Excited State Proton Transfer in 2’-hydroxchalcone derivatives

by Michael Dommett, 4/01/2017

2016 ended on a high. Our paper “Excited State Proton Transfer in 2’-hydroxchalcone derivatives” was accepted in PCCP. It was a great piece of news to receive before the Christmas holidays and was met with both joy and relief. The process of writing the paper began in January 2016 and had many versions before we were happy enough to submit. From there, the review process was quite straightforward. Thankfully, we didn’t have to perform any more calculations and the most time consuming part was reformatting some figures and analysing the dynamics data slightly differently. We are delighted with the final version and excited for the next part of the project, as this marks only the beginning of an interesting story in understanding the photochemistry of these molecules.

In our group, we are interested in modelling the photochemistry of molecules which undergo some kind of ‘switching’ process when they absorb light, for use in photochromic materials. In October 2015, we were intrigued by photochromes based on 2’-hydroxychalones. When exposed to UV-light, these systems undergo excited state intramolecular proton transfer (ESIPT), a highly interesting and useful phenomenon which can be used in optoelectronics, imaging, and solid state lasers.

The experimental data suggest that the molecular structure (electron donating substituents) and the way the molecules aggregate in the solid state determine whether the crystal fluoresces or not. In a good solvent, the molecules do not fluoresce at all.  Little theoretical data exists for these compounds, and we decided to undertake a study to understand the photochemistry of these systems. In particular, we wanted to understand why some 2’-hydroxychalcones fluoresce, and others do not. For that, we needed to decouple the effect of the molecular substituent and the effect of the packing mode.

The first step of the study is to understand the photochemistry 2’-hydroxychalcones and the influence of different groups (methyl and methoxy) on the potential energy surface. This is the work we present in the paper.  We studied 5 compounds in gas phase, focussing on the important steps in the photochemical process: UV-absorption, proton transfer, and relaxation. We also used non-adiabatic dynamics to understand the rate of proton transfer and the examine the relaxation pathways after absorption.

As discussed in the paper, we uncover two relaxation channels in the first excited state (S₁) which cause the S₀ and S₁ states to converge.  The first pathway involves the expected proton transfer process followed by intramolecular rotation. The conical intersection (point of degeneracy between S₀ and S₁) is easily accessible, resulting in non-radiative relaxation to S₀ (i.e no fluorescence).  The second pathway results in a conical intersection being reached via intramolecular rotational but without the proton transfer step.

The identity and position of the substituent determines which of these modes is preferred. We find that the para position on the ring is sensitive to a strong electron donating group, as the conjugation changes the character of the S₁ state and favours the proton transfer path, a conclusion reached thanks to the non-adiabatic dynamics. We carefully analysed the potential energy surface with different levels of theory, in particular the region of the conical intersections. This is one of the most important aspects of the paper, as the easy-to-access conical intersections are the key in determining the fluorescence.

Whenever I read a paper in the literature, it appears that all of the pieces fell beautifully in to place for the authors and the final product reflects exactly how the work was carried out. I have now learned first-hand that this is (almost never) the case. In January 2015, I started writing the paper thinking that I already had most of the data. In fact, it took another six months of calculations to finally complete the picture. The most time-consuming (and frustrating) part was understanding the effect of method and basis set, as nothing seemed to match in the beginning. It was only through really understanding the strengths and limitations of each method did the data make sense. The characterisation of the conical intersection geometries was also quite problematic. I located at least 40 conical intersections (5 compounds, 2 methods, 2 basis sets, 2 related structures dependent on +/- rotation angle), with additional points calculated based on the dynamics data. This was really tricky and quite labour intensive.

The most enjoyable part was the running and analysis of the dynamics calculations. I used Jupyter notebooks with python/matplotlib to interpret the data, which allowed me to write scripts to analyse and plot on-the-fly.  It was a workflow which was really efficient, as a slight modification to the analysis codes could easily produce new figures, something I was thankful for upon returning to the data ~4 months later to respond to the reviewers’ questions.

Going forward, we are almost ready to start writing the paper for the second step of this study. We have lots of data which I have been collecting and analysing over the last 5 months or so. It should produce a really good paper once the final holes are filled in. Watch this space.