Products of the chemical industry, such as plastics, textiles, fertilisers, pharmaceuticals, and others, are ubiquitous and greatly impact the quality of human life. While the demand for these products increases every year, so does the consumption of fossil feedstocks that sit at the start of the vast majority of all chemical value chains. With the current trends in decarbonising transportation and energy production, the chemical industry is set to be the principal consumer of fossil fuels1. It has become clear that to avert catastrophic impacts of human consumption on the planet, a significant shift of this industrial sector away from unsustainable feedstocks to a circular carbon economy and renewable energy is a pressing necessity2-4. These goals have also been formulated in the Paris Agreement7, the fulfilment of which will be crucial in order to offset climate change. Using CO2 to construct the multitude of carbon skeletons in diverse sets of molecules is a challenging imperative. For example, life cycle assessment studies on synthetic polymers revealed that besides efficient recycling, both biomass and captured CO2 have to be implemented as feedstocks, in order to achieve a sustainable economy for plastics within the planetary boundaries2. For introducing renewable carbon into chemical production chains, redesigning the complexly interconnected plethora of processes from scratch, and utilising CO2 in various chemical processes, is problematic8. Instead, rerouting access points to established productions from fossil resources to sustainable platforms would be preferable.
Green methanol, available from biomass or captured CO2, is materialising as a sustainable energy carrier. George Olah first proposed methanol as carbon neutral fuel system (Fig 1A), as it is easier to handle in storage and transportation than gaseous hydrogen5. Currently, considerable investments are driving the strong expansion of green methanol production across EU countries. Estimations predict an annual production of 135 Mt bio-methanol from biomass, and 250 Mt e-methanol from captured CO2 by 20506. The advent of the first cargo ship running purely on green methanol in 2023, both bio-methanol and e-methanol9, signals the acceleration of the energy transitions. This development is further supported by the growing implementation of power-to-X concepts using excess CO2-neutral energy to produce green hydrogen and methanol10. For carbon-neutral chemical synthesis, the increasing availability of green methanol becomes an ideal entry point to rethink value chains for the generation of sustainable chemicals. As methanol is already integrated into several large-scale chemical processes11-14, extending its utilisation to access other commodity chemicals is tangible15. For example, the implementation of methanol-to-olefin (MTO) processes on an industrial scale for accessing important alkenes, demonstrates the worth of this principle on a commercial level12. Reforming methanol into CO and H2 (syngas) is challenging16, but could be another central transformation for utilising it as a C1 building block17-19. Recently, Leitner et al. reported the first homogenous catalysed system for this valuable reaction20 (Fig. 1B).
Hydroformylations are highly relevant for accessing a variety of products from bulk to fine chemicals, including plasticisers, vitamins, fragrances, and flavours, as well as many organic building blocks. It is a central transformation as it introduces molecular complexity into simple hydrocarbon feedstocks21-24. Currently, olefins with syngas (H2/CO mixture) are reacted with molecular rhodium or cobalt catalysts to oxo-products with an annual production of >12 million metric tons21-23 As of now, the carbon skeleton of industrial oxo-products derives entirely from virgin grade fossil resources (Fig. 1C). To minimise them, Huber et al. recently reported the hydroformylation of pyrolysis oils25. This elegant approach could serve to upcycle low-value polyolefin waste to valuable aldehydes.
