Unraveling the pathways linking reactants and products in complex reaction networks is a perennial challenge for researchers, as we found while studying electrochemical CO₂ reduction (eCO₂R).

Electrocatalytic reduction of CO₂ is a promising pathway to the more sustainable production of chemicals and fuels and could, if successful, replace the petrochemical processes which are now used. However, eCO₂R has proven exceedingly difficult to control. Prototype reactors produce a mixture of C₁-C₄ hydrocarbon and oxygenate products; separation of the desired products from this mixture will be costly. To improve selectivity, we need to understand the formation pathways for the desired products so that we can create more selective catalysts. However, this is a tall order: just considering the C₂ and C₃ products, there are over 18 possible intermediate species and many different possible reaction pathways.

Nevertheless, we sought to shed some light on eCO₂R reaction network by operating a prototype reactor under industrially relevant conditions, generating, as expected, a mixture of different products. Crucial questions immediately arose.

• How can we accurately classify the many analytes?
• How to pick out the products and the corresponding major intermediates?

We turned to other complex reaction networks for clues, notably photosynthesis, which has evolved over billions of years to fix CO₂ and generate, very selectively, the molecules of life. The work of Melvin Calvin was particularly inspiring [1]. Calvin used an isotopically labeled tracer, ¹⁴C, to learn the order in which intermediates appeared in the cycle now named for him (and was awarded the Nobel Prize in 1961). We reasoned that we might be able to use isotopic labeling in an analogous way for eCO₂R.

However, we faced another problem. Electrochemical reactions are much faster than photosynthesis. Also, we did not want to use radioactive ¹⁴C. We then looked more deeply into the photosynthesis literature and learned about a very interesting phenomenon in plant primary metabolism. Photosynthesis fixes ¹²CO₂ at a slightly faster rate than ¹³CO₂  [2]. The ¹³C discrimination is quantified as δ¹³C, representing the change in the ratio of ¹³C to ¹²C (n.b. natural abundance of the stable ¹³C isotope is 1.1%). For an irreversible reaction, breaking or making a bond involving ¹³C is less preferred than ¹²C in climbing the reaction energy barriers, thus resulting in ¹³C-depleted products. Thus, triose-phosphates produced in chloroplasts are ¹³C depleted (negative values of δ¹³C) compared with atmospheric CO₂. We also learned that δ¹³C measurements have been used not only in the study of plant primary metabolism, but also in geosciences, climatology, and oceanography.

Turning to the system of interest to us, eCO₂R, we would expect and were excited to observe discrimination against ¹³C in the products (negative values of δ¹³C) as elementary steps in eCO₂R are irreversible. Moreover, as we expect the largest kinetic isotope effect at the C-C coupling steps, we were able to group products sharing a given pathway by their δ¹³C values. In simple terms, a product and its major intermediates with the same number of carbon atoms would be expected to have similar values of δ¹³C, allowing us to deduce key relationships in the chemical network.

Drawing inspiration from its use in atmospheric sampling [3], we applied operando proton-transfer-reaction time-of-flight mass spectrometry (PTR-TOF-MS) for real-time detection of intermediates and products as well as their ¹³C isotopologues during eCO₂R. We encountered a few hurdles in doing so. Surprisingly, the technique can be too(!) sensitive; we found we needed to dilute the product stream from our reactor to stay within the useful dynamic range of the instrument. Ultimately, we were able to establish unambiguous identification of dozens of C₁-C₄ species by unique m/z values by coupling the PTR-TOF-MS to gas chromatography (GC). Importantly, due to the extreme sensitivity of PTR-TOF-MS, we only need to use a natural abundance feed of CO₂ (1.1% ¹³C) rather than expensive ¹³C or ¹⁴C-labelled CO₂.

Our operando measurements were used for mechanism evaluation in two ways. First, by slowly scanning the potential to more negative values, we can precisely determine the onset potential for volatile minor and major products. Second, we tracked δ¹³C through the chemical network. For all products, we observed a negative δ¹³C, which is also the case for the CO₂ fixation reactions in photosynthesis and Fischer-Tropsch synthesis. An unexpected finding is that the values of δ¹³C in eCO₂R are much more negative. Using these two approaches to analyze the operando results of different GDEs under different electrochemical conditions, we find that formaldehyde and acetaldehyde are not the major intermediates for the formation of methanol and ethanol/ethylene, respectively, and that propionaldehyde reduction is on the major pathway for 1-propanol formation.

Looking forward, we envision that use of δ¹³C measurements can be used to determine the intermediate pathways from reactants to products in other complex reaction networks involving carbon. It may also be possible to extend our method to a wider range of elements, as well as a wider range of fields.

So, when you are baffled by the heaps of products and intermediates and the many possible reaction pathways, how about considering δ¹³C?

For more details, please read our recent publication in Nature Catalysis:

https://www.nature.com/articles/s41929-022-00891-3 

Thanks to all the co-authors, and in particular Prof. Joel Ager, for their enormous contribution to this work.

References

1. Calvin, M. Forty Years of Photosynthesis and Related Activities. Res. 1989, 21, 3–16.
2. Tcherkez, G.; Mahé, A.; Hodges, M. ¹²C/¹³C Fractionations in Plant Primary Metabolism. Trends Plant Sci. 2011, 16, 499–506.
3. Yuan, B.; Koss, A. R.; Warneke, C.; Coggon, M.; Sekimoto, K.; de Gouw, J. A. Proton-Transfer-Reaction Mass Spectrometry: Applications in Atmospheric Sciences. Rev. 2017, 117, 13187–13229.