Asymmetric epoxidation: a twinned laboratory and molecular modelling experiment

K. K. (Mimi) Hii, Henry S. Rzepa and Edward H. Smith

Department of Chemistry, Imperial College London, South Kensington Campus, Exhibition Road, London, SW7 2AZ, UK

Web-enhanced Object 1

The Electronic laboratory notebook (ELN)

We set out the following requirements for providing an environment for students to record their computational experiments

1. The tool had to be relatively simple, with a very short learning curve. We adopted the principle of markdown, whereby relatively simple formatting syntax and commands need to be learnt, and the result is human readable. A simple Wiki environment was judged ideal. The cheatsheet for the MediaWiki system (the essential commands to be learnt) is literally just one page long.
2. MediaWiki allows date-stamping for all entries, and a simple versioning system that allows all modifications, additions and their dates to be inspected.
3. MediaWiki provides an auto-numbering citation system.
4. The system is easily enhanced to provide a simple equation formatter if this is needed
5. Examples (templates) of the syntax to be used to construct tables of results are provided
6. The system accepts graphical files in SVG format (which can be produced using Gaussview for the spectral data). These are scalable, whereas the alternative bit-map screen dump of a spectrum is not so (unscalable images are unfortunately common in supporting information).
7. The addition of the JSmol extension to the Wiki¹ allows molecular coordinates to be uploaded and rendered as rotatable/reusable objects using the syntax <jmolFile text="Explanatory text for link">BCl3.log</jmolFile> (where e.g. BCl3.log is an output file from a Gaussian calculation). This feature is particularly useful for the graders of the reports, since it instantly exposes the student's stereochemistry for the models. JSmol is also conveniently compatible with modern Android and IOS tablets.
8. JSmol can also be used to incorporate rotatable molecular surfaces into the report. This might include e.g. orbitals used to illustrate reactivity (for example the anomeric effect), or an NCI surface analysis.^2,3
9. A MediaWiki template was used to provide simple DOI handling (including the DOIs the students obtain by digital publishing their data), taking the form e.g. {{DOI|10.1021/ed078p1266}}. The students use this not only to record the DOI for citing literature, but also the data-DOIs they have themselves created by deposition into a digital repository.⁴

We have operated such a MediaWiki based student ELN for five years now, passing some 600 students through the system. There have been very few failures of the system, and to our knowledge none of these students has failed to produce a report because difficulty of use. Students were asked to use the Wiki-ELN throughout their experiment and almost all took advantage of the possibility of including either the coordinates of their models directly within their report, or by citing the digital repository identifier in their citation list. Some were a little hesitant to be initially deprived of their favourite word processor for this task, but soon learnt that if they did not know how to achieve a particular effect using the Markdown Wiki syntax, a friend could probably advise. For most, the learning curve was surmounted in only a day or so. One particular advantage (to the student) of the use of a date-stamped ELN was that the experiment submission deadline was enforced not so much by the specific time that a document was submitted, but by the date-stamps present on the various entries. Only the URL of the Wiki page had to be notified to the graders, and even if this notification went missing or was itself late, full credit could still be given to any material with a date stamp predating the submission deadline. This certainly eased the student stress levels associated with rigid submission deadlines.

The use of such an authenticated and provenance date-stamped, data-rich online system for recording experiments parallels the increasing introduction of research data-management systems in industrial environments and helps create in the students a culture of preserving and curating scientific data and information.

Analysis of reaction transition states

The following transition states were located at the ω97xd/6-311G(d,p) level using a continuum solvent field corresponding to water using the Gaussian 09 program, revision D.01. A formatted log file for each transition state can be downloaded using the appropriate link to a digital repository and analysed for the free energy of that isomer.

Topological analysis of the transition state model for the Shi and Jacobsen transition states

The wavefunction each transition state can be obtained by downloading the WFN file stored at the digital repository. This can be used to obtain the electron density ρ(r) and from there two properties of ρ(r) are derived to help understand the student explore possible reasons why the reaction is stereospecific.

1. The QTAIM topology of the electron density ρ(r) is computed using the Avogadro program. This identifies regions of the electron density known as topological critical-points, in which the curvature of ρ(r) can adopt one of four types only. The most interesting of these types is the bond-critical-point (BCP). These can occur obviously in the region connecting two atoms that are formally bonded, but of more relevance are the additional BCPs that are often found in weakly interacting regions of a molecule, such as hydrogen bonds or weak dispersion zones. The value of the density and its first derivative at these BCPs can impart information about how attractive they might be. The Avogadro implementation of this analysis also offers the opportunity to locate lone pair centroids, although in fact this is quite expensive computationally. These could be of use in e.g. the anomeric analysis of the Shi catalyst.
2. Another manipulation of ρ(r) known as NCI (non-covalent-interactions)² filters the density ρ(r) to only those values associated with the weak interactions noted above, and displays an isosurface of a quantity known as the reduced density gradient (RDG). This is color coded on the basis of (λ₂)ρ(r) (the second eigenvalue of the density Hessian) to indicate either stabilisations (blue and green in the colour scheme) or destabilisations (yellow and red). This gives the student a rapid visual inspection of where such regions occur in the reaction transition state, and can be useful for providing information about how possibly to optimise the system.³ The students proceed by acquiring the computed transition state wavefunction in the form of a formatted Gaussian checkpoint file, and using a program such as Gaussview, they convert this to a Gaussian cube file of electron density. This then can be analysed for the NCI interactions noted above using an online procedure.⁴

3D printing

This rapidly developing technology allows 3D molecular coordinates (obtained from e.g. experimentally determined crystal structures or from computational experiments) to be converted to quantitatively accurate 3D full colour models. Our model library can be viewed at DOI: 10042/a3uxu along with details of how they can be manufactured. The initial material for such models was gypsum (calcium sulfate), but more recent printers can manufacture more robust and lighter models by cutting, gluing and color-printing A4 sheets of paper.