We recognised that the conversion of methanol to syngas could be interlocked with hydroformylations of terminal olefins via a sequential dual catalysis. Here in, we identify the success criteria for matching the kinetics and selectivity of both catalytic cycles, allowing access to various aldehydes in up to excellent selectivity and yields. The protocol for hydroformylations relies on a two-reactor setup, separating the catalytic reactions within a closed system. Syngas is released by a ruthenium-catalysed acceptorless dehydrogenation20 of stoichiometric amounts of methanol with respect to the alkene, while a rhodium-based catalyst consumes syngas yielding aldehydes from alkenes in the adjacent reactor. While not mimicking the conditions of industrial processes, this work represents a blueprint study on how redesigning such processes can be approached and achieved. Furthermore, by combining this concept with MTO processes using bio- or e-methanol for the synthesis of olefins, or bioethanol to ethylene processes, which also have also been implemented at scale26, oxo-products could be generated solely from CO2 as chemical feedstock. Preliminary investigations for developing the dual catalysis focused on exploring the efficiency and selectivity of the low-pressure hydroformylation regarding the H2/CO ratio. Experiments were conducted using estragole (1a) as a benchmark substrate (Fig. 1D). As allylbenzenes present challenging linear-to-branched selectivity for oxo product formation and the isomerisation to the corresponding β-methyl styrene is preferred, we considered it a suitable model olefin for selectivity-focused studies. Rh(acac)(CO)2 and 6-diphenylphosphanyl-1H-pyridine-2-one (6-DPPon) were chosen as catalytic system suitable for low-pressure hydroformylations27,28. In order to precisely control the ratio of gases, stochiometric gas releases for each CO29 and H230 were conducted in a three-reactor system. While an overshot of hydrogen resulted in alkene hydrogenation and, to a minor degree, favoured olefin isomerisation, the hydroformylation was found to be most efficient at a 2:1 H2 to CO ratio. At increasing CO concentrations, decreasing aldehyde yields were obtained, possibly due to inhibition of the catalyst by CO. Based on this, methanol should be an ideal syngas source for the tested catalytic system. However, the dependency on gas mixture also highlights the importance of matching the kinetics and selectivity of gas release to the hydroformylation, as hydrogen and carbon monoxide must be liberated close to simultaneously in the correct ratio.
The experimental investigations were initiated by attempting to match the Ru-MACHO catalysed methanol dehydrogenation with the Rh(acac)(CO)2/6-DPPon hydroformylation system on estragole 1a (Fig. 2A). In the original report, Ru-MACHO in neat methanol was heated to 150 °C in an autoclave in order to form H2/CO at the stoichiometric limit of 2:1 ratio20. However, the hydroformylation proceeds in THF at room temperature27,28. As the two methods require vastly different reaction temperatures, a two-reactor setup with a reflux condenser integrated into the syngas-releasing reactor was designed (see supplementary material).
Performing the syngas formation in neat methanol (10 equivalents with respect to olefin 1a), as reported by Leitner et al.20, the dual catalysis resulted in only 28% yield of the aldehyde after 16 h. Meanwhile, 32% of estragole isomerised to the internal olefin, while 39% was hydrogenated. The distribution of products indicates that the hydrogen release is considerably favoured over the CO release, resulting in a gas mixture unsuitable for efficient hydroformylation. The methanol to syngas reforming formally proceeds through two consecutive dehydrogenations20,31. The first one liberates hydrogen and formaldehyde, which then reacts with methanol, releasing hydrogen under the formation of methyl formate. This intermediary formate then serves as CO. We recognised that in an apolar reaction medium, the consecutive steps could potentially proceed at higher rates as releasing the polar intermediates from the catalyst into the solution should be less favoured. To evaluate this hypothesis, methanol was dissolved in toluene (1.0 mL). Using 10 equivalents of methanol (6.2 M) resulted in a slight decrease in side-product formation with further decreasing methanol amounts improving the selectivity. Gratifyingly, applying 1.5 equivalents of methanol (1.4 M) significantly increased the yield of the combined aldehydes (2a) to almost quantitative and with a linear-to-branched ratio (l:b) of 9:1. Decreasing the methanol amount further to 1.1 equivalents was found to be slightly less efficient, potentially due to overall a low partial pressure of gaseous reagents. Therefore, the studies were continued using 1.5 equivalents of methanol. Reducing the reaction temperature to 130 °C proved to be less effective, but nonetheless yielded the linear aldehyde in good yield and high selectivity.
As both methyl formate and paraformaldehyde have been reported to act as intermediates in the catalytic syngas release20,31, we were interested in testing them as syngas surrogates instead of methanol. Indeed, both compounds liberated suitable syngas mixtures in the dual catalysis set-up, although unreacted starting material 1a was detected in both cases. Lastly, alternative MACHO-derived metal complexes were considered. However, none of the tested candidates led to an overall more efficient system than Ru-MACHO.
