4-Methyltetrahydropyran as a Convenient Alternative Solvent for Olefin Metathesis Reaction: Model Studies and Medicinal Chemistry Applications.

A number of metathesis reactions were successfully conducted in 4-methyltetrahydropyran, including both standard model dienes, as well as more complex substrates, such as analogues of biologically active compounds and active pharmaceutical ingredients. To place this solvent in a context of pharmaceutical R + D, larger-scale syntheses of SUAM 1221, a prolyl endopeptidase inhibitor with potential application in Alzheimer disease treatment, and a derivative of sildenafil, an analogue of the popular Viagra drug, were executed. In the latter case, despite all the setup being made in air, the metathesis reaction at a 33 g scale proceeded very well with relatively low catalyst loading and without need of aqueous workup or column chromatography.

Introduction

Recent regulations enforced by many countries issued a challenge to chemical industry to adapt to green chemistry principles,[1] especially in the regards of the highest possible atom economy and strict environment protection. This goal most often is achieved by using carefully selected catalysts that allow for highest possible selectivity and yield and by scrupulous optimization of the reaction conditions. Importantly, reactions shall be conducted neatly or in solvents that are nontoxic and have minimal impact on the environment.[1]

One example of a catalytic reaction of increasing popularity in both basic research and in industrial context[2,3] is olefin metathesis, a process that enables the effective formation of carbon–carbon double bonds.[4,5] The most significant impact on the development of metathesis methodology had understanding of its mechanism[6] and introduction of well-defined transition metal catalysts, especially air-stable ruthenium complexes containing N-heterocyclic carbene (NHC) ligands (see Figure ).[6−8] A smaller but still profitable improvement was the development of complexes with unsymmetrical N-heterocyclic carbene (uNHC) ligands, exhibiting increased stability in the presence of ethylene and lesser isomerization properties.[9] Recently, also catalysts containing cyclic-(alkyl)(amino)-carbenes (CAAC) gained more and more popularity.[10−13] What is more, Lemcoff et al. proved that some of the latter are excellent in reducing double bond isomerization at high temperatures.[14]

Figure 1

(a) Symmetrical NHC ligands IMes, SIMes, and SIPr and (b) selected recently developed uNHC and CAAC ligands and ruthenium catalysts Ru1–Ru4 derived thereof.

In general, these Ru catalysts exhibit high functional group tolerance and low sensitivity toward moisture and oxygen. The compatibility with many functional groups (incl. polar ones) enabled the use of olefin metathesis in reactions of a plethora of densely functionalized substrates, including natural and bio-active compounds at the last stages of functionalization.[15−17]

With some exceptions, like ethenolysis of plant oils, self-cross metathesis (self-CM) of Fischer–Tropsch α-olefins or ROMP (ring-opening metathesis polymerization) performed usually under solvent-free conditions, the majority of metathesis reactions are carried out in a solution. Despite many “green solvents”, such as water,[18] ethanol,[19] dimethyl carbonate,[20−22] ethyl acetate,[23,24] supercritical carbon dioxide,[25] polyethylene glycol,[26] methyl decanoate,[27] ethyl lactate,[28] or p-cymene,[29] having been proposed in context of metathesis, the truth is that halogenated and aromatic solvents are still the most frequently used.[23] Unfortunately, they are toxic or at least harmful (dichloromethane and 1,2-dichloroethane belong to ICH class 1 and toluene to ICH class 2 solvents),[30] and the use of some of them will be, or already is, restricted or even banned. Therefore, it is crucial to identify alternative reaction media[31] compatible with existing catalytic systems that are more environmentally friendly and safer to use. This would be in line with the principles of a circular economy,[32] which is recommended not only by the European Commission but also by other countries.[33]

