See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/5394040

Rapid Development and Optimization of Tablet Manufacturing Using

Statistical Tools

Article  in  AAPS PharmSciTech · February 2008

DOI: 10.1208/s12249-008-9095-z · Source: PubMed

CITATIONS READS
6 144

13 authors, including:

Eutimio Gustavo Fernández Núñez Arturo Gil

Universidade Federal do ABC (UFABC) University Pompeu Fabra
43 PUBLICATIONS   166 CITATIONS    2 PUBLICATIONS   6 CITATIONS   

SEE PROFILE SEE PROFILE

Rodolfo Valdés Mirna Fernández Cervera

Center for Genetic Engineering and Biotechnology University of Havana
71 PUBLICATIONS   455 CITATIONS    60 PUBLICATIONS   434 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

PHARMACEUTICAL INDUSTRY View project

Obtención de quitosana y sus sales a partir de quitina de langosta (Panulirus argus) para aplicaciones farmacéuticas y alimentarias View project

All content following this page was uploaded by Mirna Fernández Cervera on 13 November 2017.

The user has requested enhancement of the downloaded file.

AAPS PharmSciTech, Vol. 9, No. 2, June 2008 (# 2008)
DOI: 10 .12 0 8 /s12 2 49 -0 0 8 -9 0 9 5-z

Research Article

Rapid Development and Optimization of Tablet Manufacturing

Using Statistical Tools

Eutimio Gustavo Fernández,1,6 Silvia Cordero,2 Malvina Benítez,1 Iraelio Perdomo,1 Yohandro Morón,1
Ada Esther Morales,1 Milagros Gaudencia Arce,1 Ernesto Cuesta,1 Juan Lugones,2 Maritza Fernández,2
Arturo Gil,3 Rodolfo Valdés,4 and Mirna Fernández5

Received 7 December 2007; accepted 24 March 2008; published online 6 May 2008
Abstract. The purpose of this paper was to develop a statistical methodology to optimize tablet
manufacturing considering drug chemical and physical properties applying a crossed experimental design.
The assessed model drug was dried ferrous sulphate and the variables were the hardness and the relative
proportions of three excipients, binder, filler and disintegrant. Granule properties were modeled as a
function of excipient proportions and tablet parameters were defined by the excipient proportion and
hardness. The desirability function was applied to achieve optimal values for excipient proportions and
hardness. In conclusion, crossed experimental design using hardness as the only process variable is an
efficient strategy to quickly determine the optimal design process for tablet manufacturing. This method
can be applied for any tablet manufacturing method.
KEY WORDS: crossed experimental design; ferrous sulfate; multivariate techniques; statistical strategy;
tablet manufacturing.

INTRODUCTION and crossed experimental designs (8 ,9 ). Multivariate design is a

combination of classical experimental designs with multivariate
In pharmaceutical industries, manufacturers of generic techniques such as principal component analysis or partial least
tablets are usually focused on the optimization of the square (10 ). Crossed experimental design describes every
excipient mixture composition to obtain a product that meet response variable as a function of the manufacturing process
established standards (1,2 ). Pharmaceutical literature, already parameters and ingredient percentages. For instance, multivar-
abounds with statistical methods considering mixture design iate designs have been used to optimize two-step process,
to optimize excipient proportions. Several tablet compositions granulation-tableting, and the excipient types and their pro-
of extended and fast release have been established with this portions in tablet formulation (11). However crossed experi-
methodology (3–7). However tablet properties do not only mental design has not been widely employed.
depend on the excipient percentage in the solid dosage. Some The application of multivariate and crossed experimental
compression and granulation process variables such as: designs is experimentally tedious but allows for finding the
compaction force, compression velocity, tableting tempera- optimal operational conditions and ingredient percentages for
ture, impeller speed and blending time can also have any given system. In general, experimental design strategies to
influences (8 ). develop pharmaceutical solid forms are complex, expensive
In order to solve the optimal tablet manufacturing in an and time-consuming because of the large number of factors
integral way, two main strategies have been used; multivariate that influence on the drug behavior (12 ). This highlights, the
need to combine technological and statistical knowledge to
reduce traditional methods for establishing optimal tablets.
1
Inorganic Chemistry Department, Center for Engineering and According to the criterion of the authors, the most
Chemical Researches, F Street, # 115 be/Ave 5th and Calzada, important variables to consider are hardness and the types
Havana, 10 40 0 , Cuba. and proportion of excipients regardless of the tablet manu-
2
“Reinaldo Gutiérrez” Pharmaceutical Laboratories, Km 5½ Avenue facture process (direct compression or wet granulation).
Independence, Boyeros, Havana, Cuba. Wet granulation variables such as: operational factors
3
Medsol Laboratories, La Lisa, Havana, Cuba. were not considered in this design because dependent
4
Monoclonal Antibody Production Department, Center for Genetic
variables such as: wet mass density and viscosity, particle size
Engineering and Biotechnology, Ave 31 be/158 and 19 0 , P.O. Box
6162 , Havana, 10 60 0 , Cuba.
distribution, flowability or tableting parameters at the end of
5
Institute of Pharmacy and Food, University of Havana, Street 2 3 this granulation are more important (13). Based on this
# 2 142 5 be/2 14 and 2 2 2 , La Coronela, La Lisa, Havana, Cuba. criterion; the reduction of experimental variables, specifically
6
To whom correspondence should be addressed. (e-mail: eutimiocu@ granulation variables, is possible and as a consequence the
yahoo.com) total experimental work to develop a new tablet.

