As a computational materials chemist studying photocatalysis, how close to realistic systems can your modeling get?

I have a sort of high-level question about the theoretical modeling of photocatalysis. Most of the computational work on photocatalysis involves DFT calculations looking at band gaps, free energy profiles, and reaction pathways. I have also seen a handful of papers that use reactive MD to probe how the surrounding water molecules get affected by adsorbed species.

As a computational chemist or materials engineer, and you want to take your work closer to a real system, what is the logical next step? Would it be device modeling, where you take inputs from DFT (such as overpotentials), and then try to calculate the I-V curves, for instance? Are there any other multi-scale modeling techniques or coarse-grain methods that can be used?

I guess this question can be extended more generally to different classes of applications, such as $\ce{CO2}$ sensors, other electrochemical reactions, and so on. I'm not really sure what the next step would be to go beyond DFT in order to model more macroscopic length and time scales.

  • 1
    $\begingroup$ At first sight the question appears overly broad for me. $\endgroup$ – Wildcat Aug 2 '15 at 18:37
  • $\begingroup$ @Wildcat: I understand, but I'm not really sure about how to make it more specific. I have edited my question a bit though. $\endgroup$ – user34801 Aug 2 '15 at 20:05

Look, I try to give you a broad answer. If you want band gaps, you need electronic structure that means DFT!

If you do MD that means classical mechanics means no electronic structure no band gaps.

You want to have solvent effects for your, I presume large system, then DFT is costly, you would be left with QM/MM or sometimes called double resolution.

Reactive forcefields such as Reaxff use some valence bond theory to treat electronic structure which is low level of theory and can not be extended to inorganic compound unless it's parametrized.

Finally all these methods are highly approximate, but for engineering purposes everything is approximate.

| improve this answer | |

Not the answer you're looking for? Browse other questions tagged or ask your own question.