However, the use of alternative solvents in the context of olefin metathesis often brings some complications. Molybdenum and tungsten alkylidenes are instantaneously decomposed by protic solvents (water and alcohols),[34] and ruthenium carbenes are easily deactivated by Brønsted bases,[35,36] thus such solvents shall be avoided. They are also easily degraded by oxidative pathways,[13,37] and a number of ether solvents, such as Et2O, i-Pr2O, and THF, are known to undergo auto-oxidation and can develop substantial amounts of peroxides during storage. On the other hand, a valuable green solvent, 2-MeTHF, was found to promote C–C double bond isomerization during the olefin metathesis course.[38] Also, the solvent price and capital outlay related to its purification, handling, and disposal, the aspects of seasonal availability, and last but not least the environmental issues (sometimes hard to be a priori predicted)[39] shall be considered in each case.

As all solvents tried till now in olefin metathesis possess some specific advantages and disadvantages, it is therefore advisable to look for new solvents for this reaction. One of the candidates can be 4-methyltetrahydropyran (4-MeTHP), a novel hydrophobic cyclic ether. According to published physicochemical data,[40] it exhibits solving properties similar to THF and 2-MeTHF; however, in contrary to the latter, it can also dissolve nonpolar materials. What is more, due to its high hydrophobicity, it can be easily separated from water (Figure ), which in comparison with THF or dioxane simplifies the purification and reduces the amount of wastes.

Figure 2

Different behavior of THF (left) and 4-MeTHP (right) mixed with water (water phase was colored blue; photo by the authors).

In addition, 4-MeTHP shows a higher boiling point compared to other popular ether solvents, therefore it can be utilized at higher temperatures. Importantly, it exhibits also a much reduced trend toward auto-oxidation and has better toxicity profile compared to THF, which makes its application safer (Table ).[41−43]

Table 1

Comparison of Selected Physical Properties of 4-MeTHP and THF[44]

solvent	bp (°C)	mp (°C)	viscosity (cP)	solubility in water (wt %)	water solubility (wt %)
4-MeTHP	105	–92	0.78	1.5	1.4
THF	65	–109	0.55	∞	∞

4-MeTHP can be utilized under acidic or alkaline conditions, thus it is being used as a solvent in esterifications, radical reactions, Grignard and Wittig reactions, halogen-metal exchange, in reductions and oxidations, including epoxidation, and many other organic reactions.[45] Due to all these advantages and green chemistry perspectives, it is expected that this solvent will find a broad range of applications, especially in industrial processes, as it was suggested in an elegant exploratory study by Kobayashi et al.[45] To the best of our knowledge, however, 4-MeTHP has not yet been tested in olefin metathesis reactions promoted by modern second-generation catalysts.[46]

Results and Discussion

Comparative Tests Using Simple Model Substrates

In present work, we decided to test 4-MeTHP as a solvent for olefin metathesis, first with standard model olefin metathesis substrates[47] then with more advanced polyfunctional reagents, such as potentially bioactive compounds, APIs (active pharmaceutical ingredients) and their analogues, also at a larger scale, typical for industrial process chemistry R + D.

For this study, we selected a known Hoveyda–Grubbs-type catalyst Ru3d containing an unsymmetrical NHC ligand (structure d in Figure ).[48] This recently introduced catalyst exhibits excellent selectivity in self-CM of α-olefins,[48] macrocyclizations performed under high-concentration conditions,[49] and in ethenolysis[12] and gave preliminary good results in a green solvent, 2-MeTHF.[48]

At the beginning of our study, we opted to compare behavior of THF and 4-MeTHP against a background composed of toluene (one of the most popular solvents for olefin metathesis) and some alternative solvents, used less frequently in this transformation.[23] In such a comparative test, we utilized a standard benchmark substrate,[47] diethyl (diallyl)malonate (DEDAM, 1). To do so, we conducted a set of model ring-closing metathesis (RCM) reactions of 1 catalyzed by 1 mol % of Ru3d at 50 °C and recorded the time/conversion curve for each solvent (Figure ) using an NMR technique.[50] Solvents used for this experiment are commercially available and were taken from freshly opened containers (with exception of 4-MeTHP and one batch of THF, which were taken from opened bottles stored for 12 months, see below for details) without further purification and drying or degassing.