1530 -9 9 32 /0 8 /0 2 0 0 -062 0 /0 # 2 0 0 8 American Association of Pharmaceutical Scientists 620

Rapid Development and Optimization of Tablet Manufacturing 621

The purpose of this paper was to develop a statistical Table II. D-optimal Mixture Design for Granules Properties and
methodology for tablet manufacturing using ferrous sulfate as Crossed Experimental Design for Tablet Properties
a model drug. This methodology was based on a crossed
experimental design considering the hardness as the only Cellulose PVP Explotab Hardness
process variable and the proportion of main ingredients of the Sample (%) (%) (%) (kgF/Monsanto)
formulation as mixture variables. 1 2 5.60 3.0 0 6.0 0 5.5
2 2 7.60 5.0 0 2 .0 0 5.5
MATERIALS AND METHODS 3 2 8 .60 4.0 0 2 .0 0 5.5
4 2 5.10 4.50 5.0 0 5.5
Materials 5 2 3.60 5.0 0 6.0 0 5.5
6 2 6.2 7 3.67 4.67 5.5
The following active pharmaceutical ingredient and exci- 7 2 9 .60 3.0 0 2 .0 0 5.5
8 2 5.60 3.0 0 6.0 0 5.5
pients were used: dried ferrous sulphate (Merck, Germany),
9 2 5.60 5.0 0 4.0 0 5.5
polyvinylpyrrolidone (PVP) Kollidon K25 (Basf, Germany), 10 2 4.60 4.0 0 6.0 0 5.5
Microcrystalline cellulose (MCC) (Blanver, Brazil), sodium 11 2 6.9 3 4.33 3.33 5.5
starch glycolate (Explotab) (Gustav Parmentier, Germany), 12 2 7.60 3.0 0 4.0 0 5.5
magnesium stearate (Otto Barlocher GmbH, Germany). All 13 2 7.60 5.0 0 2 .0 0 5.5
other chemicals and solvents were of analytical reagent grade. 14 2 9 .60 3.0 0 2 .0 0 5.5
15 2 3.60 5.0 0 6.0 0 5.5
Methods
Note: this experimental matrix was repeated two more times at
hardness values, 6.5 and 7.5 KgF/Monsanto respectively to generate
Preparation of ferrous sulphate tablets
the crossed experimental design.

