1.63	0.224
X3X1	−1.13	0.534	2.30	0.107
X12	−3.70	0.089	2.36	0.111
X22	6.10	0.018	1.88	0.183
X32	0.254	0.891	3.33	0.041
Constant	52.3		3.33
Model
R2	0.9734	p = 0.002	0.8395	p = 0.1263
Adjusted R2	0.9255		0.5506

Fig. 3. Contour Plots of Amount of Water Required for Granulation

In contrast, no significant model was obtained for the CV% (Table 3). The CV% showed large values when the vessels were small and speed was slow, however, this trend was not consistent (Table 2). In most cases, the CV% showed values below 10%, indicating that the added water was well dispersed. In the multi-purpose dispensing mixer, the vessel was rotated and revolved at a speed of 13.3–20 times/s (Table 1). This intense force might have caused rapid dispersion of water.

In general, the amount of water used as a binder solution is the most critical process parameter of wet granulation; therefore, the amount of water is adjusted using various monitoring apparatuses during granulation.¹³⁾ However, as the planetary centrifugal granulation is performed in a closed system and the granulation time is extremely short, the operator must decide the amount of water poured into the vessel before the operation. Therefore, a predictive method is required to estimate the amount of water required for granulation. It is valuable to determine the contribution of the parameters influencing the PL%.

Effect of Equipment Parameters on Granule Properties

The yield, median diameter, span, sphericity, and strength of the granules are listed in Table 4. The yield was greater than 92%, indicating that granulation was possible under a wide range of conditions. The d50 varied from 803–1470 µm and the span was in a relatively wide range of 0.38–1.54. The sphericity was greater than 0.7. The granule strength was greater than 4 N/mm², indicating that there was no problem with normal handling and dispensing.¹⁴⁾

Table 4. Yield and Granule Properties

Code	Yield (%)	d50 (µm)	Span (−)	Sphericity* (−)	Granule strength* (N/mm2)
1	92.3	1038	1.15	0.686 ± 0.086	4.49 ± 0.95
2	94.2	961	1.12	0.671 ± 0.086	5.89 ± 1.06
3	92.7	976	1.35	0.784 ± 0.053	6.56 ± 0.84
4	94.4	1108	0.64	0.731 ± 0.051	5.66 ± 0.97
5	93	1079	0.97	0.733 ± 0.060	5.34 ± 1.08
6	97.6	1002	0.82	0.823 ± 0.028	7.01 ± 1.46
7	94	1092	1.07	0.730 ± 0.060	5.45 ± 1.13
8	97	1470	0.38	0.799 ± 0.035	7.17 ± 1.44
9	96.2	1004	1.17	0.729 ± 0.075	5.63 ± 1.27
10	95.9	803	1.54	0.724 ± 0.060	4.61 ± 0.89
11	96.3	964	1.03	0.739 ± 0.042	6.01 ± 0.86
12	97.7	1143	0.63	0.774 ± 0.066	6.51 ± 1.00
13	95.4	1147	0.87	0.815 ± 0.026	6.11 ± 0.85
14	94.8	988	0.92	0.808 ± 0.032	5.72 ± 0.96
15	97.5	1021	0.91	0.774 ± 0.039	5.48 ± 0.69

* Mean ± S.D. (n = 30).

To investigate the effect of the equipment parameters on granule properties, the results listed in Table 4 were analyzed using a quadratic model. The coefficients of the terms and the determination coefficients of the model are listed in Table 5. The quality of the developed model was evaluated based on the R² values. R² values for d50 and span were 0.9072 and 0.9384 (p < 0.05), respectively. These results indicate that the applied regression model provides an explanation of the relationship between the equipment parameters and granule properties, d50 and span. A positive sign indicates a synergistic effect, whereas a negative sign indicates an antagonistic effect of the responses. In the case of d50, vessel size was a significant parameter in the main term. Additionally, the interaction terms of processing time and vessel size (X₂X₃) and those of vessel size and processing speed (X₃X₁) were also significant. These results indicate that vessel size had a greater effect on d50. This may be due to the difference in the motion distance and speed of the granules in the vessel when using a larger container. In addition, a larger container implies a longer rotation radius. Consequently, the granules in the larger vessels tended to grow larger. In the case of span, the processing speed and vessel size were the significant parameters with similar magnitudes of contribution (Table 5). The interaction terms of processing speed and processing time (X₁X₂) and those of vessel size and processing speed (X₃X₁) were also significant. The higher processing speed and larger vessel size caused a higher moving velocity of the granules in the vessel, leading to more agglomeration.

