Catalytic Conversion of Dihydroxyacetone to Lactic Acid with Brønsted Acids and Multivalent Metal Ions

Home , Dihydroxyacetone, Glyceraldehyde

S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015) 575 Catalytic Conversion of Dihydroxyacetone to Lactic Acid with Brønsted Acids and Multivalent Metal Ions

S. Lux* and M. Siebenhofer doi: 10.15255/CABEQ.2014.2110 Graz University of Technology, Institute of Chemical Engineering Original scientific paper and Environmental Technology, NAWI Graz, Received: September 11, 2014 Inffeldgasse 25C, 8010 Graz, Austria Accepted: November 26, 2015

The exploitation of by-products from chemical processes shows high potential for the development of new synthesis routes for valuable chemicals. Glycerol, provided as a by-product from the biodiesel manufacturing process, is a potential feedstock chemical. From dihydroxyacetone, a primary oxidation product of glycerol, lactic acid may be obtained. The catalytic effect of Brønsted acids and multivalent metal ions on the conversion of dihydroxyacetone to lactic acid in aqueous solutions was investigated. Lactic acid yields of 83 % were achieved when carrying out the reaction under reflux boiling conditions with the catalyst HCl in excess. High acidity of the reaction solution is essential for the dehydration of dihydroxyacetone to pyruvic aldehyde. Consecutive conversion of pyruvic aldehyde to lactic acid was accelerated by multivalent metal ions (e.g. Al3+). The

Lewis acid Al2(SO4)3 provides both acidic reaction conditions for dehydration of dihydroxyacetone to pyruvic aldehyde and acceleration of lactic acid formation from pyruvic

aldehyde. Lactic acid yields of up to 78 % were obtained with Al2(SO4)3. Key words: lactic acid, Brønsted acid, Lewis acid, dihydroxyacetone, glycerol

Introduction and cosmetics industries, growth in demand is expected due to expanding polymer markets (biode- At present, a vast majority of products such as gradable synthetics) and elevated demand in the plastics, cosmetics or pharmaceuticals are still pe- chemical sector due to ecologically friendly sol- troleum-based. A shift towards renewable bio-based vents and oxygenated chemicals.1,2 Nevertheless, feedstock will be inevitable in the future. However, commercial success depends on the development of the use of biomass for chemical production impli- an effective production method for highly pure lac- cates the redesign of existing processes and involves tic acid from abundant non-food biomass. new challenges. Handling of aqueous reaction sys- Oxidation of one of the three alcoholic func- tems and the need for adapted separation techniques tions of glycerol gives access to the synthesis of the from aqueous solutions come to the fore. Retrieval trioses dihydroxyacetone (1,3-dihydroxypropan-2-one, of biomass which does not compete with the food DHA) and glyceraldehyde (2,3-dihydroxypropanal, production draws attention to the exploitation of re- GLAH), which may further be catalytically convert- sidual materials. Glycerol (propane-1,2,3-triol) is a ed to lactic acid. versatile molecule obtained as a by-product in the biodiesel manufacturing process. Due to increasing Dihydroxyacetone from glycerol is accessible biodiesel production over the last decades, it may via different synthesis routes. Chemical synthesis be regarded as a potential renewable feedstock involves liquid phase self-condensation of formal- 3 chemical. Exploitation of glycerol for the produc- dehyde with thiazolium catalysts. Much research tion of lactic acid is a promising approach. has been done in the area of noble metal catalysed oxidation of glycerol.4–6 Partial electrochemical oxi- Lactic acid (2-hydroxypropionic acid) is emerg- dation of glycerol to dihydroxyacetone is carried ing as a building block in a new generation of bio- out with different electrode types in neutral7, alka- based materials. With both a hydroxyl and a carbox- 8–11 10,11 ylic acid group, lactic acid may undergo many line and acidic media . The first microbiological dehydration of glycerol to dihydroxyacetone reactions. It therefore represents a central feedstock 12 for the chemical industry, e.g., in the production of was reported in 1898 by Bertrand. Since then, sev- propylene glycol, acrylic acid or different conden- eral routes for microbiological synthesis of dihy- sation products. Besides applications in the food droxyacetone have been developed and patented. Mainly vinegar bacteria type Acetobacter and Glu- *Corresponding author: [email protected] conobacter are used.13,14 576 S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015)

Scheme 1 – Conversion of dihydroxyacetone (DHA) and glyceraldehyde (GLAH) to lactic acid (LA)