The composition of the fast release ferrous sulphate tablets

and the range for three excipients under study are listed in granules to measure the time of granule mass flow through the
Table I. The drug and MCC were weighed and mixed. The funnel and calculate the flow rate by the mathematical Eq. 1
PVP dissolved in Ethanol was added to make a wet mass in a (each reported value is the average of three determinations):
lab mixer. The end point of the granulation process was
m
determined using the wet mass density as criterion. Then the Fr ¼ ð1Þ
0 :78 5 " d2o " t
wet mass was granulated through a 14 mesh sieve. Granules Where:
were dried at 40 °C in an oven for 8 h, and sieved through a 16
Fr is the flow rate expressed as gram per square
mesh sieve, after that they were blended with a constant
centimeter per second, m is the granules mass (gram), do is
percentage of magnesium stearate and Explotab. Each 32 0 mg
the outlet orifice diameter of the funnel (centimeter) and t is
tablet containing 65 mg elemental iron was compressed using
the time required (second) for the granules mass to flow
5/16 in. diameter normal concave punches at three hardness
through the metal funnel (14).
values, 5.5, 6.5 and 7.5 kgF/Monsanto (Table II). Tablets were
For the determination of bulk and tap densities, 60 g of the
produced in an instrumented eight-punch press (Ronchi, Italy).
sample was poured in a 10 0 mL tared graduated cylinder. The
The upper punch force was registered in an analyzing recorder
volume was then read directly from the cylinder and used to
(Yokogawa model 3655E).
calculate the bulk density (Db) according to the mass/volume
ratio. For tap density (Dt) the cylinder was tapped 1,0 0 0 times
Methods for characterization of granules
using a tap density analyzer (Erweka SVM1, Germany).
The percent compressibility (C) was measured and
The granule mean particle size and the size distribution
calculated from Eq. 2 (15):
were measured by applying a shaking sieve with a set of sieves of
the following apertures; 8 0 0, 500 , 2 50 , 12 5, 63, 45 μm. Dt % Db
The flow rate was determined according to the fixed- C¼ & 10 0 ð2 Þ
Dt
funnel method (14). The stainless steel funnel end with a wall
angle of 45° and 2 .3 cm outlet orifice diameter was placed
Methods for characterization of tablets
10 cm above a flat base. The funnel was filled with 10 0 g of
The percentages of friability were calculated as the
Table I. Variables in the Mixture Design percentage of weight loss of 2 0 tablets after 10 0 rotations in
an Erweka Abrasion Tester (Model 1AP, Germany). Tablet
Levels
height was measured with an Ultra-Micrometer Fowler
Formulation components Low (%) High (%) (USA) sensitive 0 -1 in. The hardness was quantified by using
a tablet hardness tester of Monsanto type (Toshiba, India).
X1 = percent of MCC in tablet 2 3.6 2 9 .6 The disintegration time (second) was determined by mean of
X2 = percent of PVP in tablet 3.0 5.0
a disintegrator Erweka ZT 12 0 (Germany) apparatus. Deion-
X3 = percent of Explotab in tablet 2 .0 6.0
ized water at 37±1 °C was used as the immersion medium.
Magnesium stearate Constant
All measurements were made in triplicate.
The amount of dried ferrous sulphate was fixed at 2 0 3 mg. The tablet Chemical characterization of the tablets including disso-
weight was 32 0 mg. lution of ferrous sulphate was measured according to the USP
622 Fernández et al.

Table III. Physical–Mechanical Parameters of the Granules

Experiments Mean particle sizes (μm) Flow rate (g/cm2 s) Tap density (g/cm3) Bulk density (g/cm3) Carr’s Index (%)

1 178 .3 15.35 0 .9 1 0 .68 2 5.6

2 2 40 .3 15.35 0 .8 0 0 .63 2 1.1
3 2 0 3.6 14.2 5 0 .8 7 0 .65 2 5.5
4 2 35.9 17.18 0 .9 2 0 .69 2 4.7
5 2 30 .3 13.52 0 .8 8 0 .71 19 .0
6 19 3.0 14.2 5 0 .9 5 0 .74 2 1.8
7 18 7.1 15.35 0 .9 6 0 .69 2 8 .1
8 2 11.7 14.2 5 1.0 2 0 .75 2 5.9
9 2 30 .0 16.45 0 .8 6 0 .71 17.4
10 2 19 .5 14.2 5 0 .8 6 0 .68 2 0 .9
11 2 2 0 .0 16.45 0 .8 9 0 .68 2 3.6
12 2 0 0 .7 14.2 5 1.0 0 0 .72 2 8 .0
13 2 0 8 .9 14.2 5 0 .8 6 0 .65 2 4.4
14 18 0 .8 13.15 0 .9 0 0 .66 2 6.7
15 2 41.7 15.35 0 .8 9 0 .68 2 3.6

The number of the experiments and excipients proportions are defined in Table II.