With both the rate of the gas release and the composition of the gas mixture being crucial for interlocking the two catalytic cycles, we were interested in elucidating the kinetics of the dehydrogenation process. In order to observe the kinetic profiles, the pressure inside the reactor was monitored using a Keller pressure manometer (Fig. 2B–D). The investigation was initiated on methanol by running the reaction with and without the consecutive hydroformylation step that would consume the gaseous reagents (Fig. 2B). In the absence of the gas-consuming reaction (Fig. 2B, dark blue), a steep increase in pressure reaching approximately 4.8 bars within 2 h demonstrates efficient gas release, whereafter the pressure stabilises. If run with the hydroformylation in the adjacent reactor (Fig. 2B, light blue), the pressure tops at a lower maximum of 4.5 bars before declining again. This reveals that the hydroformylation is initiated before full gas release is achieved and stresses the need for close to simultaneous CO and H2 release from the beginning. After 6 h the pressure stagnates at 3.8 bar, signalling full consumption of the olefin. Lastly, MeOH-d4 was tested in the acceptorless dehydrogenation (Fig. 2B, violet). The gas formation was found to be significantly slower but viable with the pressure reaching its maximum after 4 h. This discrepancy can be explained by the kinetic isotope effect observed with deuterated methanol. Next, the proposed intermediates of syngas formation20,31 were tested to compare the efficiency of the gas release. Using paraformaldehyde instead of methanol (Fig. 2C, light blue) resulted in a significantly slower increase in pressure. However, using a 1:1 mixture of methanol and paraformaldehyde revealed a kinetic profile which superimposes well with pure methanol (Fig. 2C, dark blue). Methyl formate alone (Fig. 2D, light blue) was found to react considerably slower than methanol and paraformaldehyde. Here again, combining methanol together with methyl formate resulted in a profile well matched to methanol only (Fig. 2D, dark blue).
Potentially, this hints at methanol supporting the deconstruction of the following intermediates into hydrogen and carbon monoxide. In turn, using paraformaldehyde or methyl formate as syngas surrogate would be detrimental. Lastly, for methanol, paraformaldehyde, and methyl formate, the gas composition was measured for each during the pressure build up using gas chromatography. For methanol, a close to the ideal 2:1 mixture of H2 and CO was consistently observed after 1 h and 2 h. For both paraformaldehyde and methyl formate, the gas mixture was found to be less consistent over time and contained a lower ratio of hydrogen vs CO than that of methanol. Considering that these two syngas surrogates resulted in incomplete consumption of estragole in the dual catalysis set-up (Fig. 2A), these results align with the conclusion of the preliminary studies on the hydroformylation using different H2 to CO ratios (Fig. 1D), indicating that a 2:1 ratio H2 to CO is ideal. Furthermore, this study stresses that both the gas mixture and the release rate must match with the hydroformylation, with toluene as reaction medium aiding a consistent ratio between hydrogen and CO being released from methanol.
With the optimised reaction conditions at hand, we turned our attention to exploring the substrate scope of the hydroformylation reaction. Several terminal olefins (Fig. 3, 1–32) of various molecular complexity were subjected to the catalytic conditions, all of which yielded the corresponding aldehydes in good to quantitative yields and with good to excellent selectivity towards the linear (l) regioisomer with styrenes being an anticipated exception. Allylbenzene derivates with various functional groups (1–16) were used to probe functional group tolerance and sterical effects. Alkyl-substituted, as well as unsubstituted allylbenzenes, undergo hydroformylation with a comparable efficiency to 1 (2–4). Several functional groups that can be utilised for orthogonal C–C or C–X bond formations, such as (pseudo)halides and pinacol boronic esters32-34, are well tolerated (5–7). Furthermore, substrates with free and protected alcohol groups (8–10), anilines (12 and 13), as well as esters (11), nitro groups, nitriles (14 and 15), and thioethers (16) afford excellent yields of the hydroformylated products.
For all allylbenzene substrates the l:b selectivity is good and commonly at a ratio of 10:1, however, for electron-deficient systems, the selectivity is reduced (5, 7, 11, 14 and 15). For the homoallylated analogue of the benchmark olefin (17) and a paracetamol-derived homoallyl (18), the selectivity for n-aldehydes was found to be significantly better than for the allylic counterparts. Styrenes are also viable starting materials in this transformation, providing corresponding aldehydes in excellent yields with the branched isomer as the major product (19 and 20)28,35. The two regioisomers of 20 can be separated, making it possible to utilise the branched aldehyde for the synthesis of Naproxen35.
