Understanding Hydrogen in Bayer Process Emissions. 3. Hydrogen

Article pubs.acs.org/IECR

Understanding Hydrogen in Bayer Process Emissions. 3. Hydrogen Production during the Degradation of Polyols in Sodium Hydroxide Solutions Allan Costine,*,† Joanne S. C. Loh,† Francesco Busetti,‡ Cynthia A. Joll,‡ and Anna Heitz‡ †

CSIRO Process Science and Engineering, CSIRO Minerals Down Under Flagship, P.O. Box 7229, Karawara, Western Australia 6152, Australia ‡ Curtin Water Quality Research Centre, Department of Chemistry, Curtin University, G.P.O. Box U1987, Perth, Western Australia 6845, Australia ABSTRACT: This is the third in a series of related studies on the fundamentals of hydrogen gas production from specific classes of organic compounds in sodium hydroxide solutions. The alkaline degradation of 10 aliphatic C2−C6 polyols was investigated under anaerobic conditions in an autoclave. The evolution of hydrogen and low molecular weight carboxylates (lactate, formate, acetate, oxalate, glycerate, glycolate, pyruvate, and acrylate) during the degradation of glycerol, erythritol, xylitol, and sorbitol was studied at 275 °C for reaction times up to 300 min. All of the compounds investigated decomposed to produce approximately 2 mol of hydrogen gas per mole of polyol used. Within a common pathway, three main reactions to hydrogen production were identified: (1) hydroxide-induced formation of an aldehyde and hydride, followed by a hydride-induced β-elimination reaction with hydroxide as the leaving group; (2) the degradation of aldehyde intermediates such as glycolaldehyde (formed by retro-aldol condensation) through base-catalyzed oxidation by water; (3) the degradation of lactic acid. Among the stereoisomers studied, hydrogen production was found to be particularly sensitive to the relative stereochemistry of the hydroxyl groups, which is explained in terms of a common sequence of initial reaction steps in alkaline solution. These findings show that the alkaline degradation of polyols may produce significant amounts of hydrogen in Bayer process digestion, and if wet oxidation is used to remove organic compounds from the liquor, then the potential exists for the formation of explosive gas mixtures. The results also advance the fundamental understanding of the alkaline hydrothermal conversion of polyols to valuable products such as lactic acid. C3H8O3 + NaOH → C3H5O3 ·Na + H 2↑ + H 2O

1. INTRODUCTION Aliphatic polyols (or alditols) are a class of organic compound which, if present in Bayer liquors,1,2 can adversely affect the precipitation rate of aluminum hydroxide and the purity of the alumina product.3−7 In light of their potentially deleterious effects on liquor productivity, a few studies have investigated the degradation reactions of polyols in concentrated sodium hydroxide solutions.3,8,9 The production of hydrogen gas during these reactions, however, which is a potential concern during Bayer process digestion and the wet oxidation of Bayer process liquors,10 is not yet fully understood. A detailed understanding of the degradation pathways that generate hydrogen is important for refinery safety in general11 and for the development of improved wet oxidation technologies for the control and/or removal of organic compounds from the liquor. Hydrogen has previously been detected as the only gaseous product (apart from water) during the alkaline hydrothermal conversion of glycerol12,13 and sorbitol14 to sodium lactate (typically carried out at 280−300 °C, 1.25−3.85 M NaOH, and 0.33−3.5 M polyol). Of the few studies available, the work of Kishida et al.12 is the only report to our knowledge in the published literature to provide quantitative information on the evolution of hydrogen from an individual polyol in sodium hydroxide solutions. Kishida et al.12 showed that the alkaline hydrothermal conversion of glycerol to sodium lactate proceeds with stoichiometric evolution of hydrogen, as shown in eq 1: © 2013 American Chemical Society

(1)

The lack of detailed information on the production of hydrogen from aliphatic polyols is surprising, given that hydrogen would be expected to be a common major alkaline degradation product of polyols in general. In the present work, the hydrogen production from the alkaline degradation of a range of C2−C6 polyols was measured to define the reaction conditions that promote hydrogen production, to determine the effect of the stereochemistry of the polyols on hydrogen yields and to provide quantitative information on the products of the reactions. Several known reaction products of the polyols were also tested as starting substrates, to help identify the key reactions that produce hydrogen and low molecular weight (LMW) carboxylate anions.

2. EXPERIMENTAL SECTION 2.1. Autoclave Facility. Alkaline degradation experiments were carried out under anaerobic conditions in a 2 L Inconel 600 autoclave (Parr Instrument Co., Moline, IL, USA), which could be operated to a maximum temperature of 280 °C. The Received: Revised: Accepted: Published: 5572

February 6, 2013 March 25, 2013 March 30, 2013 March 30, 2013 dx.doi.org/10.1021/ie400435k | Ind. Eng. Chem. Res. 2013, 52, 5572−5581

Industrial & Engineering Chemistry Research

Article

Figure 1. Structures and Fischer projections of the polyols studied in which ω represents a primary hydroxyl−secondary hydroxyl sequence, “t” represents a threo sequence, and “e” represents an erythro sequence. The abbreviated hydroxyl sequences are written from the top carbon atom down.