2 7th ed. The dissolution test was carried out by the Paddle tions and quadratic for the hardness. The best fitting mathe-
method, at a paddle speed of 50 rpm, in 9 0 0 mL HCl 0 .1 N at matical model was selected based on the comparison of the
37±0 .5 °C, and a dissolution time of 45 min. The dissolution predicted residual sum of square (PRESS). This statistical
tester employed was the Erweka DT 60 0 (Germany). The parameter indicates how well the models fit the data, and the
iron concentration of each sample (n=6) was determined chosen model PRESS should be small relative to the other
employing atomic absorption spectrophotometry at a models under consideration (5). The goodness and lack of fit
wavelength of 2 48 .3 nm (UNICAM 9 2 9 AA Spectrometer). tests were also used to demonstrate the model statistical
The dissolution profile was made taking 5 mL samples at 10 , adjusts. Desirability function was the numeric method to op-
2 0 , 30 and 45 min while the same volume of fresh dissolution timize mixture and process variables in tablet manufacturing.
medium was returned to the vessels. The SIMCA P version 11.0 software and Statgraphics
Plus 5.0 were used to carry out the principal component
Statistical analysis analysis (PCA) and the cluster analysis for granule variables.
The aim for the application of the PCA and the cluster
The definition and data processing of experimental analysis was to detect similitude or difference among the 15
designs for granules and tablet properties were made in granules included in D-optimal mixture design.
Design Expert version 6.0 .1 software. D-optimal mixture
design for three components (MCC, PVP, Explotab percen- RESULTS AND DISCUSSION
tages) was used for modeling the granule parameters, the
number of experimental points (15) were enough to adjust The present statistical strategy includes the following stages.
special cubic models (Eq. 3): Firstly, it is necessary to define the excipients and elaboration
! " method in accordance with chemical and physical properties of
Y ¼ b1 " X1 þ b2 " X2 þ b3 " X3 þ b12 " X1 " X2 þ b13 " X1 " X3 the drug and its function in the human body. Secondly, the
þ b2 3 " X2 " X3 þ b12 3 " X1 " X2 " X3 excipients with the greatest impact on granule and tablet
ð3Þ properties should be selected for optimizing excipient propor-
tions in the formulation, together with the definition of hardness
A crossed experimental design was employed for tablet range. Then, mixture variables and hardness are combined using
properties modeling. The mixture variables were the same for a crossed experimental design (Table II). However, in spite of
granulation step and the process variable was hardness at three the few variables required for this strategy, costs could still be
levels (5.5, 6.5, 7.5 kgF/Monsanto). This second design is able to appreciable. A multivariate methodology can be used to reduce
fit mathematical models, special cubic for the excipient propor- the experimental cost. Especially in granulation–compression

Table IV. Best Models for Granules Variables and Some Statistical Parameters

p value
p value (goodness
Variable Model Press (lack of fit test) of fit)

Mean particle sizes (MPS) (μm) MPS ¼ 3:39 42 8 " X1 þ 2 3:8 0 752 " X2 þ 6:42 8 55 " X3 2 658 .0 2 0 .9 746 0 .0 0 0 3
Flow rate (FR) (g/cm2 s) FR ¼ 0 :37716 " X1 þ 0 :8 0 143 " X2 þ 0 :41376 " X3 2 5.0 3 0 .52 10 0 .510 7
Tap density (TD) (g/cm3) TD ¼ 0 :0 30 78 3 " X1 % 0 :0 18 556 " X2 þ 0 :0 39 9 9 6 " X3 0 .0 3 0 .8 59 5 0 .0 0 30
Bulk density (BD) (g/cm3) BD ¼ 0 :0 2 0 0 73 " X1 þ 6:60 516 " 10 % 3 " X2 þ 0 :0 31770 " X3 0 .0 1 0 .7445 0 .0 2 56
Carr’s Index (CI) (%) CI ¼ 1:0 560 4 " X1 % 1:58 9 16 " X2 þ 0 :538 74 " X3 8 2 .0 6 0 .49 45 0 .0 0 37
Rapid Development and Optimization of Tablet Manufacturing 623

Fig. 1. Cox trace graphs for a tap density (g/cm3) and b bulk density (g/cm3) of granules

processes, the compression of all mixtures at different hardness Specifically, the principal difficulty with wet granulation
values for subsequent chemical and mechanical tablet assess- in high-shear mixers is to decide “when to stop”: hence, the
ments would be limited due to cost restraints. importance of controlling the end point. It has been also
Finally, granule and tablet properties are described as demonstrated that the operational parameters in high-shear
mathematical functions of excipient proportions and hard- mixers do not have a significant influence on the granulation
ness. Then, the best hardness and excipient proportions can effectiveness. Thus, the manufacturers can lose time and
be obtained using the numeric multiple optimization proce- money meticulously studying granulation variables on lab
dure known as the desirability function. scale such as: mixing time, fill level, liquid spray rate, etc. (16).
This methodology could be useful for any tablet manufac-
turing method. In the direct compression method, it is further Characterization of granules
recommended because the granulation variables do not exist.
However, the present methodology was applied for the wet gran- A D-Optimal mixture design was carried to determine
ulation method where the granulation step variables were omitted the granule behavior as a function of selected excipient
because of the scale-up of granulation processes is a difficult task proportions. The results of granule variables (Table III)
and the trial and error methods have been suggested (16). allowed adjusting the best numeric model for each variable

Table V. Granule Size Compositions of the Mixtures Included in D-optimal Mixture Design

Run >8 0 0 μm 8 0 0 –50 0 μm 50 0 –2 50 μm 2 50 –12 5 μm 12 5–63 μm 63–45 μm 45 μm–collector

