3230
Ind. Eng. Chem. Res. 1994,33, 3230-3237
Recovery of Propylene Glycol from Dilute Aqueous Solutions via Reversible Reaction with Aldehydes Robert R. Broekhuis, Scott Lynn, and C. Judson King* Department of Chemical Engineering and Lawrence Berkeley Laboratory, University of California, Berkeley, California 94720
The recovery of propylene glycol from dilute aqueous solutions via reaction with formaldehyde to form 4-methyl-l,3-dioxolane or with acetaldehyde to form 2,4-dimethyl-1,3-dioxolanewas studied experimentally. The equilibrium and kinetics of the reaction with formaldehyde were studied in systems catalyzed by Amberlite IR-120 ion exchange resin. The equilibrium constant ranged from 5.9 to 8.7 in the temperature range from 25 to 85 "C,with no obvious trend with respect to temperature, The kinetics was found to be first-order in the concentrations of propylene glycol, formaldehyde, and Amberlite IR-120, with a n activation energy of 102 kJ/mol. I n the reaction with acetaldehyde, the equilibrium constant decreased from 18.1at 40 "C to 8.5 at 83 "C. The kinetics was faster than with formaldehyde. The volatilities of 4-methyl-1,3dioxolane and 2,4-dimethyl-1,3-dioxolane relative to water were 100 and 33, respectively. Of several solvents screened, aromatic hydrocarbons exhibited the highest distribution of 2,4dimethyl- 1,3-dioxolane from the aqueous into the organic phase. Recovery of propylene glycol by reactive distillation with formaldehyde or acetaldehyde is hampered by unfavorable chemical and phase equilibria. A process combining reaction and extraction into a n organic solvent appears to be more attractive and substantially reduces the energy requirement, in comparison with a triple-effect evaporation process.
Introduction Recovery of compounds bearing multiple hydroxyl groups from aqueous solutions is important in many process industries. Typical examples are recovery of fermentation chemicals from complex and dilute broths, removal of these chemicals from waste streams generated by a variety of industrial processes, and recovery of glycols from the aqueous solutions in which they are produced by petrochemical means. Recovering chemicals with low volatility and strongly hydrophilic characteristics from dilute aqueous solutions is a difficult separation problem, one that in many fermentation processes causes downstream processing to account for 40% or more of the final product cost. The heat load per amount of recovered chemical for multiple-effect evaporation is high, since all the water must be evaporated. Also, the chemical of interest is often just one of many nonvolatile chemicals in the solution; fractionation among these can be difficult and inefficient. More economical separation methods would capitalize on specific properties of the chemicals to be recovered, t o remove them selectively and efficiently from aqueous solution. Chemicals of interest include glycerol and diols (which have boiling points well above that of water) and saccharides (which are nonvolatile). The research described here focuses on propylene glycol (1,2-propanediol, 1,2-PD), which is a large-scale petrochemical and potential fermentation chemical of commercial interest. Propylene glycol is produced petrochemically in about 20% aqueous solution from propylene oxide or by fermentation of glucose using the bacterium Clostridium thermosaccharolyticum (Cameron and Cooney, 1986). The concentration of propylene glycol in the fermentation broth is about 0.8% by weight. Other glycols which can be produced by fermentation include 1,3-propanediol, using glycerol as a substrate (Cameron and Cooney, 19861, and 2,3-butylene glycol from glucose (Senkus, 1946).
Glycols do not distribute into organic solvents sufficiently to make simple extraction efficient. Randel et al. (1994) propose a separation process based on complexation of glycols with organoboronates. Another alternative is separation by reversible chemical reaction. Glycols are so difficult to recover from aqueous solution because of their hydrophilic properties, caused by the presence of two hydroxyl groups, allowing the formation of multiple hydrogen bonds with surrounding water molecules. A possible avenue toward easier recovery of glycols would therefore be to react them to form a substance that does not have hydroxyl groups or other highly polar groups. One such reaction is cyclic acetalization of vicinal diols with aldehydes t o form substituted 1,3-dioxolanes, under the influence of an acid catalyst.
A few reactions of particular interest are listed in Table 1. Senkus (1946) used an acetalization with formaldehyde to recover 2,3-butanediol from a fermentation broth containing 3.5% of the glycol by weight. The recovery was done in several batch reaction and separation steps. Lucas et al. (1950) prepared 4-methyl-1,3dioxolane from propylene glycol and paraformaldehyde. They report that the conversion to the acetal is essentially quantitative. The acetal was recovered as the azeotrope with water (bp 74 "C at 745 mmHg). Cyclic acetalization reactions are not limited to glycols. A series of studies (Tink and Neish, 1951;Tinket al., 1951; Tink and Roxburgh, 1951;Roxburgh, 1951)investigated the extraction of a variety of polyhydroxy compounds into n-butyraldehyde, in which the polyhydroxy compound was taken up as the corresponding cyclic acetal. Batch and continuous extraction of glycerol into nbutyraldehyde and the subsequent recovery of the
0888-5885/94/2633-3230$04.50/0 0 1994 American Chemical Society
Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994 3231 Table 1. Reactions of Vicinal Diols with Aldehydes no.
