Application of the central composite design to optimize the calcium carbonate-HPMC co-processed excipient prepared by co-spray drying

Abstract

To develop a novel co-processed tablet excipient based on calcium carbonate and hydroxypropyl methylcellulose (HPMC), the central composite design-response surface methodology (CCD-RSM) was applied to optimize the amount of HPMC and the solid content of feed suspensions used for co-spray drying by choosing powder and tablet properties as the evaluation indicators. Significant and adequate regression models were developed for each property. An increase in % HPMC improved powder bulk density, tablet tensile strength, and various tableting parameters. However, a high content of HPMC had a negative influence on tablet disintegration time. In addition, the median particle size of the product increased with both factors. However, they had little effect on the fast elastic stretch of tablets. Numerical optimization determined the optimal parameters as HPMC in the feed solid: 6.7% (w/w) and the solid content of the feed: 42% (w/v). The experimental values of the optimized co-processed excipient were mostly close to the model-predicted values. The tablet tensile strength and disintegration time were 3.28 MPa and 8.91 min, respectively. The combined use of CCD, RSM, and the desirability functions can provide an insight into the object studied and the results achieved are helpful for further development of novel co-processed excipients.

---

1. Introduction

Tablets are still the most commonly used dosage form because of the ease of manufacturing, convenience in administration, accurate dosing, and excellent stability. However, tablet manufacturing has changed significantly over the years because of the introduction of direct compaction and high speed machines, with these two developments increasing the functionality demand on the excipients used in terms of flowability and compatibility.¹ Single-component excipients do not always provide requisite performances or physicomechanical properties to allow active pharmaceutical ingredients to be formulated or manufactured successfully. Therefore, co-processed multi-component excipients were brought in.^2–5 A co-processed excipient is a combination of two or more established excipients designed to physically modify their properties in a manner not achievable by simple physical mixing. Many co-processing methods can be used, among which spray drying is the most widely used and successful one. A majority of commercially available co-processed excipients are produced by spray drying.^3–5

Calcium carbonate (CC), employed as a pharmaceutical excipient, is mainly used in solid dosage forms as a diluent^6–8 and also used as a calcium supplement, which is inexpensive, easily obtained, and widely used. However, CC is cohesive, and thus, has poor flowability. Moreover, it is a brittle material, which makes it exhibit relatively poor binding properties and low compactibility.^9,10 To combat these shortcomings, several CC-based co-processed excipients therefore have been developed with some success, such as Barcroft® CS90 (90% CC-10% starch), ForMaxx® (70% CC-30% sorbitol), and MCC-CC (80 : 20).⁴ All of these are produced by spray drying. However, to our knowledge, there is no co-processed excipient based on CC and hydroxypropyl methylcellulose (HPMC) yet.

In direct tableting processes, a dry binder is added generally prior to compaction in order to increase the compactibility of the formulation, and hence, the tensile strength of the resulting tablet.¹¹ Hydroxypropyl methylcellulose (HPMC) is commonly used as a binder in tablets. It is also a common material in thin-film coating.⁹ With relatively low hygroscopicity and high glass transition temperature, HPMC was found to be able to reduce particle adhesion, and thus, improve production yield during spray drying.¹²

In our preliminary experiments, the feasibility of improving both powder and tablet properties of CC by co-spray drying with a small amount of HPMC was tested. The results show that most powder properties (e.g., flowability, compactibility, and hygroscopicity) and tableting properties (e.g., yield pressure, tensile strength, and unit effective compaction work) of the co-processed product were significantly improved compared to their physical mixture. In addition, it was reported that % HPMC and the solid content of feeds could affect the performance of spray-dried composite particles.^13,14 Thus, in this study, to obtain the ideal CC-HPMC co-spray dried powders, the central composite design-response surface methodology (CCD-RSM) was applied to optimize the amount of HPMC E3 and the solid content of the feed used for co-spray drying. The reason for using HPMC E3, the lowest viscosity grade in the HPMC family, was to minimize the possible adverse effect of HPMC on tablet disintegration time.

CCD-RSM is a preferred method to further optimize the preparation with a nonlinear model to fit the experimental data.^15–17 Compared with commonly used traditional methods, such as orthogonal design and uniform design, CCD-RSM has higher accuracy, and can find the optimal value in the whole area. CCD-RSM has been widely applied in food^18,19 and pharmaceutical^20,21 fields. The objective of the method is to simultaneously optimize the levels of the variables studied to achieve the best systemic performance. In this study, a five-level and two-factor CCD-RSM was used to optimize the amount of HPMC and the solid content of the feed used for co-spray drying by choosing various powder and tablet properties as the evaluation indicators.

