Peter G. N. Neatea,
Liangliang Huang*b,
Margaret R. Jonesa,
Ruilian Wuc,
Nilusha M. Sudasinghea and
Xiaokun Yang
*a
aChemistry Division, Los Alamos National Laboratory, Los Alamos NM 87544, USA. E-mail: xiaokuny@lanl.gov
bSchool of Sustainable Chemical, Biological & Materials Engineering, University of Oklahoma, Norman, OK 73019, USA. E-mail: hll@ou.edu
cBioscience Division, Los Alamos National Laboratory, Los Alamos NM 87544, USA
First published on 17th June 2025
Alkyl furans have a variety of applications and therefore numerous approaches exist for their synthesis. While these methods can be effective, there are various drawbacks largely associated with the waste generated. Herein we report a novel method for the direct coupling of readily available furans with allylic alcohols. These include terpenoid alcohols such as geraniol and prenol, which are readily available from renewable, bio-derived feedstocks. Furthermore, the reaction is facilitated using aluminosilicates, including zeolites. It was found that Y-type zeolites were highly effective at mediating the coupling with furan, giving the 2-alkylated furan high yields. While various allylic alcohols were effective for this reaction, the identity of the furan proved to have a drastic impact on reactivity. For substituted furans, the reaction proved most effective with an amorphous aluminosilicate and calculations later revealed potential key factors in further developing and generalizing this methodology.
![]() | ||
Scheme 1 Approaches to alkylation of furan derivatives. In our approach, allyl alcohol (CH2![]() |
Entry | Catalyst (50 wt%) | Yield (%) | Catalyst details |
---|---|---|---|
a 25 wt% zeolite used. | |||
1 | CBV720CY | 92 | Y-type; SiO2/Al2O3 = 30 |
2 | CBV720CYa | 81 | Y-type; SiO2/Al2O3 = 30 |
3 | CBV760 | 93 | Y-type; SiO2/Al2O3 = 60 |
4 | CBV90A | 62 | Mordenite-type; SiO2/Al2O3 = 90 |
5 | CBV8014CY | ND | ZSM-5-type; SiO2/Al2O3 = 80 |
6 | CBV3024E | ND | ZSM-5-type; SiO2/Al2O3 = 30 |
7 | Perlkat 46–10 | 26 | Amorphous; SiO2 > 90 wt%; Al2O3 ∼10 wt% |
8 | SiO2 | ND | — |
9 | Al2O3 | ND | — |
10 | Amberlyst-15 | 78 | — |
The reactions were conducted for 18 hours at 25 °C using 50 wt% of the heterogeneous catalyst. Y-type zeolites were found highly effective for this coupling, producing the allyl furan 3 in high yields in excess of 90% (Table 1, entries 1 and 3). Reducing zeolite loading to 25 wt% led to a modest drop in yield to 81% (Table 1, entry 2). Increasing the silica-to-alumina ratio, thereby enhancing the acidity of the Y-type zeolite, had no significant impact on the reaction (Table 1, entry 3). In contrast, other types of zeolites were much less effective. A mordenite-type zeolite resulted in a significantly lower yield (Table 1, entry 4), and ZSM-5-type zeolites did not yield any product, regardless of the silica-to-alumina ratio (Table 1, entries 5 and 6). Perhaps surprisingly, Perlkat 46–10, a commercially available amorphous aluminosilicate, did produce some product, albeit in low yield (Table 1, entry 7). Control reactions confirmed that silica or alumina alone did not yield any detectable product (Table 1, entries 8 and 9). Amberlyst-15, a more conventional heterogenous acid catalyst was also effective, although less so than Y-type zeolites (Table 1, entry 10). This suggests that the reaction likely proceeds via protonation of the alcohol and elimination of water to form a stabilized allylic cation. It should be noted, however, that reactions using Amberlyst-15 were found to be less regio-selective, with complete consumption of phytol and multiple alkene signals observed in the 1H-NMR spectrum of the crude reaction mixture.
