|Posted on January 4, 2014 at 6:40 PM|
Anyone working on research knows that very seldom one has the chance to start and finish a paper within a short period of time. Usually a research project extends over a long period, with some lapses of time when the project gets stuck or halt for this or that reason. Namely, projects in collaboration with people from different institutions tend to be slow running projects. The story I explain in this post completely feels in this category. The research started on November 2004 and it got published on September 2013.
It all started back in 2004, when I was a PhD student visiting for the first time the University of Szczecin (Poland). I went to Szczecin to work with Jerzy Cioslowski, who agreed on hosting me for three months. I had read some paper of Jerzy and I was enthusiastic to have the chance to work with him. So, I got quite disappointed when I realized that he would be on a short stay in Dresden for most of my time in Szczecin. At least, he arranged someone to look after me. He was Jacek Styszynski, a physicist working on relativistic effects on molecules. Jacek took good care of me and had the patience to teach me about (relativistic) four-components calculations. Under his guidance in no more than three months we completed a research project that we published in Chemical Physics. This is the counterexample of how much time a collaborative project takes. Afterwards, Jacek suggested to work on a more ambitious project, which pretended to produce benchmark results for several carbonyl complexes and explain some controversial results obtained in the literature. The work was very appealing from both the chemical and the computational viewpoints. By then, there were very few relativistic codes and none of them could perform geometry optimization. So we would have to implement the optimization and the frequency calculations of these species. In 2004 I started working on this project for a few weeks before Jerzy came back. Then, I switched to some research project about the homogeneous electron gas. Short after Jerzy returned I went back to Girona and basically left Jacek's work aside for a while.
In 2005 I visited again Szczecin. This time I worked on both projects full time, and we got some results on the project with Jacek. However, at the end of my stay, after careful check of our results we noticed a large deviation of some vibrational frequencies as compared to the experimental ones. This result made no sense and it was Jerzy who noticed that our scheme to calculate vibrational frequencies (taken from an old German manuscript) missed some important coupling effects. On Christmas 2005 I left Szczecin with the feeling we still had plenty of work to do.
On July 2006 I defended my thesis, and right after that I went to Szczecin for another four-month stay. This time the focus of my stay in Szczecin was completing with Jerzy some paper on aromaticity and preparing the postdoc project I would carry out in Szczecin. I also spent quite some time working on Jacek's project but I left Szczecin with the feeling that the metal carbonyl project needed some re-thinking and we were far from the end...
On 2006 I was not awarded two (Spanish and Catalan) postdoctorals to go to Szczecin, so I look for another postdoc possibility. That is how I ended up in Aarhus, working with Ove Christiansen. I had a great time in Denmark and learnt a lot about vibrational calculations, potential energy surfaces, coupled-cluster methods and even some QM/MM theory. If you want to learn these techniques, visiting Aarhus (and Ove) is a must. At the beginning of 2007 I was awarded the Marie Curie fellowship to work in Szczecin but I enjoyed so much my time in Aarhus that I postponed my start in Szczecin more than one year. (I wish it could have been more!) Nonetheless, I started in Szczecin full of enthusiasm, because I was giving the chance to start doing some research on model systems, and I needed this background to later work on theory development as I really wanted to do.
Altogether, I had abandoned my project with Jacek two years and I took it back ion Spring 2009. After my experience in Aarhus I knew how to deal with the vibrational problem of metal carbonyls. Since I was mostly focused on the research about harmonium it took one year to obtain the first optimization results from Dirac (which was now capable of geometry optimization) and realize that the optimizations were neither efficient nor reliable. This finding coincided with Pedro Salvador visit to Szczecin, and he got enrolled in the project. Using a modified code of his, we interfaced a modified version of Dirac to obtain more reliable geometry optimizations and harmonic frequencies. Some months later we obtained our first final results. So, at the end of my postdoctoral stay in Szczecin we had most of our geometries and frequencies for the lightest transition metals.
On 2011 I moved to the Basque Country, where I had basically no time to work on Jacek's project. On 2012 I moved back to Girona but, once again, it took me a few months to find the time to work on the metal carbonyl work. By then, Jacek had completed most calculations. I started writing the manuscript at the beginning of 2013 and generated the chemical bonding analysis of these complexes. On Spring we submitted our manuscript, which got finally accepted on Autumn. It got published before the end of 2013.
This project took longer than expected because it was never part of my main research subject but also because of the research itself. If you are familiar with computational chemistry you may wonder how a project of three-atom molecules can take so long. The reason is four-component calculations. Unlike (regular) two-component methods, there are very few basis sets (sometime none!) available for these complexes. We had to change the basis sets when better ones were available (Prof. Dyall kindly sent us these basis set in advance). On top of this, one should take into account that there was no gradients available in Dirac for four-components CCSD(T) and, therefore, the optimization and frequency calculations are done numerically with the aid of an external software. The nature of the potential energy surface of MCO molecules also complicated both the geometry optimization and the frequency calculations. Our results included forth-row transition metals without the use of pseupotentials (they are not available); these were by far the most expensive calculations. Hopefully, we proved that Sadley's X2C methods, which are much more less demanding than pure four-component methods, give good results for geometries, dissociation energies and CO stretching frequencies (although a few bending frequencies are wrongly estimated with this method).
I am happy that despite the difficulties we could complete the paper with Jacek and Pedro. The metal carbonyl project was a good excuse to keep in touch with Jacek, who was my neighbor in Szczecin for two years and he is now also a very good friend. Jadwiga and him made my time in Szczecin a great and fruitful experience. If we ever get enrolled in a research project together with him let us hope it will take far less time. My feeling is that I will have the opportunity to see it by myself :-)