Decarboxylation of cinnamic acids using a ruthenium sawhorse

Abstract

The ruthenium sawhorse has proven effective in the conversion of trans-cinnamic acid, and substituted trans-cinnamic acids, giving an effective source of biobased styrene and styrene analogues. The reaction is especially versatile, as it achieves product without utilising co-reagents. However, the optimum conditions and substrate scope of the reaction remain unexplored. This report covers the decarboxylation of a series of cinnamates with different structures. Apparent activation energies ranging from 66 to 142 kJ mol⁻¹ were uncovered.

The development of industrially useful polymers from non-petroleum sources is a topic of seemingly everlasting interest (Biermann et al. 2000, 2011; Bozell et al. 2000). The conversion of bio-based components into materials that are identical to those already used in industry, a direct drop-in replacement strategy, may someday lead to a viable biorefinery (Yao and Tang 2013). Removing oxygen from biological materials is the key to this goal, especially for industrial use of natural oils. Currently, many different technologies are competing in this transformation (Dawes et al. 2015; Foglia and Barr 1976; Gooßen and Rodríguez 2004; Gooßen et al. 2009; Kraus 2014; Kraus and Riley 2012; Liu et al. 2014; Maki-Arvela et al. 2006; Miller, Nelson, and Byrne 1991, 1993; Miranda et al. 2012; Murzin et al. 2009; Popov and Kumar 2015; Snåre et al. 2006, 2008; Stern and Hillion 1985; Tanaka, Shimizu, and Yamamoto 1997; van der Klis et al. 2011, 2012). One of the most promising utilises only heat and a small amount of ruthenium catalyst, and does not require a co-reagent, such as an anhydride, peroxydisulphate (Murray, Doll, et al. 2014; Murray, Walter, et al. 2014) or the environmentally dreaded lead (Kochi, Sheldon, and Lande 1969). The initial report (Murray, Walter, et al. 2014) described the process on several different natural oil-based substrates including 10-undecenoic acid, 9-cis-octadecenoic acid (oleic acid) and cinnamic acid (Scheme 1); yields ranged from 20 to74%, with the best conversion being from cinnamic acid to styrene.

This high-yield reaction is of interest in the quest for a biobased production route to polystyrene or acrylonitrile butadiene styrene thermoplastics. Natural trans-cinnamic acid is commonly obtained from oil of cinnamon (Klejdus and Kováčik 2016), but also in other plants of commercial interest such as shea butter (Maranz, Wiesman, and Garti 2003), basil, storax and cocoa (Hoskins 1984). It has been widely studied as an anti-bacterial, nutraceutical and cosmetic material (Edris 2007). However, there are no significant reports of the industrial uses of this versatile precursor. This report focuses on the catalytic decarboxylation of cinnamic acid, using a systematic approach to elucidate parameters, such as the apparent activation energy of decarboxylation. Additionally, the rate of decarboxylation is observed in substituted cinnamic acid analogs.

The synthesis of polymeric ruthenium compounds of ruthenium is dated back to the 1960s (Crooks et al. 1969). Starting from triruthenium dodecacarbonyl, the orange/yellow [Ru(CO)₂(CH₃COO)]_n and analogous compounds were produced by the addition of carboxylic acid. The addition of ligands, such as pyridines, has allowed the isolation and production of high-quality crystals of the dinuclear compounds which have been termed, ‘sawhorses’ (Kepert et al. 2000). As well as in the work described earlier in this paper (Murray, Doll, et al. 2014; Murray, Walter, et al. 2014), triruthenium dodecacarbonyl and the sawhorse compounds, sometimes with other added ligands, have all been effective precatalysts in isomerisation (Castiglioni et al. 1976; Caulton et al. 1976; Salvini, Frediani, and Piacenti 2000; Sherlock et al. 1988; Sivaramakrishna et al. 2008; Sleeter 1998; Yang et al. 2016) and transvinylation (Ziriakus et al. 2013) reactions. The synthesis of the precatalyst used in this work, 2-ethylhexyl ruthenium sawhorse, [Ru(CO)₂ (CH₃(CH₂)₃C(CH₂CH₃)HCOO)]_n, was performed analogously to the literature methods (Scheme 2) where sawhorses containing propanoic and acetic acid groups were prepared (Crooks et al. 1969) from triruthenium dodecacarbonyl (Ru₃(CO)₁₂ (99%, Acros Organics, Pittsburgh, PA) or (99%, Strem Chemical, Newburyport, MA). Catalysts of this type have been previously synthesised in the literature by this method. (Doll et al. 2017; Kepert et al. 2000; Murray 1991; Yang et al. 2016; Ziriakus et al. 2013).