glycol
aldehyde
1 2 3
1,a-ethanediol 1,2-propanediol 1,2-propanediol 1,2-propanediol 2,3-butanediol
formaldehyde formaldehyde acetaldehyde propanal formaldehyde
4 5 a
Tb ("cia
reaction product 1,3-dioxolane 4-methyl-l,3-dioxolane 2,4-dimethyl-1,3-dioxolane 2-ethyl-4-methyl-l,3-dioxolane 4,5-dimethyl-1,3-dioxolane
75 84.2 cis, 90.5; trans, 93.3 cis and trans, 113-115 cis, 97.0; trans, 101.5
Sources given by Broekhuis et al., 1993.
aldehyde were studied experimentally. Astle et al. (1954) carried out acetalization reactions under the influence of a strong acid ion exchange resin, Amberlite IR-120. To avoid costs of purchase and disposal of strong mineral acids, it would be preferable to use solid catalysts in a continuous process based on cyclic acetalization. Furthermore, since conversion for such reactions is incomplete at equilibrium, it is desirable to remove the dioxolane from the reaction as it is formed to force the reaction in the forward direction. This may be achieved through the application of reactive distillation, in which the dioxolane is continuously removed from the reaction zone. Reactive distillation can have many advantages over separate reaction and separation operations (DeGarmo et al., 1992). In cases where the equilibrium conversion is limited, it avoids large recycle streams and the large equipment sizes associated with processing these streams. Solid catalyst material may be immobilized onto structured packings, allowing for more intimate contact between the liquid phase and the catalyst material. Data on reaction equilibrium, reaction kinetics and vaporfliquid equilibrium are essential for the design of a reactive distillation operation. An alternative to reactive distillation as a means of forcing the acetalization reaction t o equilibrium is the use of combined reaction and extraction. An aqueous solution containing the glycol and the aldehyde would be countercurrently contacted with an organic solvent, taking advantage of the large distribution ratio of the hydrophobic dioxolane between the organic and aqueous phases. The reaction could be carried out in a continuous countercurrent contacting device or in a series of mixer-settlers. Alternatively, the dioxolane could be removed from the reaction mixture by uptake onto a regenerable adsorbent that exhibits a high capacity for dioxolanes. It is now possible to envision various separation processes based on acetalization. Formaldehyde and acetaldehyde will be considered for the aldehyde reactant. The reaction products with propylene glycol are 4-methyl-1,3-dioxolane (4-MD) and 2,4-dimethyl-1,3dioxolane (2,4-DMD),respectively. Reactive Distillation, Reaction with Formaldehyde. Figure 1 shows a flow schematic of the process using formaldehyde. Formaldehyde solutions are complex mixtures containing hydrates and hydrated oligomers. The concentration of free formaldehyde in such solutions is very low (Walker, 1975). The binary system of formaldehyde and water exhibits an unusual vapor/ liquid equilibrium behavior (Gmehling et al., 1981). There is a minimum-boiling azeotrope a t 1atm and 97 "C for a mole fraction of formaldehyde of 0.13. At lower concentrations the volatility of formaldehyde relative to water is very close t o unity. Formaldehyde tends to polymerize out of the gas phase t o form solid paraformaldehyde, HO(CHzO),H. It is therefore impractical to take formaldehyde overhead in a distillation column. A direct reversal of the acetalization reaction results in a
tormaldehyde solution recycle dioxolane
JV
glycol solution
-31
L glycol product
Figure 1. Flow schematic of the reactive distillation process with formaldehyde. (1) Acetalization column, (2) transacetalization column, (3) methylal hydrolysis column, and (4) glycol purification column. acetaldehyde recycle
k glycol product
I
spent feed
-
Figure 2. Flow schematic of the reactive distillation process with acetaldehyde. (1)Acetalization column, (2) hydrolysis column, and (3) glycol purification column.
mixture containing formaldehyde and propylene glycol which would be difficult to separate. Senkus (1946) carried out a transacetalization reaction in which 2,4,5trimethyl-l,3-dioxolane reacts with two molecules of methanol to form 2,3-butylene glycol and methylal (dimethoxymethane), which can be easily separated. Methylal was then hydrolyzed to a formaldehyde solution and methanol. This strategy is applicable to propylene glycol as well. If formaldehyde is used in excess, the spent feed must be treated to recover the excess amount. The Distillative Acetal ehyde Process. Figure 2 shows a flow schematic o f t ,e process using acetaldehyde. Acetaldehyde has veryxlifferent properties from formaldehyde. It is very volatile (bp 21 "C), does not form an azeotrope with water, and does not readily
l l
3232 Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994 acetaldehyderecycle
1
glycol feec
raffinate
f*
-
spent feed
Figure 3. Flow schematic of the extractive acetaldehyde process. (1) Continuous countercurrent contactorheactor, (2) and (3) acetaldehyde stripping columns, (4) dioxolane recovery column, (5) hydrolysis column.