Following the initial success with phytol, we examined other allylic alcohols (Fig. 1). The length of the alkyl chain did not significantly affect the reaction, with the terpenoids prenol and geraniol both proving effective, albeit with slightly lower yields compared to phytol (Fig. 1, 4, and 5). In contrast, allyl alcohol did not produce any product, with complete recovery of the starting materials. This may be due to the lack of substitution around the alkene, which would stabilize the carbocation intermediate. It is also worth noting that cinnamyl alcohol was also highly effective in coupling with furan, although it resulted in an inseparable mixture of the two products. The major component was the intended linear product (Fig. 1, 6), while the minor product was the branched isomer, where the furan ring reacted at the benzylic position (Fig. 1, 7). The mixture of products can be attributed to the allylic cation being more stable due to conjugation with the phenyl ring and less sterically hindered compared to previous examples.
While the reaction was relatively insensitive to different allylic alcohols, substitution on the furan ring had a drastic effect. Extending the reaction protocol to 2-methylfuran resulted in negligible product formation. Consequently, we re-examined reaction parameters and catalysts using the reaction between 2-methylfuran 8 and phytol 2 as a platform (Table 2).
Entry | Catalyst | Temperature (°C) | Residual phytol (%) | 10 (%) | 9 (%) | 9![]() ![]() |
---|---|---|---|---|---|---|
a 25 wt% used.b 10 wt% used. | ||||||
1 | CBV720CY | 25 | 95 | — | 3 | — |
2 | CBV760 | 25 | 97 | — | 2 | — |
3 | CBV90A | 25 | 80 | 5 | 7 | 1.4![]() ![]() |
4 | Perlkat 46–10 | 25 | 85 | 3 | 6 | 2.0![]() ![]() |
5 | CBV720CY | 40 | 91 | — | 5 | — |
6 | CBV760 | 40 | 85 | — | 8 | — |
7 | CBV90A | 40 | 44 | 11 | 18 | 1.6![]() ![]() |
8 | Perlkat 46–10 | 40 | 34 | 9 | 30 | 3.3![]() ![]() |
9 | CBV720CY | 60 | 75 | 4 | 9 | 2.3![]() ![]() |
10 | CBV760 | 60 | 19 | 8 | 30 | 3.8![]() ![]() |
11 | CBV90A | 60 | 11 | 19 | 35 | 1.8![]() ![]() |
12 | Perlkat 46–10 | 60 | — | 14 | 51 | 3.6![]() ![]() |
13 | Perlkat 46–10a | 60 | — | 14 | 50 | 3.6![]() ![]() |
14 | Perlkat 46–10b | 60 | 21 | 9 | 36 | 4.0![]() ![]() |
15 | Amberlyst-15b | 60 | — | 12 | 41 | 3.4![]() ![]() |
At 25 °C, a temperature effective for furan as the substrate, only trace amounts of product were observed with any aluminosilicates tested (Table 2, entries 1–4). Heating to 40 °C had little effect for Y-type and ZSM-5-type zeolites (Table 2, entries 5 and 6), while mordenite-type zeolite and amorphous Perlkat 46–10 showed improved reactivity with low but appreciable product formation (Table 2, entries 7 and 8). Further increasing the reaction temperature significantly improved product formation for all aluminosilicates except the Y-type zeolite (Table 2, entries 9–11). Using Perlkat 46–10 (Table 2, entry 12), the reaction provided a combined isolated yield of 65%, composed of 51% of the 5-substituted isomer 9 and 14% of the 3-substituted isomer 10. In all cases, the 5-substituted allyl furan 9 was the major product, with small amounts of the 3-substituted analog 10 also observed. Perlkat 46–10 (Table 2, entry 12) and CBV760 (Table 2, entry 10) gave comparable selectivity for the 5-substituted isomer, with product 9:
10 yield ratios of 3.6
:
1 and 3.8
:
1, respectively. In contrast, mordenite CBV90A (entry 11) showed markedly lower selectivity for product 9, with a corresponding 9
:
10 yield ratio of 1.8
:
1. Amberlyst-15 was also effective, albeit with lower yields and significantly more side-product formation even at lower catalyst loadings (Table 2, entry 15). The loading of Perlkat 46–10 could be reduced to 25 wt% without a significant drop in yield, although any further reduction led to a marked decrease in yield (Table 2, entries 13 and 14).