2. Materials and methods

2.1. Materials

Calcium carbonate (CC) (Daheng calcium carbonate Development Co., Ltd, Yidu, China), hydroxypropyl methylcellulose (HPMC; Methocel E3, Dow Chemical, Midland, MI, USA), cross-linked polyvinylpolypyrrolidone (PVPP; Kollidon CL, BASF, Ludwigshafen, Germany), magnesium stearate, and acetone (Sinopham Chemical Reagent Co, Shanghai, China) were used as supplied.

2.2. Preparation of the spray dried particles

CC was dispersed homogeneously in water solution of HPMC to form the feed dispersion. With constant stirring, the feed dispersion was spray-dried immediately using a pilot scale spray dryer (Mobile Minor 2000, Niro, Søborg, Denmark) with a rotary atomizer operated under the following conditions: inlet temperature, 170 °C; atomizing pressure, 4.0 MPa; and feed rate, ∼22 mL min⁻¹. The dryer was allowed to reach steady state by operating it for at least 30 min with water spraying before the feed was sprayed. The particles collected were sealed into plastic bags and stored in desiccators at room temperature until further tests.

2.3. Experimental design

Preliminary experiments were carried out to establish appropriate ranges for the amount of HPMC in the feed solid and the solid content of the feed suspension. The lower limit of the solid content of the feed (19%, w/v) was chosen to obtain a minimum production capacity. And, the upper limit (44%, w/v) was selected to avoid pumping problems and blocking of the atomization device due to high viscosity of suspensions. On the other hand, the maximum amount of HPMC in the feed solid was set as 10.5% (w/w), based on the improvement efficiency and the consideration that further increasing the amount of HPMC could cause unacceptable adverse influence on tablet disintegration time.

Table 1 displays the coded and uncoded levels of the two independent variables along with their ranges, which were determined from preliminary trials, and then the central composite design (CCD) was applied to investigate the effects of independent variables on the powder properties (flowability, median particle size, bulk density, moisture content, production yield) and tablet properties (tensile strength, disintegration time, and tableting parameters) of spray-dried particles. The names and symbols of the properties studied are shown in Table 2.

Table 1 Independent variables and levels (coded and uncoded) of the central composite design

Independent variables	Ranges	Levels of variables
		−1.414	−1	0	1	1.414
A: HPMC in the feed solid (%, w/w)	3.5–10.5	3.5	4.5	7.0	9.5	10.5
B: Solid content of the feed (%, w/v)	19–44	19	23	32	40	44

---

Table 2 The names and symbols of powder and tablet properties studied

Name	Symbol
Median particle size	D(0.5)
Production yield	PY
Moisture content	MC
Bulk density	ρb
Angle of repose	AR
Tensile strength	TS
Disintegration time of tablets containing the coprocessed excipient only	DT
Disintegration time of tablets composed of 96.5% the coprocessed excipient and 3.5% PVPP	DTP
Yield pressure	Py
Fast elastic stretch	FES
Unit effective compaction work	Esp

---

Each factor was coded at five levels according to the following equation.²²

where x_i is the dimensionless coded value of an independent variable; X_i is the real value of an independent variable; X_z is the real value of an independent variable at the center point; and ΔX_i is the step change of the real value of the variable i.

In this study, the design matrix was generated by the Design-Expert® (Version 8.0.6.1) software. The total runs of the CCD were 13, which included one run for each of 4 points from a two-factor two-level full factorial 2^{λ = 2} design (−1 and +1 being the coded levels for each factor) and 4 additional axial points (2 experimental points per factor at a distance α = (2^λ)^0.25 = 1.414 from the design center) and five replicated runs for the central point (Table 3). Replication allowed to determine experimental error and to increase the precision of estimates.

Table 3 Design runs of the central composite design experiments^a

Run	Code		Value
	A (%)	B (%)	A (%)	B (%)
a A, HPMC in the feed solid (%, w/w); B, the solid content of the feed (%, w/v).
1	1.414	0	10.5	32
2	0	0	7.0	32
3	−1	−1	4.5	23
4	1	−1	9.5	23
5	0	−1.414	7.0	19
6	0	0	7.0	32
7	0	0	7.0	32
8	0	0	7.0	32
9	1	1	9.5	40
10	0	0	7.0	32
11	−1	1	4.5	40
12	−1.414	0	3.5	32
13	0	1.414	7.0	44

---

To avoid bias, all the experiments were performed in random order. The empirical second order polynomial model was shown as follows:

where Y is the response; X_i and X_j are variables (i and j ranged from 1 to λ); β₀ is the constant term; β_j is the linear coefficient, β_ij are interaction coefficient, and β_jj is the quadratic coefficient; λ is the number of independent parameters (λ = 2 in this study).^23,24

The experimental data were analyzed by multiple regression analysis through the generalized least square. Analysis of variance (ANOVA) was used to estimate the statistical parameters. The Design-Expert software was utilized for the response surface analysis. The significant model was used for fitting the response. The lack-of-fit test and a normal probability plot of the residuals were performed in order to evaluate the model and to detect outliers. The models provide several comparative measures for model selection.