polymerize in neutral or acidic solutions. Therefore, acetaldehyde is much more volatile in aqueous solution, and the process using acetaldehyde does not require a transacetalization step. The reaction products of a direct acetal hydrolysis are acetaldehyde and propylene glycol, which can be taken as distillate and bottoms product, respectively. A n excess of acetaldehyde is necessary for the distillative acetalization, since the aldehyde distributes unfavorably between the liquid and the vapor phase. A large vapor-phase concentration is necessary to reach a sufficient liquid-phase concentration. The Extractive Acetaldehyde Process. A schematic flow diagram for a process using the reaction/ extraction scheme described earlier is presented in Figure 3. Processing downstream of the reactodcontactor includes stripping acetaldehyde out of both phases, distilling the dioxolane out of the organic phase and recovering the glycol from the dioxolane.
Experimental Procedures Preparation of Various Dioxolanes. A. 4-Methyl-1,3-dioxolane(4-MD). Three moles each of 1,2-PD (Aldrich Chemical, 99%) and formaldehyde (37 wt % solution, Fisher Scientific) were reacted in a 500-mL round-bottom flask, above which was a 30-cm, l-in.diameter vacuum-jacketed column packed with 6-mm ceramic and equipped with a condenser and reflux overhead. Methanol and methylal, present in the formaldehyde solution as supplied, were distilled off, after which a batch reactive distillation was carried out with ca. 20 g of Dowex 50 or Amberlite IR-120 ion exchange resin as catalyst. Fractions rich in 4-MD were collected at a vapor temperature of 76 "C, at which the azeotrope between water and 4-MD is distilled. The condensate was saturated with potassium chloride, to decrease the mutual solubility of 4-MD and water. The aqueous phase (50-mL)and the organic phase (230-mL) were separated. The latter was dried over molecular sieves. A final distillation was carried out to remove the main impurities, methanol and methylal. Only the distillation fractions containing 99.5+ wt % 4-MD were consolidated. The yield was about 1.25 mol 4-MD (42%), boiling at 84.5 "C. B. 2,4-Dimethyl-1,3-dioxolane (2,4-DMD). A mixture of 80 g of 1,2-PD (1.05 mol), 10 g of Dowex 50 ion exchange resin, and 5 g of potassium chloride was
stirred in a 250-mL round-bottom flask immersed in an ice bath. Sixty milliliters of acetaldehyde (1.06 mol) was slowly added to this mixture, with an attempt made to avoid boiling as the exothermic reaction proceeded. After the addition was completed, stirring was continued for 10 min in the ice bath and then for 30 min at room temperature. The potassium chloride and Dowex 50 were removed by filtration, resulting in two clear liquid layers. The organic phase, which contained 2 wt % water, was dried over molecular sieves. The reverse reaction of 2,4-DMD with water proceeds at a considerable rate at higher temperatures, even when not catalyzed, so that the removal of water before final distillation is essential. Acetaldehyde was distilled off t o produce more than 100 g of 2,4-DMD (ca. 0.98 mol) with a purity of 99.5+% and a boiling point of 93.0 "C. GC/ MS shows that both the cis- and the trans-isomer are formed. Reactive Distillation. Batch reactive distillations were carried out under the influence of both strong mineral acids (hydrochloric acid is most effective) and Amberlite IR-120 Plus (H) ion exchange resin (Rohm & Haas Corp). Acetalization of aqueous solutions of 1,2PD and formaldehyde and transacetalization of 4-MD with methanol were carried out, using a 15-cmVigreux column or a 30-cm column packed with 6-mm Berl saddles. Chemical Equilibrium Measurements. The equilibria for the three reaction steps in the 4-MD process (acetalization, transacetalization, and methylal hydrolysis) were studied at various temperatures. Measured amounts of the reactants were added to 20-mL vials, along with IR-120 catalyst. When possible, the equilibrium was approached from both directions. The vials were kept in a shaker bath until the composition no longer changed. The concentrations of reactants and products were then determined by gas chromatography (GC), except for formaldehyde which was determined by the sulfite titration method (Walker, 1975). Chemical Kinetics Measurements. Chemical kinetics of the acetalization, transacetalization, and methylal hydrolysis reactions in the 4-MD process and the acetalization reaction in the 2,4-DMD process, were studied in a 250-mL three-neck round-bottom flask, equipped with a thermometer and a sampling tube with a stopcock. The reaction flask was stirred magnetically and heated by submersion in a temperature-controlled ethylene glycol bath. After the reactants weighed into the flask reached the reaction temperature, reaction was initiated by the addition of a measured amount of Amberlite IR-120. The resin was used as supplied by Rohm & Haas, in a moist form. Samples (l-mL) were taken regularly through the sampling tube and analyzed by GC. Reaction temperatures ranged from 40 to 100 "C. The final reaction mixture was analyzed to obtain additional values for the equilibrium constant at various temperatures. Simultaneous Reaction and Extraction Experiment. Equal volumes (50-mL) of toluene and an aqueous solution containing 2 wt % 1,2-PD and 20 wt % acetaldehyde were stirred vigorously in the equipment described above, resulting in a fine dispersion with a high interfacial area for mass transfer. The temperature was maintained at 47 "C. Reaction was started by the addition of 5 g of Amberlite IR-120. Samples of the two-phase mixture were taken repeatedly. Analyses were made by GC, for 2,4-DMD (both phases) and 1,2PD (aqueous phase only).
Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994 3233
Phase Equilibrium Measurements. A. LiquidLiquid Equilibria. Binary systems containing water and either 4-MD or 2,4-DMD are not miscible in all proportions. Liquid-liquid equilibria were measured over a range of temperatures. The compositions of both the aqueous and the organic phases were measured as a function of temperature (up to the atmospheric bubble point), by GC and Karl Fischer titration, respectively. Addition of methanol serves to homogenize these binary heterogeneous systems. Liquid-liquid equilibrium measurements at atmospheric pressure and 25 "C were carried out to obtain ternary equilibrium data for the system methanoll4-MD/water. The methanol contents of both phases, as well as the 4-MD content of the aqueous phase, were quantified using GC. The water content of the organic phase was determined using Karl Fischer titration. B. Vapor-Liquid Equilibria. Vapor-liquid equilibrium (VLE) data were collected in a modified atmospheric-pressure Othmer still, consisting of a roundbottom flask heated by a spherical heating mantle and magnetically stirred, a water-cooled condenser, and a vapor condensate return line with a condensate holdup loop. The section of the still between the flask and the condenser was heated and insulated to avoid condensation. The temperatures of the liquid inside the flask and the vapor before the condenser were measured. Samples of the liquid phase in the flask and the vapor condensate in the return loop were taken through septa and analyzed by GC. In cases where the water content of the system was low, Karl Fischer titration was used in addition to GC. Some equilibrium points for the ethanollwater system were reproduced to ensure that the Othmer still accurately measured VLE data (Broekhuis et al., 1993). For the system 4-MD/water, binary VLE data were collected in those composition regions where both the liquid phase and the vapor condensate phase were outside the region of immiscibility. Since this excludes a large composition region, additional experiments were carried out with methanol as a third, homogenizing component. The methanol content of the system was kept as low as possible, to allow for interpretation of the ternary data as pseudobinary data. For the system 2,4-DMD/water, only two VLE data points at low concentration of 2,4-DMD were measured. At elevated temperatures, the uncatalyzed reaction between 2,4-DMD and water proceeds at a considerable rate, making the task of measuring binary equilibrium data a difficult one. A series of VLE measurements was made for 2,4DMD in toluene. C. Extraction Equilibria. A number of solvents were screened for extraction of 2,4-DMD and acetaldehyde from an aqueous phase. Dilute aqueous solutions of 2,4-DMD were contacted with the solvents by stirring vigorously at room temperature. The 2,4-DMD contents of both phases were then determined by GC. Subsequently, acetaldehyde was added to the system. After equilibrium had once again been reached, the concentrations of both acetaldehyde and 2,4-DMD in both phases were measured. More details on the various experimental procedures are available elsewhere (Broekhuis et al., 1993). Results Reactive Distillation. Batch reactive distillation of aqueous solutions of 1,2-PDand formaldehyde produced
a distillate rich in 4-MD. Due to the inherent limitations of batch distillation through Vigreux columns or simple packed columns, the yield of 4-MD when a 20% excess of formaldehyde was used did not exceed 84% with strong mineral acids as catalysts or 60% with Amberlite IR-120. However, yields of 4-MD were as high as 97% when formaldehyde was added in 100% excess. Batch reactive distillation of small amounts of 4-MD with methanol was hindered by overheating of the solution a t high conversions. Addition of water prior to distillation alleviated this problem, so that a 50 wt % aqueous 1,2-PD solution could be obtained with a yield in excess of 90%. Both problems noted above are related to laboratory experimental conditions. Industrial applications of such reactive distillations would probably employ continuous distillation columns, and should not suffer these difficulties. Propylene glycol can therefore be recovered from dilute aqueous solution by reaction with formaldehyde to 4-MD, catalyzed by strong mineral acids in solution or by strong acid ion exchange resins. Chemical Equilibrium. Equilibrium of the acetalization reaction between propylene glycol and formaldehyde, reaction 2 in Table 1,was reached in aqueous solution, as confirmed by experiments approaching equilibrium from both directions (Broekhuis et al., 1993). The equilibrium constant K2 (eq 1)was calcu-
[4-methyl-1,3-dioxolane][waterl
K, = [total formaldehydel[propylene glycol1 (1) lated from the equilibrium concentrations a t various temperatures. All calculated values were in the range 5.9-8.7. The scatter in these values was too large for any conclusions to be drawn concerning the temperature dependence of the reaction equilibrium. The reason for the scatter in the data may be that the concentration of free formaldehyde in solution is a function of the total concentration of formaldehyde species, as well as the concentrations of methanol and water, and is a very small fraction of the total formaldehyde concentration measured by the sulfite method. The assumption, explicit in eq 1, that the equilibrium constant varies with the first power of the total formaldehyde concentration may be incorrect. The equilibrium constant for the reaction between propylene glycol and acetaldehyde (reaction 3) was determined over a range of temperatures. A plot of K3 us T is shown in Figure 4. Chemical Kinetics. Figure 5 shows plots of the conversion of 1,2-PD to 4-MD as a function of time, for various temperatures and initial concentrations. In these experiments Amberlite IR-120 was used as a catalyst. Table 2 lists the conditions of each of the experiments. In experiments at higher temperatures, smaller amounts of catalyst were used to ensure that the reaction time scale was of the same order of magnitude as the sampling time scale. For this reason, the time coordinate in Figure 5 is normalized by multiplying by the weight ratio of catalyst to solution. Included in Table 2 is the initial rate, ro, for each of these experiments, as determined from the initial slopes of the curves in Figure 5. By linear regression of ro us. all of the four variables influencing the reaction rate (Broekhuis et al., 1993) it was found that ro varies linearly with the concentrations of catalyst, formaldehyde, and 1,2-PD and that the activation energy is 102
3234 Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994
G?