Non-fermentative catalytic chemical conver- Bicker et al.22 investigated the catalytic effect sion of dihydroxyacetone and glyceraldehyde to of the metal ions Co2+, Ni2+, Cu2+, and Zn2+ on the lactic acid or lactate is currently gaining great inter- formation of lactic acid from hexoses and trioses in est in both academia and industry. The reaction pro- sub- and supercritical water. Zn2+ gave the best re- ceeds via dehydration of the triose yielding pyruvic sults with a lactic acid yield of 86 % (g g–1) from aldehyde (2-oxopropanal, PA) as intermediate dihydroxyacetone at 300 °C (p = 25 MPa). Rasren- (Scheme 1). Subsequent conversion of pyruvic al- dra et al.23 presented a study of the catalytic effect dehyde to lactic acid is an intramolecular Cannizza- of 26 metal salts on the conversion of dihydroxyac- ro type rearrangement reaction. The key reaction etone and glyceraldehyde to lactic acid in water. A step is the shift of a hydride.15 temperature range of 120–180 °C was investigated. Conversion properties of dihydroxyacetone to Al3+ salts were identified as the most promising cat- pyruvic aldehyde under acid exposure were first alysts with lactic acid yields of >90 % at 140 °C mentioned by Pinkus16 in 1898. Prey et al.15 report- and a reaction time of 90 minutes. ed the formation of lactic acid when treating dihy- These studies highlight that both acids and multi- droxyacetone, glyceraldehyde and pyruvic aldehyde valent metal ions are capable of catalysing lactic acid with acids. Acid catalysed isomerization of glycer- formation from dihydroxyacetone. However, an inves- aldehyde and dihydroxyacetone and dehydration to tigation of liquid phase synthesis of lactic acid from pyruvic aldehyde were investigated by Lookhart dihydroxyacetone at moderate temperatures (room and Feather17. Strongly acidic operating conditions temperature to reflux boiling conditions at ambient favour dehydration. The rate of dihydroxyacetone pressure) is still lacking. Apart from potential energy dehydration is higher than the rate of glyceralde- savings and equipment simplifications, by-product hyde dehydration. formation is generally reduced or even suppressed at The use of heterogeneous Brønsted and Lewis lower reaction temperatures. For this reason, an exten- acid catalysts and the use of homogeneous multiva- sive study on liquid phase conversion of dihydroxyac- lent metal salts as catalysts have been reported in etone to lactic acid in aqueous solution was conducted. literature. Heterogeneous catalysts include zeo- The Brønsted acids HCl, H2SO4, H3PO3, the strongly lites18, hydroxyapatite-supported Lewis acids19, sub- acidic ion exchange resin Lewatit®K2620, and acidic stituted mesoporous MCM-41 materials20, and tin hydrolysing metal salts were used as catalysts. The in- ion-exchanged montmorillonite21. Their application vestigation of the conversion behaviour of dihydroxy- is limited to lactate synthesis in alcoholic medium acetone in acidic aqueous medium is especially rele- as coking and irreversible structural modifications vant, as biomass derived precursors are often obtained occur in aqueous solutions. These result in deactiva- after acid treatment of biomass. Consequently, both tion of the heterogeneous catalysts when the reac- water and acid are present in the reaction solution. tion is carried out in water for lactic acid synthe- Furthermore, concurrence of homogeneous Brønsted sis.18,20 For the exploitation of biomass derived acids with multivalent metal ions was discussed. The precursors, the handling of aqueous reaction sys- reaction kinetics with different catalysts was modelled tems is crucial. and the most promising catalyst was identified. S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015) 577 Experimental tors (2 L) equipped with a reflux condenser, heating coil, and stirrer with a PTFE stirring blade. The re- Chemicals action was started by admixture of the catalyst to the preheated feed. The reaction temperature was Dihydroxyacetone (dimer, puriss.) was sup- controlled and kept constant throughout the reaction plied by Merck. Pyruvic aldehyde (~40 % in H O) 2 (± 0.5 °C). was provided by Sigma-Aldrich. Lactic acid in aqueous solution (1 mol L–1) was provided by Rie- For reuse of the aluminium-laden resin, the res- del-de Haën. Hydrochloric acid (HCl, 37 %, p.a.) in was separated from the reaction mixture, washed with deionised water, vacuum dried and used again. and sulphuric acid (H2SO4, 95 %) were purchased from Carl Roth. H3PO4 (85 %) was purchased from Experiments were replicated at least three to Merck. Metal salts Al2(SO4)3 ∙ 16H2O (≥95 %, p.a.) five times. Mean deviation of dihydroxyacetone, and ZnSO4 ∙ 7H 2O (≥99.5 %, p.a.) were supplied by pyruvic aldehyde and lactic acid concentrations Carl Roth, SnCl2 ∙ 2H 2O (≥98 %, p.a.), Al(NO3)3 ∙ 9H 2O were below 3 %, respectively. Results are given as (≥98.5 %, p.a.), CrCl2 (90 %), and ZrOCl2 ∙ 8H 2O mean values. (≥99 %, pro analysi) by Merck, AlCl3 (≥99.9 %) by Analytics Sigma Aldrich; Cr4(SO4)5(OH)2 (100 %) by Rie- del-de Haën; and CuSO4 (99 %, pur) by Fluka. The ion exchange resin Lewatit®K2620 (cation exchange Dihydroxyacetone, the intermediate pyruvic al- capacity ≥ 5.2 eq kg–1) was provided by Lanxess. dehyde and the product lactic acid were quantita- All chemicals, except the ion exchange resin, were tively analysed by high-performance liquid chroma- used without further pretreatment. The ion exchange tography (Ultimate 3000, Dionex). The substances were separated with the columns Rezex ROA Or- resin was washed with deionized water until the su- + pernatant liquid was colourless and the pH neutral. ganic Acid H (Phenomenex) and Acclaim OA Then it was vacuum dried until its mass remained (Dionex). Detection was carried out with a variable constant. wavelength UV/VIS detector at 190 nm and 210 nm (Dionex). For elution of the substances from Rezex Aqueous solutions of dihydroxyacetone were ROA 0.005 mol L–1 H SO at 50 °C was used. From prepared by dissolving dimeric dihydroxyacetone in 2 4 Acclaim OA column substances were eluted with deionized water. Dissolution results in the monom- 0.1 mol L–1 Na SO (pH 2.65) at 30 °C. The pH was erisation of the dimeric form.24 In most experiments, 2 4 adjusted with methane sulphonic acid. the concentration of dihydroxyacetone was 50 g L–1 –1 (0.56 mol L ). This value is representative for dihy- Modelling of the reaction kinetics droxyacetone solutions derived from the fermenta- tion of glycerol. Lactic acid formation from dihydroxyacetone was formally best modelled as a dual step consecu- Experimental setup tive reaction. Both reaction steps are first order with respect to the corresponding educt. According to the Homogeneous catalytic experiments were car- mechanism shown in Equation 1, k is the rate con- ried out in batch mode in reactors (500 mL) 1 stant for the conversion of dihydroxyacetone to py- equipped with a reflux condenser, heating coil, and ruvic aldehyde, k the rate constant for the consecu- magnetic stirrer (Heidolph MR 3003 control). Ho- 2 tive conversion of pyruvic aldehyde to lactic acid. mogeneous catalysts included mineral acids and soluble metal salts. The reaction was started by ad- kk12 mixture of preheated catalyst and feed solutions. DHA→→PA LA (1) The reaction temperature was controlled and kept constant throughout the reaction (± 0.5 °C). Kinetics of by-product formation was not mod- ® 3+ elled but taken into account through implementa- For loading of Lewatit K2620 with Al , the tion of the selectivity factor S. S represents the mo- sodium form of the macroreticular resin was used. lar fraction of feed dihydroxyacetone, which is The loading procedure included stirring of Lewa- converted to pyruvic aldehyde and lactic acid via tit®K2620 in aqueous solutions of Al (SO ) ∙ 16H O 2 4 3 2 dehydration and Cannizzaro rearrangement, respec- at room temperature for five hours, washing with tively. deionised water and drying in a vacuum dryer until its mass remained constant. Excess of the metal The conversion of acetic acid was modelled ac- ions Al3+ was assured. At least a threefold molar cording to Equation 2. 3+ amount of Al was provided per mole of functional −⋅kt cc=⋅e 1 (2= resin site. This procedure was carried out twice. DHA DHA,0

