Catalytic oxidative dehydrogenation of malic acid to oxaloacetic acid

Asmaa Drifa, Antonio Pinedaa, Didier Morvanb, Virginie Belliere-Bacab, Karine De Oliveira Vigiera and François Jérôme*a
aInstitut de Chimie des Milieux et Matériaux de Poitiers, CNRS-Université de Poitiers, 1 rue Marcel Doré, 86073 Poitiers, France. E-mail: francois.jerome@univ-poitiers.fr
bADISSEO, CINACHEM, 85 Rue des Frères Perret, 69192 Saint Fons, France

Received 28th May 2019 , Accepted 24th July 2019

First published on 24th July 2019


Here we report the oxidative dehydrogenation of malic acid to oxaloacetic acid, a key precursor in the fabrication of amino acids, over Pt–Bi/C catalysts. Under optimized conditions, we discovered that OAA was selectively produced with up to 60% conversion (i.e. 60% yield). The recurrent unwanted decarboxylation of OAA to pyruvic acid was circumvented by successfully conducting the catalytic reaction at 25 °C. A comparison with the classical Fenton oxidation reaction is discussed.


Oxaloacetic acid (OAA) is a strategic chemical platform to synthesize various key amino acids such as asparagine, methionine, lysine and threonine, which are naturally synthesized by living organisms1–4 or currently by chemists through the Strecker synthesis (Fig. 1).5 With the exponential growth of world population,6 the market for amino acids is expected to significantly grow in the coming years. In this context, OAA has become an important industrial target.
image file: c9gc01768b-f1.tif
Fig. 1 OAA: a chemical platform for the synthesis of amino acids.

In nature, OAA is synthesized through a biochemical route, by the oxidation of L-malate by malate dehydrogenase.7 Although highly selective, biochemical routes suffer from low space–time yields, a serious limitation with the growing demand of our society for amino acids.8 Theoretically, OAA can be produced by catalytic oxidation or dehydrogenation of malic acid, a chemical currently synthesized from crude oil by catalytic hydration of maleic anhydride.8 One should note that the recent progress in biomass processing has opened an alternative route to produce maleic acid from furfural, a low cost renewable building block, thus potentially opening a bio-based route to OAA.9

From a scientific point of view, the catalytic conversion of malic acid to OAA is a difficult reaction, mainly because OAA is thermally unstable and quickly decarboxylates to pyruvic acid or malonic acid.10–12 To date, there is no catalytic route capable of selectively synthesizing OAA from malic acid. The latter is always considered an intermediate chemical, often produced with low yields and low selectivity, in the total catalytic wet oxidation of waste, for instance.13 To date, the development of catalytic systems capable of selectively stopping the oxidation of malic acid to OAA remains a challenging scientific task, which we address in this work. One of the keys to success is the development of catalytic systems that are active at low temperatures (<40 °C) in order to limit the unwanted decarboxylation of OAA.

Using the Spartan software, we found that the oxidation of malic acid to OAA is a thermodynamically favourable reaction with a negative Gibbs free energy (ΔG) of −75 kJ mol−1. This was confirmed by our first catalytic experiments using the classical Fenton reaction. At only 2 °C, in the presence of 5 mol% of Fe(SO4)2 and 1 eq. of H2O2, malic acid (2 wt% in water) was spontaneously converted, but OAA was obtained with a very low yield of 15% at pH 2. Other detected products were mainly oxalic, malonic and formic acids and the determination of the carbon mass balance also suggested the formation of gaseous products, presumably CO and/or CO2. It is noteworthy that at a pH higher than 2, the selectivity towards OAA was negligible. To better control the selectivity towards OAA, the catalytic reaction was deliberately slowed down by the dropwise addition of diluted portions of H2O2 (Fig. 2). The kinetic profile of the reaction showed that, under these conditions, OAA was formed with a maximum yield of 29% at 65% conversion, which corresponds to an OAA selectivity of only 45% (Fig. 2). Other products formed were oxalic acid (21% yield) and malonic acid (5% yield). At higher conversions, the yield of OAA decreased gradually to 16%, with oxalic acid and malonic acid now being produced in 58% and 12% yield, respectively, at 93% conversion. All our attempts to increase the OAA yield and selectivity failed and oxalic acid was always the major product formed.


image file: c9gc01768b-f2.tif
Fig. 2 Catalytic oxidation of malic acid in the presence of Fe(SO4)2, 1 eq. H2O2, 2 °C.