In chemical industries, a central role of hydroformylations is introducing molecular functionality into acyclic alkyl olefin platforms acquired from naphtha feedstock36. Among certain others, oxo products made from C8-olefins are of high economic importance. Using the here-developed dual catalysis approach, n-octene was converted to n-nonanal (21) in close to quantitative yield with excellent selectivity using methanol. More complex aliphatic compounds also undergo hydroformylation smoothly and selectively, affording the desired n-aldehydes in excellent yields (22–24). To demonstrate the value of this protocol for research and development laboratories, the hydroformylation of various natural product derivatives (25–27) and drug precursors (28–32) containing a diverse range of functionalities commonly displayed in drug-like compounds was conducted. The natural product quinine, containing an N-heterocycle, underwent the reaction seamlessly with excellent selectivity (25). A protected sugar and an allylated estradiol steroid derivative presented themselves as suitable substrates and afforded the corresponding aldehyde (26) and lactol (27), respectively.
Finally, five drug precursors (28–32) were prepared in excellent yields but with various selectivity using the methanol-based hydroformylation. It should be noted that the linear regioisomers of 28–32 can in a single transformation be converted into the respective drug molecules (28 to buspirone, 29 to diphenidol, 30 to mebeverine, 31 to vilanterol, and 32 to crispine A), thus this protocol may be of interest for late-stage isotopic labelling35.
In line with evaluating the potential application of this protocol in pharmaceutical R&D laboratories, we conducted stable isotope labelling of compound 1a by simply substituting the methanol source for commercial methanol-13C or methanol-d4 (Fig. 4A). As expected, utilising methanol-13C in the syngas-releasing reactor with the optimised conditions (Fig. 2A) did not affect either the yield or regioselectivity of the hydroformylation reaction, and aldehyde 33 was isolated with excellent incorporation of carbon-13 of 98% at the carbonyl position. We reasoned that quantitative incorporation is not achieved as both precatalysts contain unlabelled carbonyl ligands, totalling a theoretical 2.3 mol% unlabelled carbon monoxide available in the reaction. Next, we investigated the use of methanol-d4 as an enriched isotope source. Once again, the hydroformylation reaction proceeded well though with a slightly lowered yield of 92% of the product, but with excellent regioselectivity towards the linear aldehyde 34 and with a deuterium incorporation of 1.99. As isotopically labelled methanol is widely available and introduction of stable isotopes into new pharmaceutically active entities is of high importance for drug metabolism studies, this protocol could represent an interesting contribution to drug development programs.
Finally, to investigate the potential for accessing renewable carbon-based oxo-products, we turned to testing samples of industrial grade e-methanol for syngas release. Vulcanol™ is the first commercial e-methanol produced from Carbon Recycling International (CRI) in Iceland from the hydrogenation of flue gas with e-hydrogen37. To our delight, utilising 1.5 equiv of fuel-grade Vulcanol™ in the hydroformylation of olefin 1a directly from a bottle obtained from CRI without any form for purification except degassing with argon afforded 2a in an excellent yield of 97% and with a satisfactory l:b-selectivity of 13:1, using the optimised conditions (Fig. 2A). With the potential for large-scale applications in mind, we were interested in exploring the scalability of the here-developed method. Once again, our benchmark reaction using estragole (1a) was now performed on a gram-scale (3.71 g, 25.0 mmol) in a two-reactor system with a volume of 430 mL in total (Fig. 4B), representing an upscaling of 25 times. With Vulcanol™ as the syngas source, 2a was isolated in a gratifying 84% yield with a l:b-selectivity of 14:1. This preliminary upscaling attempt applying e-methanol highlights the robustness of the interlocking catalysis.
In summary, we report the successful union of two valuable catalytic cycles, the acceptorless dehydrogenation of methanol to syngas and hydroformylations of olefins. We demonstrate the importance of the rate and selectivity of the syngas release, and how matching it with a low-pressure rhodium-catalysed hydroformylation results in an efficient methodology for accessing oxo products. Furthermore, it is possible to replace coal- or natural gas-derived syngas with fuel-grade e-methanol accessed from CO2 capture and hydrogenation on gram scale. While these conditions do not mimic those applied in industrial settings producing bulk chemicals, we consider this dual catalysis a proof-of-concept for the possibility of synthesising oxo-products entirely from CO2 as renewable carbon feedstock and integrating this important transformation into a methanol economy. It is our expectation that redesigning the chemical value chains to extend from renewable platforms such as methanol may be an important part of establishing a sustainable chemical industry.