2.4. Characterization of particles and tablets

2.4.1. Moisture content. The moisture content was determined by a fast infrared moisture analyzer (Sartorius, Germany). About 2 g of each testing sample was distributed into a disc and the result was recorded until a constant reading was achieved.

2.4.2. Median particle size. The median particle size (D_0.5) was determined by laser diffraction (Malvern 2000, Malvern Instruments Ltd, England). Three replicates were carried out for each measurement.

2.4.3. Flowability. The angle of repose, determined by a powder property tester (Baite Corporation, China), was used to indicate the flowability of the materials. The samples were poured through a vibrating metal funnel onto a platform until a stable and height-fixed heap was formed. The angle of repose was measured as the angle made by the inclined plane of the heap with the horizontal.

2.4.4. Bulk density. The powder property tester was used to determine the bulk density (ρ_b). The samples were poured through a vibrating metal funnel into a measuring cylinder, which was exactly 100 mL and weighted in advance as m₀ (g), until it was full. After the excessive powder was scraped off, the weight of the cylinder was recorded as m₁ (g). ρ_b (g mL⁻¹) was calculated by eqn (3).

2.4.5. Compactibility. Materials were compacted on a fully instrumented press (Korsch XP1, Germany) using 8.5 mm, round flat-faced tooling with a compaction pressure of 176 MPa. For lubrication, magnesium stearate suspension (0.5%) in acetone was spread on the upper punch, lower punch, and die before compaction. The thickness (T; mm), diameter (D; mm), and crushing force (F; N) of five compacts were determined immediately by a crushing force tester (Sotax HT10, Switzerland). The tensile strength (TS) of each compact in MPa was calculated using eqn (4):²⁵

The works during compaction were also recorded by the press. The E_sp value was calculated using eqn (5):

where W_net is the net work (J), i.e., the energy that remained after the punch left the compact, and M is the tablet weight (g).

2.4.6. Yield pressure (P_y). The “in-die” method was used. The compaction was carried out as above. The Athy–Heckel eqn (6)²⁶ was used to analyze the volume reduction mechanism during compaction.where D is the relative density of the compact at pressure P; K is a material constant and is the slope of the linear portion of the plot. P_y reflects plasticity and was calculated from the reciprocal of the slope K by performing linear regression in the compaction phase, where the R² value is above 0.995.

2.4.7. Tablet disintegration time. The disintegration time was determined on 6 compacts per sample according to the USP method using a disintegration apparatus (LB-2D, Shanghai Huanghai Ltd, China). Distilled water was used as disintegration medium, which was maintained at 37 °C ± 1 °C. The disintegration time for each compact was recorded separately.

2.4.8. Fast elastic stretch (FES). FES describes the axial elastic recovery in a die and was calculated from the following eqn (7):²⁷where FES is given as a percentage (%), T₁ is the edge thickness of the tablet during unloading, and T₂ represents the edge thickness of the tablet at the time of maximum pressure.

3. Results and discussion

Co-spray drying uses a one-step continuous process to dry and agglomerate multi-component powders, which are generally homogeneous, subglobular, and porous, with improved physicochemical and mechanical properties. In addition, it is a technique that can be easily automated, scaled up, and equipped for in-line product analysis. These features altogether offer obvious benefits for the development of composite materials with various applications.^28–32

In our preliminary experiments, co-spray drying of calcium carbonate (CC) suspension with dissolved hydroxypropyl methylcellulose (HPMC) had demonstrated the efficiency of HPMC to improve various powder and tableting properties of CC.²⁸ The results showed that co-spray drying with HPMC led to a remarkable improvement in flowability for CC, and via reducing particle adhesion, achieved a production yield as high as 81.2%. The tensile strength of tablets made from the co-spray dried product was several times higher than that from their physical mixture. The co-spray dried product also exhibited lower lubrication sensitivity and appropriate tablet disintegration time. It was reported that the content of binders and the solid content of feeds probably affected the performance of the spray dried composite particles by virtue of influencing their size and bulk density, and the distribution of the components.^13,14 In light of the above, the central composite design-response surface methodology (CCD-RSM) was therefore carried out in this study to investigate the effects of two factors, i.e., % HPMC and the solid content of the feed, on the powder and tableting properties of the co-spray dried product. The chosen responses were powder flowability, median particle size, bulk density, moisture content, production yield, tablet tensile strength and disintegration time, and tableting parameters (including fast elastic stretch, yield pressure, and E_sp).