Low reaction yields are often managed quite easily on a large scale, for instance by recirculating the unreacted starting materials into the reactor; however, the lack of selectivity in a catalytic reaction is much more problematic since it requires a valorization of the co-product and, often, a complex downstream process. Considering the instability of OAA, it is clear that a balance between yield and selectivity should be found. Ideally, being able to maximize the yield of OAA while keeping the selectivity close to 100% is a major objective.

In this context, we next investigated the catalytic dehydrogenation of malic acid to OAA. In contrast to the oxidation reaction, our calculations revealed that the dehydrogenation of malic to OAA is thermodynamically unfavourable, with a ΔG value of +33 kJ mol−1 (vs. −75 kJ mol−1 for oxidation), indicating that in situ produced hydrogen should be rapidly converted in order to shift the reaction to the formation of OAA. Upon looking at the different catalysts previously reported in the oxidative dehydrogenation of alcohols,14 Pt–Bi/C catalysts attracted our attention.15,16 As previously described, in such catalytic systems, Pt ensures the dehydrogenation of alcohol, while Bi2O3 reacts with the adsorbed hydrogen on Pt to produce water, a reaction accompanied by the reduction of Bi2O3. The catalytic system is then regenerated (re-oxidation of Bi) by bubbling oxygen or air into the solution. Conceptually, it occurred to us that the in situ conversion of H2 to H2O should be a good means to shift the dehydrogenation of malic acid to OAA.

Various Pt–Bi/C catalysts, differing in the Pt/Bi molar ratio, were prepared according to the procedure previously reported by Besson (ESI).16 These catalysts are known, and the characterization of the as-synthesized Pt–Bi/C catalysts is provided in the ESI (Fig. S2, S4 and S5). In a typical procedure, an aqueous solution of malic acid (4 wt%) was heated at 50 °C in the presence of 2 wt% of a Pt–Bi/C catalyst. As described in the literature, 2 eq. of NaOH (pH = 8) were added in order to convert the –CO2H group into its carboxylate form, a known strategy to avoid poisoning of the catalyst. Reactions were first conducted under an air flow of 15 mL min−1. Kinetic profiles are shown in Fig. 3. All tested Pt–Bi/C were capable of converting malic acid into OAA, although with notable differences in terms of OAA yield. The maximum yield of OAA (∼40%) was obtained in the presence of Pt20–Bi20/C and Pt20–Bi10/C. However, independent of the catalytic system, the yield of OAA reached a maximum and then decreased, which is due to the thermal decarboxylation of OAA to pyruvic acid. Pyruvic acid is the only detected by-product but its formation occurred as soon as the reaction began, making the reaction selectivity difficult to control.


image file: c9gc01768b-f3.tif
Fig. 3 Catalytic oxidation of an aqueous solution of malic acid (4 wt%) in the presence of various Pt–Bi/C catalysts (0.1 g), pH = 8, air flow = 15 mL min−1, 50 °C.

As can be pointed out from Fig. 3, the decarboxylation of OAA occurred at a nearly similar rate, independent of the Pt–Bi/C catalysts, suggesting that the decarboxylation of OAA is not a catalytic reaction and spontaneously took place at 50 °C. The thermodynamic calculations we performed indeed revealed that the decarboxylation of OAA to pyruvic is highly favourable with a calculated Gibbs free energy of −249 kJ mol−1. To further support this claim, OAA was treated in an alkaline solution (pH = 8) heated at 50 °C without the addition of Pt–Bi/C. In line with the above assumption, under these conditions, OAA was quickly (0.9 mol% per min) and selectively decarboxylated to pyruvic acid at 50 °C. No product other than pyruvic acid was detected, confirming that the decarboxylation of OAA to pyruvic acid is mostly a “thermal” reaction. Very interestingly, when the temperature of the reaction was brought down to 25 °C, OAA was found to be stable and no further decarboxylation occurred. Catalytic experiments were then performed again at 25 °C instead of 50 °C. Kinetic profiles are shown in Fig. 4. Under these conditions, only Pt20–Bi2/C was active, affording OAA in 25% yield. As expected, at 25 °C, the side decarboxylation of OAA to pyruvic acid was nearly completely inhibited, which is also highlighted in the kinetic profile by a very slow decrease of the OAA yield with an extended reaction time. Although this yield is lower than the one obtained at 50 °C (40%), in this case, the reaction was fully selective towards OAA.


image file: c9gc01768b-f4.tif
Fig. 4 Catalytic oxidation of an aqueous solution of malic acid (4 wt%) in the presence of various Pt–Bi/C catalysts (0.1 g), pH = 8, air flow = 15 mL min−1, 25 °C.

