2608
Ind. Eng. Chem. Res. 1993,32, 2608-2613
Conversion of Lactic Acid to Acrylic Acid in Near-Critical Water Carl T.Lira* a n d P e r r y J. McCrackin Department of Chemical Engineering, Michigan State University, East Lansing, Michigan 48824-1226
The dehydration of lactic acid is studied in a Hastelloy C-276 annular reactor in near-critical water. Experiments are performed at a pressure of 310 bar, a t temperatures of 320-400 "C, a t residence times of 25-110 s, and with various catalysts, including disodium hydrogen phosphate, phosphoric acid, and sodium hydroxide. Three main reaction pathways are investigated and evaluated. A temperature of 360 "C optimizes the formation of acrylic acid with molar yields as high as 5 8 % , based on conversion. The presence of phosphate salts and/or base increases the yields of acrylic acid by suppressing the competing pathways. The aging of the Hastelloy reactor for 60-70 h decreases the degradation reactions resulting in higher yields of acrylic acid. Acetic Acid. Methane, Acetone, etc.
Introduction Lactic acid (a-hydroxypropionic acid) is a commercial fine chemical used in food and medical applications which is readily available via fermentation of biomass. Lactic acid's attractiveness as a chemical feedstock is enhanced by the nearly unlimited renewable sources of biomass. Dehydration of lactic acid to acrylic acid has been an objective of recent research. Acrylic acid and its esters are the primary building blocks of all acrylate polymers and plastics. Mok et al. (1989) studied the dehydration of lactic acid to acrylic acid in supercritical water at 385 "Cand 340 bar primarily with initial lactic acid concentrations of 0.1 M and residence times of approximately 30 s. They report that in the presence of a strong acid catalyst, HzS04, lactic acid exists primarily in the free acid form and the reaction is fast and produces acetaldehyde and carbon monoxide in equal proportions. This reaction is designated as pathway one, or the decarbonylation pathway shown in Figure 1. As the pH of the reactant solution is increased, the rate of conversion decreases, accompanying a major shift in the product distribution. The relative amount of carbon monoxide decreases with an increase in the levels of carbon dioxide and hydrogen. These products are indicative of pathway two-the decarboxylation pathway. The final major pathway (pathway three) is the dehydration of the lactic acid to acrylic acid, which is enhanced as the acid catalyst is removed. The yields of the acrylic acid and propionic acid reach a maximum when a small amount of base is added, but additional amounts suppress the dehydration pathway as well as the overall conversion. Propionic acid is shown experimentally to be the result of a hydrogenation of acrylic acid. Increasing temperature enhancesthe rates of all three pathways with no particular pathway favored. The lactic acid dehydration,heterogeneously catalyzed in the vapor phase, has been studied more widely. Phosphate impregnated metal oxides catalyze the dehydration between 200 and 400 "C at atmospheric pressure (Sawicki, 1988). The best yields and selectivities occur with catalysts prepared from basic solutions. The reported yields are typically in the range of 40 mol 76 with a high of 58 mol % based on lactic acid fed to the reactor. A European patent application (Paperizos, 1985) involves the catalytic conversion of lactic acid and ammonium lactate to acrylic acid in a vapor phase reaction. The catalyst is base-treated, calcined aluminum phosphate. The dehydration occurs at temperatures between 320 and 375 "C and contact times of 2-4 s. The highest reported yield of acrylic acid is 61 mol 76, with the majority of results below 40%. Several examples show very good ratios OS8S-5885/93/2632-2608$04.00/0
t CO t CH&HO
t
H20
-
WateriGas Shift
c CO2 t CH,CHO + H2
Lactic Acid !
Pathway 3
I
- ~~0
Acrylic Acid
+ C2H4+ CO?