3.1. Summary statistics for the models

Based on the design matrix of CCD for two factors at five levels, a total of 13 experiments were performed and the observed responses are summarized in Tables 4 and 5. RSM was used to identify the optimum levels of the variables that had significant effects on powder and tablet properties of the co-spray dried product. All the responses observed for the 13 products prepared were fitted to various models using the Design-Expert software. It was observed that the best-fit model was the quadratic or cubic model for each response. Analysis of variance of the responses (Table 6) indicated that for the optimization of the co-spray dried excipient, the surface models developed for powder flowability, median particle size, bulk density, tablet tensile strength, disintegration time, yield pressure, E_sp, and fast elastic stretch were significant, without significant lack of fit. In addition, the model summary statistics for the selected significant models were also detailed in Table 6. It can be observed that for powder flowability, median particle size, bulk density, tablet tensile strength, disintegration time, yield pressure, and E_sp, R², predicted R², and adjusted R² were in good agreement, resulting in reliable models. Although the response surface model for fast elastic stretch was significant, R², predicted R², and adjusted R² were not in good agreement. Thus, the developed regression model for fast elastic stretch did not show acceptable statistical measures. Moreover, fast elastic stretch values of all design runs were within acceptable ranges. However, for the moisture content, since the ability to spray dry a product to a specific moisture content at a given outlet drying air temperature depends upon the humidity of the air leaving the drying chamber, daily changes of ambient humidity conditions could affect the moisture content in the spray dried powder.³³ As a result, the model estimating the moisture content of spray dried powders was not significant. For the production yield, it is mainly affected by spray drying parameters (atomizing air pressure, inlet and outlet drying air temperatures, and feed rate). Thus, both of the factors have little effect on it. In addition, the anti-adhesion effect of HPMC has already approached the maximum when its amount was 3.5%. As a result, the model estimating the production yield was also not significant.

Table 4 Results of the central composite design: powder properties of the coprocessed particles prepared (n = 3)^b

Run	A (%)	B (%)	D(0.5) (μm)	PY (%)	MC (%)	ρb (g mL^−1)	AR (°)
a Standard deviation.b A, HPMC in the feed solid (%, w/w); B, the solid content of the feed (%, w/v). D(0.5), median particle size; PY, production yield; MC, moisture content; ρb, bulk density; AR, angle of repose.
1	10.5	32	39.61 (0.01)^a	81.21	0.67 (0.02)	0.298 (0.04)	33 (0.31)
2	7.0	32	29.50 (0.02)	79.05	0.53 (0.04)	0.379 (0.01)	32 (0.38)
3	4.5	23	21.98 (0.04)	83.25	0.37 (0.04)	0.498 (0.08)	31 (0.28)
4	9.5	23	34.41 (0.02)	83.52	0.53 (0.05)	0.308 (0.09)	31 (0.33)
5	7.0	19	25.97 (0.05)	80.09	0.66 (0.02)	0.362 (0.06)	32 (0.23)
6	7.0	32	29.89 (0.01)	79.98	0.56 (0.06)	0.365 (0.11)	32 (0.27)
7	7.0	32	29.42 (0.03)	80.13	0.56 (0.03)	0.372 (0.05)	31 (0.21)
8	7.0	32	30.04 (0.01)	79.51	0.55 (0.02)	0.374 (0.01)	32 (0.31)
9	9.5	40	40.85 (0.06)	73.50	0.51 (0.03)	0.341 (0.04)	33 (0.34)
10	7.0	32	29.07 (0.01)	80.08	0.47 (0.06)	0.352 (0.07)	32 (0.25)
11	4.5	40	24.67 (0.02)	76.01	0.47 (0.07)	0.476 (0.06)	30 (0.22)
12	3.5	32	21.14 (0.01)	83.25	0.29 (0.01)	0.606 (0.08)	25 (0.26)
13	7.0	44	31.49 (0.04)	85.23	0.56 (0.09)	0.398 (0.03)	34 (0.33)

---

Table 5 Results of the central composite design: tableting and tablet properties (n = 3)^e