Although many different saturated carboxylic acids could have been chosen, 2-ethylhexanoic was used as the ligand for several reasons. First, 2-ethylhexanoic acid is widely available, being derived from a hydroformylation of propene followed by condensation and oxidation. Next, it is soluble in organic solvents which allowed homogeneous catalysis to take place. Finally, its branched structure has proven useful in the use of agriculturally based industrial products (Cermak, Brandon, and Isbell 2006; Doll, Sharma, and Erhan 2007). Because 2-ethylhexanoic acid (99%, Sigma-Aldrich, St. Louis, MO) is more difficult to remove than the smaller acids used in the literature, only a single equivalent compared to ruthenium was used, toluene solvent (99.8%, Fisher Scientific, Fairlawn, NJ) was utilised and a nitrogen atmosphere was maintained during the reaction using Schlenk techniques. The sawhorse was characterised by IR spectroscopy (Thermonicolet, Madison, WI,) Nexus 470 spectrophotometer equipped with a Smart Orbit accessory and diamond plate, 32 scans. Peaks in the spectrum are in the expected 2870–2965 cm⁻¹ region, and definitively show the signs of successful sawhorse preparation from the carbonyl signals, 2064, 1994 1959 cm⁻¹ (Figure 1).

Figure 1. The infrared (IR) spectra of the 2-ethylhexyl ruthenium sawhorse compound.

The protocol selected for this study involved using glass pressure vessels which was done for several reasons. First, the clear glass allowed direct observation of the solution during reaction. Next, the pressure vessels allow sampling during the reaction. Finally, pressure build-up could be safely monitored in this experimental set-up. In this study, reactions typically used ~30 mg of the sawhorse precatalyst which was weighed into a 3 oz glass pressure vessel, (Lab Crest-Andrews Glass, Vineland, NJ). Trans-cinnamic acid (99%, Sigma-Aldrich, St. Louis, MO), ~1 g, was dissolved in ~9 g of 2-ethylhexanoic acid (99%, Sigma-Aldrich, St. Louis, MO), and heated to 65 °C to ensure dissolution. The mixture was transferred to the pressure vessel containing the precatalyst and the vessel was purged with argon. The reactor was then pressurised to 25 lbs in⁻² with argon, magnetic stirring was set to 200 RPM and heat was introduced using a silicone oil bath. Aliquots were taken at 1 h intervals using Vici (Baton Rouge, LA) precision sampling pressure lock syringes and analysed by gas chromatography (two different Agilent 7890 (Santa Clara, CA) gas chromatographs were employed. For product identification, a 5975 mass spectrometry detector was used, whereas flame ionisation detection was used for quantitation. Identical columns, Agilent/J&W (Santa Clara, CA) DB35-MS, 30 m x 320 um, 0.25 um film thickness; sample preparation, injections of 1 uL of samples made by diluting ~10 uL of reaction solution in 1 mL of acetone were done by autosampler using a 50:1 split ratio; and temperature/injection programmes, starting at 40 °C for 3 min, then was raised 10 °C min⁻¹ to 190 °C and held for 5 min, then raised 25 °C min⁻¹ to 340 °C; were used in each instrument. A pressure build-up of 35–55 lbs in⁻² was observed for reactions with significant conversion.