+I +HZ Propionic Acid
Figure 1. Reaction network for lactic acid conversion in supercritical water as proposed by Mok et al. (1989).
of acrylic acid to propionic acid, as high as l O : l , while others are very poor. Paperizos shows that when the catalyst is not treated with base prior to use, the products are mainly acetaldehyde and propionic acid. The literature led us to believe that lactic acid could be converted to acrylic acid in supercritical water by utilizing catalysts other than simple acids and bases. Experimental Section Experimental investigations are performed using an annular flow reactor shown in Figure 2. The reactor is composed of a 6.4 mm 0.d. Hastelloy C-276 tube with a 1.6 mm 0.d. inner Hastelloy C-276 tube extending through the reactor. The inner tube is used as a thermocouple sheath, accessible from either end. Hastelloy is found to offer the most resistance to the supercritical water. The furnace used three zones. The first zone is a 5 cm long preheater fabricated from two half-moon-shaped blocks of aluminum clamped to the reactor tube which are heated with cartridge heaters. The middle zone consists of a 30.5 cm long 5 cm diameter cylindrical furnace. The third zone is constructed of a bead heater sandwiched between two 10-cm-squarebrass plates and it is used to compensatefor heat loss at the outlet end of the cylindrical furnace. The temperature of the preheater is controlled by a 0.6 mm diameter thermocouple inside the inner reactor tube at a position approximately 15mm past the preheater into the 1993 American Chemical Society
Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993 2609
-
Figure 2. fiperimental apparatua LLB described in the text.
furnace zone. The furnace temperature is controlled with a thermocouple at a position approximately 10 cm into the furnace as measured from the outlet end. A thermocouple placed just inside the outlet end of the furnace is usedtocontrol the temperatureofthe thirdzone. Cooling water jackets on the reactor tube at each end maintain a constant inlet temperature and cool the effluent to approximately room temperature. A temperature profile of the reactor, at typical reaction conditions, shows the temperature to be i 2 O of the desired set-point along a 28-cm section of the reactor. A Milton-Roy HPLC pump provides the pressure for the system. It is gravity fed from a 2.0-L flask containing the reactant solution. A surge vessel positioned after the sample valve, composed of two 15 cm long stainless steel tubes in series, dampens the pressure fluctuations in the system. Duringstart-up, the surge vessels are pressurized to 135 bar with helium gas and then one end of the vessel isexposedto thepressurizedaqueouseffluentstream. The pressure is maintained in the reactor system with abackpressure regulator rated to 340 bar. The liquid effluent is collected in a 50-mL buret to measure the flow rate through the system. Gas and liquid samples are extracted from the effluent at high pressure through the use of a Valco Instruments six-portswitchingvalve containinga 1.304-mLsamplelmp. The sample loop is emptied into an evacuated, stoppered test tube of known volume. The test-tube volume is determined by introducing an exact volume of air at atmospheric pressure into the evacuated test tube and recording the pressure with a calibrated pressure transducer. With the volume of the test tube known, the moles of gas before and after sampling are determined. The sample tube, with both gas and liquid samples, is removed from the sampling loop. By use of a PerkinElmer 8500 gas chromatograph (equipped with a 80/100 mesh Spherocarbm packed bed column), the composition of the gas phase is determined. The liquid phase of the sample is analyzed on a Waters 600 HPLC (equipped with a 10-fim Hypenil C-18 column and refractive index detector with a mobile phase of pH 4.2.0.1 M KlHPO, (Baker Analyzed) in water at a flow rate of 2.0 mL/min). The standardization of lactic acid requires the solution to be heated at 85 "C for 12 h before use to hydrolyze the naturally occurring oligomersof lactic acid to the free acid, otherwise inconsistent results are obtained. The data
gathered from both GC and HPLC are transferred to a spreadsheet where response factors for individual components are incorporated into the results. All experimental runs used 0.4 M lactic acid feed. The solution was prepared by mixing 40.78 g of 88.4 w t 9% lactic acid (Columbus Chemical Industries, Inc.) with HPLC grade water (Baker) in a 1.0-L volumetric flask. Catalyst (if any) was added and HPLC grade water was added to 1.0 L. Catalysts included phosphoric acid, sulfuric acid, nitric acid, sodium hydroxide, and disodium hydrogen phosphate (Baker Analyzed/Reagent Grade). The reactant solution was then sparged with argon gas 30 min before and during the experiments. Preheating was not required to hydrolyze the oligomers of lactic acid since consistent liquid analysis results were achieved after passing the liquid through the heated reactor.