Table 5. Multiple Regression Analyses of Yield and Granule Properties

Term	Yield		d50		Span		Sphericity		Granule strength
	Coefficient	p-Value	Coefficient	p-Value	Coefficient	p-Value	Coefficient	p-Value	Coefficient	p-Value
X1	1.40	0.017	44.5	0.147	−0.198	0.006	0.0114	0.512	0.486	0.107
X2	0.213	0.616	8.04	0.770	−0.037	0.428	0.0235	0.205	0.165	0.534
X3	0.288	0.502	97.6	0.013	−0.173	0.010	0.0041	0.808	0.319	0.254
X1X2	−0.050	0.933	52.3	0.215	−0.170	0.038	−0.0095	0.694	−0.575	0.161
X2X3	−0.400	0.509	114	0.027	−0.133	0.080	−0.0053	0.827	0.0125	0.973
X3X1	0.425	0.484	95.2	0.049	−0.196	0.023	0.0100	0.679	0.380	0.327
X12	−1.81	0.027	75.4	0.106	−0.057	0.411	−0.0256	0.330	0.216	0.578
X22	−0.688	0.293	−107	0.038	0.225	0.016	−0.0554	0.067	−0.336	0.398
X32	1.31	0.075	33.3	0.425	−0.030	0.657	−0.0021	0.932	0.256	0.513
Constant	95.9		1052		0.897		0.799		5.77
Model
R2	0.8622	p = 0.092	0.9072	p = 0.039	0.9384	p = 0.015	0.6526	p = 0.510	0.7003	p = 0.406
Adjusted R2	0.6142		0.7400		0.8275		0.0272		0.1609

Three-dimensional response surface plot is a graphical representation of the estimated regression equation, leading to a better understanding of the interactions between the parameters. Figure 4 shows the response surface plots of d50 and span for each vessel. In Figs. 4a, b, d, and e, the change in speed is observed to be larger than the change over time. With larger vessels, the surfaces became diagonal to the front direction (Figs. 4c and f), showing interaction effects.

Fig. 4. Response Surface Plots of d50 (Upper Layer) and Span (Lower Layer) Using 12-mL (a, d), 35-mL (b, e), and 58-mL (c, f) Vessels

However, in the cases of sphericity and granule strength, no significant models were obtained (Table 5). These results are derived from the granulation mechanism of planetary centrifugal granulation. Circularity generally depends on the method of granulation.¹⁵⁾ In this study, the generated granules were rolled on the wall of the vessel during the process, leading to high sphericity (0.686–0.823) irrespective of the equipment parameters. Granule strength generally depends on the equipment parameters that influence the degree of liquid saturation of ingredients.¹⁶⁾ However, in this study, the granules showed a similar granule strength (4.49–7.17 N/mm²). This can be explained by the lack of significant parameter affecting CV% (Table 3). The added water was rapidly and consistently dispersed in the diluents to form liquid bridges between particles, eventually increasing the granule strength.

Relationship between Amount of Water and Granule Properties

Figure 5 shows the relationship between the PL% and granule properties. The d50 increased and span decreased with a decrease in PL%. PL% represents the amount of water added at the point where the ungranulated fraction disappears. Thus, when the PL% is low, the amount of water added is low, and the granulation process must be more advanced until the ungranulated fraction is eliminated. Progression of the granulation process is considered to have resulted in a larger d50 and smaller span.¹¹⁾ In addition, as the PL% decreased, the sphericity and granule strength tended to increase. This was due to the progress of spherization and consolidation of the granules as the granulation process progressed.¹⁷⁾

Fig. 5. Correlation between Amount of Water (PL%) and Granule Properties

(a) d50, (b) span, (c) sphericity, and (d) granule strength.

Predictability of the Mathematical Model

Finally, the predictability of the mathematical model obtained in this study was examined. Here, granules with a median diameter of approximately 1000 µm and a narrow particle size distribution width were set as the granules of interest. We then used all vessel sizes in these experiments. Table 6 lists the estimated processing time and speed for each vessel size based on the model. Granulation was performed according to Table 6, and the PL% and particle size distribution were measured. The results are shown in Table 7, along with the values predicted using the model. All the experimental values were within the 95% confidence interval of the predicted values, indicating good predictability.