Run	A (%)	B (%)	TS^a (MPa)	DT^b (min)	DTP^b (min)	Py^c (MPa)	FES^a (%)	Esp^a (J g^−1)
a These values were determined under the compaction force of 176 MPa.b Tablets having a breaking force of 60 N were used for the determination.c The yield pressure was determined under the compaction force range of 50–110 MPa.d Standard deviation.e A, HPMC in the feed solid (%, w/w); B, the solid content of the feed (%, w/v). TS, tensile strength; DT, disintegration time of tablets containing the coprocessed excipient only; DTP, disintegration time of tablets composed of 96.5% the coprocessed excipient and 3.5% PVPP; Py, yield pressure; FES, fast elastic stretch; Esp, unit effective compaction work.
1	10.5	32	4.40 (0.04)^d	28.41 (0.02)	17.04 (0.01)	298.33 (0.21)	8.16 (0.21)	13.73 (0.19)
2	7.0	32	3.33 (0.05)	17.07 (0.03)	9.70 (0.04)	316.26 (0.18)	8.97 (0.12)	13.52 (0.12)
3	4.5	23	1.85 (0.05)	15.74 (0.05)	5.65 (0.01)	355.49 (0.17)	8.79 (0.22)	11.13 (0.23)
4	9.5	23	4.09 (0.04)	23.63 (0.09)	15.73 (0.09)	308.56 (0.22)	8.08 (0.18)	14.64 (0.27)
5	7.0	19	2.98 (0.07)	16.75 (0.02)	10.88 (0.01)	316.95 (0.29)	8.49 (0.11)	12.49 (0.21)
6	7.0	32	3.28 (0.05)	15.70 (0.02)	9.90 (0.03)	322.86 (0.11)	8.73 (0.21)	12.98 (0.18)
7	7.0	32	3.32 (0.04)	16.13 (0.01)	9.61 (0.05)	315.56 (0.16)	8.68 (0.28)	13.51 (0.16)
8	7.0	32	3.35 (0.03)	16.03 (0.04)	9.66 (0.05)	316.21 (0.21)	8.87 (0.23)	13.45 (0.24)
9	9.5	40	4.13 (0.04)	21.35 (0.03)	14.91 (0.02)	303.88 (0.18)	8.18 (0.17)	13.90 (0.16)
10	7.0	32	3.31 (0.03)	16.90 (0.08)	9.31 (0.02)	316.04 (0.24)	8.84 (0.16)	13.09 (0.13)
11	4.5	40	2.35 (0.03)	15.29 (0.12)	5.18 (0.02)	354.38 (0.23)	8.39 (0.22)	12.00 (0.21)
12	3.5	32	1.59 (0.04)	14.26 (0.06)	4.34 (0.03)	358.66 (0.19)	8.56 (0.10)	11.01 (0.25)
13	7.0	44	3.23 (0.04)	15.86 (0.08)	8.71 (0.01)	313.36 (0.31)	8.81 (0.15)	13.31 (0.17)

---

Table 6 Model summary statistics and ANOVA – influence of the HPMC level and the solid content of the feed on the response^a

Responses	Model F-value	Prob > F	Lack of fit F-value	Prob > F	St. dev.	R^2	Adjusted R^2	Predicted R^2
a DT, disintegration time of tablets containing the co-processed excipient only; DTP, disintegration time of tablets composed of 96.5% the co-processed excipient and 3.5% PVPP; Esp, unit effective compaction work.
Angle of repose (°)	41.50	0.0005	1.36	0.35	0.47	0.9765	0.9529	0.8370
Median particle size (μm)	248.97	<0.0001	4.77	0.08	0.65	0.9920	0.9880	0.9687
Bulk density (g cm^−3)	112.64	<0.0001	4.95	0.07	0.02	0.9575	0.9490	0.8932
Tensile strength (MPa)	874.25	<0.0001	5.30	0.07	0.04	0.9984	0.9973	0.9904
DT (min)	113.76	<0.0001	0.48	0.65	0.54	0.9913	0.9826	0.9646
DTP (min)	476.18	<0.0001	4.28	0.09	0.36	0.9937	0.9917	0.9812
Yield pressure (MPa)	97.39	<0.0001	3.49	0.12	5.92	0.9512	0.9414	0.8710
Fast elastic stretch (%)	6.72	0.0113	3.60	0.12	0.17	0.7706	0.6559	0.1878
Esp (J g^−1)	25.60	0.0001	2.77	0.17	0.35	0.9275	0.8913	0.7312

---

3.2. Combined effects of the solid content of the feed and % HPMC

The prediction equations of the developed response surface models in function of the amount of HPMC in the feed solid and the solid content of the feed suspension are mentioned in Table 7, while the corresponding contour plots and three-dimensional response surfaces plots generated by the regression models are shown in Fig. 1 and 2, which show the relationship between the independent variables and the responses graphically. Among the symbols used, A, B, A², and B² represent the single effect on the responses by the two independent variables (i.e., % HPMC in the feed solid and the solid content of the feed), with A and B referring to the interaction effects.

Table 7 Prediction equations in terms of coded factors for reliable and adequate regression models^a

a A, HPMC in the feed solid (%, w/w); B, the solid content of the feed (%, w/v). DT, disintegration time of tablets containing the coprocessed excipient only; DTP, disintegration time of tablets composed of 96.5% the coprocessed excipient and 3.5% PVPP; E_sp, unit effective compaction work.