Results Table I presents results of the studies. The reaction products are tabulated on a basis of grams obtained per liter of reactant solution. The water produced in the reaction is calculated from the total moles of lactic acid reacted via pathway one and pathway three. The moles of carbonmonoxide plus themolesof acrylic acid, propionic acid,andethyleneareequaltothemolesofwaterproduced. Hydrogen gas is a product of the pathway two reaction and it is produced in such small mass quantities that it is neglected in the overall mass balance. The Table I column labeled "reaction effluent (g/L)" is the sum of the mass of all reaction effluents per liter of feed solution. The molar yield, based on feed (BOF) is calculated by the moles of acrylic acid produced, divided by the theoretical moles of lactic acid fed into the reactor. The molar yield, based on conversion (BOO is calculated by the moles of acrylic acid divided by the moles of lactic acid reacted to products (calculated from "reaction effluent (g/L)" minus "lactic acid"). The percentage of lactic acid reacting by pathway three is calculated by dividing the number of moles of lactic acid reacted via pathway three (acrylic propionic ethylene) by the moles of lactic acid reacted. The remaining percentage of lactic acid reacted is divided among the other two pathways based on the ratio of the moles of carbon monoxide and carbon dioxide produced (pathway one and pathway two, respectively). Carbon dioxide associated with ethylene formation is excluded in the last calculation. The yields of certain products (acetic acid, methane, etc.) are neglected in the calculated yields because they are negligible in comparison to the specified pathways.
+
+
Discussion Aging of the Hastelloy C-276 reactor has a dramatic effecton the threepathwaysdescribedinFigure1. During aging, there is a dramatic shift in carbon monoxide and carbon dioxide yields. Toexplore the aging phenomenon, we collected data for a reactor in the first 4 h of use and again after 70 h of use when virtually no further change in pathways were detected. The data for comparison were collected during two 3-h periods where the residence time was varied between 30 and 120 8 at 360 OC and 310 bar. Data in Table I show a comparison of the absolute yield of carbon dioxide and carbon monoxide for the two data sets (runs 1-7 and 8-12), The carbon dioxide yield decreases with aging while the carbon monoxide yield is largely unaffected. The table shows the acrylic acid pathway isnotsignificantlyaffected,buttheoverallacrylic
2610 Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993
-r n-e -4 N-l -- Oq q
ossqs
W8S88
888868
$qqRsqq
d d d d d o o
a 0 0 0 0
d d d d d
d d d d d d
0 0 0 ~ 0 0 0 0 0 0 0
clr(c(-HH-
8 6888 d d d d d
$$sa8
88tltlS8
6652a8
0000000
d d d d d d d
00000
d d d d d d
d d d d d d
==?a8
68868 d d d d d
329 fpj
Ind. Eng. Chem. Res., Vol. 32, No. 11,1993 2611 70 AA + PA + Ethy
60
--ae 50
8z-40 .-
>
3 30 5
8
201
320
AA Molar Yield, BOF
340
360
380
400
Reactor Temperature PC)
Figure 3. Acrylic acid (AA)molar yields and pathway distributions for lactic acid reactions at various temperatures using 0.02 M NazHPO, (rune 13-17). Pathway percentage calculations are explained in the results section (PA, propionic acid; Ethy, ethylene).