Figure 3

Time/conversion curves (monitored by NMR) for RCM of 1 (c = 0.2 M) with 1 mol % of Ru3d at 50 °C in various solvents conducted under otherwise identical conditions. 4-MeTHP and THF2 were taken from opened bottles stored for 12 months, the other solvents were taken from freshly opened containers. Lines are visual aids only.

While the highest conversions in the studied transformation were achieved in EtOAc, MTBE, toluene, and anisole, the reaction in 4-MeTHP was also proceeding very well, and visibly faster than in THF (a sample from the freshly opened bottle, Figure , THF1). The results of the RCM reaction conducted in an aged THF sample (from a bottle opened 12 months ago and stored in air) were even worse (Figure , THF2). As the purity of these two samples of THF was similarly high (≥99.8 ± 1%, see Table ), the difference must be related to some other factors, for example, to contaminants present in THF in a much smaller amount. Using Karl Fischer titration, the water content was measured, exhibiting much higher number in the case of the old THF bottle (879 versus 152 ppm, respectively). In addition, a simple peroxide test (see the Supporting Information) showed—as one might expect—that the aged sample of THF contains much higher amount of peroxides than the solvent taken from a freshly opened bottle. Although not quantitative, these results show that 4-MeTHP is more resistant to “aging”, thus it can be a valuable alternative to THF due to its lower tendency toward auto-oxidation and lower hygroscopicity.

Table 2

Solvents Effects in RCM Reaction of 1

entry	solvent	purity (%)a	water content (ppm)b	conversion at 60 min (%)
1	4-MeTHPc	≥99.0	65	95
2	EtOAc	≥99.7	185	99
3	Toluene	≥99.9	91	98
4	Anisole	99.7	759	97
5	MTBE	≥99.8	203	98
6	THF1	≥99.9	152	91
7	THF2c	≥99.8	879	81

Purity declared by the producer.

Measured by Karl Fischer titration.

Sample taken from previously opened bottle stored for 12 months.

As the catalyst Ru3d is known to provide high selectivity even in the case of products sensitive toward C–C double bond isomerization,[48] we were curious if the same trait is exhibited also in a 4-MeTHP solvent. To test this, we selected Ru3d and two commercial general purpose catalysts, Ru3c and Ru4b. Gratifyingly, in an experiment presented in Figure , it was observed that in RCM of N,N-diallyl tosylamide (DATA, 3) conducted in the 4-MeTHP complex, Ru3d provided excellent selectivity toward the expected metathesis product (4), even at higher temperatures (Table ). This result supports the usefulness of 4-MeTHP as a medium for selective olefin metathesis utilizing Ru3d. It shall be noted that under the same temperature, the general purpose catalysts, Ru3c and Ru4b, led to C–C double bond shift, producing substantial amounts of isomerized product 4′ (Table ), which, however, is significantly lower than this produced in the reaction of 3 catalyzed by Ru4b in 2-MeTHF (Table , entry 9).[38] Catalysts featuring symmetrical NHC ligands, such as SIPr and SIMes, are known to decompose under demanding reaction conditions to form various ruthenium species that are responsive for C–C double bond isomerization (shift), thus eroding the selectivity of the reaction.[26,51−55]

Figure 4

RCM of 3 conducted in 4-MeTHP with various catalysts.

Table 3

RCM of 3 Conducted in 4-MeTHP with Selected Catalysts at Various Temperatures

entry	catalyst	temperature (°C)	conversion (%)	4/4’
1	Ru3d	80	>99	99:1
2		105a	>99	95:5
4	Ru3c	80	>99	100:0
5		105	>99	29:71
7	Ru4b	80	>99	99:1
8		105	>99	57:43
9		80	>99	5:95b

Boiling point of 4-MeTHP.