an individual autoclave run carried out for the nominated reaction time in the range 0−300 min. 2.6. Solution Analysis. The concentrations of the LMW carboxylates, glycolate, glycerate, and pyruvate, in the reaction product mixtures were measured by direct injection liquid chromatography−mass spectrometry. Erythritol was also detected in the reaction product mixtures and its concentration was measured. Sample preparation involved neutralization with 37% hydrochloric acid solution and 1:2 sample dilution (v/v) with ultrapure water before injection. The water used for sample preparation was purified using an IBIS Technology ion exchange system (Perth, Australia) followed by an Elga Purelab Ultra System (Sydney, Australia). Aliquots (1 μL) of each prepared solution were injected into an Agilent 1100 LC system (Palo Alto, CA, USA) coupled to a Micromass Quattro Ultima triple quadrupole mass spectrometer (Manchester, U.K.) fitted with an electrospray interface operated in negative mode. For separation, a Gemini C18 column (250 mm × 3 mm i.d., 3 μm particle size) from Phenomenex (Torrance, CA, USA) was employed. The mobile phase composition was H2O/ MeOH (90/10 v/v) with 0.1% formic acid and 5 mM ammonium formate. Separation was achieved in isocratic mode at a flow rate of 150 μL/min. For detection, the following settings were used on the mass spectrometer: capillary 2500 V; cone 25 V; front hexapole 0.0 V, aperture 0.1 V, and exit hexapole 0.0 V; desolvation gas 550 L/h; cone gas 25 L/h; source temperature 135 °C; and desolvation temperature 325 °C. Detection and quantification was achieved in single ion monitoring (SIM) mode, at unit mass resolution. Ions monitored were [M−H]− for each analyte with a dwell time of 100 ms. To account for matrix effects,17 matrix matched calibration curves (linear range 1−75 mM, R2 > 0.998) were used for quantification. Matrix matching was achieved by spiking increasing amounts of analytical standards in 3.77 M NaOH solution neutralized with concentrated hydrochloric acid solution and diluted with ultrapure water to a final dilution ratio of 1:2 (v/v). Capillary electrophoresis (CE)18 and ion chromatography (IC)16 were used to measure the concentrations of the following LMW carboxylate products: formate, acetate, oxalate, acrylate, and lactate. The starting polyols, NaOH pellets, and water used in this study were found to be free of organic contaminants that could potentially affect the accurate measurement of the LMW carboxylates in the reaction product mixtures. The carbonate content of the solutions was measured using an automated potentiometric titration system.19 Molar yields have been calculated as the ratio of the number of moles

oxygen and hydrogen concentrations in the headspace were measured using online electrochemical sensors. The repeatability of hydrogen measurements was found to be ±0.05 mol % H2, which gave an error of less than ±0.01 mol of H2/mol of organic compound when expressed as molar yields. A detailed description of the autoclave facility is given elsewhere.15 2.2. Organic Compounds. The structures and Fischer projections of the 10 polyols tested in alkaline degradation experiments are shown in Figure 1. The polyols were selected from among those already known to affect aluminum hydroxide nucleation and precipitation rates,3,5−7 and they encompass a range of carbon chain lengths and hydroxyl group stereochemistries. Erythro and threo sequences have two adjacent hydroxyl groups on either the same side or opposite sides of a polyol, respectively, when it is presented as a Fischer projection. Ethylene glycol, glycerol, DL-threitol, erythritol, xylitol, Darabinitol, ribitol, dulcitol (galactitol), D-mannitol, and Dsorbitol were all analytical grade reagents procured from Sigma-Aldrich (Sydney, Australia). Erythritol, xylitol, ribitol, and dulcitol are all meso isomers. The alkaline degradation of five carboxylates, identified in the reaction products of glycerol, erythritol, xylitol, and sorbitol, were also investigated under similar conditions. These carboxylates were glycolate (from 99% purity glycolic acid), pyruvate (from 98% purity pyruvic acid), glycerate (from 99% purity D-glyceric acid calcium salt dihydrate), lactate (from 98% purity L-lactic acid), and acrylate (from 99% purity acrylic acid inhibited with 200 ppm monoethyl ether hydroquinone). 2.3. Solution Preparation. Sodium hydroxide solutions were prepared by dissolving the appropriate mass of NaOH pellets (>97%, 1% Na2CO3, Rowe Scientific, Perth, Australia) in deionized water. The autoclave was charged with 1 L of a 40 mM polyol solution and then sealed and inserted into the heating cavity. 2.4. System Preparation. The solutions were degassed by sparging with nitrogen until the concentration of residual oxygen in the headspace was below 0.01 mol %. The solutions were heated at a rate of approximately 5 °C/min, with stirring (60 rpm), to the designated temperature. The time at which the required temperature was reached was taken as t = 0 min, the nominal starting point of the reaction. 2.5. Gas Analysis. The procedure for gas analysis was the same as previously described.16 For each analysis, it was necessary to stop the reaction by cooling the autoclave to enable the hydrogen content of the headspace gas to be measured. Each data point in Figures 2−5 therefore represents 5573

dx.doi.org/10.1021/ie400435k | Ind. Eng. Chem. Res. 2013, 52, 5572−5581


Article

For the C4−C6 sets of stereoisomers, it was found that hydrogen production was particularly sensitive to the relative stereochemistry of the hydroxyl groups (see Figure 1). Specifically, it was observed that polyols with at least one erythro sequence adjacent to a primary carbon (ω-e) had a higher rate of hydrogen production than their same carbon length counterparts without this sequence (threitol, xylitol, and dulcitol). The magnitude of this effect was greatest for the tetritols, with erythritol (ω-e-ω) producing significantly more hydrogen than threitol (ω-t-ω) after 15 min, and was found to diminish with increasing carbon chain length and reaction temperature. Among the pentitols studied, xylitol (ω-t-t-ω) had the lowest rate of hydrogen production. The hexitols, dulcitol (ω-t-e-t-ω) and sorbitol (ω-t-t-e-ω), have the same number of sequences but in different orders, highlighting that interior erythro sequences had slightly less influence on hydrogen production. Considering these trends, allitol (ω-e-e-e-ω) would be expected to have a higher rate of hydrogen production than iditol (ω-t-t-t-ω); however, solutions of these polyols could not be prepared at the desired concentration (40 mM) for testing because of their limited solubilities at room temperature. The effect of the relative stereochemistry of the hydroxyl groups on hydrogen production could be related to the degree to which these polyols are able to form complexes with metal ions in solution. It is known that metals leached from the autoclave construction materials have the potential to affect the reactions of polyols in sodium hydroxide solutions,13,14,20 as shown for the alkaline conversion of glycerol to lactate in Hastelloy-C and titanium autoclaves.21 This is particularly relevant to the results shown in Figure 2 because the C4−C6 stereoisomers that had a lower rate of hydrogen production were also those with hydroxyl sequences that most readily complex with metal ions22 (threitol, xylitol, and dulcitol). Three vicinal hydroxyl groups can form complexes with metal ions with varying affinities depending on the number and types of hydroxyl sequences present, as follows: t-t > ω-t > e-t > ω-e > e-e. The stronger metal ion complexation for the polyols with hydroxyl group sequences that favor metal complexation may protect these groups from alkaline degradation. To test this, the degradation of two sets of stereoisomers, threitol and erythritol, and xylitol and ribitol, was studied in a zirconium autoclave under the conditions reported in Figure 2. Hydrogen production was the same from each stereoisomer in both the Inconel autoclave and the zirconium autoclave within experimental error, indicating that the apparent effect of the relative stereochemistry of the hydroxyl groups on hydrogen production was not induced by the presence of trace metals leached from the Inconel autoclave. This aspect is discussed further in section 3.3, in relation to the proposed degradation pathways of the C3−C6 polyols. 3.2. Detailed Study of the Degradation of Selected Polyols. Glycerol (C3), erythritol (C4), xylitol (C5), and sorbitol (C6) were selected for more detailed investigation to determine the effect of carbon chain length on the nature and distribution of their alkaline degradation products. These polyols were chosen because they have been the subject of previous studies (except erythritol) to develop new routes to lactic acid by alkaline hydrothermal chemistry;12−14,20,23 however, little information is available on the production of hydrogen during these reactions. 3.2.1. Evolution of Products. The evolution of the principal degradation products of the polyols was studied at 275 °C for up to 300 min to determine the total hydrogen produced when