Experiments with aluminium-laden Lewa- The rate constant k1 was derived by minimizing tit®K2620 were carried out in batch mode in reac- the mean squared deviation of the calculated and 578 S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015) the measured dihydroxyacetone concentrations. The Table 1 – Catalytic effect of different Brønsted acids on the conversion of dihydroxyacetone (X ) and the yields of pyruvic formation and conversion of pyruvic aldehyde was DHA aldehyde (YPA) and lactic acid (YLA) after a reaction time of 1 h modelled according to Equation 3. After determina- –1 –1 and 6 h (cDHA,0 = 0.56 mol L , cH+ = 1 mol L , reflux boiling tion of k1, k2 and SPA were derived by the method of conditions) least squares with respect to calculated and experi- X Y Y X Y Y mental pyruvic aldehyde concentrations. DHA PA LA DHA PA LA Catalyst % % % % % % cS⋅⋅k 1 h 6 h DHA,0 PA 1 −−kt12⋅⋅kt cPA = ⋅−()ee (3) Lactic acid[a] 9 7 2 39 26 4 kk21− [a] Lactic acid formation was modelled according HCl 100 87 7 100 54 30 [a] to Equation 4. H2SO4 100 87 4 100 54 19

[a] H3PO4 32 23 ≤ 0.1 85 65 1 ccLA = DHA,0 ⋅⋅SLA Lewatit®K2620[b] 88 88 ≤ 1.1 100 72 28   (4) 1 −−kt12⋅⋅kt [a] [b] ⋅⋅1+ ⋅−()ke21ke⋅  100 °C 98 °C  kk12− 

With known rate constants k1 and k2, SLA was With HCl and H2SO4 as the strongest acids, di- derived by minimizing the mean squared deviation hydroxyacetone was completely converted within of the calculated and the measured concentrations one hour. The yields of lactic acid were 7 % and of lactic acid. 4 %, respectively. After a reaction time of six hours, Temperature dependence of k and k was based lactic acid yields were 30 % with HCl and 19 % 1 2 with H SO . From the results in Table 1, it can be on Arrhenius’ law (Equation 5). 2 4 depicted that after six hours, 16 % of dihydroxyac- E − A etone must have reacted to unfavourable by-prod- =⋅RT⋅ kAe (5) ucts when using HCl as catalyst. With H2SO4, 27 % of dihydroxyacetone (on a molar basis) must have been converted to by-products after six hours. Results and discussion In contrast to complete conversion within one hour with HCl and H2SO4, only 32 % of dihydroxy- Catalysis with Brønsted acids acetone were converted with H3PO4 and 88 % with Lewatit®K2620. Lactic acid was not formed in re- The catalytic effect of the homogeneous Brøn- markable amounts (≤ 1 %) during that time. With sted acids HCl, H SO , and H PO on the formation 2 4 3 4 H3PO4, the yield of lactic acid was still low (1 %) of lactic acid was investigated. The ion exchange after six hours, whereas with Lewatit®K2620 yields resin Lewatit®K2620 was used as a heterogeneous of 28 % could be achieved. Limited lactic acid Brønsted acid catalyst. Moreover, the autocatalytic yields of 28 % are dedicated to the lower rates of effect of lactic acid was screened. reaction with Lewatit®K2620. According to the material balance, no by-product formation was ob- Comparison of different Brønsted acids served during a reaction time of six hours. Further- more, easy separation from the reaction mixture and For comparison of the catalytic effect of the reusability of Lewatit®K2620 are advantageous. No Brønsted acids HCl, H2SO4, and H3PO4 on the for- loss of catalytic activity was observed after repeated mation of lactic acid, experiments were performed use under the conditions specified (six times for six at reflux boiling conditions with dihydroxyacetone hours). –1 concentrations of 0.56 mol L and acid concentra- The autocatalytic effect of lactic acid on the –1 tions of cH+ = 1 mol L , respectively. For investiga- conversion of dihydroxyacetone was also investi- tion of the autocatalytic effect of lactic acid, exper- gated. After one hour, 9 % of dihydroxyacetone iments were performed under the same reaction were converted when lactic acid was present in the –1 conditions (cDHA,0 = 0.56 mol L , reflux boiling feed solution. Due to its lower acidity, the yields of conditions) and an initial lactic acid concentration lactic acid were limited to 2 % after a reaction time of 1 mol L–1. of one hour. High acidity of the reaction solution results in This shows the high impact of acidity on the fast conversion of dihydroxyacetone. As depicted rates of conversion. High acidity of the reaction from Table 1, the rate of dihydroxyacetone conver- solution favours both conversion of dihydroxyace- sion increases with increased acidity. tone and consecutive rearrangement of pyruvic al- S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015) 579 dehyde to lactic acid. Highest lactic acid yields were achieved with HCl due to its high acidity. Since the highest reaction rates and higher selectiv- ities towards lactic acid in contrast to the catalyst

H2SO4 were achieved with HCl, further investiga- tions were based on HCl as catalyst.