1 3.48 1.9 6 10 .40 57.9 2 13.56 9 .56 3.9 2

2 8 .52 4.9 6 2 9 .9 2 46.12 7.76 3.8 8 0 .2 4
3 4.60 3.40 18 .12 52 .0 8 12 .8 4 7.0 0 2 .16
4 11.56 10 .9 6 2 0 .44 40 .16 10 .44 5.44 1.8 4
5 8 .0 0 6.8 4 2 3.72 45.64 8 .8 8 5.40 2 .8 4
6 5.2 4 3.36 14.36 52 .48 16.60 6.0 4 2 .0 8
7 3.40 2 .76 12 .76 53.56 17.64 8 .0 4 2 .32
8 6.36 7.40 17.16 42 .64 16.44 7.12 3.0 8
9 7.64 5.60 2 5.60 45.8 0 9 .56 5.0 8 1.40
10 6.40 5.0 0 2 2 .64 47.68 9 .9 2 6.32 2 .9 2
11 9 .32 5.76 2 3.32 44.12 9 .2 8 6.8 4 1.72
12 4.40 6.0 0 16.0 0 44.9 2 13.68 8 .52 7.44
13 7.48 4.0 8 19 .64 51.2 4 11.2 4 3.8 0 0 .32
14 2 .58 1.53 10 .47 59 .43 16.2 4 9 .0 7 0 .68
15 9 .8 0 9 .68 2 4.52 39 .68 10 .48 4.8 0 0 .8 8
624 Fernández et al.

Fig. 2. Principal component analysis for granule properties without Carr’s Index inclusion (Score plot). Cluster analysis by Nearest Neighbor
method and Squared Euclidean metric distance

(Table IV). The flow rate was independent of the excipient different (Table IV). The equations listed in Table IV are
percentage because the goodness of fit and lack of fit tests Scheffé models that can conduce to false interpretations of
were not significant for a linear model (p>0 .0 5), the best components influence, especially in mixture problem with
adjusted model, suggesting that the values for flow rate in the constrains. For solving this problem, Cox models are prefer
experimental region are oscillating around a central value. (17). Figure 1a and b illustrate Cox models for tap and bulk
The mean particle size showed a significant linear depen- densities respectively. As predictable, the PVP percentage is
dence with excipient proportions (Table IV). The positive inversely proportional to tap and bulk densities due to the
influence of PVP percentage was expected because its function particle size, and the increase in free space among particles
is binding the excipients and favoring the mechanical resistance causing an increase of the volume per mass unit.
of the formed particles in the granulation process. The Carr’s Index showed a behavior similar for bulk and
Tap and bulk densities also confirmed a linear depen- tap densities (Table IV). The direct measurement of the
dence with excipients proportions. It is important to emphasize granule flowability (flow rate) is not in concordance with the
that the influences of ingredient percentages are similar for indirect measurement (Carr’s Index). Many studies suggest
both variables, although the sign of the coefficients for PVP are that flow properties are manner test performance dependent

Fig. 3. Biplot graph (scores and loading) for granules properties not including Carr’s Index
Rapid Development and Optimization of Tablet Manufacturing 625