Table 6. Equipment Parameters for Preparing Granules Targeting to 1000 µm for d50 and Minimizing Span

Formulation	Equipment parameter
	Processing speed (rpm)	Processing time (s)	Vessel size (mL)
1	1150	79	12
2	1050	66	35
3	960	66	58

Table 7. Checkpoint Analysis of Targeted Formulations for d50, Span, and PL%

Formulation	PL% (%)		d50 (µm)		Span
	Observed value	Predicted value median (95% CI)	Observed value	Predicted value median (95% CI)	Observed value	Predicted value median (95% CI)
1	0.50	0.45 (0.40–0.56)	1090	975 (826–1123)	1.05	1.00 (0.76–1.24)
2	1.39	1.41 (1.30–1.52)	953	1005 (903–1107)	1.04	1.00 (0.83–1.17)
3	2.23	2.30 (2.06–2.54)	1068	1043 (909–1177)	1.10	0.99 (0.78–1.21)

CI: confidence interval.

Conclusion

The effects of equipment parameters on the amount of water and its in-vessel dispersion in the planetary centrifugal granulation process were investigated by generating a response surface using an experiment design method. The amount of water required for granulation was significantly affected by processing speed, processing time, and vessel size. In contrast, the dispersion state of water was rarely influenced by the equipment parameters. The effects of the equipment parameters on the granule characteristics were also examined. The d50 was mostly affected by the vessel size, and the span was affected by the vessel size and processing speed. In addition, a correlation was observed between the amount of water required and the granule characteristics. Thus, the equipment parameters and the amount of water required for the preparation of granules with desired characteristics were obtained from the generated model. The methodology and results of this study would contribute to the establishment of extemporaneous compounding procedures.

Acknowledgments

This study was supported by the Japan Society for the Promotion of Science KAKENHI (Grant number JP19K07169). We also thank Yukino Ikeuchi and Risa Miyaki for their assistance with the experiments.

Conflict of Interest

Y. Miyazaki received a research Grant from BeatSensing Co., Ltd.

References

• 1) Ishikawa Y., J. Pharm. Sci. Tech. Jpn., 75, 28–31 (2015).
• 2) Sasaki M., Sano M., Tanaka Y., Yamamoto I., Jpn. J. Pharm. Health Care Sci., 32, 420–423 (2006).
• 3) Matsumoto M., Igarashi N., Ohtani M., Namiki M., Yamamura Y., Uchino K., Sugiyama K., Jpn. J. Pharm. Health Care Sci., 37, 395–402 (2011).
• 4) Eda T., Miyazaki Y., Uchino T., Kagawa Y., Chem. Pharm. Bull., 68, 251–257 (2020).
• 5) Miyazaki Y., Tsuboi A., Maruyama S., Aoshima H., Uchino T., Kagawa Y., Chem. Pharm. Bull., 68, 989–995 (2020).
• 6) Miyazaki Y., Uchino T., Kagawa Y., Eur. J. Hosp. Pharm., 30, 10–13 (2023).
• 7) Morino K., Matsuno M., Miyazaki Y., Reito, 97, 480–483 (2022).
• 8) Miyazaki Y., Uchino T., Kagawa Y., Pharmazie, 77, 103–106 (2022).
• 9) Nguyen T. H., Shen W., Hapgood K., Chem. Eng. J., 164, 330–339 (2010).
• 10) Adams M. J., Mullier M. A., Seville J. P. K., Powder Technol., 78, 5–13 (1994).
• 11) Kanda Y., Bone Marrow Transplant., 48, 452–458 (2013).
• 12) Iveson S. M., Litster D. J., Hapgood K., Ennis B. J., Powder Technol., 117, 3–39 (2001).
• 13) Suresh P., Sreedhar I., Vaidhiswaran R., Venugopal A., Chem. Eng. J., 328, 785–815 (2017).
• 14) Ochiai Y., J. Pharm. Sci. Tech. Jpn., 71, 17–20 (2011).
• 15) Morin G., Briens L., AAPS PharmSciTech, 15, 1039–1048 (2014).
• 16) Rahmanian N., Ghadiri M., Ding Y., Chem. Eng. Sci., 63, 915–923 (2008).
• 17) Rahmanian N., Ghadiri M., Jia X., Stepanek F., Powder Technol., 192, 184–194 (2009).