Catalysis with HCl Figure 1 representatively shows the conversion of dihydroxyacetone to lactic acid via the intermediate pyruvic aldehyde with HCl. Dihydroxyace- tone was completely converted within 10 minutes when carrying out the reaction with 3 mol L–1 HCl and boiling under reflux conditions. Subsequent conversion of the intermediate pyruvic aldehyde to lactic acid was rate determining. The rate of lactic Fig. 2 – Effect of the reaction temperature on the yield of lactic acid (LA) with HCl (c = 0.56 mol L–1, acid formation was low at the beginning. This peri- DHA,0 c = 3 mol L–1) od of induction is characteristic for consecutive re- HCl actions. Isomeric glyceraldehyde was not detected in remarkable amounts. depicted from Figure 2. The effect of the reaction temperature on intermediate formation of pyruvic aldehyde and its consecutive conversion can be seen in Figure 3. The initial dihydroxyacetone concentration was 0.56 mol L–1 in those experiments; HCl concentration was 3 mol L–1.

Fig. 1 – Conversion of dihydroxyacetone to lactic acid via the –1 intermediate pyruvic aldehyde with HCl (cDHA,0 = 0.56 mol L , –1 cHCl = 3 mol L , boiling at reflux conditions, ambient pressure)

With HCl, the transparent reaction solution changed colour to yellow and brownish with prog- Fig. 3 – Effect of the reaction temperature on the yield of –1 ress in reaction. Polymerized pyruvic aldehyde is pyruvic aldehyde (PA) with HCl (cDHA,0 = 0.56 mol L , –1 assumed to be responsible for the change in colour. cHCl = 3 mol L ) Dark brown to black precipitates segregated at high acid concentrations and high reaction temperatures. The yield of lactic acid increased significantly This was also observed for catalysis with H2SO4 25 with increasing reaction temperatures. No lactic and H3PO4. Hahn and Schales described precipitates similar to the species found in this project as acid formation was observed for six hours when segregated acids with humic acid-like properties. carrying out the reaction below 50 °C. As expected from the negligible lactic acid formation, formation Effect of reaction temperature of pyruvic aldehyde was suppressed at low temperatures (25 °C), too. The yield of pyruvic alde- The liquid-phase conversion of dihydroxyace- hyde was less than 6 % after six hours at 25 °C. At tone to lactic acid with HCl was investigated in a 50 °C, 70 % of dihydroxyacetone was converted temperature range of 25 °C to 100 °C. The yield of after six hours. A yield of 70 % of pyruvic aldehyde lactic acid at different reaction temperatures can be after six hours showed that by-product formation 580 S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015) from dihydroxyacetone was negligible at these re- Effect of HCl and feed dihydroxyacetone concentrations action conditions resulting in complete conversion of dihydroxyacetone to pyruvic aldehyde. However, At a fixed feed dihydroxyacetone concentration –1 lactic acid formation was suppressed due to the low of 0.56 mol L HCl concentrations were varied –1 –1 reaction temperature. between 0.5 mol L and 5 mol L , respectively. According to the HCl concentrations in the solu- At 75 °C, the yield of lactic acid was still low tion, a boiling point temperature between 98 °C and (16 %) after six hours. A remaining pyruvic alde- 104 °C was observed. hyde concentration of 0.46 mol L–1 and complete conversion of dihydroxyacetone within 2.5 hours Figure 5 shows the strong impact of the acidity indicated that by-product formation was still of mi- of the reaction solution on the rate of lactic acid nor relevance (below 2 %). formation again. Complete conversion of dihydroxyacetone was achieved within 45 minutes at 85 °C and within 10 minutes at 100 °C. At 100 °C, the peak pyruvic aldehyde concentration was obtained within 5 minutes. Consecutive rearrangement to lactic acid was completed within 6 hours. The yield of lactic acid was 78 %. The temperature dependency was modelled according to Arrhenius’ law (Figure 4). Table 2 lists the activation energies EA and pre-exponential factors A for both reaction steps. Activation energies and the pre-exponential factors confirm that lactic acid formation from pyruvic aldehyde is rate-determining. From the Arrhenius’ plot, no change of reaction mechanism within the investigated temperature range could be deduced. Fig. 5 – Effect of HCl concentrations on the yield of lactic acid –1 at operation under reflux boiling conditions (cDHA,0 = 0.56 mol L ; –1 –1 98 °C (cHCl = 0.5 mol L ), 99 °C (cHCl = 1 mol L ), 100 °C (cHCl –1 –1 –1 = 2 mol L ), 101 °C (cHCl = 3 mol L ), 103 °C (cHCl = 4 mol L ) –1 and 104 °C (cHCl = 5 mol L ))

HCl concentrations of 0.5 mol L–1 resulted in complete conversion of dihydroxyacetone within three hours. The results indicated that dihydroxyacetone was entirely converted to pyruvic aldehyde. Increasing HCl concentrations to 1 mol L–1 reduced the time for complete conversion of dihydroxyacetone to one hour. With HCl in a concentration of 5 mol L–1 complete dihydroxyacetone conversion was achieved within 5 minutes. Apart from higher reaction rates, increasing HCl concentrations resulted in increasing lactic acid Fig. 4 – Arrhenius’ plot for the first (dihydroxyacetone to selectivity. This indicates that by-product formation pyruvic aldehyde) and second (pyruvic aldehyde to lactic acid) from pyruvic aldehyde is suppressed at high acid reaction step (c = 3 mol L–1, c = 0.56 mol L–1) –1 HCl DHA,0 concentrations. With 5 mol L HCl, the final lactic acid yield was 83 %. Subsequent degradation of lactic acid was not observed.