The use of solvents did not improve the reaction yields (see Table 3). Ethereal solvents such as THF and 1,2-dimethoxyethane resulted in significantly lower yields (Table 3, entries 1 and 2). While toluene and 1,2-dichloroethane did not significantly affect the overall yield or selectivity (Table 3, entries 3 and 4), they provided greater reproducibility. Under solvent-free conditions (Table 2, entry 12), approximately one in three reactions exhibited a roughly 15% lower yield, however, this issue was mitigated when using 1,2-dichloroethane or toluene. Additionally, reducing the loading of 2-methylfuran 8 when using 1,2-dichloroethane negatively impacted the yield (Table 3, entries 7 and 8). Although the decrease was not drastic, the yield dropped to 52% when using 3 equivalents of 2-methylfuran, and even lower yields were obtained with further reductions (Table 3, entries 5, 6, and 9). The higher reproducibility with toluene and 1,2-dichloroethane suggests that these solvents provide a more stable reaction environment, potentially by better solubilizing the reactants and intermediates. In contrast, the detrimental effect of ethereal solvents could indicate that these solvents interact unfavorably with the catalytic sites or intermediates.
Entry | Solvent | Equivalents 2-methylfuran | Solvent volume (mL) | Residual phytol (%) | 10 (%) | 9 (%) | 9![]() ![]() |
---|---|---|---|---|---|---|---|
1 | THF | 10 | 2 | 66 | 6 | 15 | 2.5![]() ![]() |
2 | DME | 10 | 2 | 66 | 5 | 15 | 3.0![]() ![]() |
3 | Toluene | 10 | 2 | — | 13 | 48 | 3.7![]() ![]() |
4 | DCE | 10 | 2 | — | 15 | 50 | 3.3![]() ![]() |
5 | DCE | 5 | 2 | — | 12 | 44 | 3.7![]() ![]() |
6 | DCE | 3 | 2 | — | 10 | 42 | 4.2![]() ![]() |
7 | DCE | 3 | 1 | — | 10 | 40 | 4.0![]() ![]() |
8 | DCE | 3 | 0.5 | — | 9 | 39 | 4.3![]() ![]() |
9 | DCE | 1.5 | 2 | — | 10 | 33 | 3.3![]() ![]() |
With effective reaction conditions for 2-methylfuran 8 established, we assessed the general applicability of this system to different allylic alcohols and substituted furans (Fig. 2). Once again, the chain length of the allylic alcohol had little effect on the overall yield, with phytol, geraniol, and prenol giving similar yields (Fig. 2, 9/10, 11, and 12, respectively). Notably, prenol exhibited higher selectivity for the 5-substituted allyl furan product compared to phytol and geraniol. Cinnamyl alcohol, while effective, resulted in an inseparable mixture of isomers (Fig. 2, 13). Three isomers were obtained with the 5-substituted allyl furan the major product and 3-substituted the secondary product in similar ratios to reactions with phytol and geraniol. The minor component was the branched isomer, with the furan ring attached at the benzylic position, as was observed previously for the reaction between furan and cinnamyl alcohol (Fig. 1, 6 and 7). Perhaps unsurprisingly, the reaction between 2-ethylfuran and phytol 2 gave a similar yield and selectivity to the analogous reaction with 2-methylfuran 8 (Fig. 2, 14). Of note is that both methyl-2-furoate and benzofuran reacted to give only the 5-substituted furan, with no other isomers detected (Fig. 2, 15–17). These results demonstrate the versatility of the reaction conditions across various allylic alcohols and substituted furans. The higher selectivity observed with prenol, along with the mixture of isomers resulting from cinnamyl alcohol, suggest that steric and electronic effects on the allylic alcohol influence product distribution. As mentioned previously, allyl alcohol did not produce any observable product and remained unreacted (Fig. 2). Disubstituted furans were also ineffective in this reaction system, with both 2,3-dimethyl- and 2,5-dimethylfuran giving poor yields or no detectable product, respectively (Fig. 2, 18 and 19). Unlike methyl-2-furoate, both furfural and 2-acetylfuran resulted in poor yields of the intended product (Fig. 2, 20 and 21, respectively). This could be due to side reactions as all have similarly electron-withdrawing substituents. The lack of reactivity with allyl alcohol suggests that substitution around the alkene plays a crucial role, potentially in stabilizing the proposed carbocation intermediate. Understanding these limitations is essential for further optimizing the reaction conditions and expanding the range of compatible substrates.