The reaction progress was followed over 4 h at temperatures of 150, 175 and 200 °C. As expected, the observed conversion increased over time, and was more rapid at higher temperatures (Figure 2). The conversion was also studied with varied catalyst amounts, where conversion could be increased significantly by increasing catalyst from 1wt.% to 3wt.%, but further increase to 5 wt.% only gave further reaction nearly within experimental deviation of the 3 wt.% results (Figure 3). Therefore, the 3 wt.% was used in the substrate studies.

Figure 2. The conversion of trans-cinnamic acid by the 2-ethylhexyl ruthenium sawhorse, 3% wt., at 150, 175 and 200 °C.

Figure 3. The conversion of trans-cinnamic acid by various concentrations of the 2-ethylhyxyl ruthenium sawhorse, at 175 °C, error bars were calculated as standard deviations with n = 3.

In order to test the substituent effects, cinnamic acid analogues were employed, where methyl, methoxy and trifluoromethyl groups, each on the para position of the cinnamic acid, were called on to give a variety of targets (Figure 4). The methyl group, had a limited effect, yielding mildly increased conversion. The trans-4-methoxy cinnamic acid (99%, Sigma-Aldrich, St. Louis, MO) system converted much more rapidly than the other cinnamates. Further, trans-4-trifluoromethyl cinnamic acid (99%, Sigma-Aldrich, St. Louis, MO) system displayed almost no conversion.

Figure 4. The conversion of substrated cinnamic acids and other substrates by 2-ethylhexyl ruthenium sawhorse, 3% wt., at 175 °C over 3 h, error bars were calculated as standard deviations with n = 3.

Scheme 1. Decarboxylation of cinnamic acids with ruthenium sawhorse catalyst.

Scheme 2. The synthesis of ruthenium sawhorse compounds. The carboxylic acids employed here include propanoic acid and 2-ethylhexanoic acid.

A few other interesting but expected facts were observed during this study. First, reactions with the 25 lbs in⁻² of added argon gave results within error of non-pressurised reactions ran in the same reaction vessel. Next, there was significant polymerisation of the 4-methoxy styrene, of up to 17% of the GC-FID area at the highest conversions. To further expand the scope of this reaction, 2-furoic acid (99%, Sigma-Aldrich, St. Louis, MO) was also decarboxylated. This substrate, while not containing a linear olefin, does have a double bond in conjugation with the carboxylic acid function. It decarboxylated rapidly (Figure 4), but reaction was not observed in benzoic acid (98%, Sigma-Aldrich, St. Louis, MO) which did not decarboxylate at all (Figure 4). Levulinic acid (98%, Sigma-Aldrich, St. Louis, MO) also did not decarboxylate, indicating the importance of a double bond in the appropriate proximity to the carboxyl. Attempts to use triosmium dodecacarbonyl (99%, Strem, Newburyport, MA) in an analogous system fashion yielded only 7% conversion of cinnamic acid to styrene.

In order to check for potential side reactions such as decarbonylation, the aldehyde, octenal (99%, Lancaster, Ward Hill, MA), was subjected to the reaction conditions, and the only reaction observed was condensation. Aromatic aldehydes such as trans-cinnamaldehyde (99%, Sigma-Aldrich, St. Louis, MO) and benzaldehyde (99%, Sigma-Aldrich, St. Louis, MO) also did not react significantly. In other words, the sawhorse is not an effective decarbonylation catalyst under these conditions.