Air flow plays an important role in the catalytic reaction by ensuring the re-oxidation of Bi and thus the removal of hydrogen from Pt sites. At low air flow (<15 mL min−1), the removal of H2 from Pt sites is quite slow and the thermal decarboxylation of OAA to pyruvic acid is the major reaction (Fig. 5). In contrast, at a high air flow (>15 mL min−1), it is well known that the Pt surface is partially covered by chemisorbed oxygen which poisons the catalytic sites.15 In this case, trace amounts of oxalic acid were detected, supporting the in situ formation of PtOx species. Indeed, based on the PtOx/Pt and malic/oxalic redox potential (EPtOx/Pt = 1.00 V; Emalic/oxalic = −0.102 V), PtOx can be reduced to Pt(0) by malic acid which is concomitantly oxidized to oxalic acid (similar to the Fenton process described above). Clearly a balance should be found. In perfect agreement, Fig. 5 shows that a maximum yield of OAA of 25% was obtained at an air flow of 15 mL min−1.


image file: c9gc01768b-f5.tif
Fig. 5 Effect of air flow on the catalytic oxidation of an aqueous solution of malic acid (4 wt%) in the presence of Pt20–Bi2/C (0.1 g), pH = 8, 25 °C.

By comparing data from previous studies with our study, it seems clear that the partial coverage of the Pt surface with oxygen is a limiting factor in the catalytic synthesis of OAA. In order to improve the OAA yield, the catalytic reaction was performed using methanol as a solvent instead of water. Methanol can be oxidized (EMeOH/CO2 = 0.02 V) over PtOx and thus used as a sacrificial agent, which could be a great means to quickly regenerate in situ active Pt.17 As expected, when the reaction was conducted at 25 °C in methanol, the OAA yield was increased to 40% and even 60% at an air flow of 15 and 50 mL min−1, respectively, while the selectivity towards OAA remained close to 100% (Fig. 6). A further increase of the air flow to 75 and 200 mL min−1 did not significantly impact the OAA yield (60%), indicating that the coverage of the platinum surface with oxygen was not a problem anymore in methanol (Fig. 6). This was further confirmed by XPS analysis (Fig. S3). While before the reaction metallic Pt and PtOx were clearly detected, only metallic Pt was found after the catalytic reaction. Our attempts to obtain a higher OAA yield unfortunately failed, suggesting that the reaction has reached a thermodynamic equilibrium. To support this claim, once the OAA yield plateaued at 60%, the used Pt–Bi/C catalyst was filtered off and freshly prepared Pt–Bi/C was added to the reaction mixture. No change in the OAA yield and OAA/malic acid ratio was observed, confirming that the reaction has reached thermodynamic equilibrium, presumably due to a competition between hydrogen transfer from platinum to bismuth oxide and re-hydrogenation reactions. The reversibility of the dehydrogenation reaction was previously shown to be feasible in many reactions, for instance in the isomerization of isosorbide to isomannide and isoiodide.18


image file: c9gc01768b-f6.tif
Fig. 6 Catalytic dehydrogenation of a methanolic solution of malic acid (4 wt%) in the presence of Pt20–Bi4/C catalysts (0.1 g), pH = 8, 25 °C.

