Design and optimisation in organic synthesis:Optimisation of telscope reactions and methodologies base on orthogonal experiments

Telescope reactions: Assume that a three-step reaction is to be optimised: Lay out a statistical design to map the first step. Run the experiments an determine the responses. The response matrix Y=[y1 y2 ....ur] summarises the results. A principal compone nt model will determine the number of independent events yielding Y, i.e. the number of independent reactions yielding the desired product and byproducts. By fitting regression models to the PCA score vectors it is possible to identify those variables th at are necessary to control for favouring the desired reaction, and which variable to control to suppress the parasite reactions. The important variables from the first reaction step are now added as extra design variables for the next step. The outcome o f the first steps is the input for the second step, By perturbing teh variables of the first step we introduce variation of the input of the second step. The evaluation as depicted above will now reveal the important variable of the first two steps. The p rocedure in then iterated for the subsequent step. This will give quantitative models for multi-step reactions. Orthogonal experiments: With new synthetic reaction it is important to determine as early as possible if the reaction can be developed into a new synthetic method. A new method for designing such explorative experiments is developed. The feature is that the search space defined by variable settings, interactions between variables can be explored by peeling of one dimension at a time of the sear ch space. The model space is factored by singular value decomposition (SVD) and the experiments that are most parallel to the singular vectors are selected. This yields a A-Optimal design. These principles are being tested with new reactions of synthons f rom small brominated unsaturated ketones.