Reaction in 2-MeTHF, see ref (38).

With this promising result in hand, we started systematic examination of catalytic activity of Ru3d in a set of metathesis reactions performed in air in 4-MeTHP distilled from sodium benzophenone ketyl.

Obviously, catalyst loading as high as 1 mol % does not correspond to the current standards, when sometimes it is possible to convert simple substrates into products in the presence of as little as a few ppm of the complex.[11] Moreover, such high loading is most often economically unviable, especially when the use of olefin metathesis is considered in production of low-price commodity chemicals. Therefore, we first examined the reaction of the already tested substrate, 1, in the presence of 0.1 and 0.05 mol % of Ru3d (Table , entry 1). In the first case, the desired product 2 was provided almost quantitatively, while after twofold decrease in the catalyst loading, lower but still satisfactory 76% conversion was observed after 2 h. Next, two rather simple derivatives of diethyl malonate were tested (Table , entries 2 and 3), leading to compounds 4 and 6 with almost quantitative yields in the presence of only 0.1 mol % of Ru3d. The other recommended[47] model substrate 7 bearing a more challenging gem-disubstituted double C–C bond required a slightly higher amount of catalyst (added in two portions), but also in this case, the conversion was very high (Table , entry 4). Furthermore, it was also possible to synthesize a much more demanding product 10, containing a tetrasubstituted double C–C bond (Table , entry 5), but this time, significantly more forcing conditions, namely, 5 mol % of catalyst at 110 °C for 48 h, were needed to achieve about 50% of conversion, which is the expected result for a catalyst such as Ru3d.[56] Furthermore, Ru3d in 4-MeTHP worked also well, however in slightly higher catalyst loading, with 2,2-diallyl-2,3-dihydro-1H-inden-1-one (11) and 2,2-diallyl-1H-indene-1,3(2H)-dione (13), giving the spiro-compounds 12 and 14, respectively (Table , entries 6 and 7). Similarly, when pharmaceutically relevant barbituric acid derivative 15 was used, the corresponding product 16 was obtained in good yield (Table , entry 8). High activity and gentleness of Ru3d in 4-MeTHP was also witnessed in reaction of (S)-N,N-diallyl-1-tosylpyrrolidine-2-carboxamide (17), providing the proline derivative 18 in good yield and with typical for catalysts with unsymmetrical NHC ligands high selectivity—migration of double bond was not observed (Table , entry 9).[57]

Table 4

Model RCM Reactions Conducted in 4-MeTHP

Calculated based on 1H NMR measurement.

Second portion added after 1 h.

Reaction performed at reflux.

Conditions: Ru3d, 4-MeTHP, 70 °C, c = 0.5 M.

Having obtained satisfactory results in the model RCM reactions, the activity of Ru3d was tested in selected cross-metathesis (CM) reactions (Scheme ). Again, high product yields were achieved in the presence of reasonable amounts of the catalyst. For example, after mixing 1-allyl-4-methoxybenzene (19) with three equivalents of cis-1,4-diacetoxy-2-butene (20) in the presence of as little as 0.3 mol % of Ru3d, 96% of the desired product 21 was obtained. The reaction performed in the presence of the same amount of SIPr analogue of the Hoveyda–Grubbs second generation catalyst Ru3c under the same conditions produced compound 21 in 52% yield. Similarly, good result was found in the next reaction as well, yielding the expected product 24—milk lactone, ingredient found in many dairy products—in very good isolated yield. The slightly more challenging CM reaction using cis-6-nonenal (28), a compound with a fresh, citrus scent, was slightly less effective; nevertheless, the product of reaction with allylbenzene (29) was obtained in 60% isolated yield.

