Electrophilic Aromatic Substitution Reactions Revisited: The Development of Statistically Designed Experiments as a Learning Tool in the Organic Chemistry Laboratory
Jason Mann, Advisor Dr. G. Lutz
Typical undergraduate students receive little exposure to statistically designed experiments, though such experiments are used extensively in research laboratories. Thus, this research had two purposes: the development of an experimental design exercise, and the determination of the effects of several variables on the electrophilic aromatic bromination (EAB) of phenol.
Discussion resulted in the identification of possible variables that would likely influence the outcome of electrophilic aromatic bromination. The final choice of variables were: reactant concentration, mole ratio of bromine to phenol, temperature, and water content of the solvent. The selected response variables were the amounts of each product produced and the amount of phenol remaining.
Calculations and Logistics:
Calculations were carried out to determine the amount of each substance that would be used to conduct sample reactions at three levels for each variable: high, low, and the average of high and low. Determining the amounts of reactants and solvent to be used ensured that performing multiple iterations of the reaction would remain manageable.
Test reactions were performed to confirm the viability of our variables, and to produce samples of the products for Nuclear Magnetic Resonance (NMR) analysis. By varying the mole ratio of bromine to phenol we were able to selectively produce products with differing levels of substitution. After identifying the products using NMR, samples containing known products were analyzed by gas chromatography (GC) to enable their identification later in samples containing multiple products.
Design Expert 7 experimental design software was used to create a statistically designed experiment. The resulting design was used as a guide to carry out reactions, in a random order, while simultaneously changing all four variables.
Data on reaction time was collected by visually monitoring the reactions, while data on the final reaction composition were collected via GC analysis of the crude reaction mixture.
Analysis of the data was carried out using Design Expert 7, which produced statistics and response surfaces.
It was found that the concentration of reactants had no significant affect on any response variable. However, the composition of the solvent had a significant affect on all response variables, while mole ratio of bromine to phenol had a significant affect on all variables except the amount of 2,4-dibromophenol produced. Reaction time alone was significantly influenced by temperature. The laboratory exercise designed through this research corresponds to the methods section of this poster.
Given the exercise outline derived from this research, and some instruction with design software, a student of chemistry should be able to create and implement a statistically designed experiment. This would result in experience with experimental design and greater confidence and independence in ordinary laboratory settings. This confidence would be the result of the trial and error nature of unscripted experimentation, which would show the student that careful application of their chemical knowledge in considering what causes an experiment to "fail" often leads not only to an experiment that will "work", but a better understanding of the knowledge used to solve the problem.
A useful function of statistically designed experiments, such as this is the optimization of some aspect of a reaction's outcome. For example, using the model to predict conditions for obtaining the maximum amount of a specific product, or reducing byproducts.
The effects of water on the outcome of the EAB of phenol could be the result of phenol's acidity and the formation of phenoxide ion, the solvent's stabilization of the transition state, or the rate of abstraction of a proton form the sigma complex. Further experiments using different solvent systems and different substrates should provide further insight into this effect.