Using over 30 physical, chemical and computation properties of solvents a database of over 500 diverse solvents has been amassed. Where available, information pertaining to the safety, health, environmental impact and availability has been collected. Using Principal Component Analysis (PCA) the 500 solvents and 30 properties has been reduced to 3 or 4 principal components (principal properties) which are approximated to polarity, polarisability and H-bonding.
For any given reaction a suitable subset of solvents can be chosen to investigate the desired design space. The subset can be diverse including solvents across the whole map or restricted to, for instance, consideration of only polar solvents for a given reaction. The selection of solvents typically begins with 9 solvents, but 5 or 17 solvent selections are also available. The solvent is investigated in a Design of Experiments looking at all other appropriate factors simultaneously.
The first set of experiments will focus the selection of solvents into a region, typically ¼ or less of region of space from the original selection while also reducing the number of potential factors to consider in subsequent experimentation. The factors are eliminated as they have been shown not to be significant for the specific reaction. In general the subsequent selection of 4 to 12 additional solvents will provide a detailed understanding of the requirements the reaction has on the solvent and clearly identify the sweet spot for the reaction. This selection process will identify a suitable solvent, either directly from the screening results or from the identified 'optimum' sweet spot for the specific reaction.
The use of further data modelling can be undertaken to identify the specific solvent properties which are important for the reaction under investigation. Through the development of a model the suitability of all solvents in the database can be predicted, within the confines of the conditions investigated. The use of solvent properties enables the mapping of solvent space. The selection of a small representative set of solvents enables the rapid focus of effort to suitable solvents in the sweet spot for the reaction enabling informed decision making around solvent choice.
Paul Murray Catalysis Consulting will provide our clients with:
- The selection of solvents from our PCA maps to enable the efficient understanding of solvent effects and the requirements of the chemical reaction.
- The selection of appropriate designs, additional factors and ranges for their experimental investigation.
- The analysis and interpretation of the experimental data.
- Partial Least Squares (PLS) modelling to understand the properties of materials that play a significant role in optimum reaction development and prediction of suitability of additional materials for the chosen reaction.
- The prediction of suitable alternative solvents to meet the needs of the client.
For an example of using PCA for solvent selection in catalysis see the following case study: