First International Congress on Catalysis for Biorefineries (CatBior), Torremolinos-Mlaga, Spain, 2011.

Chemicals from biomass: Synthesis of lactic acid from glycerol and sorbitol catalyzed by NaOH under hydrothermal conditions
Jos R. Ochoa-Gmeza,b*, Camilo Ramrez-Lpeza, Silvia Gil-Roa, Beln Mestre-Madurgaa, Olga Gmez-Jimnez-Aberasturia, Jess Torrecilla-Soriaa, Mikel Belsuea a TECNALIA RESEARCH AND INNOVATION, ENERGY DIVISION, Department of Bioenergy, 01510 Miano, Spain. b Universidad Alfonso X el Sabio, Department of Industrial Technology, 28696 Villanueva de la Caada, Madrid, Spain *Corresponding author:jramon.ochoa@tecnalia.com; jrochoag@telefonica.net

Abstract
The synthesis of lactic acid from sorbitol under hydrothermal conditions using NaOH as catalyst at industrially suitable sorbitol concentrations is reported, and results are compared to those obtained from glycerol under the same conditions. The influence of temperature, NaOH/starting chemical molar ratio (MR), initial starting chemical concentration (SC) and reaction time on conversions and yields is studied. A 100% sorbitol conversion and a maximum 39.5% yield of lactic acid on a carbon basis are obtained at 280C, 50 min, 1.0 M SC and 2.0 MR. Starting from glycerol and under similar conditions a 100% conversion and a 84.5% yield are achieved. Temperature, MR and reaction time are the more influent variables. Glyceraldehyde was the only identified intermediate, while formic acid, acrylic acid, acetic acid, oxalic acid and sodium carbonate were identified as over-oxidation products by HPLC, all of them in very low yields with the exception of formic acid (16% yield at a MR of 4 and 280C) from sorbitol. Hydrogen was the only chemical detected in the gas phase other than gaseous water. A significant polyol fraction is converted into a complex mixture of chemicals structurally similar to those of bio-oil as shown by GC/MS. The higher the content of hydroxyl moieties in the polyol the higher the bio-oil-like fraction and the lower the lactic acid yield are. Several plausible conversion routes are proposed. Keywords: lactic acid; glycerol; sorbitol; alkaline catalysis; hydrothermal conversion.

Introduction
In recent years there is a growing interest in the chemical industry for developing new synthetic routes based on renewable raw materials. Glycerol and Sorbitol are two main candidates to be used as building blocks in chemical synthesis [1]. Lactic acid is one of the industrially valuable chemicals which can be obtained from them. Most of its world production is based on fermentation of glucose, process of low productivity and in which 1 ton of calcium sulfate per ton of lactic acid is generated accounting for up to 50% of the production cost [2]. Then, more efficient processes for lactic acid production are desired. The synthesis of lactic acid by hydrothermal conversion of biomass wastes has been reported by several research groups [3,4] although with low yields. A much better 53% yield is obtained starting from sucrose using Co(II), Ni(II), Cu(II) and Zn(II) catalysts [5]. Yuksel et al. have reported the synthesis of lactic acid by hydrothermal alkaline electrolysis of glycerol at low glycerol concentration (0.1 M) and 280C both in a batch reactor [6] and in a continuous flow one [7]. A maximum 34.7% lactic acid yield was reported. A much higher 90% yield is obtained by alkaline hydrothermal conversion of glycerol at

First International Congress on Catalysis for Biorefineries (CatBior), Torremolinos-Mlaga, Spain, 2011. 300C and 0.33 M glycerol concentration [8,9]. Similar results were reported by Ramrez-Lpez et al. [10] at 10-fold higher glycerol concentrations. Zhou et al. [11] have reported a 25.6% lactic acid yield on a carbon basis starting from manitol, a C6-polyol, under alkaline hydrothermal conditions at 300C. To our knowledge, few studies related to the alkaline hydrothermal conversion of C-6 polyols have carried out and none with sorbitol. In this report, the synthesis of lactic acid from sorbitol under hydrothermal conditions using NaOH as catalyst at industrially suitable sorbitol concentrations is investigated, and results are compared to those obtained from glycerol under the same conditions.

Experimental
Materials All chemicals were chemical pure reagent and were supplied by Sigma-Aldrich (Madrid, Spain) and Panreac (Madrid, Spain). Deionized water was used in all experiments. CH 4, CO, CO2 and H2, all of them 99.99% pure, were supplied by Air Liquide (Spain). Experimental set up and reaction procedure Reactions were carried out in a 400 mL stainless steel high-pressure reactor fitted with temperature and pressure measurement devices, a magnetic stirrer, a Watlow Band Heater ThinBand of 110 mm in diameter and 80 mm in width (Watlow Ibrica, Madrid, Spain) made of stainless steel with mica insulation able to work up to 540C, and an EZ-ZONE temperature controller (Watlow Ibrica, Madrid, Spain). 100 mL of an aqueous solution composed of either glycerol or sorbitol and sodium hydroxide at the concentrations and molar ratios desired were introduced into the reactor. The reaction mixture was heated to the selected temperature and kept under vigorous stirring for the desired time under autogeneous pressure ranging from 35 to 70 bar depending on mixture composition and temperature. Then, the reaction mixture was quickly cooled and an aliquot was taken for analysis. Analysis All chemicals were identified by HPLC with a Varian apparatus Model 920-LC, fitted with both a Diode Array UV detector Varian series 920 (210 nm and 280 nm) and a Refractive Index detector Varian 920-LC (for glycerol and sorbitol determination), and a reverse phase Synergi C12 Phenomenex column, using a column temperature of 35C and an aqueous solution of H3PO4 pH 2.2 as the mobile phase at a flow rate of 1 mL/min. Alternatively, a Hewlett Packard HPLC apparatus Model 1100, fitted with a Diode Array UV detector Agilent series 1100, a Synergi C18 column, a UV detector (210 nm) and a refractive index detector Agilent Series 1100 was also used, using a column temperature of 25C and an aqueous solution of H3PO4 pH 2.0 as the mobile phase at a flow rate of 1 mL/min. Yields are on a carbon basis. Glycerol and sorbitol conversions were calculated as a percentage of the initial moles reacted. Gas products were analyzed by Mass Spectrometry using an OmniStar GSD 301 C mass spectrometer (Pfeiffer Vacuum) previously calibrated with CH4, CO, H2 and CO2. Samples of the gaseous phase were taken by means of a Tedlar bag of 3 L at the end of reactions.

First International Congress on Catalysis for Biorefineries (CatBior), Torremolinos-Mlaga, Spain, 2011. Carbonate content was analyzed by titrimetry and carbon recovery in the reaction mixtures was determined by a Total Organic Carbon Analyzer (Shimadzu Co., Ltd., Model TOC-5050A). In all cases, recoveries were between 96 and 105%. Volatile compounds in liquid reaction mixture were qualitatively analyzed by GC/MS with an Agilent HP 6890 gas chromatograph and a HP 5973 mass spectrometer detector, using a 30 m x 0.25 mm I.D. x 0.25 m SLBTM 5ms column. Samples were adjusted at pH 6 with concentrated HCl and extracted with methyl isobutyl ketone prior analysis. Organic phase was directly injected in the split mode (1:59). Operating conditions were: injector, oven (temperature ramp), interphase and source temperatures were 200C, 40C (3 min) 7C/min up to 200C, 280C and 230C, respectively; carrier gas was He at 1 mL/min; acquisition mode was scan 27250 amu up to 1.5 min, and 33-300 amu up to 28 min; Running time was 28 min. Chemicals were identified with The National Institute of Standards and Technologies, NIST MS search 2.0 library.

Results and Discussion

The influence of reaction temperature and NaOH/polyol molar ratio (MR) is depicted in Figure 1.

Figure 1. Influence of temperature (T) and NaOH/polyol molar ratio on conversions (() sorbitol; () glycerol), lactic acid yields (() from sorbitol; () from glycerol), formic acid yields (() from sorbitol; () from glycerol) and carbonate yields (() from glycerol). Common reaction conditions: 90 min; initial polyol concentration: 1 M for sorbitol and 2.5 M for glycerol. Left plot: NaOH/polyol molar ratio: 2.2 for sorbitol and 1.1 for glycerol. Right plot: 280C. Lactic acid yields are very low below 210C despite that sorbitol conversions are as high as 8592% between 150C and 210C. Then, yields increase rapidly reaching a peak of 36.5% at 280C, temperature at which conversion is 100%. Yields of the other HPLC identified chemicals (formic acid, oxalic acid, acetic acid, acrylic acid glyceraldehyde and sodium carbonate) remain below 2% in the whole range of temperature studied. It is apparent that, up to 210C, sorbitol is mainly converted into intermediates with a number of carbon atoms higher than or equal to 3 and some of them are precursors of lactic acid because between 210C and 280C conversion increases only 8 points whereas yield increases 32 points. Above 280C lactic acid yield decreases slightly while yields of acetic acid and

First International Congress on Catalysis for Biorefineries (CatBior), Torremolinos-Mlaga, Spain, 2011. carbonate increase from 0.5% to 2%, meaning that thermal decomposition of intermediates and lactic acid are beginning to be predominant. Interestingly, the optimum temperature for lactic acid synthesis (280C) is independent of the starting polyol, although a much higher yield (84.5%) is obtained from glycerol than from sorbitol (36.5%). As both have a different number of carbon atoms, this probably means that they are first converted into a common intermediate which subsequently evolves to lactic acid. This key chemical may be glyceraldehyde or pyruvaldehyde [8,10]. As glycerol has half the carbon atoms that sorbitol the number of competing reactions of conversion is much lower starting from the former than from the latter, which explains the great difference in yields. Referring now to MR, conversions were 100% in all the range studied. Lactic acid yield increases with MR, reaching a peak of 39.5% around a MR of 2. This result is in agreement with conversion route of sorbitol depicted below in Scheme 1 in which 2 moles of lactic acid per mol of sorbitol are obtained according to the stoichiometry of the reaction. As NaOH acts as a catalyst, it should be used in catalytic amounts but, however, NaOH is neutralized by lactic acid and, consequently, it must be used in a stoichiometric amount with respect to this acid to allow the reaction to proceed to 100% conversion. Above a MR of 2, yield decreases because the OH- in excess favors the scission of sorbitol by different C-C bonds resulting in an increasing number of competing reactions. Furthermore, NaOH concentration increases with MR which may favor the reactions of decomposition of both lactic acid and intermediates, as it is illustrated by the remarkable increase in formic acid yield above a MR of 2.5. Further experiments showed that optimum reaction times were 50 min and 90 min from sorbitol and glycerol, respectively, for the initial glycerol concentration herein reported. The only gaseous product identified was H2. Then, according to the chemicals identified by HPLC above reported and the reaction mechanism proposed by Kishida et al. for the alkaline hydrothermal synthesis of lactic acid form glycerol [8], a plausible conversion route of sorbitol into lactic acid and over-oxidation products is depicted in Scheme 1.In the first step, sorbitol 1 is converted into the intermediate 2 by reaction with OH- through a concerted mechanism wherein a dehydrogenation reaction and a dehydration reaction occur simultaneously with regeneration of catalyst OH-. In the second step, 2 is converted into 3 by keto-enol tautomerization. Then, cleavage of 3 at C3-C4 bond by reverse aldol condensation leads to glyceraldehyde 4 (detected) and pyruvaldehyde 5 (no detected at 280 nm and a reaction time of 90 min). Next, glyceraldehyde 4 is dehydrated by reacting with OH- to yield 2hydroxypropenal 6 which by keto-enol tautomerization gives pyruvaldehyde 5 which finally yields lactic acid (sodium lactate) 9 in basic medium through an intramolecular Cannizzaro reaction. On the other hand, glyceraldehyde 4 may also undergo a reverse aldol condensation to give formaldehyde 7 and glycolaldehyde 8. The latter may also be converted into lactic acid 9 in a 28% yield on a carbon basis under alkaline hydrothermal conditions at 300C as reported by Kishida et al. [12]. However, neither glycolaldehyde nor erythrose have been identified in the conversion of sorbitol herein reported, perhaps because both chemicals undergo a very fast conversion under our stringent experimental conditions. In fact, Kishida et al [12] only were able to identify erythrose at a reaction time as low as 1 min, and Yuksel et al. [7] detected glycolaldehyde in the hydrothermal electrolysis of

First International Congress on Catalysis for Biorefineries (CatBior), Torremolinos-Mlaga, Spain, 2011. glycerol, a C-3 polyol, at reaction times similar to the ones used herein but with NaOH concentrations 0.1M, much lower than the ones used in this work. Nevertheless, we detected and quantified oxalic acid 12 which is an oxidation product of glycolic acid 11 which in turn is an oxidation product of glycolaldehyde 8. Acrylic acid 10 can be obtained by lactic acid dehydration. Pyruvaldehyde 5 may also undergo a Cannizzaro reaction to yield pyruvic acid 15 and hydroxyacetone 16 (not detected). Decomposition of hydroxyacetone, and pyruvic, lactic and acrylic acids would lead to acetic acid 13 (detected) and carbonate 17 (detected). Formic acid 14 (detected) can be formed by oxidation of formaldehyde 8, oxalic acid 12 and acetic acid 13, and sodium carbonate 17 by oxidation of formic acid 14. Likewise, formic acid and sodium carbonate can be obtained in relatively high yields from pyruvaldehyde 5 under alkaline hydrothermal conditions as shown by Ramrez-Lpez et al. [10]

Scheme 1. Plausible conversion route of sorbitol to lactic acid and over-oxidation products Now, the much lower lactic acid yield from sorbitol than from glycerol can easily be explained considering that, unlike with glycerol, reaction between sorbitol and NaOH (the first step in Scheme 1) can proceed not only through the hydroxyl group linked to the C1 carbon atom leading to the scission of the C3-C4 bond of intermediate 3 but also through the remaining hydroxyls moieties. Thus, sorbitol

First International Congress on Catalysis for Biorefineries (CatBior), Torremolinos-Mlaga, Spain, 2011. cleavage will occur at C4-C5 bond when sodium hydroxide reacts with the hydroxyl group linked to the C2 carbon atom, whereas the scission will occur through both C5-C6 and C4-C5 bonds when it reacts with the hydroxyl moiety linked to the C3 carbon atom. Consequently, a lot of intermediates can be produced, which may subsequently undergo a cascade of reactions (aldol condensation, reverse aldol condensation, decomposition) boosted by the strong alkaline environment and the high temperature, resulting in molecules of both higher and lower molecular weights other than lactic acid. To confirm the formation of these chemicals resulting from the high number of possible conversion routes of sorbitol a reaction sample was neutralized with concentrated HCl and extracted with methyl isobutyl ketone, and the extract was qualitatively analyzed by GC/MS. More than 35 compounds were identified in this fraction with the main ones being ethanol, 4-methyl-2-pentanol, cyclopentanol, 2-methyl-cyclopentanol, 1,6,6-trimethyl-cyclohexene, 2,5-dimethyl-cyclopentanone, 1,2-benzenediol, 3-methyl-1,2-benzenediol, and isosorbide. This fraction could be used as a fuel because, generally speaking, a large number of its components are chemically similar to the ones contained in bio-oil [13]. Its proportion increases with the number of carbon atoms in the polyol being lower than 15% from glycerol and as high as 50% from sorbitol.

Conclusions
Hydrothermal conversion of polyols catalyzed by NaOH is a good synthetic procedure for obtaining lactic acid. A 100% conversion is achieved at 280C within 90 min independently of the starting material, although a higher yield is obtained from glycerol (84.5%) than from sorbitol (39.5%) under the optimum conditions. Results indicate that polyols are converted into four fractions: a) Lactic acid; b) Hydrogen; c) a bio-oil-like fraction; d) A minor inorganic fraction (carbonate) which could be practically eliminated by chosen carefully the experimental conditions. The higher the content of hydroxyl moieties in the polyol the lower the lactic acid yield and the higher the bio-oil-like fraction are. As all fractions are of industrial interest, the alkaline hydrothermal conversion of polyols can be considered as a technology allowing the full valorization of these chemicals.

Acknowledgements
This research has been funded by the Departamento de Industria y Comercio of the Basque Country (Project SOSTENER Exp:IE06-171). The contribution of Leire Lorenzo-Ibarreta and Aitor Godoy to the experimental work is also gratefully acknowledged. TOC and Total Carbon analysis were carried out by Thaiss Prez.

References
1. Werpy T and Petersen G (Eds). Top value added chemicals from biomass, Vol. I-Results of screening for potential candidates from sugars and synthesis gas; N DOE/GO-102004-1992 (2004). http://www1.eere.energy.gov/biomass/pdfs/35523.pdf (accessed, February 20, 2010). 2. A. Corma, S. Iborra, A. Velty, Chem. Rev. 107 (2007) 2411-2502. 3. A.T. Quitain, M. Faisal, K. Kang, H. Daimon, K. Fujie, J. Hazard Mater. 93 (2002) 209-220. 4. H. Yoshida, M. Terashima, Y. Takahashi, Biotechnol Prog 15 (1999) 1090-1094. 5. M. Bicker, S. Endres, L. Ott, H. Vogel, J. Mol. Catal. A: Chem. 239 (2005) 151-157.

First International Congress on Catalysis for Biorefineries (CatBior), Torremolinos-Mlaga, Spain, 2011.
6. A. Yuksel, H. Koga, M. Sasaki, M. Goto, J. Renewable Sustainable Energy 1 (2009) 033112, online; DOI: 10.1063/1.3156006. 7. A. Yuksel, H. Koga, M. Sasaki, M. Goto, Ind. Eng. Chem. Res. 49 (2010) 1520-1525. 8. H. Kishida, F. Jin, Z. Zhou, T. Moriya, H. Enomoto, Chem. Lett. 34 (2005) 1560-1561.. 9. Z. Shen, F. Jin, Y. Zhang, B. Wu, A. Kishita, H. Kishida, Ind. Eng. Chem. Res. 48 (2009) 8920-8925. 10. C. Ramrez-Lpez, J.R. Ochoa-Gmez, M. Fernndez-Santos, O. Gmez-Jimnez-Aberasturi, A. Alonso-Vicario, J. TorrecillaSoria, Ind. Eng. Chem. Res. 49 (2010) 6270-6278. 11. H. Zhou, F. Jin, B. Wu, J. Cao, X. Duan, A. Kishita, J. Phys.: Conf. Series 215 (2010) 012125. 12. H. Kishida, F. Jin, X. Yan, T. Moriya, H. Enomoto, Carbohyd. Res. 341 (2006) 2619-2623. 13 A. Demirbas, Energy Convers. Manage. 50 (2009) 2782-2801.