18
-
16
-
14
-
12
-
0
"
10-
3
2
M
0
01
20
I
I
4
30
40
50
4
I
I
I
60
70
80
90
d
0 I
I
T ("C) Figure 4. Experimental equilibrium constant data for the formafor 1,2-propanedioi and acettion of 2,4-dimethyl-1,3-dioxolane aldehyde us temperature. 0.16 0.14
250
0.12
g
-P E
$
-
0
o^
2
0 .
0.10
E
0.08
P
v
0.06
B
200
-
150
-
100
- 8
0 0
i
E
0.04
=c
0.02
0
ep
e 0.00 0
1
2
0
3
normalized time: t(min)*mcat(g)/mtot(g)
Figure 6. Kinetics of the reaction between 1,2-propanediol and formaldehyde to form 4-methyl-1,3-dioxolane. Conditions are given in Table 2.
0 0
5
10
15
20
25
time (hours) Table 2. Reaction Conditions for 4-MDAcetalization Kinetics Experiments ~
expt A B C
D E F G
temp ("C) 80 90 98.5 90 90 90 92
[glyc~ll [HCHOI (mol/kg) (rnolkg) 0.788 0.781 0.782 0.783 0.778 0.782 0.781 1.560 1.586 0.903 0.390 0.798 0.382 0.382
IR-120 (w/w) 0.0248 0.0200 0.0160 0.0160 0.0159 0.0160 0,100
ro (moY (kgCatmin)) 0.034 0.095 0.190 0.202 0.253 0.061 0.027
kJ/mol. The expression correlating the reaction rates of all the experiments described above is then given by eq 2, in which r is the rate in mollkg min, w,t the weight
r = 4.73 x 1013w,,,[glycoll[HCHOle-101600'RT (2) ratio of catalyst to solution and the concentrations of propylene glycol and formaldehyde are given in molkg. Transacetalization of 4-MD in methanolic solution was carried out at a temperature of 57 f 2 "C. Besides the desired reaction, side reactions occurred, leading to
Figure 7. Kinetics of the methylal hydrolysis reaction at 46 f 1 "C.
the formation of several unidentified high-molecularweight byproducts, amounting to a combined GC peak area equal to 30% of the 1,2-PG peak area. Figure 6 shows the amount of 1,2-PD formed as a function of time. The initial rate calculated from the slope of the plot is 29 mmol/(kgIR-120min). Attempts t o identify the byproducts by mass spectrometry coupled with gas chromatography were unsuccessful. The kinetics of the methylal hydrolysis reaction was determined in one experiment at 45 "C. Higher temperatures were impractical due to the low boiling point of methylal. The extent of reaction is plotted against time in Figure 7. The initial rate of this reaction is 1.0 mOl/(kgIR-laomin). The acetalization reaction between acetaldehyde and 1,2-PD to form 2,4-DMD was studied at 60 f 1"C. The amount of 2,4-DMD formed us time is shown in Figure 8. If it is assumed that the reaction is first-order in each of the reactants, as was the case for the formaldehyde reaction, then the equivalent rate expression is given
Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994 3235
0
0
.