Angle of repose = 31.80 + 2.83 × A + 0.48 × B + 0.75 × A × B − 1.34 × A² + 0.66 × B² − 2.08 × A × B²

The median particle size = 29.43 + 6.81 × A + 2.18 × B + 1.05 × A × B + 0.64 × A²

Bulk density = 0.37 − 0.095 × A + 0.039 × A²

Tensile strength = 3.32 + 1.00 × A + 0.11 × B − 0.11 × A × B − 0.15 × A² − 0.09 × B²

DT = 16.37 + 5.00 × A − 0.50 × B − 0.46 × A × B + 2.53 × A² + 0.02 × B² − 1.52 × A × B²

DTP = 9.71 + 4.72 × A − 0.54 × B + 0.54 × A²

Yield pressure = 318.13 − 22.84 × A + 7.60 × A²

E_sp = 13.22 + 1.16 × A + 0.16 × B − 0.40 × A × B − 0.38 × A²

---

Fig. 1 2D contour plots of the developed models. The dots in each graph represent the experimental runs, the number 5 represents the number of repetition of the central run, and the star is the selected optimized point ((A), 6.7%; (B), 42%). D_(0.5), the median particle size; DT, disintegration time of tablets containing the co-processed excipient only; DTP, disintegration time of tablets composed of 96.5% the co-processed excipient and 3.5% PVPP; E_sp, unit effective compaction work.

Fig. 2 3D response surface plots of the developed models. D_(0.5), the median particle size; DT, disintegration time of tablets containing the co-processed excipient only; DTP, disintegration time of tablets composed of 96.5% the co-processed excipient and 3.5% PVPP; E_sp, unit effective compaction work.

3.2.1. Effects on the powder properties. Angle of repose (AR) is a measure that is used to investigate the flowability of spray-dried powders. Fig. 1a1 and 2b1 are the contour and response surface plots that show the effect of the independent variables on AR. The results showed that when the amount of HPMC was constant, the solid content of the feed had little effect on AR, and when the solid content of the feed was constant, AR increased as the amount of HPMC increased. This could be explained by the fact that the bulk density of spray dried powders decreased with increasing the content of HPMC. HPMC itself has a low bulk density,⁹ and granules prepared by using HPMC as a wet binder are always porous and fluffy.

The results in Fig. 1a2 and 2b2 show that when the solid content of the feed was constant, the median particle size of the spray dried powders increased as the amount of HPMC increased. This is because that there is more and more HPMC available to agglomerate the suspended CC particles. In addition, a higher content of HPMC produced more viscous sprayed droplets, and then a layer of more completely thin film formed on the surface of composition particles, wrapping more primary CC particles inside. When the amount of HPMC was constant, the median particle size of the spray dried powders increased with increasing the solid content of the feed. This may be due to the larger volume occupied by the solid fraction, resulting in more particle collisions and agglomeration. In addition, the effect of the two factors on median particle size exhibited certain interaction (runs 1, 4, and 9) (Table 4).

The decrease in bulk density with increasing % HPMC was due to the low bulk density of HPMC. In addition, in the spray drying process, two raw materials were intimately and homogeneously combined to form the composite particles and the surfaces of the primary and composite particles were covered by HPMC macromolecules. Thus, the outer and inner surface properties of the composite particles become identical with those of the pure binder material.³⁴ In contrast, the solid content of the feed had much less influence on the bulk density of the spray dried powders (run 5 vs. 6 and 3 vs. 11) (Table 4).

3.2.2. Effects on the tablet properties and tableting parameters. At a constant solid content of the feed, runs 1, 4, and 9 with a relatively high content of HPMC (10.5% or 9.5%, w/w) resulted in a higher tablet tensile strength and disintegration time than the others. This is reasonable in terms of the specific distribution form of HPMC (homogeneous distribution on the surface of the primary and composite particles) induced by the co-spray drying process. With increasing the amount of HPMC, the outer and inner surface properties of the composite particles gradually approached to those of the pure binder HPMC, resulting in correspondingly increased binding and cohesion among and within particles during tableting.³⁴ In addition, the increase in tensile strength and disintegration time was correlated with decreasing tablet porosity. Tablets with a low porosity are prone to slow disintegration due to the slow water penetration into the network. In addition, the increase in tablet tensile strength leveled off when the amount of HPMC reached to certain value (Fig. 2b4). It has also been reported in literature that the tensile strength increased with binder polymer level up to a maximum, beyond which the tensile strength decreased with further increases in binder polymer levels.¹⁴ By contrast, at a constant amount of HPMC, the solid content of the feed had much less influences on the tablet tensile strength and disintegration time (run 6 vs. 13 and 4 vs. 9).

In this study, as shown in Table 5, two sets of disintegration time were determined, which corresponded to tablets containing 3.5% PVPP or not. The compacts, tableted with the co-processed excipient alone and having a breaking force of ∼60 N, exhibited a disintegration time of 14.26–28.41 min. Whereas, the disintegration time of most tablets with the same breaking force but containing 3.5% PVPP decreased to 4.34–14.91 min, less than the 15 min requirement for rapid release tablets in the European pharmacopoeia. Therefore, disintegration appears not to be a limiting factor for the application of the co-processed excipient developed here.

