Planning formulation experiments: screening

Experimental design techniques are essential tools in formulation. Virtually all methods are applicable at one time or another: choosing an "ingredient" from a set of products, fine-tuning the composition of a mixture or manufacturing conditions, optimizing a property or finding a compromise between several properties, and so on. However, despite a growing number of publications devoted to experimental design, many tools are still little-known or little-used.

Having decided on the nature of the various families of ingredients to be used in the final formula (antioxidant, hardener, binder, etc.), the experimenter often has a greater or lesser number of products to choose from for each of them. In this article, we will focus on their selection. We will distinguish two cases. In the first, which we'll call "independent factor screening", the experimenter wants to know which products in each family have the same behavior, which are the best performers, so as to select just one. At the same time, he can identify which of the factors characterizing the manufacturing process have a real effect on the properties of the final product. In the second case, which we will refer to as "mixed component screening", the experimenter considers the possibility of introducing into the formula a mixture of several products from the same family (a mixture of binders, for example), provided that each has a specific interest.

A second article of this section will be devoted to the fine-tuning of a formula, which corresponds to its optimization and is often the next step.

For the sake of pedagogy, we have deliberately chosen to illustrate the various techniques presented in these two articles using a single problem from the field of galenics: tablet manufacturing. Some of the experimental results are real, while others are simulated on the basis of real experiments, the essential aim of the examples being to show the implementation of experimental planning tools and not to determine real manufacturing conditions. The main reasons for this choice are as follows:

• keep the introduction of each new example to a minimum;

• show that the same problem can be approached and solved in many different ways, and that an apparently slight change in the statement can lead to a very different choice of strategy;

• introduce continuity between different possible stages of the same study. It is indeed exceptional for a formulation problem to be solved in a single step.

As mentioned in the introduction, this presentation consists of two articles:

[J 2 240]: Design of formulation experiments:...