(18 ,19 ). Results can also vary with the prior measurement Table VI. Results of Crossed Experimental Design (Tablet Properties)
sample manipulation (2 0 ). Therefore, a direct method to
evaluate the fluidity is recommended. In the present study, Height Friability Disintegration Press force
the Carr’s Index was eliminated in subsequent analysis. Run (mm) (%) (min) (kN)
The granule size composition is very useful to reproduce the 1 4.412 0 .14 2 2 .0 0 2 .9 6
granulation process on large-scale (Table V). In this study the 2 4.438 0 .2 3 40 .0 0 1.61
experimentation cost was not significant and all formulations 3 4.58 2 0 .2 5 38 .0 0 1.49
were compressed at three hardness values, under the crossed 4 4.558 0 .51 55.0 0 2 .8 3
experimental design definition. 5 4.2 33 0 .2 8 45.0 0 3.63
The practical alternative of this methodology for more 6 4.2 49 0 .33 54.0 0 4.2 3
complicated and expensive tablet manufacturing problems is 7 4.42 0 0 .42 48 .0 0 1.9 4
the PCA to reduce the variable number through the formula- 8 4.2 48 0 .57 2 9 .0 0 3.18
9 4.2 9 6 0 .45 42 .0 0 2 .9 6
tion scores in principal components (PC) and to combine them
10 4.2 34 0 .40 37.0 0 3.55
with the hardness employing a central composite design, Box– 11 4.2 48 0 .48 55.0 0 2 .19
Behnken design, or a three level factorial design. 12 4.42 4 0 .32 65.0 0 3.71
In this work the PCA was used to determine the 13 4.511 0 .60 56.0 0 1.44
differences and similarities among granules and the influenc- 14 4.576 0 .45 35.0 0 2 .50
ing variables. The PCA was applied from the Tables III and V 15 4.778 0 .2 3 62 .0 0 2 .33
without including the Carr’s Index. 16 4.311 0 .14 2 7.0 0 3.77
In Fig. 2 are represented the 15 granules of the D- 17 4.38 4 0 .17 40 .0 0 2 .79
Optimal design in a score plot. This figure allows visualizing 18 4.68 6 0 .30 44.0 0 2 .56
with fewer dimensions the data included in Tables III and V 19 4.2 35 0 .2 5 76.0 0 5.62
20 4.12 0 0 .2 4 42 .0 0 4.55
facilitating the detection of similarities among granules. In
21 4.2 43 0 .32 50 .0 0 4.39
summary, three clusters were detected. 22 4.2 16 0 .2 5 77.0 0 4.9 8
The Biplot graphic (Fig. 3) allowed detecting the variables, 23 4.18 6 0 .30 45.0 0 5.10
which cause the differences among granules. The granules 24 4.173 0 .2 0 48 .0 0 3.69
corresponding to experiment 12 had high percentages of 25 4.18 7 0 .2 6 51.0 0 3.9 5
small particle fractions whereas in experiment 2 a contrary 26 4.315 0 .2 2 70 .0 0 3.41
effect for granules was observed. The rest of the granules had 27 4.412 0 .2 6 65.0 0 4.64
similar properties (Fig. 3). 28 4.454 0 .49 60 .0 0 2 .59
According to the loading proximity of granule descriptor 29 4.52 3 0 .2 9 65.0 0 3.49
variables, tap and bulk densities increase in proportion to the small 30 4.8 41 0 .2 1 61.0 0 3.2 0
31 4.2 10 0 .15 50 .0 0 5.76
particle fraction percentage. This relationship is explained by the
32 4.2 0 0 0 .13 75.0 0 4.0 1
reduction in void space among particles and as a consequence the 33 4.10 0 0 .2 0 60 .0 0 4.8 3
volume decreases. In addition flow rate, mean particle size and 34 4.140 0 .2 4 62 .0 0 6.74
large size fractions are positively correlated because particle size 35 4.2 0 0 0 .18 49 .0 0 5.2 7
increase generally produces high flow rates (Fig. 3). 36 4.2 50 0 .2 5 52 .0 0 5.12
37 4.130 0 .14 10 9 .0 0 5.47
Characterization of tablets 38 4.0 9 0 0 .2 7 69 .0 0 6.36
39 4.130 0 .16 68 .0 0 5.0 4
The 15 granules were compressed at three hardness 40 4.0 8 0 0 .16 57.0 0 4.8 3
values: 5.5, 6.5, and 7.5 kgF/Monsanto (Table VI). The 41 4.130 0 .0 9 70 .0 0 4.8 8
42 4.360 0 .11 8 0 .0 0 5.70
models, which describe the behavior of four tablet parame-
43 4.340 0 .14 76.0 0 3.18
ters: height (Eq. 4), friability (Eq. 5), disintegration (Eq. 6) 44 4.40 0 0 .18 111.0 0 5.2 2
and press force (Eq. 7) are: 45 4.730 0 .34 68 .0 0 4.46

Height ¼ 0 :0 52 579 " X1 % 2 :62 130 " X2 % 3:52 454 " X3

þ 0 :11666 " X1 " X2 þ 0 :152 55 " X1 " X3

Disintegration ¼ 2 :69 8 37 " X1 þ 147:7650 0 " X2
% 3:468 8 4:10 % 3 " X1 " Hardness þ 1:0 8 370 " X2
% 69 :340 2 5 " X3 % 5:9 69 2 9 " X1 " X2
" X3 % 0 :42 157 " X1 " X2 " X3 ð4Þ
Friability ¼ 0 :459 58 " X1 þ 14:9 8 142 " X2 þ 11:419 12 " X3 þ 2 :1110 3 " X1 " X3 ð6Þ
% 0 :60 478 " X1 " X2 % 0 :48 654 " X1 " X3 % 0 :0 658 43X1 " þ 0 :48 545 " X1 " Hardness
Hardness % 3:74444 " X2 " X3 % 2 :2 1416 " X2 " Hardness
% 1:62 39 2 " X3 " Hardness þ 0 :139 2 5 " X1 " X2 " X3 Press force ¼ % 0 :0 9 0 549 " X1 þ 6:30 9 55 " X2 % 4:2 710 0
þ 0 :0 8 9 2 0 3 " X1 " X2 " Hardness þ 0 :0 69 2 62 " X1 " X3 % 0 :2 58 71 " X1 " X2 þ 0 :169 78 " X1 " X3
" X3 " Hardness þ 0 :540 43 " X2 " X3 "
Hardness % 0 :0 2 0 0 2 9 " X1 " X2 " X3 " Hardness
þ 0 :0 48 19 0 " X1 " Hardness % 0 :0 158 2 3 " X3