Tableting parameters, including yield pressure (P_y), E_sp, and fast elastic stretch, for the excipients studied were recorded and are summarized in Table 5. P_y is a parameter related to the ability of a material to deform under pressure.^35,36 Typically, lower values of P_y indicate the onset of deformation at lower applied pressures. The results in Fig. 1a7 and 2b7 indicate that when the amount of HPMC was constant, the solid content of the feed had no effect on P_y. When the solid content of the feed was constant, P_y decreased as the amount of HPMC increased. For example, the powders prepared by runs 1 and 4 with a high content of HPMC (10.5% and 9.5% w/w, respectively) had a significantly lower P_y than those with low amounts of HPMC (3.5 w/w for run 12 and 4.5 w/w for run 3) (Table 5). This is attributed to the plastic deformation nature of HPMC.

The E_sp values, which represent the energy retained in the tablet after unloading and are related to the deformation properties of tested materials as well as their binding properties, were calculated based on the works recorded by the press. When the amount of HPMC was constant, E_sp decreased as the solid content of the feed decreased. The lower the solid content is, the lower the HPMC concentration and the system viscosity are, and thus, the higher percentage of small atomized droplets and small dried particles are produced in order. During tableting, small particles tend to increase the friction, and thus, decrease the unit effective compaction work E_sp. When the solid content of the feed was constant, E_sp increased with HPMC level up to a maximum, beyond which the value decreased with further increases in HPMC levels (Fig. 1a8 and 2b8). In addition, the effect of the two factors on E_sp exhibited certain interaction. As to fast elastic stretch, no clear trends were observed, suggesting that the effects of the two factors are both insignificant or are offset from each other.

3.3. Optimization and model validation

Tablet tensile strength and disintegration time are two key considerations for excipients used in tablets. Tensile strength is important for assessing the binding capacity and compactibility of the excipient. And, tablet disintegration time can affect the drug release rate, which is related to the drug absorption and efficacy. Therefore, maximum tensile strength and minimum DTP were both chosen as the optimization criteria in this study. Numerical optimization was performed using the Design-Expert software to find the optimal values of the independent variables. Finally, the optimized feed composition was achieved, i.e., the HPMC level in the feed solid: A = 6.7% (w/w) and the solid content of the feed: B = 42% (w/v). These values predict tablet tensile strength of 3.21 MPa and DTP of 8.48 min. An additional experiment was performed to verify the choice feed composition. The results are shown in Table 8. The low value of prediction error (not more than 6.24%) indicates that the observed responses are in close agreement with the predicted values and that the model is valid. It can also be concluded that the optimization technique is appropriate for optimizing the feed composition to produce the co-processed excipient of CC and HPMC by co-spray drying.

Table 8 Observed responses and point prediction of the optimal feed composition (% HPMC, 6.7%; the solid content of the feed, 42%) (n = 3)^b

Responses	Observed values	Predicted values	Prediction errors (%)
a Standard deviation.b DT, disintegration time of tablets containing the coprocessed excipient only; DTP, disintegration time of tablets composed of 96.5% the coprocessed excipient and 3.5% PVPP; Esp, unit effective compaction work.
AR (°)	34.86 (0.32)^a	33.32	4.62
Bulk density (g mL^−1)	0.39 (0.03)	0.39	0
Median particle size (μm)	31.94 (0.02)	30.97	3.13
Tensile strength (MPa)	3.28 (0.02)	3.21	2.18
DT (min)	16.52 (0.01)	15.55	6.24
DTP (min)	8.91 (0.02)	8.48	5.07
Yield pressure (MPa)	319.20 (0.34)	321.03	−0.57
Esp (J g^−1)	13.69 (0.19)	13.33	2.70

---

4. Conclusions

To develop a novel co-processed tablet excipient based on calcium carbonate and HPMC, the central composite design-response surface methodology (CCD-RSM) was applied in this study to optimize the amount of HPMC E3 and the solid content of the feed used for co-spray drying. The results show that the surface models developed for powder flowability, median particle size, bulk density, tablet tensile strength, disintegration time, yield pressure, E_sp, and fast elastic stretch were significant, without significant lack of fit. However, modeling of production yield and moisture content was not significant. By variance and regression analysis, the quadratic regression equation was established as a predicted model. The optimal conditions were obtained at 6.7% (w/w) for the HPMC amount in the feed solid and 42% (w/v) for the solid content of the feed. The experimental values of the co-processed excipient prepared under the optimized conditions were mostly close to the model predicted values. Under these conditions, the tablet tensile strength and disintegration time were 3.28 MPa and 8.91 min, respectively. The combination use of CCD, RSM, and the desirability function can provide an insight into a lab-scale pharmaceutical study, and quickly obtain the optimal conditions, saving experimental time and materials as well as personal costs. It can provide theoretical guidance for further development of co-processed excipients based on calcium carbonate and HPMC.

Conflict of interest

The authors report no conflicts of interest.