of each degradation product to the initial number of moles of polyol in each case. The repeatability of the molar yield measurements of the LMW carboxylate products was generally ±0.05 mol/mol. High performance liquid chromatography (HPLC) with a photodiode array detector was used to analyze the solutions for glycolaldehyde, pyruvaldehyde, and glyceraldehyde. Only glycolaldehyde was detected (from all polyols studied) due to the high reactivity of these short-lived intermediates. An Alltech Prevail organic acid column (250 mm × 4.6 mm i.d.) was used. The mobile phase was acetonitrile−10 mM KH2PO4 (60/40 v/ v) with a flow rate of 1 mL/min. Carbon recoveries, defined as the ratio of the number of moles of organic and inorganic carbon in the solution products to the initial number of moles of carbon in the starting polyols (expressed as a percentage), were determined by a Shimadzu TOC-V CPH/CPN total organic carbon (TOC) analyzer. For all polyols studied, recoveries of carbon in the aqueous phase were between 97 and 106%, indicating that the formation of gaseous hydrocarbons was insignificant. Inductively coupled plasma atomic emission spectrometry (ICP-AES) was used to measure the concentrations of metals leached from the Inconel autoclave construction materials. Typical concentrations after 300 min at 275 °C were 24.2 mg L−1 Fe, 11.6 mg L−1 Cr, and 1.8 mg L−1 Ni. The autoclave was refurbished (through surface skimming) by the Parr Instrument Co. in 2008. Since then, approximately 4000 h of operation have been accumulated with alkaline solutions or slurries. To determine any possible effect on hydrogen yields of the materials of construction of the Inconel autoclave, four duplicate experiments were also performed in a 2 L Parr zirconium (Zr 705) autoclave (see section 3.1).

3. RESULTS AND DISCUSSION 3.1. Screening Study. The production of hydrogen during alkaline degradation of the 10 polyols was measured after holding at 250 and 275 °C for 15 min to provide an indication of their initial rates of hydrogen formation. The number of moles of hydrogen produced per mole of polyol for the 10 polyols is shown in Figure 2. The polyols are categorized based on their carbon chain length. Under the same conditions, hydrogen production increased with increasing carbon chain length for the shorter chain polyols (C2−C4) and was similar for the longer chain polyols (C5−C6).

Figure 2. Number of moles of H2 produced per mole of polyol. Polyol names are abbreviated to their first three letters. Conditions: temperature = 250 and 275 °C; time at temperature = 15 min; [NaOH] = 3.77 M; [polyol] = 40 mM. 5574



Article

Figure 3. Number of moles of H2 and LMW carboxylates produced per mole of (A) glycerol, (B) erythritol, (C) xylitol, and (D) sorbitol. Conditions: temperature = 275 °C; time at temperature = 0−300 min; [NaOH] = 3.77 M; [polyol] = 40 mM. Symbols: ● = H2; ▽ = lactate; □ = formate; △ = acetate; ○ = oxalate; blue ■ = glycerate; purple + = glycolate; ∗ = erythritol; red ● = pyruvate.

the reaction was close to completion and to provide information on the rates of formation of the products over time. These results are presented in Figure 3 in terms of molar yields. Acrylate and carbonate were detected at less than 0.05 mol/mol in the products of all polyols studied (not shown). For all four of the polyols tested, hydrogen was the principal degradation product (approximately 2 mol/mol after 300 min), with lower amounts of carboxylate salts also formed. For glycerol degradation (Figure 3A), lactate was the major carboxylate formed (0.48 mol/mol) followed by formate, acetate, and oxalate as stable products. Among the less abundant products were glycolate, glycerate, and pyruvate, all of which degraded to be present in only trace amounts upon completion of the reaction. The formation of small amounts of erythritol from glycerol reveals that C−C bond-forming aldol condensation reactions can occur under these conditions. These organic products (along with acrylate and carbonate) accounted for 93% of the initial carbon after 300 min. Similarly, for erythritol (Figure 3B), hydrogen was the dominant product and its initial rate of formation was faster than in the case of glycerol degradation. Glycolate was the major carboxylate produced at short reaction times, its concentration approximately twice that observed in glycerol degradation. The lactate yield was relatively low (0.19 mol/ mol) and acetate and formate were main products, followed by lower amounts of oxalate. Only minor amounts of glycerate and pyruvate were formed. The carbon recovery from these products was 67% after 300 min, indicating that unidentified degradation products must have formed within the reaction mixture. In the degradation of xylitol (Figure 3C), significant concentrations of glycerate and glycolate formed rapidly at short reaction times but degraded during the course of the