Under these conditions, malic acid was selectively converted into OAA and no pyruvic or oxalic acid was detected. The high selectivity of the reaction towards OAA was further confirmed by running an experiment in CD3OD and analyzing the crude reaction medium, at the plateau, by using 1H and 13C NMR spectra (Fig. S7 and S8). It is noteworthy that NMR investigations showed that, under basic conditions, OAA is in its enol form. Methanol was titrated by gas chromatography. At an air flow of 50 mL min−1, only 1.5% of methanol was converted, presumably to CO2 since no trace of formic acid was detected, a claim in line with the redox potential of the system. Methanol can also be used as an additive in water up to 20% (i.e. MeOH/H2O: 20/80) without impacting the OAA yield. However, with a methanol content lower than 20%, oxalic acid was detected (5% yield), highlighting a competitive reduction of PtOx with malic acid, as discussed above. A general scheme of the catalytic cycles is presented in Fig. 7 to summarize the different reactions taking place on the catalyst surface and also to clearly distinguish the oxidative dehydrogenation reactions from the oxidation reactions.


image file: c9gc01768b-f7.tif
Fig. 7 Summary of the different reactions taking place on the catalyst surface.

Although with methanol the side formation of CO2 needs to be managed, the recovery of a clean mixture composed of unreacted malic acid and OAA in methanol is of high interest with regard to the manufacturing chain of methionine for instance. Indeed, according to the findings described in ref. 19, this methanolic mixture of malic acid and OAA can be directly converted into an intermediate of methionine which precipitates from methanol, providing a means to recirculate unreacted malic acid and methanol.

The Pt20–Bi4/C catalyst was found to be very robust under our optimized experimental conditions. After the first catalytic run, the catalyst was filtered, washed with water, dried at 60 °C and reused without any further purification. As shown in Fig. 8, the Pt20–Bi4/C catalyst was successfully recycled at least eight times without any decrease in OAA yield. No appreciable leaching of Pt and Bi in the solution was detected by ICP, further confirming the robustness of the catalytic procedure.


image file: c9gc01768b-f8.tif
Fig. 8 Recycling of the Pt20–Bi4/C catalyst (4 wt% malic acid in methanol, 4 h, 0.1 g of catalyst, 25 °C).

Conclusions

In this work, we report the catalytic conversion of malic acid into OAA, a strategic building block in the fabrication of amino acids. Although the classical oxidation reaction is thermodynamically favourable, it affords OAA in low yield (<29%) and with low selectivity (<45%), due to the quasi spontaneous oxidation of malic to oxalic acid, a result in line with the redox potential of the system. To circumvent this problem, we showed that the oxidative dehydrogenation of malic acid over Pt–Bi–C/catalysts was much more selective towards OAA. Although this reaction is clearly not thermodynamically favourable, the formation of OAA was driven by the in situ conversion of H2 to H2O over the Pt–Bi/C catalyst. The unwanted decarboxylation of OAA to pyruvic acid was avoided by successfully conducting the catalytic reaction at 25 °C. The coverage of the Pt surface by chemisorbed oxygen can be circumvented by adding methanol as a sacrificial agent into the reaction medium, yielding a mixture composed of unreacted malic acid, OAA and aqueous methanol, a mixture that can be directly used in the manufacturing chain of methionine for instance. Under optimized conditions, OAA was obtained with 100% selectivity at up to 60% conversion (i.e., 60% yield), opening an alternative pathway to the classical biocatalytic routes. The coupling of this catalytic process with the manufacturing process of amino acids is the topic of current investigations.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors are grateful to the CNRS, the University of Poitiers and the region Nouvelle-Aquitaine for funding this work. The authors are also grateful to ADISSEO for the funding of the postdoc grants to AD and AP.

