Date of Award

Spring 2015

Degree Type


Degree Name

Doctor of Philosophy (PhD)


Pharmacy Practice

First Advisor

James D. Litster

Committee Chair

James D. Litster

Committee Member 1

Gregory Knipp

Committee Member 2

Jon Hilden

Committee Member 3

Carl Wassgren


A mechanistic understanding of the relationship among the granule composition, individual granule property, downstream robust processing, and the final tablet attributes is critical for rational development of a quality tablet. However, in the dry granulation literature, milled granules that are polydisperse in solid fraction, size, and shape are extensively used for compaction studies. Thus, the effect of an individual granule property or an individual component on the compaction properties of dry granules was not adequately separated. To advance the mechanistic understanding of the effect of dry granulation on tensile strength of tablets, individual granule properties such as size, solid fraction, and granule composition need to be decoupled, precisely controlled, and independently varied. To accomplish this, in this thesis, small cylindrical biconvex compacts of powder were used as model dry granules for compaction studies, which have the advantage of being monodisperse in both size and solid fraction. In addition, size and solid fraction of monodisperse granules and their composition were independently varied and precisely controlled. The novel use of monodisperse granules brings a new perspective on understanding the compaction properties of deformable dry granules. The effect of granule size and solid fraction on tensile strength of tablets was deconvoluted using monodisperse granules as well as milled granules of microcrystalline cellulose (MCC). A strong linear relationship (with negative slope) exists between the tablet tensile strength and granule solid fraction. In contrary to popular perception, granule size has no statistical impact on the tablet tensile strength. A smooth or rough fracture surface of tablets prepared from low or high solid fraction granules, respectively indicates differences in fracture of the tablets. ^ Subsequently, a novel method was developed to map the fracture surfaces of tablets. The proportion of intra-granular versus extra-granular fracture of tablets was quantified by image analysis. Low solid fraction granules deform extensively, neighboring granule surfaces intermingle closely, and produce homogeneous tablet matrices. At a high deformation potential (defined as, tablet solid fraction - initial solid fraction of the packed granule bed), tablets fracture indiscriminately both intra-granularly (fracture of individual granules) and extra-granularly (separation of neighboring granules). In contrast, at a low deformation potential, tablets preferentially fracture extra-granularly. The proportion of intra-granular fracture is a function of the deformation potential only. Tensile strength of tablets increase nearly linearly with the deformation potential and the slope of the linear relationship is larger for higher tablet solid fraction. At or below the critical deformation potential the tablet structure is not coherent and it does not fracture intragranularly. ^ Calibration of the Drucker Prager Cap (DPC) model parameters provides a means for a deeper understanding of the impact of dry granulation, granule SF, and granule composition (MCC/mannitol ratio) on the compaction properties of granules. MCC of any granulation status requires the same in-die compaction stress state for densification to a given tablet solid fraction. Only cohesion of materials and tensile strength of tablets are a strong linear function of MCC granule solid fraction. However, properties such as cohesion and diametrical tensile strength go through a maximum as the mannitol level increases in the binary granules, and clearly do not follow the linear mixing rule. Other properties either approximately follow the linear mixing rule (e.g., hydrostatic yield strength, young's modulus and Poisson’s ratio) where some interactions between the constituents are present, or not sensitive to the composition (e.g., internal angle of friction). In general, the properties of a multicomponent system may not be precisely estimated from the properties of individual components, simply by using the linear mixing rule.