reaction. Lactate was detected as a major product at longer reaction times (0.42 mol/mol), followed by formate, acetate, and oxalate. Minor amounts of erythritol and pyruvate were also formed. After 300 min at 275 °C, the observed products accounted for 73% of the initial carbon. For sorbitol degradation (Figure 3D), lactate was the major carboxylate produced throughout the reaction period (0.49 mol/mol). Significant amounts of formate and acetate were also formed, along with some oxalate. Glycerate, glycolate, pyruvate, and erythritol were all detected at short reaction times but degraded over time. Carbon recovery in the analyzed products was 66% after 300 min. These results show that aliphatic polyols, if present in Bayer liquors,1,2 may produce significant amounts of hydrogen gas during Bayer process digestion, which is a particular concern for wet oxidation processes used to remove organic compounds from the liquor.10 A general explanation of the degradation pathways for polyols, based on the nature and amounts of products observed in the above reactions, is proposed in section 3.3. 3.2.2. Proposed Degradation Pathways for Glycerol. The proposed degradation pathways for glycerol are illustrated in Scheme 1. While these pathways are not unique or complete, they provide a reasonable explanation for the formation and distribution of the main reaction products, which account for 93% of the initial carbon in glycerol (Figure 3A). To help elucidate the key pathways that produce hydrogen, five identified productsglycolate, glycerate, pyruvate, lactate, and acrylatewere also tested as starting substrates under similar conditions. Table 1 shows the product yields for each substrate tested, which will be discussed separately in relation to the degradation pathways shown in Scheme 1. 5575



Article

Scheme 1. Proposed Degradation Pathways for Glycerol To Produce Hydrogen and LMW Carboxylatesa

a

The highlighted pathway is from the work of Kishida et al.12 Identified products are marked with asterisks. R1, keto−enol tautomerization; R2, benzilic acid rearrangement; R3, elimination of water; R4, base-catalyzed oxidation by water; R5, retro-aldol condensation; R6, aldol condensation; R7, cross-Cannizzaro reaction.

generates hydrogen gas and 2-hydroxypropenal (route A, Scheme 1). By keto−enol tautomerization, 2-hydroxypropenal converts to pyruvaldehyde, which then readily undergoes the benzilic acid rearrangement in alkaline solution to form lactate. In the current study, the ratio of the molar yields of hydrogen to lactate was about 4:1 after 300 min, which points to the existence of several parallel and/or consecutive pathways that produce hydrogen. Glyceraldehyde is an established intermediate in the alkaline degradation of glycerol21 and sorbitol14 and can also react to produce formaldehyde, together with glycolaldehyde, by a retro-aldol condensation (route B, Scheme 1). Formaldehyde may then react by an ionic degradation mechanism involving base-catalyzed oxidation by water (R4, Scheme 1) to produce formate and hydrogen, as we have previously proposed.3 This pathway is consistent with earlier studies in which hydrogen was observed as a product of the alkaline degradation of formaldehyde24,25 and other aldehydes.26 Formaldehyde may also undergo the Cannizzaro reaction, which proceeds without hydrogen formation, to yield a mixture of formate and methanol (not shown in Scheme 1). The detection of small amounts of methanol in an earlier study on the alkaline conversion of glycerol21 suggests that this reaction is likely under the current conditions. The formation of the aldehyde intermediates can be inferred from the nature of the carboxylates detected. For example,

Table 1. Molar Yields of H2 and LMW Carboxylates from Sodium Glycolate, Glycerate, Pyruvate, Lactate, and Acrylatea molar yields (mol of product/mol of substrate) substrate

time (min)

H2

formate

acetate

oxalate

lactate

C bal (%)

glycolate glycerate pyruvate lactate acrylate

300 300 300 300 120

0.42 0.01 0.08 0.42 1.04

0.03 0.91 0.85 0.36 0.94

0.00 0.86 0.90 0.38 0.98

0.28 0.00 0.03 0.03 0.02

0.00 0.00 0.06 0.59 0.00

30 88 96 98 98

Conditions: temperature = 275 °C; [NaOH] = 3.77 M; [substrate] = 40 mM.

a

Kishida et al.12 showed that, in the alkaline hydrothermal conversion of glycerol, the ratio of the molar yields of hydrogen to lactate was approximately 1:1, consistent with the highlighted pathway in Scheme 1. Those authors proposed that the reaction is initiated by deprotonation of a primary hydroxyl group of glycerol to produce a glyceroxide ion intermediate, which then loses a H− ion to yield glyceraldehyde. Abstraction of the α-hydrogen of glyceraldehyde by the H− ion and βelimination of the hydroxyl group (by an E2 mechanism) 5576



Article

quantify hydrogen production during the alkaline degradation of lactate. The hydrogen production is probably a result of the base-catalyzed dehydration of lactate to acrylate (R3, Scheme 1) which then degraded to hydrogen, acetate, and formate (Table 1). We previously showed that, above 200 °C, all α,βunsaturated carboxylates30 decomposed in alkaline solution to produce hydrogen, as follows:

glycolate and oxalate are undoubtedly consecutive oxidation products of glycolaldehyde (detected), and are formed along with hydrogen by base-catalyzed oxidation by water. Glycolaldehyde was also detected in an earlier study of the alkaline hydrothermal electrolysis of glycerol at 280 °C.27 In the degradation of glycolate (Table 1), hydrogen and oxalate were the dominant products, consistent with the observation that all of the polyols studied showed a slow decrease in glycolate concentrations and a corresponding increase in oxalate concentrations (Figure 3). The relative stability of glycolate (in comparison with glycerate and pyruvate) was in agreement with an earlier study on the long-term stability of various carboxylates in 6 mol kg−1 NaOH, which showed that 83% of the original glycolate remained unreacted after 36 days at 180 °C.9 Those authors also presented an ionic degradation mechanism for glycolate involving base-catalyzed oxidation by water, explaining the formation of hydrogen and oxalate as major products. The retro-aldol condensation of glyceraldehyde (route B, Scheme 1) can account for, at least in part, the higher hydrogen yields observed here in glycerol degradation compared with the work of Kishida et al.12 Although this reaction proceeds only to a small extent, as indicated by the relatively low oxalate yields, it produces up to 4 mol of hydrogen per mole of glyceraldehyde consumed. Glycolaldehyde may undergo aldol condensation (R6, Scheme 1) to form erythrose, which is subsequently converted to lactate, with the overall yield being 28% (based on the starting carbon mass of glycolaldehyde).28 We also observed lactate in the degradation products of ethylene glycol, a precursor to glycolaldehyde, under the alkaline conditions reported in Figure 2. The presence of small amounts of erythritol in the glycerol degradation products could be the result of a cross-Cannizzaro reaction of formaldehyde with erythrose to yield formate and erythritol as oxidation and reduction products, respectively.29 The glyceraldehyde intermediate may also undergo basecatalyzed oxidation by water to yield glycerate and hydrogen (route C, Scheme 1). The same reaction can be envisaged for pyruvaldehyde to yield pyruvate and hydrogen, although the absence of α-hydrogens in this aldehyde intermediate could favor the formation of pyruvate and hydroxyacetone by the Cannizzaro reaction (not shown in Scheme 1). Kishida et al.12 have previously identified pyruvaldehyde and pyruvate in the alkaline degradation products of glycerol at 300 °C. For the degradation of glycerate and pyruvate (Table 1), formate and acetate are the dominant products and hydrogen yields are small. The similar product distributions observed for these two substrates implies the existence of a common degradation pathway, as suggested in Scheme 1. Abstraction of the acidic αhydrogen in glycerate by hydroxyl ion and β-elimination of the hydroxyl group produces 2-hydroxypropenoate, which then undergoes keto−enol tautomerization to form pyruvate. Subsequent nucleophilic attack by hydroxyl ion on the αcarbonyl group of pyruvate results in an intermediate species that reacts with water to produce formate, acetate, and hydroxyl ion. In the degradation of lactate (Table 1), hydrogen, acetate, and formate were the dominant products. The ratio of the molar yields of these products was close to 1:1:1. The observed degradation products accounted for 39% of the initial carbon after 300 min at 275 °C, with undegraded lactate accounting for the remainder. To our knowledge, this is the first report to

R·CHCH·CO2− ⇌ R·CH(OH)·CH 2 ·CO2− → H 2 + CH3·CO2− + R·CO2−

(2)

where R is a hydrogen atom, an alkyl group, or a carboxylate group. The nature and amounts of products are the same when either the unsaturated carboxylates30 or their corresponding βhydroxycarboxylates16 are used as starting substrates. Earlier studies on the conversion of lactic acid to acrylic acid in high-temperature water (320−450 °C) reported the formation of propionic acid as a minor hydrogenation product of acrylic acid.31−33 In the alkaline degradation of lactate and acrylate (Table 1), the carbon recoveries in the analyzed products were almost complete (98%), which indicated that the hydrogenation of acrylate to propanoate was negligible. The absence of a significant hydrogen-consuming reaction was supported by the stoichiometries of the formation of hydrogen, acetate, and formate observed in the degradation of lactate and acrylate. The present results may explain the observation of Kishida et al.12 that, in the alkaline hydrothermal conversion of glycerol, the yields of hydrogen were slightly higher than those for lactate (see Figure 4). The higher yields of hydrogen can be

Figure 4. Number of moles of H2 and lactate produced per mole of glycerol at different reaction times. Data from the work of Kishida et al.12 included as red dots and triangles (300 °C, [NaOH] = 1.25 M, [glycerol] = 0.33 M). Conditions: temperature = 275 °C; time at temperature = 0−300 min; [NaOH] = 0.5, 1, 3.77 M; [glycerol] = 40 mM; NaOH/glycerol molar ratio = 12.5 (dotted lines), 25 (dashed lines), and 94 (solid lines).

attributed to minor degradation of lactate, consistent with the presence of small amounts of acrylate, acetate, and formate in the reaction products reported in that study. We previously showed that formate, acetate, and oxalate produced no hydrogen under similar conditions,16 consistent with their formation as stable products in Figure 3. The decarboxylation of C2−C6 carboxylates in NaOH solution to produce hydrogen and various gaseous hydrocarbons was reported by Oakwood and Miller,34 albeit at much higher temperatures (∼370 °C) than employed here. In the current 5577



Article

xylitol, and sorbitol, and follows the same sequence of steps that were described in the degradation of glycerol (Scheme 1). For the longer chain polyols, however, the retro-aldol reactions that proceed via routes A and B can produce a variety of aldehyde intermediates, the nature of which depends on the starting polyol. Although the formation and consecutive oxidation of these aldehydes is fast,35,36 their presence in the reaction mixtures can be inferred from the nature and distribution of their secondary carboxylate products. In this context, it is worthwhile to review the product distributions of the C3−C6 polyols investigated (Figure 3), in relation to the degradation pathways proposed in Schemes 1 and 2. For erythritol degradation (n = 0, Scheme 2), the molar yield of lactate was small in comparison with the yield obtained in glycerol degradation (0.19 and 0.48 mol/mol, respectively). This finding is in agreement with an earlier study showing lower lactate yields in alkaline hydrothermal reactions that proceed via erythrose rather than via glyceraldehyde.28 As discussed above, the retro-aldol reaction of glyceraldehyde to produce formaldehyde and glycolaldehyde is an important parallel pathway in the degradation of glycerol (route B, Scheme 1). If this reaction proceeds in the degradation of erythritol, then 2 mol of glycolaldehyde would be produced per mole of erythrose intermediate converted (route B, Scheme 2). Consistent with this, the initial molar yield of glycolate (an immediate oxidation product of glycolaldehyde) during erythritol degradation was about twice that observed in glycerol degradation. The formation of significant amounts of glycolaldehyde by the retro-aldol reaction is also consistent with the high initial rate of hydrogen production observed in erythritol degradation because up to 3 mol of hydrogen is produced per mole of glycolaldehyde that reacts via glycolate to form oxalate, as shown in Scheme 1. These interpretations are supported by the finding of Srokol et al.,37 who identified glycolaldehyde as the main product in the hydrothermal degradation of erythrose after 204 s at 340 °C. In the degradation of xylitol (n = 1, Scheme 2), the retroaldol reaction of xylose intermediate produces glyceraldehyde and glycolaldehyde (route B, Scheme 2). Consistent with this, Sasaki et al.38 observed these aldehydes as the major products in the hydrothermal degradation of xylose after 1 s at 340 °C. In the current work, the presence of significant amounts of glycerate and glycolate at early reaction times implies that the retro-aldol reaction is also important in xylitol degradation. The higher glycolate concentrations observed in xylitol degradation compared with glycerol degradation can be attributed to the formation of glycolaldehyde as a retro-aldol product of xylose and also the keto intermediate formed via route A. The degradation pathways proposed in Scheme 2 are consistent with the observation of Zhou et al.,23 who identified similar alkaline degradation products when either xylitol or xylose was used as the starting substrate. In the alkaline conversion of sorbitol (n = 2, Scheme 2) to lactate, Ramirez-Lopez et al.14 were unable to detect the presence of glucose as an intermediate. On this basis, those authors proposed that sorbitol converts to the enol intermediate (route A, Scheme 2) by a concerted mechanism involving simultaneous dehydrogenation and dehydration reactions. If this is the case, then pyruvaldehyde and glyceraldehyde are the expected retro-aldol products formed via route A. As both of these aldehydes can evolve to lactate (see Scheme 1), this may account for the high lactate yields observed in sorbitol degradation, particularly at early reaction