r-----
0
0'3 0 0
1,2-PD in aqueous phase 4-MD in aqueous hase 4-MD in toluene piase
1
0
20
40
60
80
100
time (min) Figure 8. Kinetics of the reaction between 1,2-propanediol and a t 60 f 1 "C. acetaldehyde to form 2,4-dimethyl-l,3-dioxolane,
0.0 0
25
50
75
100
125
150
Table 3. Distribution Ratios of 2,4-DMDand time (min) Acetaldehyde Figure 9. Aqueous- and toluene-phase concentrations of 12boiling aqueous during simultaneous solvent point ("C) solubility (wt S ) D2,4DMD DM~CHO propanediol and 2,4-dimethyl-l,3-dioxolane reaction and extraction a t 38 "C. 0.51 0.047 10.1 toluene 110.6 0.42 0.013 7.7 o-xylene 144.4 Table 4. Composition (Weight Fractions) of Equilibrium 7.1 0.36 136.2 0.014 ethylbenzene Liquid Phases in WaterlDioxolaneSystems 2.8 0.11 125.7 0.002 n-octane 0.26 0.3 4.4 Wdioxolane dibutyl ether 142 0.90 3 7.4 methylcyclo168 temp ("C) aqueous organic hexanone A. 4-MDiWater 28 0.279 0.957 in eq 3. The value of the reaction rate constant K at 50 0.269 0.934 63 0.270 0.924 = k 0e-E~iRT~ , . & ~ ~ C O ~ I [ C H ~ C H O I (3) 70 0.274 0.919 B. 2,4-DMD/Water 60 "C, the product of the pre-exponential factor K O and 25 0.21 0.962
the exponential expression at 60 "C, is 0.905 kg2/(mol min). The equivalent k(60 " C )calculated from eq 2 is 0.165 kg2/(mol min), which shows that the acetalization reaction with acetaldehyde is considerably faster than with formaldehyde. An explanation of this difference may again be the fact that the concentration of free formaldehyde in a formaldehyde solution is very small. Formaldehyde may therefore not be as readily available for reaction as acetaldehyde is. From the rapid rate of 2,4-DMD formation and the moderate equilibrium constant for this reaction, it may be inferred that the reverse reaction, the hydrolysis of 2,4-DMD, also proceeds at a high rate. Screening of Solvents for Extraction. Selected solvents were screened for effectiveness in a reaction/ D DM~CHO, extraction process. Table 3 lists D ~ , ~ - D Mand the distribution ratios (volume basis) between the organic and aqueous phases for 2,4-DMD and acetaldehyde, respectively. All solvents screened except 4-methylcyclohexanone have low solubilities in water. To minimize the solvent flow rate in an extraction process a high value of D2,4-DMD is needed. Also, a low value Of DMulecHOis desirable, to keep as much acetaldehyde as possible in the aqueous phase (where the reaction takes place), and to minimize costs associated with downstream separations. Judging by these criteria, aromatic hydrocarbons appear to be the best solvents of those examined. Also, aromatic solvents would be stripped out of the aqueous solution in the acetaldehyde stripping column, resulting in an extremely low effluent concentration. Simultaneous Reaction and Extraction. As noted above, reaction of 1,2-PD to 2,4-DMD in an aqueous
Table 6. Liquid-Liquid Equilibrium Data in the Ternary System Methanol W-4-MD (2)-Water (3) at 24 "C bottom phase top phase w1' w2' Wl" w 2" 0.050 0.380 0.019 0.88 0.071 0.393 0.027 0.96 0.087 0.479 0.042 0.583 0.064 0.93 0.105 concentration of first homogeneous mixture 0.101 0.652
phase, with simultaneous extraction into a toluene phase, was carried out at 38 "C. The aqueous-phase concentrations of 1,2-PDand 2,4-DMD and the organicphase concentration of 2,4-DMD are plotted as functions of time in Figure 9. Extraction equilibrium is attained on a time scale of ca. 10 min; the reaction proceeds to a near-equilibrium state in ca. 40 min. The overall conversion was 88%, and the recovery into the toluene phase, 80%. Liquid-Liquid Equilibria in DioxolanelWater Systems. The compositions of the dioxolane-rich and water-rich phases in the liquid-liquid equilibria between (A) 4-MD and water and (B) 2,4-DMD and water are shown in Table 4. The measurements in system A were done for a range of temperatures, from room temperature up to around the atmospheric bubble point. Liquid-liquid equilibria were measured in the ternary system methanol/4-MD/water, to investigate how the addition of methanol affects the liquid-liquid equi-
3236 Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994 binary x4.MDY4.MD 0.8 0.7
:0.5
0 0
-
0.6 -
A*
pseudo-binary
-
0
e 0
Ef
0
h
-0
2
0.4 8 6
0.3
0.1
/
0 . o v ' 0.0 0.1
6
0.2
0.3
I
1
0.4 0.5 X
1
,
0.6 0.7
I
0.8 0.9
I
1.0
4-MD
0.0
0.2
0.4 X
0.6
0.8
1.o
24-DMD
and water: binary data (xy) and pseudobinary data (x*y*).
