Knowing why something works can be quite a complicated affair. For chemists, that's an everyday challenge that we embrace as part of research.
For some reactions, we could draw curly arrows and think about molecular orbitals, which tells us how a reaction might proceed and how we could modify that process.
This might look a little alien to some, but it allows chemists to show how a reaction works in a concise way that also bypasses language barriers. However, it only really works for reactions in organic chemistry and part of inorganic. Take, for example, an alkali metal like lithium when added to water.
Curly arrows can’t really describe this reaction, a class of reactions called redox, but there’s a different system for that: half-equations. With this, we describe each species in the reaction as undergoing its own reaction and then combine them to get the overall. It's not quite the same, but it works.
So at this point, we’ve got down the route in which a reaction proceeds. Now we need some way of describing the rate at which these reactions happen. Not to worry, we’ve got a whole branch of chemistry called kinetics for that. Provided it’s possible to measure the rate at which a reaction goes under a few different conditions, the necessary values we need can be calculated.
I’ve got curly arrows, half-equations, rate kinetics and every other way of describing a chemical process. I’m set, surely? I would be, if the chemical models I’m using were robust enough to work under every conceivable circumstance and account for every single factor. Perhaps unsurprisingly, they don’t.
Chemists love models. We love coming up with ways to describe a chemical system, be that curly arrows of the organic chemists to the more complex and abstract wavefunction theories that the quantum chemists adore. They work incredibly well for the majority of chemistry I could think of, but it doesn't mean that they're foolproof. These models don’t, and often can’t, explain everything about a reaction, always containing some assumptions which don’t always hold up. Even if they do, the results out of them aren’t always meaningful or useful to us. If a model doesn't work, it's tempting to throw it out and make a new one afresh.
The trouble is, models can get complex quickly. That’s the reason we have these assumptions to begin with: to reduce the complexity. It’s frustrating when a model doesn’t work, your experiment fails and you’re taken back to the drawing board because a critical assumption didn’t quite hold true. It’s times like that where your only option is to look back at what you’ve done and ask the big question.
Why didn’t it work?
Credit to Nuri Morad and Otilia Rose-Marie Meden for inspiring this post from a discussion we had during the programme. My thanks to my supervisors, Dr Renald Schaub and Dr Iain Smellie, for their support during the project, even when the models did fail.
Thank you to Lord Laidlaw and the Laidlaw Foundation for supporting my work.