ð5Þ " Hardness ð7Þ

626 Fernández et al.

Fig. 4. Dissolution profile for tablets elaborated according to run 1 belongs to crossed experimental design. 1, 2 ,….,6 dissolution vessels

In order to determine if tablet properties are affected by In sample 1 the minimum disintegration time corre-
hardness in equal magnitude for all mixtures, elimination of a sponded with the minimum values of binder proportion and
mixture variable of the models is suggested. If the trans- hardness and the maximum vale of the disintegrant within
formed model considers hardness as an isolate variable, the each respective range studied. Finally it is important to
hypothesis is accepted (2 1). The Height (Eq. 8 ), Friability verify the chemical quality and in vitro dissolution of the
(Eq. 9 ) and Press force (Eq. 10 ) showed this behavior: tablets. The dissolution profile of experiment 1 is illustrated
in Fig. 4. The percentage of the released drug was greatest
Height ¼ 4:8 8 32 5 þ 0 :0 2 7479 " X2 % 0 :0 178 14 " X3 than the required (8 0 %) (2 2 ). The rest of chemical parame-
ters satisfied the USP XXVII requirements. In this way the
% 0 :0 9 0 567 " Hardness ð8 Þ
ferrous sulphate tablet was optimized rapidly and more cost-
effectively.

Friability ¼ 0 :9 0 718 þ 3:62 143 " 10 % 3 " X2 % 3:8 8 0 78 CONCLUSIONS

" 10 % 3 " X3 % 0 :0 9 7333 " Hardness ð9 Þ Crossed experimental design using hardness as the only
process variable is an efficient strategy to quickly determine
the optimal design process for tablet manufacturing. This
Press Force ¼ % 3:370 68 % 0 :42 8 0 4 " X2 þ 0 :2 8 0 9 0 " X3 method can be applied for any tablet manufacturing method.
þ 1:2 10 65 " Hardness ð10 Þ
ACKNOWLEDGMENTS
Negative values for hardness in height and friability and
positive in press force were expected according to the relation Authors would like to thank the financial support
among the variables. Nevertheless, the disintegration showed a granted by Quimefa Group and Merck for supplying dried
complex relation among the mixture variables and hardness. ferrous sulphate. Authors kindly thank Mrs. Relma Tavares
de Oliveira for the inspiration and Yohana and Brenda
Morón for their kind contributions. Authors are also grateful
Tablet optimization to Mena Cayetana Ramos from “Salvador Allende” hospital
for the carefully reviewing of the paper.
Since height and friability showed acceptable values, the
only criterion to optimize the tablet manufacturing was the
minimum disintegration time using desirability function with REFERENCES
an importance value of 5. In more complex solid pharmaceu-
tical systems like extended release tablets, more response 1. J. Gabrielsson, N. O. Lindberg, and T. Lundstedt. Multivariate
variables should be included, principally drug release at methods in pharmaceutical applications. J. Chemom. 16:141–160
different times. Even in some fast release solid systems in (2 0 0 2 ).
2 . B. Campisi, D. Chicco, D. Vojnovic, and R. Phan-Tan-Luu.
which granule flow properties show a significant relationship Experimental design for a pharmaceutical formulation: optimi-
with the excipient percentage should be considered them in sation and robustness. J. Pharm. Biomed. Anal. 18:57–65 (19 9 8 ).
the desirability function. 3. A. Bodea, and S. E. Leucuta. Optimization of hydrophilic matrix
The proportions that guarantee the minimum disintegra- tablets using a D-optimal design. Int. J. Pharm. 153:2 47–2 55
tion time were: MCC (2 5.47%), PVP (3.13%) and Explotab (19 9 7).
4. Y.-B. Huang, Y.-H. Tsai, W.-C. Yang, J.-S. Chang, P.-C. Wu, and
(6.0 0 %) and a hardness value of 5.5 kgF/Monsanto. Hence, K. Takayama. Once-daily propranolol extended-release tablet
the experimental conditions are very close to experiment dosage form: formulation design and in vitro/in vivo investiga-
number 1 of the crossed experimental design (Table II). tion. Eur. J. Pharm. Biopharm. 58:60 7–614 (2 0 0 4).
Rapid Development and Optimization of Tablet Manufacturing 627