Figure 11. Vapor-liquid equilibrium data in the system 2,4dimethyl-l,3-dioxolane and toluene.
librium of 4-MD/water. The concentrations of the water-rich phase could be quantified considerably more accurately than those of the dioxolane-rich phase. Ternary results are given in Table 5. From the data it can be inferred that the addition of ca. 10 wt % methanol t o any 4-MD/water system will serve to homogenize the mixture. Vapor-Liquid Equilibria. A 4-MDIWater and MethanoV4-MDIWater. Binary VLE data were obtained for those compositions where the liquid and vapor condensate phases were homogeneous. At intermediate compositions, methanol was added as a homogenizer. Figure 10 shows an x-y diagram of the binary system, with ternary data presented as pseudobinary data, using X*4-MD = X4-MD/(Xq-MD XHZO) and an equivalent expression for y*4-MD. The system shows a heterogenous, minimum-boiling azeotrope a t a composition of 60 mol % 4-MD. At low concentrations of 4-MD in water, the relative volatility a12 is about 100. B. 2,PDMDIWater. The two equilibrium data points at low concentrations of 2,4-DMD indicated a relative volatility a12 of about 33. C. 2,CDMDffoluene. VLE data points of 2,4-DMD in toluene are shown in Figure 11. They are well represented by the curve for an ideal mixture with a 1 2 = 1.76. The choice of a high-boiling solvent such as o-xylene or ethylbenzene rather than toluene should make for an easier separation of 2,4-DMD from the organic phase. Process Evaluation. A. The Formaldehyde Process. In the process shown in Figure 1,the first column processes a large volume of aqueous solution; the remaining columns process much smaller organic streams. Energy expenditure is therefore expected t o be largest in the first column. This reactive distillation column carries out the acetalization step, with a solid ion exchange resin as a catalyst. If the feed is stoichiometric, eq 2 becomes second-order, with very slow kinetics at low concentrations. Preliminary calculations indicate that this would lead to prohibitively long reaction times. An excess of formaldehyde of the order of 100%would be necessary to carry out the reaction to a satisfactory conversion. However, removing the formaldehyde from the effluent aqueous phase then becomes
difficult. Senkus (1946)treated the spent fermentation broth by reacting it with an excess of methanol, converting the formaldehyde to methylal which was then recovered by distillation. If the excess of formaldehyde is small, the effluent might be biologically treated to remove the formaldehyde without attempting to recover it. But although these solutions to the engineering problem are available, they are not likely to lead to an economically attractive process. B. The Acetaldehyde Process. 1. Distillative Route. With acetaldehyde some of the problems with formaldehyde are avoided. Excess acetaldehyde is recovered at the top of the column rather than at the bottom, so that contamination of the spent broth with the aldehyde would not be a problem. A large excess is needed since in distillation most of the acetaldehyde would distribute into the vapor phase. The process modeling software ChemCAD (Chemstations Inc.) was used to model somewhat crudely the reactive distillation (Broekhuis et al., 1993). Feeds are a 1.5 wt % solution of 1,2-PD, fed near the top of the column, and acetaldehyde vapor, fed near the bottom of the column. The stages between the two feeds constitute the reaction zone, modeled by a number of reaction and phase equilibrium stages. Even with a mass flow of acetaldehyde as large as that of the aqueous feed, recovery of glycol as 2,4-DMD in the distillate does not exceed 80% of the theoretical yield. An excessive flow rate of acetaldehyde would be needed to achieve an acceptably high yield. Thus, reactive distillation with acetaldehyde does not appear to be economically attractive. Higher-boiling aldehydes such as propionaldehyde or butyraldehyde, which distribute more favorably between the liquid and the vapor phase, are not useful because of their tendencies to participate in aldol condensation reactions (Roxburgh, 1951). 2. Extractive Route. An operation without a vapor phase, e.g., extraction, would eliminate the problems attributable to the high volatility of acetaldehyde. Reaction combined with extraction, either in a continuous countercurrent mode or in a series of mixer-settler reactors, was investigated by a mathematical model (Broekhuis et al., 1993). Given a residence time of 30 min, such an operation can recover more than 99% of the propylene glycol as 2,4-DMD.
Figure 10. Vapor-liquid equilibrium for 4-methyl-l,3-dioxolane
+
Ind. Eng. Chem. Res., Vol. 33, No. 12, 1994 3237 Downstream processing in this extractive process, including stripping acetaldehyde out of both the spent broth and the organic phase, recovering the 2,4-DMD out of the organic phase, and recovering propylene glycol by reactive distillation, was modeled using ChemCAD. Ethylbenzene was chosen as the organic solvent. In the last step, in which 2,4-DMD is hydrolyzed, high yields can be attained without excessive flows. The energy input for downstream processing corresponds to 42% of the energy input of a triple-effect evaporation process. The energy input for the actual reactiodextraction equipment should be moderate, which means that the reactiodextraction process with acetaldehyde has a considerable energy advantage over direct distillation. This advantage is offset by higher equipment and chemical costs. However, the purpose of this project was to identify processes that might be technically feasible. Time did not permit a complete economic evaluation. C. Formation of Byproducts. The only reaction step in which significant byproduct formation was observed was the methanolysis reaction for the formaldehyde process. In the acetaldehyde processes, byproducts could arise from aldol condensation of acetaldehyde, e.g., 3-hydroxybutanal. This compound and any further polymerization products would not build up in the process but would emerge in the glycol product, or in the spent feed if they are water-soluble. However, in the acidic conditions present in the reactive columns, aldol condensation would not be expected to occur.