Scheme 1

Preparative CM Reactions in 4-MeTHP (Isolated Yields of Analytically Pure Products)

The examples provided so far, although interesting from a scientific point of view, are still not of high structural complexity. So, in order to test if 4-MeTHP can be indeed a solvent of interest for a broad range of synthetic chemists including industrial process chemists, we focused on biologically active compounds, APIs (active pharmaceutical ingredients) and related analogues (Scheme ). First, we attempted formation of cyclohexene fragments of two β-lactams 33 and 34. Both were obtained in good yields, although lactam 32 that contains a trisubstituted C–C double bond needed four times larger amount of the catalyst to be formed. Next, we attempted the RCM reaction of sulfide 35, producing 2,5-dihydro-1H-pyrrole-based compound 36, an analogue of modafinil, an API commonly used in treatment of sleep disorders.[58] The reaction was carried out in the presence of only 0.5 mol % catalyst and delivered the expected product in 85% yield. Similar reactivity of Ru3d in 4-MeTHP was observed for substrate 37, which effectively formed compound 38, being an analogue of UR-144, a selective full agonist of the peripheral cannabinoid receptor invented in Abbott Laboratories[59] (recently used rather as a recreational drug).[60] An isolated yield of 70% was observed in RCM of substrate 39 leading to an unsaturated product 40, which in just one step can be transformed directly into SUAM 1221, a prolyl endopeptidase inhibitor with potential application in Alzheimer disease treatment.[61] However, an almost quantitative yield (96% isolated) was observed when the same reaction was performed on 10 times bigger scale performed with 0.1 mol % more of the catalyst.

Scheme 2

Preparative RCM Reactions of Complex Substrates in 4-MeTHP (Isolated Yields of Analytically Pure Substances)

Experiments reported above have shown that 4-MeTHP can act as a solvent of choice also in the case of polyfunctional and relatively complex compounds. As a final test of this promising solvent, we decided to perform the RCM reaction on a larger scale, choosing sulfonamide 42, a close relative of sildenafil—a drug sold inter alia under the trade-name Viagra—as a target product (Scheme ). Despite diene 41 containing a number of polar or Lewis basic groups, a test RCM reaction run at a small scale (95 mg of 41) catalyzed by 2 mol % of Ru3d (added in four portions) underwent smoothly at 70 °C, leading to 42 in an isolated yield of 86%. To test the same transformation on a 0.70 mol scale, a glass OptiMax reactor was charged in air with 33 g of diene 41 and 720 mL of 4-MeTHP, freshly distilled under nitrogen. Next, Ru3d (1 mol %, 463 mg, 0.72 mmol) was weighed in air and added to the reactor in one portion as a solid. After stirring for 2 h at 70 °C, the reaction was complete according to TLC, so the reactor content was slowly cooled to 10 °C and stirred for 60 min. This operation caused crystallization of a solid, which was then filtered off, washed with cold 4-MeTHP, and dried in a vacuum dryer giving 42 in 88% yield (27.17 g). Purity of this crude material obtained as an off white solid was ≥99% according to NMR and HPLC analysis. It is worth mentioning that use of 4-MeTHP allowed for isolation of the product without need for classical aqueous extraction, thus completely eliminating production of water wastes, and along with the scale-up, it was possible to decrease the catalyst loading by a factor of two. Importantly, use of 4-MeTHP allowed for synthesis of 42 with the Ru content as low as 88 ppm only after a simple filtration of the reaction mixture, thus significantly reducing the metal amount in the product (2297 ppm of ruthenium was used in this reaction, see the Supporting Information for calculations).

Scheme 3

Larger-Scale Preparation of Sildenafil Analogue 42

Photographs: (a) weighing of the substrate, (b) charging the reactor with 4-MeTHP, (c) adding the catalyst in air, (d) crystallization of the product after cooling, (e) filtration, and (f) isolated crude 42 containing 88 ppm of trace metal (acc. to ICP MS).