Notes and references

  1. T. T. Lillich and G. H. Elkan, Can. J. Microbiol., 1971, 17, 683–688 CrossRef CAS PubMed.
  2. L. Gou, J. Lee, H. Hao, Y. D. Park, Y. Zhan and Z. R. Lü, Int. J. Biol. Macromol., 2017, 101, 59–66 CrossRef CAS PubMed.
  3. D. L. Nelson and M. M. Cox, in Principles of Biochemistry, W. H. Freeman, New York, 4th edn, 2005, ISBN: 0-7167-4339-6 Search PubMed.
  4. M. D. Hatch and H. W. Heldt, Anal. Biochem., 1985, 145, 393–397 CrossRef CAS PubMed.
  5. H. Groger, Chem. Rev., 2003, 103(8), 2795–2827 CrossRef PubMed.
  6. https://www.un.org/en/sections/issues-depth/population. Consulted on May 21st, 2019.
  7. M. V. Panchenko and A. D. Vinogradov, FEBS Lett., 1991, 286, 76–78 CrossRef CAS PubMed.
  8. W. Deng, Y. Wang, S. Zhang, K. M. Gupta, M. J. Hülsey, H. Asakura, L. Liu, Y. Han, E. M. Karp, G. T. Beckham, P. J. Dyson, J. Jiang, T. Tanaka, Y. Wang and N. Yan, Proc. Natl. Acad. Sci. U. S. A., 2018, 115(20), 5093–5098 CrossRef CAS PubMed.
  9. (a) J. Lan, Z. Chen, J. Lina and G. Yin, Green Chem., 2014, 16, 4351–4358 RSC; (b) N. Alonso-Fagúndez, M. L. Granados, R. Mariscal and M. Ojeda, ChemSusChem, 2012, 5(10), 1984–1990 CrossRef PubMed; (c) X. Li, J. Ko and Y. Zhang, ChemSusChem, 2018, 11(3), 612–618 CrossRef CAS PubMed; (d) N. Araji, D. D. Madjinza, G. Chatel, A. Moores, F. Jérôme and K. De Oliveira Vigier, Green Chem., 2017, 19, 98–101 RSC.
  10. D. L. Rohlfing, Arch. Biochem. Biophys., 1967, 118, 468–474 CrossRef CAS PubMed.
  11. C. S. Tsai, Can. J. Chem., 1967, 46, 873–880 CrossRef.
  12. E. Gelles, J. Chem. Soc., 1956, 4736–4739 RSC.
  13. L. S. Arvanitoyannis, A. Kassaveti and D. Ladas, in Waste management for the food industries, ed. L. S. Arvanitoyannis, Elsevier Academic Press, Amsterdam, 2008, pp. 345–410 Search PubMed.
  14. (a) R. Anderson, K. Griffin, P. Johnson and P. L. Alsters, Adv. Synth. Catal., 2003, 345(4), 517–523 CrossRef CAS; (b) N. Chodankar, in Industrial Catalytic Processes for Fine and Specialty Chemicals, ed. S. S. Joshi and V. V. Ranade, 2016, Elsevier, Amsterdam, pp. 509–594 Search PubMed; (c) T. Mallat and A. Baiker, Chem. Rev., 2004, 104(6), 3037–3058 CrossRef CAS PubMed; (d) M. Besson, P. Gallezot and C. Pinel, Chem. Rev., 2014, 114(3), 1827–1870 CrossRef CAS PubMed; (e) M. Sankar, N. Dimitratos, P. J. Miedziak, P. P. Wells, C. J. Kiely and G. J. Hutchings, Chem. Soc. Rev., 2012, 41, 8099–8139 RSC; (f) C. Della Pina, E. Falletta and M. Rossi, Chem. Soc. Rev., 2012, 41, 350–369 RSC.
  15. (a) M. Besson, F. Lahmer, P. Gallezot, P. Fuertes and G. Fleche, J. Catal., 1995, 152, 116–121 CrossRef CAS; (b) W. Hu, B. Lowry and A. Varma, Appl. Catal., B, 2011, 106(1–2), 123–132 CAS; (c) M. Besson, G. Flèche, P. Fuertes, P. Gallezot and F. Lahmer, Recl. Trav. Chim. Pays-Bas, 1996, 115(4), 217–221 CrossRef CAS; (d) W. Hu, D. Knight, B. Lowry and A. Varma, Ind. Eng. Chem. Res., 2010, 49(21), 10876–10882 CrossRef CAS; (e) T. Mallat and A. Baiker, Catal. Today, 1994, 19, 247–284 CrossRef CAS.
  16. A. B. Crozon, M. Besson and P. Gallezot, New J. Chem., 1998, 22, 269–273 RSC.
  17. D. Young Chung, K.-J. Lee and Y.-E. Sung, J. Phys. Chem. C, 2016, 120(17), 9028–9035 CrossRef.
  18. (a) J. Le Nôtre, J. van Haveren and D. S. van Es, ChemSusChem, 2013, 6(4), 693–700 CrossRef PubMed; (b) L. W. Wright and J. D. Brandner, J. Org. Chem., 1964, 29(10), 2979–2982 CrossRef CAS.
  19. R. Huet, V. Henryon, J. Monbrun, E. Airiau, S. Aubry and P. Rey, WO 2017/125673, 2017 Search PubMed.

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c9gc01768b

This journal is © The Royal Society of Chemistry 2019