Science-Based Formulation

Who could be against S-BF? Or if you are for it, what are the best approaches? Here are my current thoughts on S-BF, DoE and HT.

Feel free to disagree!

Don't we all do S-BF?

The answer is, sadly, an overwhelming No! The most usual way to formulate is "How we've always done it". This is especially the case when there is a wise, experienced formulator who has the magic hands and who is the one everyone goes to for advice. These people are very often wrong, especially if they have become comfortable with their position and knowledge. They can actively resist new ways of thinking because this threatens their position.

Or, to remove personalities from the story; what happens when your company's XY95 product needs to be changed in order to accommodate a change in raw materials (e.g. greener) or to head off a competitive threat? The common answer is that no one has any good idea of what each of the (say) 10 components in the formulation are doing. So they end up adding an 11th component - which means that at the next cycle there will be 12. Those who've had the courage to re-create their XY95 from scratch have often found that 10 ingredients can be replaced by 5 to give a cheaper, greener, better product.

How should we do S-BF?

A rather modest number of core scientific principles underlie a large number of formulation issues. So before formulating, work out what the principles are and find whatever tools can help you formulate from first principles rather than rules of thumb. Here is my personal list, based on the work around the 160+ apps on my site:

  1. Solubility, compatibility, dispersibility need a versatile and robust practical theory behind them. Hansen Solubility Parameters do a good job. Kirkwood-Buff, though still little-known, will increasingly be used for more complex solubility issues.
  2. Surfactant-based formulations need a robust tool for guiding us through the complexities of oils, salts, temperature and surfactants. HLD-NAC is the best-available tool for this.
  3. For adhesion, the mantra "Adhesion is a Property of the System" is a great start. The ideas of entanglement and dissipation are (literally) 1000x more important than surface energy.
  4. Rheology is under-rated and under-used. The links to ideas of entanglement are powerful, and there is a common theme about rheology dependence on the fraction of particles or emulsion drops.
  5. Surface energy is of very little importance while subtle effects at the interface are of huge importance. The focus on contact angles has meant we have not used the right tools for understanding the interface.

Don't go with the principles that are conveniently to hand in your organisation; many of these will be mythology or bad science.

What about Design of Experiments?

I've been involved in, read about, studied and analysed a large number of DoE work programs and the vast majority have been a waste of precious resource. This view has been confirmed by numerous colleagues from a wide variety of industries.

The relatively small percentage of DoE that have been done in the right way at the right stage of the process have delivered all the good things that DoE promise. So I am not against DoE but against them being used at the wrong stage of the process. I've seen lab staff who've spent a couple of days carrying out DoE work when their very first test tube should have told them that something was wrong. If we think of three phases of research, DoE should usually figure in only one of them

  1. Scouting Stage: definitely not
  2. Development Stage: probably not
  3. Refinement Stage: yes!

I'm equally against them being used in the context of "the whole thing is much too complex so we'll at least try some plausible DoE things". Let's take printing as an example. It's complicated! So let's vary speed, viscosity, pressure and temperature. Wow, look, some effects are big and some are small. But we have made zero advance in our understanding and almost invariably cannot apply the correlations to any other situation. If the DoE had been done against some scientific hypotheses (rather than vague ideas like "viscosity will be important"), then they might have been more insightful, but it's probable that some scientific tests would have been more suitable.

High throughput

As someone who rapidly gets bored, and then careless, in a lab, I fully appreciate how wonderful it can be when a robot does, accurately, in a few hours what I would have done unreliably in a few days. But they have to do experiments that are worth doing. It's like DoE. 100 robotic experiments which, from observation of the first, are all going to fail, are not a good idea.

So HT is a way to do lots of carefully thought-out experiments against some clear hypotheses, and not some form of magic where quantity can overcome quality. Formulation space is mostly barren, and getting a robot to wander through a barren space does no good to anyone, even if the robot doesn't complain.

Science-Based HT is a preferred methodology where a few key "good enough" scientific criteria can be used to massively reduce the amount of tests to be carried out - in other words to reduce the barren space that is such a waste of precious resource.