6
s sqqq ddooo
88888$8 d d d d d d d
*0
i6b E
.M
88866
ddddd
8888866
d d d d d d d
3
acid yield benefits due to less reaction along competing pathways. We have aged other reactors in our laboratory in the presence and absence of phosphate salts (discussed later) with nearly identical results. All aging was performed in the presence of lactic acid under flow conditions of 310 bar and 360 "C. All subsequent runs were performed in an aged reactor. Ramayya et al. (1987) and Torry et al. (1991) noticed differences in reaction rates between old and new reactors at supercritical conditions. Torry and co-workers conducted bibenzyl ether hydrolysis reactions in both stainless steel and titanium reactors. No differences in hydrolysis reaction rates were noted between the two types of reactors, but the yield of pyrolysis products was an order of magnitude greater in new stainless steel reactors than in used stainless steel or titanium reactors. Torry and coworkers propose that this is due to passivation of active wall sites during the reaction and that the hydrolysis reaction, the reaction of interest, is not affected by the passivation. Figure 3 shows the effect of reactor temperature on the reaction pathways and on the acrylic acid molar yields (runs 13-17). A temperature of 360 "C provides maximum yields of acrylic acid. Carbon monoxide is minimized while the carbon dioxide yield does not change appreciably. The shift in the pathway yields over the temperature range is a result of two effects: (1) the decrease in the density of the reaction fluid as the temperature increases, promoting more gaseous products; (2) changes in the selectivity due to different activation energies of the pathways. Based on this temperature study, all subsequent studies were performed at 360 OC. Based on previous literature sources which suggest that phosphate salts may provide catalytic dehydration effects on hydroxy acids, homogeneous phosphate salts (NazH P 0 3 were added to the reactant mixture. Varying levels of the phosphate salt were studied up to 20 % of the lactic acid on a molar basis (runs 18-23]. The pH at room temperature of the buffered reaction solutions varied from ayalue of 2.0 for the blank run without phosphate up to 3.0 for the 0.08 M experiment. Figure 4 shows the most noticeable effects from the phosphate salt addition are the abrupt changes in the acrylic acid and carbon monoxide pathways from a small addition of the salt. Additional amounts of phosphate over 0.02 mol/L did not have significant effects on the reaction pathways.
2612 Ind. Eng. Chem. Res., Vol. 32,No. 11, 1993 70
AA + PA + Ethy
AA Molar Yield, BOC
I
'.irbi 0
0
0.02
0.04
0.06
0.08
Na2HPQ (mol/L)
Figure 4. Effect of phoephate concentration on pathway distribution and acrylic acid molar yield at 360 O C (runs 18-23). 60
li
AA
AA
+ PA + Ethv
Mala-
// "
2
2.2
2.4
2.6
2.8
3
3.2
3.4 3.6
Feed Solution pH (25OC)
Figure 5. Effect of NaOH on pathway distribution and acrylic acid molar yield at 360 O C (runs 24-29).
The effect of base on the product distribution is shown in Figure 5 (runs 24-29). The feed pH was increased with NaOH to compare with the pH range associated with the basic phosphate salts in the earlier runs. The pH values of the solutionswere measured at room temperature before the runs. Although the pH and dissociation constants will be different at supercritical conditions, the effect is not investigated in this work and we refer to the pH of the reactant solution at room temperature for convenience. The molar yield of acrylic acid increases to a value of about 43% and then starts to decrease because of increased reaction along the secondary pathway (propionic acid, ethylene, and carbon dioxide). The carbon monoxide pathway decreases steadily while the carbon dioxide pathway increases. The general trend for the increase in pH of the reactant solution is to increase the acrylic acid pathway including secondaryproducts. Phosphate salt is a better catalyst than a simple base for acrylic acid production (56% yield vs 44%). Pathway two is considerably smaller when phosphate salt is used rather than base. Phosphoricacid was added and compared to the sulfuric acid studies of Mok and co-workers. Table I shows a considerable increase in the total lactic acid conversion using an acid catalyst, while the acrylic acid yield (BOF) remains fairly constant (runs 30-34). Low concentrations of phosphoric acid have a negligible effect on the acrylic acid pathway compared to the similar concentration of
Table 11. First-Order Rate Constants for Lactic Acid Reactions In Near-Critical Water Droduct Dathwavs overall rate k (8-9 k (8-1) lactic acid CO + CO? AA + PA + Ethy 0.0020 0.0011 new 0.0031 0.0018 0.0012 aged 0.0030 0.0012 0.0012 NaOH 0.0024 0.00083 0.0012 Na2HPO4 0.0020 ~
~~
phosphate salt shown in Figure 4. The dramatic effect with the phosphate salt may be due to complex formation of the carboxylic acid with the salt or deprotonization of the carboxylic acid. Phosphoric acid decreases the molar yield of acrylic acid (BOC), because of the large increase in pathway 1. Runs 35-37 did not show appreciable differences between sulfuric acid and phosphoricacid, but nitric acid increases the lactic acid conversion while decreasing the acrylic acid yield. These runs were prepared in the following manner: 0.04 mol of the acid catalyst was added to each liter of reactant solution and the pH was adjusted with NaOH up to 2.8to correspond to the pH of previous phosphate salt (Na2HP04) catalyzed runs.