Table 2 – Activation energies EA and pre-exponential factors A for the HCl catalysed conversion of dihydroxyacetone to lac- The validity of the rate law for different dihy- –1 –1 droxyacetone concentrations was investigated from tic acid (cHCl = 3 mol L , cDHA,0 = 0.56 mol L ) 25 g L–1 (0.28 mol L–1) to 75 g L–1 (0.83 mol L–1) A (h–1) E (kJ mol–1) –1 A and a constant HCl concentration of 3 mol L at 1. DHA → PA (7.29 ± 0.02)·1015 102.4 ± 0.3 101 °C. In a second series of experiments, the valid-

18 ity of the rate law was tested for varying feed dihy- 2. PA → LA (2.35 ± 0.01)·10 132.2 ± 0.4 droxyacetone concentrations but constant 5.4-fold S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015) 581 molar excess of H+ with respect to feed dihydroxy- Table 3 – Catalytic effect of Brønsted acids with multivalent metal ions on lactic acid formation from dihydroxyacetone acetone at 101 °C. –1 –1 (cDHA,0 = 0.56 mol L , cH+ = 1 mol L , for metal/HCl experi- –1 Again, it was confirmed that the HCl concen- ments: cMe = 0.06 mol L , 99 °C, reaction time: 1 h). tration in the aqueous solution has a major impact X Y Y Y [c] acid Me DHA PA LA LA on the rate of lactic acid formation. Acidity of the % % % % reaction mixture consequently has a high impact on HCl Zn2+ 100 70 5 23 lactic acid formation. Highest lactic acid formation 2+ rates were obtained at high HCl concentrations. HCl Sn 100 ≤ 0.1 75 75 With 75 g L–1 dihydroxyacetone and 4.5 g L–1 HCl HCl Fe2+ ≥ 99.9 84 7 26 in the feed, the yield of lactic acid was 50 % after HCl Fe3+ ≥ 99.9 77 4 23 30 minutes and a final lactic acid yield of 78 % was achieved within three hours. After three hours, the HCl Al3+ ≥ 99.9 74 24 51 –1 yield of lactic acid was 71 % with 50 g L dihy- HCl Cr2+ ≥ 99.9 48 46 91 droxyacetone and 3 mol L–1 HCl in the feed, and HCl Cr3+ ≥ 99.9 45 40 91 36 % with 25 g L–1 dihydroxyacetone and 1.5 mol L–1 HCl. Lewatit®K2620[a] Al3+ [b] 42 20 8 73 At constant HCl concentration and varying Lewatit®K2620[a] Cr3+ [b] 67 13 31 89 feed dihydroxyacetone concentrations, the yield of [a]98 °C [b]doped [c]reaction time: 6 h lactic acid decreases with increasing dihydroxyacetone concentration. By material balance, the portion of dihydroxyacetone which was converted to pyru- With Al3+/HCl, the lactic acid yield was 24 % vic aldehyde and lactic acid, respectively, was de- after one hour, and 51 % after six hours. Al3+ had an termined. With 25 g L–1 feed dihydroxyacetone 87 accelerating effect on lactic acid formation without % of dihydroxyacetone were converted to pyruvic the negative side effect of precipitation. aldehyde and lactic acid after a reaction time of six Admixture of Zn2+, Fe2+, and Fe3+ had a nega- hours. In contrast to this, the percentage was lower tive impact on lactic acid formation. Even though, with 50 g L–1 (77 %) and 75 g L–1 (75 %). conversion of dihydroxyacetone was comparable for Al3+ and Zn2+, admixture of Zn2+ resulted in in- Concurrence of Brønsted acid catalysts creased formation of unidentified by-products. Low and multivalent metal ions yields of pyruvic aldehyde with Zn2+ indicated that

2+ by-product formation was favoured. Yields of lactic Experiments with multivalent metal ions Zn , acid were higher with Fe2+ than with Fe3+ or Zn2+. Sn2+, Fe2+, Fe3+, Al3+, Cr2+, and Cr3+ revealed differ- The highest catalytic effect on lactic acid for- ent catalytic effects on the acid catalysed conver- mation was achieved with Cr2+/HCl and Cr3+/HCl. sion of dihydroxyacetone to lactic acid. Experi- ® 3+ 3+ ments were performed with 0.56 mol L–1 feed Doping of Lewatit K2620 with Al and Cr concentrations of dihydroxyacetone, acid concen- showed a high catalytic effect on lactic acid forma- trations of 1 mol H+ L–1 plus catalytic amounts of tion, too. The lactic acid yields were 73 % and 89 % after six hours with Al3+ and Cr3+-doped Lewa- metal ions at 99 °C. The conversion of dihydroxy- tit®K2620. The yield was limited to 14 % with non- acetone and yields of pyruvic aldehyde and lactic doped Lewatit®K2620. Conversion rates of dihy- acid are compared in Table 3 for the various metal droxyacetone were lower with both non-doped and salts after a reaction time of one hour. doped Lewatit®K2620 with respect to the HCl cata- For combinations of Brønsted acids and metal lysed reaction. This confirms that dihydroxyacetone ions, the highest catalytic effect on lactic acid for- dehydration requires strongly acidic reaction condi- mation was observed with Sn2+. With HCl/Sn2+ both tions, not provided by the specific ion-exchange dihydroxyacetone and pyruvic aldehyde were com- resin. pletely converted within one hour. A final lactic The higher yields with Al3+-doped Lewa- acid yield of 75 % was achieved after one hour. In tit®K2620 (73 % after six hours) with respect to contrast, the yield of lactic acid after one hour was Al3+/HCl (51 %) may be attributed to the higher 7 % for HCl catalysis without Sn2+. It was shown concentrations of metal ions due to doping. Howev- that the addition of catalytic amounts of Sn2+ to the er, reusability of metal-doped Lewatit®K2620 was HCl catalysed reaction had a significant catalytic unsatisfactory. Lactic acid yields after six hours impact on lactic acid formation. Nevertheless, Sn2+ dropped to 61 % after reuse of Al3+-doped Lewa- in aqueous solution hydrolyses and precipitates. tit®K2620. Aluminium, detected in the reaction 3+ Hydrolysis was not suppressed with HCl or H2SO4 solution, confirmed that the stability of the Al -lad- in the corresponding concentration range. en resin was unsatisfactory. Limited stability of 582 S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015) metal-doped Lewatit®K2620 and toxicity of chro- Table 4 – Catalytic effect of various metal salts on lactic mium were unfavourable for Cr3+ catalyst. acid formation from dihydroxyacetone (reaction time: 1 h; c = c = 0.56 mol L–1, 99 °C) In Figure 6, dihydroxyacetone conversion, for- DHA,0 Me X Y Y Y [b] mation of the intermediate pyruvic aldehyde and the Metal salt DHA PA LA LA product lactic acid are compared for HCl catalysed % % % % 3+ [a] and HCl/Al catalysed reaction. Dihydroxyacetone Al2(SO4)3 63 21 22 73 conversion was comparable for both catalysts. Ad- Al (SO ) 97 15 57 77 mixture of catalytic amounts of Al3+ had a high ac- 2 4 3 Al(NO ) 78 28 50 77 celerating impact on the rate determining conver- 3 3 sion of pyruvic aldehyde. This selective acceleration ZnSO4 87 1 3 4 of lactic acid formation from pyruvic aldehyde may ZrOCl [a] 100 2 18 21 be attributed to chelat complexes formed of pyruvic 2 3+ 26 22 aldehyde and Al . Bicker et al. proposed that an CuSO4 n.a. – – – inversion of the polarity of the Lewis acid to a Lew- [a]c = 0.14 mol L–1 [b]reaction time: 6 h is base for metal ions could occur. The latter is able Me to form a complex between pyruvic aldehyde and the metal ion. Then rehydration of this complex Minor amounts of isomeric glyceraldehyde leads to the formation of lactic acid and the free were detected for aluminium salt catalysts. Glycer- catalyst. aldehyde was not detected with CuSO4, ZnSO4 and ZrOCl2.