work, the full carbon recoveries recorded in the solution products and the small carbonate yields observed are in accord with the formation of formate, acetate, and oxalate as stable products. 3.2.3. Effect of NaOH/Glycerol Molar Ratio. It follows from Scheme 1 that, if lactate were the only carboxylate produced in the alkaline degradation of glycerol (route A), then the ratio of the molar yields of hydrogen to lactate would be unity. As virtually all of the parallel and consecutive degradation pathways in Scheme 1 are base-catalyzed reactions, the effect of the NaOH/glycerol molar ratio on hydrogen and lactate yields was studied at 275 °C. The number of moles of hydrogen and lactate produced per mole of glycerol at different times is shown in Figure 4. At low molar ratio (12.5), lactate was produced along with stoichiometric evolution of hydrogen according to route A. These results are in accord with the work of Kishida et al.,12 who observed near-stoichiometric formation of lactate and hydrogen at 300 °C with a low molar ratio of 3.8 (Figure 4). With increasing molar ratio the decomposition of intermediates is promoted by the excess hydroxide ion concentration. At a high NaOH/glycerol molar ratio (94), the ratio of the molar yields of hydrogen to lactate was about 4:1 after 300 min. These results are consistent with an increasing prevalence of the retro-aldol condensation of glyceraldehyde (route B) and some minor decomposition of lactate at high NaOH/glycerol molar ratios. 3.2.4. Effect of Temperature. The conversion of glyceroxide ion to glyceraldehyde requires the cleavage of a strong primary C−H bond. The extent of this reaction below 225 °C was insignificant as shown in Figure 5, which shows the effect of

Figure 5. Number of moles of H2 and lactate produced per mole of glycerol at different reaction times. Conditions: temperature = 225 (dotted lines), 250 (dashed lines), and 275 °C (solid lines); time at temperature = 0−300 min; [NaOH] = 3.77 M; [glycerol] = 40 mM.

temperature on hydrogen and lactate yields at a high NaOH/ glycerol molar ratio of 94. Up to 250 °C, the ratio of the molar yields of hydrogen to lactate was about 2:1, which increased to about 4:1 at 275 °C. These results are consistent with the increasing rate of the retro-aldol condensation of glyceraldehyde producing hydrogen as a main product (route B, Scheme 1) and some minor degradation of lactate at higher temperatures. 3.3. Proposed Degradation Pathways for C 3−C6 Polyols. In this section, a general degradation pathway is proposed for the initial reactions of C3−C6 polyols in alkaline solution to produce hydrogen, lactate, and various aldehydes. The proposed pathway is illustrated in Scheme 2 for erythritol, 5578



Article

Scheme 2. Proposed Common Degradation Pathway for Erythritol, Xylitol, and Sorbitol To Produce Hydrogen, Lactate, and Various Aldehydesa

a

The relative stereochemistry of each starting polyol is not shown for simplicity. R1, keto−enol tautomerization; R2, retro-aldol condensation; R3, benzilic acid rearrangement.

hydrogen production observed for these polyols in comparison with their corresponding stereoisomers (Figure 2).

times. The absence of a significant retro-aldol pathway via glucose is also suggested by the relatively low glycolate yields (an immediate oxidation product of glycolaldehyde) observed in sorbitol degradation because up to 3 mol of glycolaldehyde could be produced per mole of glucose converted in route B. The longer chain polyols can react not only through deprotonation of a primary hydroxyl group but also through the secondary hydroxyl groups, resulting in a variety of other retro-aldol intermediates besides those shown in Scheme 2. This is consistent with the lower carbon balances recorded in the degradation of the longer chain polyols (67−73%) in comparison with the glycerol degradation (93%). While Scheme 2 does not fully describe all the possible reactions of the polyols studied, it nevertheless provides a useful rationalization for the formation and distribution of many of the observed products. The production of hydrogen via route A of Scheme 2 would be expected to be particularly sensitive to the relative stereochemistry of the hydroxyl groups of the starting polyols, as shown in Figure 2. The mechanism involves loss of hydride, which acts as a Lewis base and removes a proton from the neighboring carbon in a β-elimination reaction with hydroxide as the leaving group. The removal of the two substituents in the β-elimination can occur by either syn- or anti-elimination, in which the proton and hydroxide are lost from either the same side or opposite sides of the developing CC double bond, respectively. As anti-elimination is greatly favored over synelimination due to the lower energy of the anti transition state,39 the relative stereochemistry of the hydroxyl groups of the starting polyols is likely to have a significant effect on the relative rates of hydrogen production for the stereoisomers studied. If a primary hydroxyl group is deprotonated in the first step, then hydrogen production would be favored in those polyols having a ω-e sequence of hydroxyl groups. Similarly, if the initial deprotonation occurs at a secondary hydroxyl group, then hydrogen production would be favored by the presence of an e-e sequence of hydroxyl groups. The absence of either of these sequences in threitol (ω-t-ω), xylitol (ω-t-t-ω), and dulcitol (ω-t-e-t-ω) is consistent with the slower rates of