Kinetic Studies Global kinetic studies were performed to investigate effects of catalysts and reactor aging. Kinetic data are presented for four different reaction conditions: (1) no catalyst (runs 8-12);(2)NaOH at pH 2.7 (runs 38-42);(3) 0.04 M Na2HP04 at pH 2.7 (runs (43-49);(4)new reactor (runs 1-7). The NaOH and phosphate runs are in an aged reactor and the aged, new, and NaOH series are without phosphate catalysts. The overall reaction of lactic acid is assumed to be irreversible and first order to evaluate the rate constants. The rates were calculated from the slope of a plot of In [LA] versus time, where [LA] is the lactic acid concentration. The plots were linear within experimental error over the range of concentrations studied. The overall reaction rates are presented in Table 11. From the same series of runs the relative rates of the three pathways are also determined using the method of Levenspiel (1972). Secondary reaction products are included in the pathway calculations but are assumed to have no effect on the first reaction step. The carbon monoxide and carbon dioxide yields are combined for comparison with the acrylic acid pathway because the potential water-gas shift involved between pathway one and two products prevents the use of the individual pathway data. The plots used for kinetic analysis were linear within experimental error over the range of experimental conversions. Table I1 shows that the addition of phosphates and NaOH to the reactant mixture suppresses the reaction of lactic acid as does the aging of the reactor. The overall rate constant of the acrylic acid pathway is insensitive to the investigated catalyst conditions, while the combined rate constant for carbon monoxide and carbon dioxide decreases significantly by one-half from the worst to the best case. The addition of salts to supercritical reactions is also studied by Torry et al. (1991).They investigated the effect of salt concentration (NaC1) on the rate of hydrolysis of dibenzyl ether and benzylphenylamine in supercritical water. Mok et al. (1989)studied effects of NaCl addition to lactic acid reactions in supercritical water. They found that all three pathways were catalyzed. Our work shows a beneficial effect with addition of phosphate salts, but
Ind. Eng. Chem. Res., Vol. 32, No. 11,1993 2613 there are differences with the other studies. The molar ratio of salt to reactant used by Torry et al. is as high as 6.25:1.0, and that of Mok et al. is 10.01.0, compared to the ratio in this work of 0.2:l.O. Torry et al. emphasize that the hydrolysis reaction in supercritical water is catalyzed by the addition of salt, while we have shown the addition of phosphate salts (NazHPOr) does not catalyze the dehydration reaction, but suppresses the competing reactions (see Table 11),thereby increasing the acrylic acid yield.
Conclusions The development of a specially design reactor and analysis techniques to investigate the conversion of lactic acid in supercritical water has led to several conclusions. Aging the Hastelloy C-276 reactor approximately 70 h at reaction conditions increases the yields of acrylic acid by decreasing the alternate pathway conversions. This may be due to passivation of the reactive wall sites. Maximum acrylic acid yield based on the conversion of 0.40 M lactic acid feed a t a residence time of 70 s occurs a t a temperature of 360 "C. Small amounts (