Even though the use of ZnSO4 resulted in dihydroxyacetone conversion of more than 85 % after one hour, selectivity for lactic acid remained below

4 %. ZnSO4 dissolves in aqueous solution under acidic hydrolysis providing acidic media for dihydroxyacetone conversion to pyruvic aldehyde. The fact that pyruvic aldehyde was not detected in remarkable amounts accounts for a fast subsequent reaction. But side reactions were dominating and lactic acid formation was suppressed.

With ZrOCl2, dihydroxyacetone was converted within one hour. Lactic acid selectivity remained rather low at 21 %. Since pyruvic aldehyde was not detected in remarkable amounts, again by-product formation could be assumed by the material balance.

Fig. 6 – Comparison of HCl (blank data points/dashed lines) Lactic acid was not formed with CuSO4. and HCl plus Al3+ (full data points/solid lines) catalysed lactic –1 acid formation from dihydroxyacetone (cDHA,0 = 0.56 mol L , Catalysis with the Lewis acid aluminium sulphate –1 –1 cHCl = 1 mol L , cAl = 0.06 mol L , 99 °C) With respect to the high catalytic effect, high selectivity for lactic acid and advantageous dissolu- Catalysis with metal salts tion properties, further experiments were performed

Comparison of different metal salts with Al2(SO4)3 as catalyst. Two complementary effects are observed with For a catalyst screening, the soluble metal salts Al2(SO4)3. Firstly, Al2(SO4)3 dissolves under strong- Al2(SO4)3, CuSO4, ZnSO4, and ZrOCl2 were com- ly acidic hydrolysis providing acidic reaction condi- pared (Table 4). Experiments were performed with tions. High acidity of the aqueous reaction solution 0.56 mol L–1 feed dihydroxyacetone at 99 °C. is essential for dehydration of dihydroxyacetone to

Whereas Al2(SO4)3 showed high catalytic activ- pyruvic aldehyde. Secondly, the multivalent metal ity and selectivity for lactic acid, the selectivity to- ion Al3+ catalyses the intramolecular Cannizzaro re- wards lactic acid with ZnSO4 and ZrOCl2 was low. action of pyruvic aldehyde to form lactic acid. This The yield of lactic acid after six hours was 54 % may be attributed to complex formation of the inter- with Al2(SO4)3 compared to 21 % with ZrOCl2 when mediate pyruvic aldehyde and the metal ion. Rehy- a metal ion concentration of 0.06 mol L–1 was used. dration of this complex leads to the formation of

It was 73 % with Al2(SO4)3 compared to 4 % with lactic acid and then the free catalyst. Lactic acid ZnSO4 when stoichiometric metal ion concentration formation from pyruvic aldehyde is rate determin- of 0.56 mol L–1 was used. ing under acid catalysis with HCl (Table 2). S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015) 583 Similar to the aluminium-catalysed formation temperature. Whereas the lactic acid yield after six of hydroxyacetic acid from glyoxal26, lactic acid hours was low (10 %) at 70 °C, high yields of 78 % formation from pyruvic aldehyde is a first order re- were achieved at reflux boiling conditions (99 °C). action with respect to pyruvic aldehyde. Experi- Complete conversion of dihydroxyacetone was ments were performed with pyruvic aldehyde and achieved within six hours at 99 °C. At 70 °C, resid- the reaction kinetics was confirmed. ual dihydroxyacetone was still 22 % after six hours. Representative conversion graphs for the Al3+ At 99 °C, maximum yields of pyruvic aldehyde of catalysed reaction of dihydroxyacetone to lactic 30 % after 15 minutes indicated a fast consecutive acid are shown in Figure 7. reaction to lactic acid. No change in reaction mechanism was indicated from the Arrhenius’ plot within the investigated temperature range. Table 5 lists the activation ener-

gies EA and pre-exponential factors A for the first (DHA → PA) and second (PA → LA) reaction step.