5. Y.-B. Huang, Y.-H. Tsai, S.-H. Lee, J.-S. Chang, and P.-C. Wu. process by wet granule sieving method. II. Utility of wet granule
Optimization of pH-independent release of nicardipine hydro- sieving method. Drug Dev. Ind. Pharm. 232 :2 0 3–2 15 (19 9 7).
chloride extended-release matrix tablets using response surface 14. A. Iraizoz, O. Bilbao, and M. A. Barrios. Conferencias de
methodology. Int. J. Pharm. 289:8 7–9 5 (2 0 0 5). Tecnología Farmacéutica II. Facultad de Farmacia y Alimentos.
6. B. Perissutti, F. Rubessa, M. Moneghini, and D. Voinovich. Departamento de Tecnología y Control de Medicamentos.
Formulation design of carbamazepine fast-release tablets pre- Universidad de La Habana, ENPES, Ciudad Habana, 19 9 0 , pp.
pared by melt granulation technique. Int. J. Pharm. 256:53–63 170 –18 1.
(2 0 0 3). 15. R. L. Carr. Evaluating flow properties of solids. Chem. Eng.
7. T. Martinello, T. M. Kaneko, M. V. Robles Velasco, M. E. Santos 18:163–168 (19 65).
Taqueda, and V. O. Consiglieri. Optimization of poorly compact- 16. A. Faure, P. York, and R. C. Rowe. Process control and scale-up
able drug tablets manufactured by direct compression using the of pharmaceutical wet granulation process: a review. Eur. J.
mixture experimental design. Int. J. Pharm. 322:8 7–9 5 (2 0 0 6). Pharm. Biopharm. 52:2 69 –2 77 (2 0 0 1).
8 . R. Bergam, M. E. Johansson, T. Lundstedt, E. Seifert, and J. 17. L. Eriksson, E. Johansson, and C. Wikströn. Mixture design-
Åberg. Optimization of granulation and tabletting process by design generation, PLS analysis, and model usage. Chemom.
sequential design and multivariated analysis. Chemom. Intell. Intell. Lab. Syst. 43:1–2 4 (19 9 8 ).
Lab. Syst. 44:2 71–2 8 6 (19 9 8 ). 18 . J. Swarbrick. Encyclopedia of Pharmaceutical Technology.
9 . C. A. A. Duineveld, A. K. Smilde, and D. A. Doornbos. Designs Boylan, James. Marcel Dekker Inc., New York. Vol. XII. Tome
for mixture and process variables applied in tablet formulations. 6. 19 9 7; pp. 165.
Anal. Chim. Acta. 2772 :455–465 (19 9 3). 19 . A. Schϋssele, and A. Bauer-Brandl. Note on the measurement of
10 . S. Wold, M. Sjöström, R. Carlson, T. Lundstedt, S. Hellberg, B. flowability according to the European Pharmacopoeia. Int. J.
Skagerberg, C. Wikström, and J. Öhman. Multivariate design. Pharm. 257:30 1–30 4 (2 0 0 3).
Anal. Chim. Acta. 191:17–32 (19 8 6). 20. O. Antikainen. New Methods to Evaluate Applicability of
11. J. Gabrielsson, Å. Nyström, and T. Lundstedt. Multivariate Powders and Granules for Tablet Compression, University of
methods in developing and evolutionary strategy for tablet Helsinki, Faculty of Science, Finland, 2 0 0 3, pp. 3–4, ISBN:ISBN
formulation. Drug Dev. Ind. Pharm. 26:2 75–2 9 6 (2 0 0 0 ). 9 52 -10 -10 61-4.
12 . S. Furlanetto, M. Cirri, F. Maestrelli, G. Corti, and P. Mura. 2 1. J. A. Cornell. Experiments with mixtures. Design, Models and
Study of formulation variables influencing the drug release rate the Analysis of Mixture Data. Chapter 7. The Inclusion of
from matrix tablets by experimental design. Eur. J. Pharm. Process Variables in Mixtures Experiments. Wiley, Inc., United
Biopharm. 62:77–8 4 (2 0 0 6). States of America, 19 9 0 , pp. 38 3–38 9 .
13. H. Emori, Y. Sakuraba, K. Takahashi, T. Nishihata, and T. 22. USP Pharmacopeia XXVII, 2 0 0 3. Pharmacopeia Convention.
Mayumi. Prospective validation of high-shear wet granulation US Pharmacopeia XXVII, Rockville, MD, USA, pp. 79 2 .

View publication stats