STATISTICAL APPROACH TO COMMON VARIABLES IN EMULSION PREPARATION By IL R. BENSON, W. C. G•,•FFIN, and H. M. TP, u^x* Presented May $, 1967, New York City IT HaS been long known that the type and stability of an emulsion formed from a given oil-emulsifier combination are related in part to method of preparation. Techniques which are more or less generally applicable to the formation of stable o/w and w/o emulsions are described in treatises on emulsion technology (1, 5). Thus it is frequent practice to add the oil to a solution of the emulsifier in water in order to obtain a stable o/w emulsion. Similarly a usual procedure for preparing a w/o emulsion is to add water to an oil solution of the emulsifier. In the inver- sion technique an initially formed w/o emulsion is inverted to an o/w emul- sion by the addition of further quantities of water. Except in the instance of inversion techniques, vigorous dispersion, such as that obtained with a colloid mill, usually generates finer particles and therefore more stable emulsions than does propeller stirring. It is usually considered that the greater the amount of proper emulsifier the more stable will be the resulting emulsion. Relative quantities of internal and external phases also con- tribute to stability although there are exceptions, the internal phase in stable emulsions usually does not exceed 70 per cent. For different oils and emulsifiers, however, the validity of these considerations, which are part of the art and theory of emulsion technology, is not always certain, nor is the relative importance of these preparational variables capable of direct deduction. The investigation reported herein was undertaken to deter- mine the importance of these common preparative variables on the type and stability of emulsions of water and different nonaqueous liquids ,vith a series of surfactants. In the present study, the importance of six preparative variables has been investigated by means of a statistically designed experiment. In this type of experiment, several factors are varied simultaneously according to a systematic plan, thereby enabling screening of the importance of each variable with an appreciably smaller number of experiments than would be required if each factor were varied one at a time. A fractional, two-level * Atlas Chemical Industries, Inc., Wilmington 99, Del. 437

438 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS factorial design (4) was used to screen the following factors at the indicated levels: Levels Order of addition o --• w w --• o Emulsifier location w o Emulsifier concentration, % 1 2.5 Proportion of water, % 90 50 30 Temperature, øC. 25 70 Agitation Propeller Homo-Mixer* * Trademark of Eppenbach, Inc. Each preparation factor was studied at two values or procedural conditions except the proportion of water, which was investigated at three levels. Although for each emulsifier-oil pair there are 96 combinations, only twelve separate experiments in this design are required to screen the importance of the six factors. The twelve methods are indicated in Table 1. TABLE 1--ExPERIMENTAL DESIGN OF EMULSION PREPARATIVE VARIABLES STUDY Method Order of Emulsifier Emulsifier Water, Temp., Agitation Number Addition Location Conc., % % øC. Method 1 o-w w 1 90 25 Propeller 2 w-o w 2.5 90 70 Propeller 3 o-w o 1 30 70 Propeller 4 w-o o 2.5 30 25 Propeller 5 o-w o 1 50 70 Propeller 6 w-o w 1 30 25 Homo-Mixer 7 o-w w 2.5 30 70 Homo-Mixer 8 w-o o 2.5 50 25 Propeller 9 o-w w 2.5 50 70 Homo-Mixer 10 w-o w 1 50 25 Homo-Mixer 11 w-o o 1 90 70 Homo-Mixer 12 o-w o 2.5 90 25 Homo-Mixer In the experimental design, as shown in Table 1, the experiments may be divided into three groups of four methods each according to the proportion of water used. Thus methods 1, 2, 11 and 12 utilize 90 per cent water, methods 5, 8, 9 and 10 utilize 50 per cent water and methods 3, 4, 6 and 7 employ 30 per cent water. Comparisons for each of the variables are made with two of these groups of four experiments at a time. For convenience, these combinations of groups are designated in this article by the quantity of water in the groups under comparison as 90-30 or 90-50. The nature of the present experimental design is such that valid comparisons may be made within the 90-30 groups and within the 90-50 groups. It was expected that the preparative variables studied might have differ- ent degrees of importance for different oils and surfactants. The chemical and physical nature of the nonaqueous liquids and the emulsifiers used in this investigation constitute additional, nonstatistical variation. The