4. CONCLUSIONS This study shows that the alkaline degradation of polyols may produce significant amounts of hydrogen in Bayer process digestion, and if wet oxidation is used to remove organic compounds from the liquor, then the potential exists for the formation of explosive gas mixtures. The results also advance the fundamental understanding of the alkaline hydrothermal conversion of polyols to valuable products such as lactic acid. The main findings from this study are that, in the alkaline degradation of a range of C2−C6 polyols, the following occur: • All of the polyols studied decomposed by a common main reaction pathway to produce approximately 2 mol of hydrogen gas per mole of polyol used. The following three main reactions to hydrogen production were identified: (1) hydroxide-induced formation of an aldehyde and hydride, followed by a hydrideinduced β-elimination reaction with hydroxide as the leaving group; (2) the parallel degradation of aldehyde intermediates (formed by retro-aldol reactions) through base-catalyzed oxidation by water; (3) the degradation of lactate. • The alkaline degradation of lactate produces hydrogen, acetate, and formate, which is consistent with base-catalyzed dehydration of lactate to acrylate, which then degrades according to the mechanism detailed in part 2 of this series of papers.30 • The relative stereochemistry of the hydroxyl groups was found to have a significant effect on the rate of hydrogen production among the polyol stereoisomers studied. Specifically, hydrogen production was favored by the presence of erythro−erythro sequences of hydroxyl groups, consistent with the existence of a hydrogen-producing pathway involving loss of hydride, which acts as a Lewis base and removes a proton from the neighboring carbon in a β-elimination reaction with hydroxide as the leaving group, the β-elimination being favored in an anti arrangement. 5579



■

Article

(15) Costine, A.; Loh, J.; McDonald, R.; Power, G. Unique hightemperature facility for studying organic reactions in the Bayer process. Light Met. 2010, 47−52. (16) Costine, A.; Loh, J. S. C.; Power, G.; Schibeci, M.; McDonald, R. G. Understanding hydrogen in Bayer process emissions. 1. Hydrogen production during the degradation of hydroxycarboxylic acids in sodium hydroxide solutions. Ind. Eng. Chem. Res. 2011, 50 (22), 12324−12333. (17) Jessome, L. L.; Volmer, D. A. Ion suppression: A major concern in mass spectrometry. LCGC North Am. 2006, 24 (5), 498−510. (18) Chovancek, M.; Choo, P.; Macka, M. Development of a fully buffered molybdate electrolyte for capillary electrophoresis with indirect detection and its use for analysis of anions in Bayer liquor. Electrophoresis 2004, 25 (3), 437−443. (19) Connop, W. L. A new procedure for the determination of alumina, caustic, and carbonate in Bayer liquors. 4th International Alumina Quality Workshop, June 2-7, 1996; AQW Inc.: Darwin, Australia, 1996; pp 321−329. (20) Shen, Z.; Jin, F.; Zhang, Y.; Wu, B.; Kishita, A.; Tohji, K.; Kishida, H. Effect of alkaline catalysts on hydrothermal conversion of glycerin into lactic acid. Ind. Eng. Chem. Res. 2009, 48 (19), 8920− 8925. (21) Roy, D.; Subramaniam, B.; Chaudhari, R. V. Cu-based catalysts show low temperature activity for glycerol conversion to lactic acid. ACS Catal. 2011, 1 (5), 548−551. (22) Angyal, S. J.; Greeves, D.; Mills, J. A. Complexes of carbohydrates with metal cations. 3. Conformations of alditols in aqueous-solution. Aust. J. Chem. 1974, 27 (7), 1447−1456. (23) Zhou, H.; Jin, F.; Wu, B.; Cao, J.; Duan, X.; Kishita, A. Production of lactic acid from C6-polyols by alkaline hydrothermal reactions. J. Phys: Conf. Ser. 2010, 215, 012125. (24) Kapoor, S.; Barnabas, F. A.; Sauer, M. C.; Meisel, D.; Jonah, C. D. Kinetics of hydrogen formation from formaldehyde in basic aqueous solutions. J. Phys. Chem. 1995, 99 (18), 6857−6863. (25) Kapoor, S.; Naumov, S. On the origin of hydrogen in the formaldehyde reaction in alkaline solution. Chem. Phys. Lett. 2004, 387 (4−6), 322−326. (26) Ashby, E. C.; Doctorovich, F.; Liotta, C. L.; Neumann, H. M.; Bàield, E. K.; Konda, A.; Zhang, K.; Hurley, J.; Siemer, D. D. Concerning the formation of hydrogen in nuclear waste. Quantitative generation of hydrogen via a Cannizzaro intermediate. J. Am. Chem. Soc. 1993, 115 (3), 1171−1173. (27) Yuksel, A.; Koga, H.; Sasaki, M.; Goto, M. Hydrothermal electrolysis of glycerol using a continuous flow reactor. Ind. Eng. Chem. Res. 2010, 49 (4), 1520−1525. (28) Kishida, H.; Jin, F.; Yan, X.; Moriya, T.; Enomoto, H. Formation of lactic acid from glycolaldehyde by alkaline hydrothermal reaction. Carbohydr. Res. 2006, 341 (15), 2619−2623. (29) Tambawala, H.; Weiss, A. H. Homogeneously catalyzed formaldehyde condensation to carbohydrates. 2. Instabilities and Cannizzaro effects. J. Catal. 1972, 26 (3), 388−400. (30) Costine, A.; Loh, J. S. C.; Power, G. Understanding hydrogen in Bayer process emissions. 2. Hydrogen production during the degradation of unsaturated carboxylic acids in sodium hydroxide solutions. Ind. Eng. Chem. Res. 2011, 50 (22), 12334−12342. (31) Aida, T. M.; Ikarashi, A.; Saito, Y.; Watanabe, M.; Smith, R. L., Jr.; Arai, K. Dehydration of lactic acid to acrylic acid in high temperature water at high pressures. J. Supercrit. Fluids 2009, 50 (3), 257−264. (32) Lira, C. T.; McCrackin, P. J. Conversion of lactic acid to acrylic acid in near-critical water. Ind. Eng. Chem. Res. 1993, 32 (11), 2608− 2613. (33) Mok, W. S. L.; Antal, M. J.; Jones, M. Formation of acrylic acid from lactic acid in supercritical water. J. Org. Chem. 1989, 54 (19), 4596−4602. (34) Oakwood, T. S.; Miller, M. R. The decarboxylation of simple fatty acids. J. Am. Chem. Soc. 1950, 72 (4), 1849−1849. (35) Kabyemela, B. M.; Adschiri, T.; Malaluan, R.; Arai, K. Degradation kinetics of dihydroxyacetone and glyceraldehyde in