Conclusion Propylene glycol may be converted to a cyclic acetal by reaction with an aldehyde. The acetal is much less hydrophilic than the glycol, enabling its separation from aqueous solution by distillation or extraction. Chemical equilibrium in the acetalization reaction does not correspond to a high degree of conversion. Therefore, it is necessary to remove the product acetal during the reaction, by simultaneous distillation or extraction. The product of the reaction between formaldehyde and propylene glycol, 4-methyl-l,3-dioxolane,has a high relative volatility in dilute aqueous solutions. However, the equilibrium in such a dilute system is such that an excess of formaldehyde would be needed for effective recovery of propylene glycol in a reactive distillation process. The residual formaldehyde would be difficult to separate from the aqueous solution, adding considerable cost and complexity to such a process. Acetaldehyde reacts with propylene glycol to form 2,4dimethyl-l,3-dioxolane. In a reactive distillation process, contamination of the effluent aqueous solution would not be a problem, but due to unfavorable equilibrium and unfavorable distribution of acetaldehyde between the liquid and the vapor phase, an excessive flow of acetaldehyde would be required to achieve an acceptable level of glycol recovery, and that level of recovery would, in any event, be incomplete. A process that combines reaction and extraction is more attractive. The energy input required for the necessary separations after the extraction step was calculated by process simulation to be 42%of the energy required for a tripleeffect evaporation process. The reduced energy load in a reactiodextraction process would be offset at least in part by increased capital and chemical costs.
The concept of cyclic acetalization as part of a separation route can be extended to other polyhydroxy compounds. Tink and Neish (1951) reacted a variety of chemicals, including glycols, glycerol, saccharides, and sugar alcohols with butyraldehyde in an extractive separation process. In their work the butyraldehyde served as both an extractant and a reactant, and butyraldehyde losses posed a significant problem due to side reactions. A process using acetaldehyde as a reactant combined with a separate solvent for the extraction would avoid losses of this type, while increasing complexity and separation costs of the process.
Acknowledgment This work was supported by the Chemical and Biochemical Technology Research (BCTR) Program, Advanced Industrial Concepts Division, Office of Industrial Processes, Assistant Secretary of Energy, under Contract No. DE-AC03-76SF00098. Literature Cited Arenson, D. R.; King, C. J. Separation of low molecular weight alcohols from dilute aqueous solutions by reversible chemical complexation. Report LBL-24944; Lawrence Berkeley Laboratory: Berkeley, CA, 1989. Astle, M. J.;Zaslowsky, J. A.; Lafyatis, P. G. Catalysis with cationexchange resins. Preparation of 1,3-dioxolanes and 1,3,6-trioxocanes. Znd. Eng. Chem. 1954,46,787. Broekhuis, R. R.; King, C. J.; Lynn, S. L. Recovery of propylene glycol from dilute aqueous solutions via reversible reaction with aldehydes. Report LBL-35155; Lawrence Berkeley Laboratory: Berkeley, CA, December 1993. Cameron, D. C.; Cooney, C. L. A novel fermentation: The production of R(-)-1,2-propanediol and acetol by clostridium thermosaccharolyticum. Biotechnology 1986,4,651. DeGarmo, J. L.; Parulekar, V. N.; Pinjala, V. Consider reactive distillation. Chem. Eng. Progr. 1992,(3), 43. Gmehling, J.,Onken, U., Arlt, W., Eds. Vapor-liquid equilibrium data collection, Dechema: Flushing, NY,1981; Vol. 1. Lucas, H. J.; Mitchell, F. W.; Scully, C. N. Cyclic phosphites of some aliphatic glycols. J . Am. Chem. SOC.1950,72, 5496. Randel, L. A,; Chow, T. K.-F.; King, C. J. Ion-pair extraction of multi-OH compounds by complexation with organoboronate. Solvent Extr. Zon Exch. 1994 12, 765. Roxburgh, J. M. Extraction of polyhydroxy compounds from dilute aqueous solutions by cyclic acetal formation. IV.Recovery of n-butyraldehyde. Can. J . Technol. 1951,30,72. Senkus, M. Recovery of 2,3-butanediol produced by fermentation. Znd. Eng. Chem. 1946,38,913. Tink, R. R.; Neish, A. C. Extraction of polyhydroxy compounds from dilute aqueous solutions by cyclic acetal formation. I. An investigation of the scope of the process. Can. J. Technol. 1951, 29, 243. Tink, R. R.; Roxburgh, J. M. Extraction of polyhydroxy compounds from dilute aqueous solutions by cyclic acetal formation. 111. The continuous extraction of glycerol. Can. J. Technol. 1951, 29, 269. Tink, R. R.; Spencer, E. Y.; Roxburgh, J. M. Extraction of polyhydroxy compounds from dilute aqueous solutions by cyclic acetal formation. 11. The Batch extraction of glycerol. Can. J . Technol. 1951,29,250. Walker, J. F. Formaldehyde, 3rd ed.; ACS Monograph Series; R. E. Krieger: Huntington, NY,1975.
Received for review March 15, 1994 Revised manuscript received July 25, 1994 Accepted August 9, 1994@ Abstract published in Advance ACS Abstracts, October 15, 1994. @