The adsorption energies of the product allylic furan 3 also provided insight into the high zeolite loading required for these reactions. The adsorption of the product was significantly more favorable than that of phytol. For faujasite and mordenite zeolites, adsorption energies for product 3 were −238 kcal mol−1 and −26 kcal mol−1, respectively, compared to −117 kcal mol−1 and −119 kcal mol−1 for phytol. These results suggest that the strong adsorption of the product allylic furan 3 within the zeolites might inhibit further catalytic activity by occupying active sites, thereby necessitating higher catalyst loadings to maintain reaction efficiency. Those computational results also reveal a potential limitation of the system, where strong product adsorption could lead to catalyst deactivation over time. As reviewed recently, tailoring the local properties of active sites in a zeolite to balance adsorption strengths and enhance the desorption of products for active site regeneration is critical for zeolite catalysis.22,23
Further supporting the superior reactivity of Y-type zeolite is its adsorption capacity for furan 1, which is 0.336 g g−1—more than twice that of ZSM-5 (0.142 g g−1) and over three times that of mordenite (0.0944 g g−1). Adsorption energies for furans, however, did not show significant differences among the three zeolites (Table 4). Notably, all substituted furans had higher adsorption energies than furan 1, especially those with carbonyl groups (Table 4, entries 4–6). This suggests that substituted furans, due to their excess and higher adsorption energies, outcompete phytol 2 for adsorption into the pores. Consequently, the reactivity of phytol 2 is hypothesized to occur on the outer surface, which, in return, explains why amorphous Perlkat 46–10 was the most effective aluminosilicate for substituted furans. Although this remains speculative, these observations underscore the complex interplay between adsorption capacities and reactivity. The higher adsorption capacity of Y-type zeolite for furan 1 likely contributes to its superior performance. However, the higher adsorption energies of substituted furans indicate that they dominate the adsorption sites, likely forcing reactions to occur on the outer surface of the zeolite catalyst.
We examined the rates of self-diffusion for both allyl alcohol and prenol within a faujasite cell containing 15 molecules of furan 1. Extremely high diffusion rates can lead to a lack of reactivity, as the substrate may not interact effectively with the solid catalyst due to rapid movement.24 The self-diffusion rates for allyl alcohol and prenol were 5.271 × 10−9 m2 s−1 and 1.307 × 10−9 m2 s−1, respectively. Similarly, the self-diffusion rates of furan did not significantly change in the presence of either allyl alcohol or prenol, with rates of 4.789 × 10−9 m2 s−1 and 2.054 × 10−9 m2 s−1, respectively. The self-diffusion rates highlight that physical movement within the zeolite pores does not significantly affect the reaction. Instead, the electronic properties of the reactants play a crucial role. The greater substitution of prenol likely stabilizes the cationic intermediate more effectively than allyl alcohol, enhancing reactivity.
For practical industrial applications, an integrated approach combining several of these strategies might prove most effective. For example, initially synthesizing or choosing zeolites with controlled acidity and hierarchical porosity followed by targeted surface modifications (e.g., hydrophobic surface silylation) could achieve an optimal balance between catalytic efficiency and long-term stability. Periodic regeneration steps should also be integrated into reaction protocols, informed by continuous catalyst performance monitoring.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cy01320d |
This journal is © The Royal Society of Chemistry 2025 |