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: +61 8 9334 8031. Fax: +61 8 9334 8001. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank Milan Chovancek and Tuyen Pham for conducting the CE, IC, TOC, and ICP-AES analyses. We thank Greg Power of Arriba Consulting Pty Ltd. for his useful suggestions to improve this paper. The financial support of the Parker CRC for Integrated Hydrometallurgy Solutions (established and supported under the Australian Government’s Cooperative Research Centres Program) and the CSIRO Minerals Down Under Flagship is gratefully acknowledged.

■

REFERENCES

(1) Power, G.; Loh, J. Organic compounds in the processing of lateritic bauxites to alumina: Part 1: Origins and chemistry of organics in the Bayer process. Hydrometallurgy 2010, 105 (1−2), 1−29. (2) Power, G.; Loh, J. S. C.; Niemelä, K. Organic compounds in the processing of lateritic bauxites to alumina: Addendum to Part 1: Origins and chemistry of organics in the Bayer process. Hydrometallurgy 2011, 108 (1−2), 149−151. (3) Loh, J. S. C.; Brodie, G. M.; Power, G.; Vernon, C. F. Wet oxidation of precipitation yield inhibitors in sodium aluminate solutions: Effects and proposed degradation mechanisms. Hydrometallurgy 2010, 104 (2), 278−289. (4) Power, G.; Loh, J. S. C.; Vernon, C. Organic compounds in the processing of lateritic bauxites to alumina: Part 2: Effects of organics in the Bayer process. Hydrometallurgy 2012, 127−128, 125−149. (5) Smith, P. G.; Watling, H. R.; Crew, P. The effects of model organic compounds on gibbsite crystallization from alkaline aluminate solutions: Polyols. Colloids Surf., A 1996, 111 (1−2), 119−130. (6) van Bronswijk, W.; Watling, H. R.; Yu, Z. A study of the adsorption of acyclic polyols on hydrated alumina. Colloids Surf., A 1999, 157 (1−3), 85−94. (7) Watling, H. Gibbsite crystallization inhibition2. Comparative effects of selected alditols and hydroxycarboxylic acids. Hydrometallurgy 2000, 55 (3), 289−309. (8) Beach, E. S.; Duran, J. L.; Horwitz, C. P.; Collins, T. J. Activation of hydrogen peroxide by an Fe-TAML complex in strongly alkaline aqueous solution: homogeneous oxidation catalysis with industrial significance. Ind. Eng. Chem. Res. 2009, 48 (15), 7072−7076. (9) Machold, T.; Laird, D. W.; Rowen, C. C.; May, P. M.; Hefter, G. T. Decomposition of Bayer process organics: Phenolates, polyalcohols, and additional carboxylates. Hydrometallurgy 2011, 107 (3−4), 68−73. (10) Arnswald, W.; Kaltenberg, H. G.; Guhl, E. Removal of organic carbon from Bayer liquor by wet oxidation in tube digesters. Light Met. 1991, 23−27. (11) Laird, D. W.; Rowen, C. C.; Machold, T.; May, P. M.; Hefter, G. Volatile products from the degradation of organics in a synthetic Bayer liquor. Ind. Eng. Chem. Res. 2013, 52 (10), 3613−3617. (12) Kishida, H.; Jin, F. M.; Zhou, Z. Y.; Moriya, T.; Enomoto, H. Conversion of glycerin into lactic acid by alkaline hydrothermal reaction. Chem. Lett. 2005, 34 (11), 1560−1561. (13) Ramirez-Lopez, C. A.; Ochoa-Gomez, J. R.; Fernandez-Santos, M.; Gomez-Jimenez-Aberasturi, O.; Aonso-Vicario, A.; TorrecillaSoria, J. Synthesis of lactic acid by alkaline hydrothermal conversion of glycerol at high glycerol concentration. Ind. Eng. Chem. Res. 2010, 49 (14), 6270−6278. (14) Ramirez-Lopez, C. A.; Ochoa-Gomez, J. R.; Gil-Rio, S.; GomezJimenez-Aberasturi, O.; Torrecilla-Soria, J. Chemicals from biomass: Synthesis of lactic acid by alkaline hydrothermal conversion of sorbitol. J. Chem. Technol. Biotechnol. 2011, 86 (6), 867−874. 5580



Article

subcritical and supercritical water. Ind. Eng. Chem. Res. 1997, 36 (6), 2025−2030. (36) Kabyemela, B. M.; Adschiri, T.; Malaluan, R. M.; Arai, K. Kinetics of glucose epimerization and decomposition in subcritical and supercritical water. Ind. Eng. Chem. Res. 1997, 36 (5), 1552−1558. (37) Srokol, Z.; Bouche, A.-G.; van Estrik, A.; Strik, R. C. J.; Maschmeyer, T.; Peters, J. A. Hydrothermal upgrading of biomass to biofuel; studies on some monosaccharide model compounds. Carbohydr. Res. 2004, 339 (10), 1717−1726. (38) Sasaki, M.; Hayakawa, T.; Arai, K.; Adichiri, T. Measurement of the rate of retro-aldol condensation of D-xylose in subcritical and supercritical water. Hydrothermal Reactions and Techniques: Proceedings of the 7th International Symposium on Hydrothermal Reactions, Changchun, China, Dec 14−18, 2003; Feng, S. H., Chen, J. S., Shi, Z., Eds.; World Scientific: Singapore, 2003; pp 169−176. (39) Smith, M. B.; March, J. March’s Advanced Organic Chemistry, 6th ed.; Wiley: New York, 2007.

5581


Understanding Hydrogen in Bayer Process Emissions. 3. Hydrogen

Recommend Documents