(560it) DESI-MS Based High Throughput Reaction Screening to Guide Bulk/Microfluidic Chemistry: A Correlation Analysis between Droplet and Bulk/Microfluidic Reaction Systems | AIChE

(560it) DESI-MS Based High Throughput Reaction Screening to Guide Bulk/Microfluidic Chemistry: A Correlation Analysis between Droplet and Bulk/Microfluidic Reaction Systems


Szilagyi, B. - Presenter, Purdue University
Jaman, Z., Purdue University
Logsdon, D., Purdue University
Ewan, H. S., Purdue University
Ferreira, C. E., Purdue University
Sobreira, T. J., Purdue University
Thompson, D. H., Purdue University
Cooks, R. G., Purdue University
Nagy, Z. K., Purdue University
High throughput screening (HTS) technologies are intensively used in various field of chemistry, from reaction pathway optimization to molecule discovery. Desorption electron ionization mass spectrometry (DESI-MS) is a recent technology that enables to carry out reactions in volumes as small as 50 nL with throughputs up to ~3000 reactions/h [1]. The essence of DESI-MS based HTS is that small volumes of reaction mixtures are spotted by a pipetting robot into high density (up to 6144 spots/plate) porous polytetrafluoroethylene (PTEF) based reaction plate, where each spot can represent and individual reaction. Then, the plate is analysed by the DESI-MS, which scans the whole ETFE plate surface by spraying it with a spray-solvent. The secondary droplets generated by the sprayer desorbs the reaction mixture from the plate, and these droplets are fed into the MS. Hence, the whole mass spectrum of each spot is collected, which enables to analyse the reaction mixture. The time the reagents spends in the droplet is in the order of milliseconds, in which the reactions happens. It is hypothesised that the reactions in the droplets are accelerated for the much reduced dimensions of droplets and their greatly increased surface areas [2] often by 104-106 times. This droplet acceleration enables to reach the throughputs of ~3000 reactions/h. However, the same droplet acceleration mechanisms raises significant questions when one tries to use the HTS data to guide batch/microfluidic reactions [3]: how the DESI-MS outcomes are expected to translate to larger volume scales, where there are no droplet effects anymore? This important question brings us to the topic of this work.

In order to analyse the correlation between the reactions observed in DESI and bulk, comparative studies were carried out for several representative reaction classes. For simplicity, in the abstract we present the correlation data for nucleophilic aromatic substitution reactions (SNAr). SNAr reactions are a versatile transformation in the modern organic chemistry arsenal and an important reaction class used for synthesizing pharmaceutically and biologically relevant molecules. The SNAr reaction mechanism have been extensively investigated. It involves stepwise addition–elimination process wherein the first step involves nucleophilic attacks of the substrate to provide a Meisenheimer complex followed by the loss of the leaving group. The reaction typically involves an amine as the nucleophile. In this work, SNAr reaction were carried out using amines and aryl halides in presence of bases for both DESI and bulk microtiter formats. The reactions used in the current analysis are published in the literature [3], the statistical interpretation of agreement between the DESI and bulk microliter reactions is in the focus of the current work.

Two SNAr datasets are available: the first set (SNAr1) involves 12 aryl halides, 8 amines, 2 reaction solvents and 4 basic conditions (3 bases and a ‘no base’ condition), which were carried out in DESI and in bulk microliter format at 150 °C with 15 h reaction time. The second dataset (SNAr2) involves the same 11 aryl halides except one as SNAr1, but 8 different bases, with one reaction solvent (from SNAr1) and the same 4 basic conditions. In contrast with SNAr1, the second dataset contains bulk reactions at 150 °C and 200 °C with 1,4 and 15 h reaction time. Hence, SNAr1 covers broader space of reaction conditions, whereas SNAr2 has richer bulk information content for the individual reactions. A simple yet effective comparison is provided by correlation plots, which is a representation of the product intensity of a given reaction recorded in the bulk as a function of product intensity in DESI. The intensity threshold, which is a critical intensity that has to be reached in order to count it as positive reaction – typically used in the DESI-MS based HTS technology - splits the correlation plot into four important quadrants. Q2 and Q3 are the regions of good agrement between DESI and bulk. The reactions in Q1 are positives in bulk and negative in DESI: in the context of DESI guided bulk reaction desing these reactions would have been missed. Q4 points are positives in DESI but negatives in bulk, which are misguiding, since these positive outcomes does not translates as potitives in bulk reaction.

In the SNAR1 the correlation in the yes-no information space between the two reactor systems is greater than 70 %. Although, 22.5 % of points were false negatives (worked in bulk but not in DESI), and 3.6 % were false negatives. Depending on the nature of reactions and the applied bulk reaction conditions, the bulk outcome can often be both “yes” and “no”. A correlation plot that compares the DESI outcome with the best bulk outcome based on the SNAr2 data revealed that there are practically no false positives. Hence, in this case, the DESI does not misguide, and the true question to answer is: how to find the best reaction conditions for bulk and microfluidics (reaction time and temperature)? These questions can be addressed by the existing reaction engineering techniques, and therefore are out of objective of the current analysis. There are several surprising conclusions that can be drawn from the rich bulk information content of SNAr2 reactions which will be discussed in details in the presentation. These are related to the thermal degradation as well as the importance of kinetic and thermodynamic control in the two reactor systems that can lead to either good agreement of discrepancy between the two platforms.


[1] M. Wleklinski, B.P. Loren, C. Ferreira, Z. Jaman, L. Avramova, T.J.P. Sobreira, D.H. Thompson, R.G. Cooks, High Throughput Reaction Screening Using Desorption Electrospray Ionization Mass Spectrometry, Chem. Sci. 0 (2018) 1–7. doi:10.1039/C7SC04606E.

[2] X. Yan, R.M. Bain, R.G. Cooks, Organic Reactions in Microdroplets: Reaction Acceleration Revealed by Mass Spectrometry, Angew. Chemie - Int. Ed. 55 (2016) 12960–12972. doi:10.1002/anie.201602270.

[3] M. Wleklinski, C.E. Falcone, B.P. Loren, Z. Jaman, K. Iyer, H.S. Ewan, S.H. Hyun, D.H. Thompson, R.G. Cooks, Can Accelerated Reactions in Droplets Guide Chemistry at Scale?, European J. Org. Chem. 2016 (2016) 5480–5484. doi:10.1002/ejoc.201601270.

[4] Z. Jaman, D. L. Logsdon, B. Szilágyi, T. J. P. Sobreira, D. Aremu, L. Avramova, R. G. Cooks, D. H. Thompson, High Throughput Experimentation and Continuous Flow Validation of Nucleophilic Aromatic Substitution Reactions, manuscript is in preparation