The scope of this reaction system has been further explored. Nitrogen containing substrates were utilised, including trans-4-dimethylamino cinnamic acid (99%, Sigma-Aldrich, St. Louis, MO) and nicotinic acid (98%, Sigma-Aldrich, St. Louis, MO), the latter being decarboxylated in a stainless steel reactor. In the dimethylamino case, insolubility prevented the reaction from occurring, again demonstrating that this is a truly homogeneous reaction. Nicotinic acid was decarboxylated in 29% yield. As a final comparison, cis-9-octadecenoic acid (oleic acid, (Nu-Check Prep, Elsyian, MN > 99%), trans-9-octadecenoic acid (elaidic acid > 99%, Nu-Check Prep, Elsyian, MN) and cis-6-octadecenoic acid (petroselinic acid (>99%, Nu-Check Prep, Elsyian, MN) were found to have identical reactivity using this catalyst in a stainless steel reactor.

Changing the reaction solvent from 2-ethylhexanoic acid to tetraglyme (99%, Sigma-Aldrich, St. Louis, MO) diminished the reaction conversion by a factor of 12. Mixtures of the two solvents gave results intermediate between the two solvent systems. In other words, ruthenium is the best choice for performing this conversion and it is best used in 2-ethylhexanoic acid.

In order to quantify these results, the initial rate of the first hour of each substrate was used. This method has been shown to give the most accurate value for pseudo-first order rate constants (Woodward and Mesrobian 1953), since at the low conversions, the substrate concentration can be considered a constant. The maximum value observed in this work was 0.000447 s⁻¹, for the 4-methoxy substrate at 200 °C. For the three cinnamic substrates where the applicable data were available at a variety of temperatures, apparent activation energies can be calculated using an Arrhenius type calculation (Table 1). The values, consistent with the limited available literature (Ortuño, Dereli, and Cramer 2016; Popov and Kumar 2015), show that ordinary cinnamic acid is more difficult to decarboxylate than the 4-methoxy substituted substrate could be decarboxylated with less than half of the energy.

Table 1. A calculation of apparent activation energies using the Arrhenius equation of rates obtained from the initial rate method, of the decarboxylation of trans-cinnamic acid and 4-substituted trans-cinnamic acids with the addition of 3% wt., 2-ethylhexyl ruthenium sawhorse.

Substrate	Apparent activation energy calculated from initial rates of decarboxylation reactions (kJ mol ^−1)
Cinnamic acid	142
4-methyl cinnamic acid	115
4-methoxy cinnamic acid	66

Conclusion

Utilisation of biobased oils to produce high-quality polymers is important. However, even some of the most interesting reports in this area remain mysterious. This report clears some of the obscurity in the sawhorse-based decarboxylation of natural aromatic oils, and also highlights the need for an understanding of the interesting chemistry behind these bio-refinery relevant conversions.

Funding

This work was supported by the Agricultural Research Service [grant number 41000-170-00D].

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Kenneth M. Doll, PhD, is a lead scientist in the Bio-Oils research unit at the National Center for Agricultural Utilization Research. His research focuses on the use of biologically derived oils in industrial applications, specifically focusing on chemical modification of oils through catalytic reactions. He has published 70 articles in journals such as Green Chemistry, Industrial and Engineering Chemistry Research, and the International Journal of Sustainable Engineering.

Erin L. Walter, BS, is a physical science technician in the Bio-oils Research Unit at the National Center for Agricultural Utilization Research in Peoria, IL. The research project she is currently involved in identifies and develops new uses for plant based oils, most commonly soybean oil. Previous publications have appeared in Journal of the American Oil Chemists’ Society, ACS Catalysis, Chemical Engineering Communications, Industrial Crops and Products, and others.

Rex E. Murray, PhD, is a research leader of the Bio-Oils research unit at the National Center for Agricultural Utilization Research. He leads a team of researchers on several different projects including: Industrial Monomers and Polymers from Plant Oils, and Value added Bio Oil Products and Processes. He has published articles in Industrial and Engineering Chemistry Research, ACS Sustainable Chemistry and Engineering, and the Journal of the American Oil Chemists Society.

Acknowledgements

This work was a part of the in-house research of the Agricultural Research Service of the United States Department of Agriculture.