Can Quantum Chemistry Predict, or Just Explain?
Within academic research, quantum chemistry has become a well-established tool and its application an increasingly expected standard analysis for organic chemistry publications focused on small molecules. Akin to providing the NMR or IR spectrum, quantum chemical analysis, especially using density functional theory, is now de rigour in the supporting information and often is used as a key element to the explanatory and phenomenological aspects of a scientific discussion and argument. Indeed, as most universities have adequate enough computing facilities and flexible minded students, along with a large choice of open-source software, why wouldn`t computational and quantum chemistry be applied? – If it is not there, there is something lacking.
However, much like other analytical methods, quantum is often applied post-facto, to guide understanding, but less often used to predict and guide new experiments de novo. This is partially diagnosed as a consequence of novice efforts in the field to use the tool of quantum but tethered, as a crutch, to verifiable experimental results. Post-facto analysis can always be more confidently carried out, as one can pick the methodology that agrees with the experiment. There is great utility to this approach, as the computations reveal and enliven the experimental results testifying to the strength and central utility of theory. Another reason quantum is applied post-facto is that the workflow does not jive with that of experimental research. Running a reaction is often quicker than computing it (one needs the full reaction path and all elementary steps, etc). In addition, performing quantum chemistry calculations is done at the computer terminal, at the same time the paper is beginning to be written and the rotovap and two-stage pump are still and silent.
However, is quantum chemistry at the point today of being a reliable predictive tool? If so, is this predictive tool qualitative, or quantitative? Other than accuracy, what barriers exists to its implementation at the front-end of an experimental endeavour?
Case Studies in Predictive Quantum Chemistry
ChemAlive asked the CCL (computational chemistry list) community to send references and stories about predictions that were later confirmed by experiment to answer the above questions to survey the state-of-the-art. We received 50 references in 1 hour and hundreds following. In addition, we asked a series of survey questions with the results found here:
With one hundred respondents the general summary is as follows. Quantum chemistry is widely viewed as superior to classical mechanics, especially for electronics but also in general. However, the quantitative nature of the results is often overkill considering the substantial time and CPU power cost. Most importantly, many industrial researchers had no access or no experience in quantum chemistry due to software and hardware barriers. In addition to the survey work a number of communications on the CCL thread revealed strong opinions that provide a sliver of here with redacted names and associations:
…we need to apply methods to sets of eg 20 to 1000 molecules. We are happy with reliable relative numbers, because we are often concerned with filtering out the molecules that will not work.
The biggest hurdle for QM in industry still is the zoo of methods hindering people with modelling and pharmaceutical background from even trying to use QM, softwares with non-intuitive input and chaotic output files (end of the 90s there were already attempts with chemical markup language to standardize outputs), stability, how to run multiple jobs etc.
I consider it as one of the greatest successes and values of
computational chemistry in industry to rule out the impossible and leave
a smaller set of potentially doable alternatives.
Working with a good experimental person is also essential because no one
believes (really) model predictions without at least comparative agreement
with “accepted” experiments.
Modeling is an enabling technology; it cannot automate processes nor will
it will replace or eliminate (all) experimentation. It is definitely not a
stand alone protocol for problem resolution (at least at the chemistry
industrial scale!). However, in conjunction with experimentation, it can
help prioritize experiments, improve experimental design, and predict
experimental participants & results, sometimes ones that are not expected
(by chemical engineers — but that’s a whole other story!).
Finally, something that shouldn’t be discounted (and has already been
mentioned in this thread), modeling constantly improves; the questions that
“can’t” be answered today most assuredly will be answerable within a few
years. And more cheaply too.