To compare 4-MeTHP with a “classical” solvent used previously in similar R + D studies, the same RCM reaction of 41 (10 g scale) was repeated in dichloroethane (DCE). After reaction was completed according to TLC (2 h), we attempted to isolate the product. Unfortunately, due to higher solubility of 42 in DCE, it was not possible to precipitate it as effectively as it was the case where 4-MeTHP was used as a solvent. Instead, the reaction mixture was concentrated under reduced pressure to dryness and then crude 42 (containing 2227.6 ppm Ru acc. to ICP-MS) was dissolved in 10% aqueous solution of NaOH at 80 °C, treated with activated charcoal, and precipitated by a drop-wise addition of concentrated HCl. The precipitated product was filtered and dried in a vacuum drier, yielding new sildenafil analogue 42 as a cream solid (7.43 g, 79% of yield). Such an obtained sample had 51.9 ppm of residual Ru according to ICP-MS analysis.

Importantly, due to different product solubilities, the reaction performed in DCE required additional purification steps due to higher content of ruthenium in the crude product. We decided to quantify this difference with the help of green chemistry metrics. Green chemistry metrics serve to quantify the efficiency or environmental performance of chemical processes and allow changes in performance to be measured. They include effective mass yield, carbon efficiency, atom economy, reaction mass efficiency, environmental (E) factor, and the EcoScale.[62] As in our case, both reactions involved the same reagents, which differed only in the solvent and purification method, two green chemical metrics were considered (see the Supporting Information for details).

Environmental (E) Factor[63]

Sheldon’s environmental factor E is defined by the ratio of the mass of waste per mass of the product (E = total waste/product). In the case of the pharmaceutical industry, the E factor is between 25 and 100. In our case, for the reaction performed in 4-MeTHP, a value of 22.94 was calculated, while the process using DCE gave an E factor of more than two times higher (56.22; for details of these calculations, see the Supporting Information).

EcoScale Score[64]

The EcoScale allows the evaluation of the effectiveness of a synthetic reaction. It gives a score from 0 to 100, but it also takes into account cost, safety, technical setup, energy, and purification aspects. It is obtained by assigning a value of 100 to an ideal reaction and then subtracting penalty points for non-ideal conditions. These penalty points take into account all possible disadvantages of specific reagents, setups, and technologies, including the risk for the operator and a possible negative impact on the environment. Usually scores of >75 are excellent; >50 are acceptable, and scores <50 are inadequate. The EcoScale score calculated for RCM made in 4-MeTHP was 67 and for DCE only 51 (this calculation does not include the obvious operational risk of using hot 10% sodium hydroxide solution and the environmental/legal factors related to use of a chlorinated solvent).

Conclusions

A recently introduced solvent, 4-MeTHP has been tested in olefin metathesis and compared with other solvents, such as EtOAc, TBME, THF, and anisole. Interestingly, due to lower tendency to form peroxides, even aged samples of 4-MeTHP gave good results in olefin metathesis, which was in contrast to THF. Next, a number of metathesis reactions were tested in 4-MeTHP, including standard model dienes and more complex substrates, such as analogues of biologically active compounds and APIs. To place this solvent in a context of pharmaceutical research, the larger-scale synthesis of sildenafil derivative 42, an analogue of the popular drug Viagra, was conducted. Despite the reaction setup being made entirely in air, the reaction at a 33 g scale in 4-MeTHP gave an 88% yield in addition to allowing for two times reduction of the catalyst loading while ruthenium content in the crude unpurified product was as low as 88 ppm. The same reaction conducted in the previously developed conditions using DCE, a solvent typically utilized in metathesis, gave a comparable final yield (79%), but due to different solving properties exhibited by the chlorinated solvent, additional purification steps were required in order to reduce high content of ruthenium in the crude product. Based on these results, we believe that 4-MeTHP can find numerous applications as an alternative solvent for olefin metathesis in academic research and in the pharmaceutical industry.