Student Feedback

After the course, we asked students for comments on what they felt the experiment had taught them.

• Unlike a safety-conscious experimental laboratory, the in-silico version can have a liberating effect on students. One student commented "The computational sandbox gives you a freedom to play with ideas, related not just to the exercise in hand, but to lectures, and other labs in general".
• Just like mis-behaving apparatus in an experimental laboratory, computer programs can give mystifyingly obscure error messages. The students "appreciated the skills one has to learn to track down such errors"
• "Rubbish-in, rubbish out" is an axiom often used for computer experiments. To avoid analysing what turns out to be rubbish, students noted that "it reminds you that reality checks in the form of effective searches of the literature are a useful skill"
• Students can fall into a mindset where they "are expected to get the right answer" (indeed that there is one "correct" answer for each problem they are set). In exploring for example the conformational possibilities of taxol, the students appreciated that the correct answer might not actually be known, and that they would have to deal with this.
• In the initial version of this experiment, where the students were presented with a small library of alkenes for them to epoxidise and also compute the outcomes, there are few unknowns. The expected results can be found in the literature, and hence the task may simply default to obtaining the "correct answer". Some students expressed a yearning for a more open experiment, where the library of alkenes was extended possibly to those for which no experimental outcomes have been reported in the literature. In fact, they were encouraged to suggest such new candidates. Whilst some were in fact rather impractical (due to expense), many interesting alkenes were the epoxidation outcome is unknown did emerge, and no doubt will be attempted in the future.
• If one sets out to find literature on specific compounds, it can be quite a revelation for students to discover the disparity in the reported values of properties such as optical rotation. It is often tempting for a student to accept the first value they track down, or indeed to quote selectively the value that agrees best with their calculation. It can take a little while for the light to dawn that actually literature values for optical rotations can be alarmingly variable, not just in the magnitude, but sometimes even the sign. Some students did appreciate that this was a reflection of the real world, and not just a reflection on perhaps their incomplete mastery of the computational techniques.
• The synergy that can be achieved by combining computational spectroscopy and experimental laboratory technique took a little while to become appreciated by students. On balance, in the sequential approach we adopted, those students who did the computational module before the experimental one felt they learnt more from the whole than those students who did it in the reverse order.
• It takes a little while for a student to realise that the experiment can be about making compromises. Very quickly, they learn to balance the advantages of a faster calculation with selective need for a more accurate but slower calculation. The cost- benefit decisions have to be learnt.

Other Supporting Information

1. Details of the computational experiment: PDF, 488K (DOI: 10042/a3uws) and MediaWiki Source (Text file, 48K to be used to create a Wiki page).
2. Details of the computational toolbox: PDF, 247K (DOI: 10042/a3v0o) and MediaWiki Source (Text file, 34K to be used to create a Wiki page).
3. Details of the electronic laboratory notebook: PDF, 369K (DOI: 10042/a3v1n) and MediaWiki Source (Text file, 28K to be used to create a Wiki page).

References

1. J. Prilusky, http://proteopedia.org/support/JSmolExtension/
2. E. R. Johnson, S. Keinan, P. Mori-Sánchez, J. Contreras-García, A. J. Cohen and W. Yang, "Revealing Noncovalent Interactions", J. Am. Chem. Soc., 2010, 132, 6498-6506. DOI: 10.1021/ja100936w; J. R. Lane, J. Contreras-Garcia, J. P. Piquemal, B. J. Miller and H. G. Kjaergaard, J. Chem. Theory. Comput., 2013, 9, 3263-3266, DOI: for details of the theoretical procedure.
3. A. Armstrong, R. A. Boto, P. Dingwall, J. Contreras-García, M. J. Harvey, N. Mason and H. S. Rzepa, "The Houk-List Transition states for organocatalytic mechanisms revisited", Chem. Sci., 2014, DOI: 10.1039/C3SC53416B
4. H. S. Rzepa, "Emancipate your data", Chemistry World, 2013, 23 September issue. URL http://www.rsc.org/chemistryworld/2013/09/open-repository-data-sharing-rzepa -figshare, DOI: 10042/a3uxk; Harvey, M. L.; Mason N. L.; and Rzepa, H. S.; Digital data repositories in chemistry and their integration with journals and electronic laboratory notebooks, J. Chem. Inf. Mod, 2014, ASAP, DOI: 10.1021/ci500302p
5. H. S. Rzepa, "Script for creating an NCI surface as a JVXL compressed file from a (Gaussian) cube of total electron density", Figshare, 2013. DOI: n5b