Acknowledgements

This work was supported by the Innovation Program of Shanghai Municipal Education Commission (13ZZ098); the funds from Shanghai Municipal Commission of Health and Family Planning (ZY3-CCCX-3-5001) and Science and Technology Commission of Shanghai Municipality (15DZ2292000); the Science and Technology Development Fund of Pudong New Area (PKF-2013-003); and the natural science fund within budget of Shanghai university of traditional Chinese medicine (2013JW27).

References

1. P. Shivanand and O. L. Sprockel, Powder Technol., 1992, 69, 177–184.
2. S. K. Nachaegari and A. K. Bansal, Pharm. Technol., 2004, 28, 52–65.
3. P. Gupta, S. K. Nachaegari and A. K. Bansal, in Excipient Development for Pharmaceutical: Biotechnology and Drug Delivery Systems, ed. A. Katdare and M. V. Chaubal, Informa Healthcare, USA, 2006, pp. 109–126.
4. J. Rojas, I. Buckner and V. Kumar, Drug Dev. Ind. Pharm., 2012, 38, 1159–1170.
5. S. M. Ambore, J. Tekale and S. G. Gattani, World Appl. Sci. J., 2014, 31, 801–810.
6. C. Bacher, P. M. Olsen and P. Bertelsen, et al., Int. J. Pharm., 2007, 342, 115–123.
7. M. Delour desgarzonserra and L. Villafuerterobles, Int. J. Pharm., 2003, 258, 153–163.
8. L. V. Allen, Int. J. Pharm. Compd., 2000, 4, 306–310.
9. R. C. Rowe, P. J. Sheskey and M. E. Quinn, Handbook of Pharmaceutical Excipients, the Pharmaceutical Press, London, UK, 6th edition, 2009, pp. 326–329.
10. R. J. Roberts and R. C. Rowe, J. Pharm. Pharmacol., 1985, 37, 377–384.
11. S. Mattsson and C. Nystrom, Eur. J. Pharm. Sci., 2000, 10, 53–66.
12. Y. Wang, Y. Xie and D. Xu, et al., Drying Technol., 2014, 32, 557–566.
13. Y. Gonnissen, E. Verhoeven and E. Peeters, et al., Eur. J. Pharm. Biopharm., 2008, 69, 320–334.
14. X. Lin, C. W. Chyi and K. F. Ruan, et al., Eur. J. Pharm. Biopharm., 2011, 79, 406–415.
15. C. Huang, H. Wu and R. F. Li, et al., New Biotechnol., 2012, 29, 372–378.
16. M. Dutta and K. B. Jayanta, Int. J. Environ. Sci. Technol., 2012, 5, 42–53.
17. M. Jain, V. K. Garg and K. Kadirvelu, Bioresour. Technol., 2011, 102, 600–605.
18. N. F. Rahmati, M. M. Tehrani and K. Daneshvar, Food Biophys., 2015, 10, 39–50.
19. X. L. Liu, T. H. Mu and H. N. Sun, et al., Food Chem., 2013, 141, 3034–3041.
20. R. Chawla, S. Jaiswal and B. Mishra, Expert Opin. Drug Delivery, 2014, 11, 31–43.
21. Y. Y. Bei, X. F. Zhou and B. G. You, et al., Int. J. Nanomed., 2013, 8, 1795–1808.
22. J. Prakash Maran, V. Mekala and S. Manikandan, Carbohydr. Polym., 2013, 92, 2018–2026.
23. V. Samavati, Int. J. Biol. Macromol., 2013, 61, 142–149.
24. J. Prakash Maran, V. Sivakumar and R. Sridhar, Ind. Crops Prod., 2013, 42, 159–168.
25. J. T. Fell and J. M. Newton, J. Pharm. Sci., 1970, 59, 688–691.
26. P. J. Denny, Powder Technol., 2002, 127, 162–172.
27. N. A. Armstrong and H.-N. Rf, J. Pharm. Pharmacol., 1972, 24, 135P–136P.
28. J. Li, F. Wu and X. Lin, et al., RSC Adv., 2015, 5, 69289–69298.
29. Z. Ma, B. Gao and P. Wu, et al., RSC Adv., 2015, 5, 21042–21049.
30. Q. Wang, J. Zhang and A. Wang, RSC Adv., 2013, 3, 23423–23431.
31. G. D. Park and Y. Kang, RSC Adv., 2014, 4, 44203–44207.
32. S. Wang, J. Li and X. Lin, et al., Int. J. Pharm., 2015, 486, 370–379.
33. K. Masters, SprayDryConsult Intl, ApS, Denmark, 2002.
34. S. Adolfsson, C. Caramella and C. Nystrom, Int. J. Pharm., 1998, 160, 187–195.
35. P. Paronen, Drug Dev. Ind. Pharm., 1986, 12, 1903–1912.
36. M. Duberg and C. Nystrom, Powder Technol., 1986, 46, 67–75.

---