Table 5 – Activation energy EA and pre-exponential factor A for aluminium-catalysed conversion of dihydroxyacetone to –1 lactic acid (catalyst: Al2(SO4 )3, cDHA,0 = cAl = 0.56 mol L )

–1 –1 A (h ) EA (kJ mol ) 1. DHA → PA (4.39 ± 0.01)·1013 93.5 ± 0.2 2. PA → LA (2.68 ± 0.01)·1012 91.8 ± 0.2

Effect of aluminium sulphate concentration

Fig. 7 – Course of dihydroxyacetone, pyruvic aldehyde, glyce The effect of the Al2(SO4)3 concentration on the raldehyde, and lactic acid concentrations for Al3+ catalysed con rate of conversion was investigated by variation of 3+ version of dihydroxyacetone to lactic acid (catalyst: Al2(SO4)3, the molar ratio of Al to dihydroxyacetone from –1 –1 cDHA,0 = 0.56 mol L , cAl = 0.83 mol L , 99 °C) 0.01 to 1.5. Feed concentration of dihydroxyacetone was 0.56 mol L–1 in the experiments. As expected, increasing Al3+ concentration resulted in the accel- Isomeric glyceraldehyde was detected at the eration of lactic acid formation. Figure 8 shows the 3+ beginning of the reaction. Both dihydroxyacetone lactic acid yields for varying Al /dihydroxyacetone and glyceraldehyde contributed to lactic acid for- feed ratio. The maximum lactic acid yield was mation. Opposite to the HCl-catalysed reaction, de- 73 %. No difference in selectivity towards pyruvic hydration of dihydroxyacetone to pyruvic aldehyde aldehyde and lactic acid was observed for varying Al3+ concentration of 0.006 to 0.83 mol L–1. was rate-determining with the catalyst Al2(SO4)3. The initially transparent reaction solution changed colour to yellowish and finally dark brown with ongoing reaction time. This was also observed when performing the conversion with the catalyst HCl. Oligo- and polymerization of pyruvic aldehyde and dihydroxyacetone may be indicated as the governing reason for colour change and formation of insoluble by-products.23

Effect of the reaction temperature

The effect of the reaction temperature on the aluminium-catalysed lactic acid formation was investigated in a temperature range of 70 °C to 99 °C (reflux boiling conditions) with 0.56 mol L–1 feed dihydroxyacetone. The molar ratio of feed Al3+/dihydroxyacetone was 1. Fig. 8 – Effect of the molar feed ratio Al3+/dihydroxyacetone As expected from the HCl-catalysed reaction, on the formation of lactic acid from dihydroxyaceto –1 lactic acid yields increased with increasing reaction ne (catalyst: Al2(SO4)3, cDHA,0 = 0.56 mol L , 99 °C) 584 S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015)

Rate constants k1 and k2 for different Al2(SO4)3 ACKNOWLEDGEMENTS concentrations are given in Table 6. An almost lin- This research was partially supported by Sonja ear correlation between k1 and k2 and the Al2(SO4)3 concentration was observed. With increasing Larissegger, Jutta Krischan and Stefan Naderer during their project work at TU Graz. The authors Al (SO ) concentration, the difference between k 2 4 3 1 also thank Herta Luttenberger for assistance with the and k reduced. This may be explained by increasing 2 HPLC, and Peter Stehring for comments that greatly acidity of the solution resulting from high Al (SO ) 2 4 3 improved the work. The authors furthermore grate- concentrations. fully acknowledge the support from NAWI Graz.

Table 6 – Rate constants k1 and k2 of the aluminium-ca talysed conversion of dihydroxyacetone to lactic Nomenclature acid (catalyst: Al (SO ) , c = 0.56 mol L–1, 2 4 3 DHA,0 A – pre-exponential factor, h–1 99 °C) c – concentration, mol L–1

cAl k1 k2 X – conversion, % (mol L–1) (h–1) (h–1) DHA – dihydroxyacetone 0.006 0.10 0.30 –1 EA – activation energy, kJ mol 0.056 0.42 0.94 k – rate constant, h–1 0.14 0.89 1.20 LA – lactic acid Me – metal 0.56 3.10 3.20 PA – pyruvic aldehyde 0.84 4.70 4.80 S – selectivity t – time, h Y – yield, % Conclusions

Synthesis of lactic acid from aqueous dihy- References droxyacetone solution is a promising approach for 1. Werpy, T., Petersen, G., Technical Report No. DOE/GO- chemical production from bio-based precursors. 102004-1992, National Renewable Energy Lab, Golden, Handling of aqueous reaction systems and integrat- CO, 2004. ed approaches including product separation are of 2. Malveda, M. P., Blagoev, M., Kumamoto, T, Lactic Acid, Its Salts and Esters. CEH Marketing Research Report, SRI prime importance for exploitation of biomass. Ca- Consulting, 2009. talysis of the liquid phase reaction was investigated 3. Matsumoto, T., Yamamoto, H., Inoue, S., Selective forma- with Brønsted acids, acidic hydrolysing metal salts, tion of triose from formaldehyde catalyzed by thiazolium and a combination thereof. Brønsted acids are often salt, J. Am. Chem. Soc. 106 (1984) 4829. present in aqueous solutions of bio-based precur- doi: http://dx.doi.org/10.1021/ja00329a031 sors. 4. Kimura, H., Tsuto, K., Wakisaka, T., Kazumi, Y., Inaya, Y., Selective oxidation of glycerol on a platinum-bismuth cata- Lactic acid formation favours high tempera- lyst, Appl. Catal. A 96 (1993) 217. tures and strongly acidic operating conditions. Lac- doi: http://dx.doi.org/10.1016/0926-860X(90)80011-3 tic acid yields of 83 % were achieved when carry- 5. Kimura, H., Selective oxidation of glycerol on a plati- ing out the reaction under reflux boiling conditions num-bismuth catalyst by using a fixed bed reactor, Appl. Catal. A 105 (1993) 147. and the catalyst HCl in excess. Lactic acid forma- doi: http://dx.doi.org/10.1016/0926-860X(93)80245-L tion from dihydroxyacetone proceeds via two con- 6. Demirel, S., Lehnert, K., Lucas, M., Claus, P., Use of re- secutive reaction steps with pyruvic aldehyde serv- newables for the production of chemicals: Glycerol oxida- ing as intermediate. Conversion of pyruvic aldehyde tion over carbon supported gold catalysts, Appl. Catal. B 70 is rate controlling under acid catalysis. The rate of (2007) 637. doi: http://dx.doi.org/10.1016/j.apcatb.2005.11.036 pyruvic aldehyde conversion is accelerated with 3+ 2+ 7. Stehring, P., Lux, S., Hilber, T., Letonja P., Siebenhofer, M., catalytic amounts of multivalent Al and Sn . Oxidation von Glycerin an Diamantelektroden, Chem. Ing. The Lewis acid Al2(SO4)3 provides both acidic Tech. 80 (2008) 1387. reaction conditions for dehydration of dihydroxyac- doi: http://dx.doi.org/10.1002/cite.200750475 etone to pyruvic aldehyde and acceleration of lactic 8. Tkaczuk, P., Elektrochemische Oxidation von Glycerin und acid formation from pyruvic aldehyde. Both acid Glycerinderivaten an einer Platin-Anode in verschiedenen 3+ Zelltypen, Shaker, Aachen, 1994. and Al catalysis is promising for the synthesis of 9. Weber, U., Die elektrochemische Oxidation von Glycerin an lactic acid from aqueous dihydroxyacetone solu- unterschiedlichen Elektrodenmaterialien, Shaker, Aachen, tions. 1994. S. LUX and M. SIEBENHOFER, Catalytic Conversion of Dihydroxyacetone to Lactic…, Chem. Biochem. Eng. Q., 29 (4) 575–585 (2015) 585

