Factorial Experimental Design Approach in Semicontinuous Emulsion Polymerization of Methyl Methacrylate to Study the Effect of Process Variables Sujata Mishra, 1 Jagbir Singh, 2 Veena Choudhary 1 1 Centre for Polymer Science and Engineering, Indian Institute of Technology, New Delhi 110016, India 2 Jubilant Organosys Limited, Noida, Uttar Pradesh 201301, India Received 20 November 2008; accepted 11 February 2009 DOI 10.1002/app.30237 Published online 14 May 2009 in Wiley InterScience (www.interscience.wiley.com). ABSTRACT: This article describes the effect of various process variables in the semicontinuous emulsion poly- merization of methyl methacrylate. A series of poly(methyl methacrylate) (PMMA) emulsions were prepared using ammonium persulphate as initiator in absence and pres- ence of Dowfax 2AI as surfactant. The effect of process variables such as initiator concentration, monomer concen- tration (solid content), surfactant concentration, reaction temperature, monomer feeding time, and holding time were systematically studied on monomer conversion, par- ticle size, gel content, and molecular weight using a two- level fractional factorial experimental method. Analysis of fractional factorial design revealed that surfactant concen- tration, monomer concentration, initiator concentration, and monomer feeding time affect all the properties. How- ever, the surfactant concentration and the interaction effect of initiator and monomer feeding time are the key varia- bles influencing the properties of PMMA latex. V VC 2009 Wiley Periodicals, Inc. J Appl Polym Sci 113: 3742–3749, 2009 Key words: emulsion polymerization; fractional factorial design; methyl methacrylate INTRODUCTION Emulsion polymerization is a complex heterogene- ous process in which lot of components such as monomers, surfactant, initiator, buffer, and chain transfer agents are present in the formulation. These components affect the rate of polymerization as well as the final properties of latex. In addition, the reac- tion variables such as temperature, reaction system design, and polymerization process, i.e., batch, semi- continuous, and continuous methods also affect the polymerization. In industry, the end-use applications mainly depend upon the final properties of the emulsion. Hence, a systematic study on emulsion polymerization to optimize the reaction variables, i.e., solid content, particle size, molecular weight, and its distribution that can produce polymer with predetermined properties is of greater industrial im- portance. So design of experiments to achieve these properties is very important. The statistical method in designing the experiments is very useful. In this case, poly(methyl methacrylate) (PMMA) la- tex was taken for study because it can be used as a seed material in the synthesis of a seeded emulsion polymerization of acrylic copolymers. To investigate the effect of seed properties on the final latex proper- ties, the synthesis of the seed latex is very important. A factorial experimental design was used to synthe- size PMMA latex by varying the reaction variables such as initiator, surfactant, monomer concentration, temperature, feeding time, and holding time. The variation of one factor at a time keeping the others at a constant level is a tedious process when a large number of factors have to be investigated. But statis- tically based experimental designs provide a more ef- ficient approach to deal with a large number of variables. 1 A two-level factorial design was used here, as the factorial design with more than three var- iables significantly increases the complexity of the experiment. Again a two-level full factorial design requires 2 N experiments where N factors have to be investigated. So the full factorial design of six varia- bles with two levels requires at least 64 experiments. The number of experiments can be reduced by using fractional factorial design without loss of information about the main effect. So the number of experiments reduced to 16 using a quarter factorial designs (1/4) with two levels and six variables. The analysis of responses leads us to identify their influences. Thus, the optimal values of these variables can be estab- lished to obtain the required response, diminishing the variability of the response, and the effects of the number of controlled variables. Journal of Applied Polymer Science, Vol. 113, 3742–3749 (2009) V VC 2009 Wiley Periodicals, Inc. Correspondence to: V. Choudhary (veenac@polymers.iitd. ernet.in).