10. Roquet, L., Belgsir, E. M., Leger, J.-M., Lamy, C., Kinetics 19. Zhang, Z., Zhao, Z., Hydroxyapatite Supported Lewis Acid and mechanisms of the electrocatalytic oxidation of glycer- Catalysts for the Transformation of Trioses in Alcohols, ol as investigated by chromatographic analysis of the reac- Chin. J. Catal. 32 (2011) 70. tion products: potential and pH effects, Electrochim. Acta doi: http://dx.doi.org/10.1016/S1872-2067(10)60162-3 39 (1994) 2387. 20. Li, L., Stroobants, C., Lin, K., Jacobs, P. A., Sels, B. F., doi: http://dx.doi.org/10.1016/0013-4686(94)E0190-Y Pescarmona, P. P., Selective conversion of trioses to lac- 11. Stehring, P., Lux, S., Letonja, P., Siebenhofer, M., Ano- tates over Lewis acid heterogeneous catalysts, Green Chem. dische Oxidation von Glycerin in einer Dünnschichtzelle, 13 (2011) 1175. Chem. Ing. Tech. 81 (2009) 1224. doi: http://dx.doi.org/10.1039/c0gc00923g doi: http://dx.doi.org/10.1002/cite.200950037 21. Wang, J., Masui, Y., Onaka, M., Conversion of triose sugars 12. Charney, W. (Schering Corporation), US Patent Appl. with alcohols to alkyl lactates catalyzed by Brønsted acid 4,076,589,28, Feb 1978. tin ion-exchanged montmorillonite, Appl. Catal. B 107 (2011) 135. 13. Ohrem, H.-L., Westmeier, F. (Merck Patent GmbH.), US doi: http://dx.doi.org/10.1016/j.apcatb.2011.07.006 Patent Appl. 5,770,411, 23 Jun 1998. 22. Bicker, M., Endres, S., Ott, L., Vogel, H., Catalytical con- 14. Wethmar, M., Dechwer, W.-D., Semisynthetic culture me version of carbohydrates in subcritical water: A new chem- dium for growth and dihydroxyacetone production by ical process for lactic acid production, J. Mol. Catal. A 239 Gluconobacter oxydans, Biotechnol. Techniques 13 (1999) (2005) 151. 283. doi: http://dx.doi.org/10.1016/j.molcata.2005.06.017 doi: http://dx.doi.org/10.1023/A:1008978903231 23. Rasrendra, C. B., Fachri, B. A., Makertihartha, I. G. B. N., 15. Prey, V., Berbalk, H., Steinbauer, E., Über die Bildung von Adisasmito, S., Heeres, H. J., Catalytic conversion of dihy- Milchsäure aus Triosen durch Einwirkung von Säuren, 2. droxyacetone to lactic acid using metal salts in water, Mitteilung, Chemical Monthly 93 (1962) 237. Chem. Sus. Chem. 4 (2011) 768. 16. Pinkus, G., Über die Einwirkung von Benzhydrazid auf doi: http://dx.doi.org/10.1002/cssc.201000457 Glucose, Ber. dt. chem. Ges. 31 (1898) 31. 24. Kobayashi, Y., Igarashi, T., Takahashi, H., Higasi, K., In- doi: http://dx.doi.org/10.1002/cber.18980310108 frared and Raman studies of the dimeric structures of 17. Lookhart, G. L., Feather, M. S., Acid-catalyzed isomeriza- 1,3-dihydroxyacetone, D-(+)- and DL-glyceraldehyde, J. tion and dehydration of DL-glyceraldehyde and 1,3-dihy- Mol. Struct. 35 (1976) 85. droxy-2-propanone, Carbohydr. Res. 60 (1978) 259. doi: http://dx.doi.org/10.1016/0022-2860(76)80104-4 doi: http://dx.doi.org/10.1016/S0008-6215(78)80033-0 25. Hahn, G. C., Schales, O., Selbstkondensation des Methyl- 18. West, R. M., Holm, M. S., Saravanamurugan, S., Xiong, J., glyoxals, Ber. dt. chem. Ges. 67 (1934) 1816. Beversdorf, Z., Taarning, E., Christensen, C. H., Zeolite 26. Kiyoura, T., Kogure, Y., Synthesis of hydroxyacetic acid H-USY for the production of lactic acid and methyl lactate and its esters from glyoxal catalyzed by multivalent metal from C-3-sugars, J. Catal. 269 (2010) 122. ions, Appl. Catal. 156 (1997) 97. doi: http://dx.doi.org/10.1016/j.jcat.2009.10.023 doi: http://dx.doi.org/10.1016/S0926-860X(96)00414-0