Trimethylolethane from Propionaldehyde and Formaldehyde

G. J. Laemmle, J. G. Milligan, and W. J. Peppel. Ind. Eng. Chem. , 1960, 52 (1), pp 33–36. DOI: 10.1021/ie50601a032. Publication Date: January 1960...
1 downloads 0 Views 454KB Size
I

G. J. LAEMMLE, J. G. MILLIGAN, and W. J. PEPPEL Jefferson Chemical Co., Inc., North Austin Station, Austin 51, Tex.

Trimethylolethane from Propionaldehyde and Formaldehyde A Practical Manufacturing Process A bench-scale process investigation has been used in proposing an attractive minufacturing process

Tm

tion from concentrated aqueous solution, but not the much more soluble trimethylolethane. The high melting point of trimethylolethane, 199' C., also precludes practical recovery by distillation, possible in the case of the lower melting trimethylolpropane. Recovery by extraction with organic solvents, from which the trimethylolethane is then crystallized, has been suggested most frequently (3,

reactions for making pentaerythritol are useful for obtaining many additional polyols-e.g., formaldehyde can be condensed with higher homologs of acetaldehyde to obtain an interesting series of triols. Trimethylolethane and trimethylolpropane, from propionaldehyde and n-butyraldehyde, respectively, have recently become available commercially. Economical production is possible, as propionaldehyde and n-butyraldehyde can now be made directly from low-cost ethylene and propylene by the oxo reaction. Trimethylolethane is obtained in yields ranging from 51% (2) to over 90% (4). Little has been disclosed about side reactions or by-products. Pure product is not easily obtained. Pentaerythritol can be recovered in good purity by crystalliza-

6,9). The reactions employed in obtaining trimethylolethane are well known. 2CH20.

+ CHaCHzCHO

-F

CHzOH

CHzOH CHaLCHO CHaOH I

CH&(CH20H)a

+

I

CH+!LCHO CHzOH I

(1)

(2)

The base MOH, generally sodium or calcium hydroxide, both catalyzes the initial condensation reaction and brings about the subsequent Cannizzaro reaction. I n practice the reactions proceed concurrently. A number of side reactions are possible; one observed in particular in the similar preparation of pentaerythritol gives condensed polyols. CHzOH HIOCHa--(C-cH~~.OH CHLOH I

n = 2,3.

..

'b

4 ACID

4

HCOOM

WATER

FORMALDEHYDE AOUEOUS CAUSTIC PROPIONALDEHYDE

+ CHiO + MOH

-

WATER

LPYER

L

2

HOLMNG

-

TANK

3

t

- COLD WATER - STEAM MIBK - METHYL ISOBUTYL KETONE TME - TRIMETHYLOLETHANE

WATER (WASTE1

C.W. S

SODIUM FbRMATE __

This simulated five-stage countercurrent extraction scheme was used to check purification of trimethylolethane F.

Feed solution.

K.

Fresh MIBK.

W.

Water.

KP.

Ketone phase.

WP.

Water phase

VOL. 52, NO. 1

JANUARY 1960

33

( 4 T H . CYCLE

( S T H . CYCE RETAI NED)

RETAINED)

Flow diagram illustrating proposed trimethylolethane process

Technical pentaerythritol may contain 8 to 20% dipentaerythritol, with traces of higher polyols. Surprisingly, the formation of condensed polyols is not a complication in the preparation of trimethylolethane. Reaction conditions can be selected that result in yields of high-purity product approaching 90% of theory based upon propionaldehyde.

Table I. Sodium Hydroxide Is Superior to Calcium Hydroxide in Preparing Trimethylolethane Laboratory Run Data Crude trimethylolethane yield, %

M.P., O C. Sublimed trimethylolethane yield, % M.P., C.

Ca(0H)p

NaOH

85.7 184

95.3 184

79.7 189

89.3 197.5

The preparation and recovery of trimethylolethane involve process operations particularly suited to study at the laboratory bench. T h e results obtained in this investigation enabled an attractive manufacturing process to be proposed. Subsequently, several critical operations were checked in pilot plant tests. laboratory Investigation of Reaction Variables

Essentially, trimethylolethane is made by reaction of propionaldehyde with formaldehyde and aqueous caustic and separation of the product from by-product sodium formate. In the preferred procedure adopted after exploratory experiments, 10% aqueous sodium hydroxide

34

is charged to a stirred reaction vessel and cooled to about 15' C. Uninhibited 30y0formaldehyde (3.2 moles per mole of NaOH) is added while cooled to hold the temperature below 25" C. Addition of propionaldehyde (0.91 mole per mole of NaOH) follows without delay at a rate adjusted to allow completion of the step in about 1 hour. T h e temperature is maintained at about 25" C. during the addition and for another 2 hours while the reaction proceeds to completion. LVith a few drops of phenolphthalein solution added as indicator, the unreacted caustic is finally neutralized with 90% formic acid. Water is distilled from the aqueous product at atmospheric pressure until the residual sirup is brought to 115' C. This clear, straw-colored concentrate contains 30 to 40% water by weight and shows no tendency to deposit solid. T h e trimethylolethane is recovered by transferring it to hot methyl isobutyl ketone (MIBK) in an operation involving complete removal of water from the residual sirup by azeotropic distillation in a vessel equipped with a reflux condenser and a water trap. Using 2 parts by weight of iMIBK per part of sirup, the two-phase mixture is stirred and heated until solvent refluxes; the water in the overhead condensate is collected and removed through the trap. .4s this operation progresses, the sodium formate separates as coarse crystals. When separation of water is complete, the hot MIBK solution is decanted from the sodium formate, clarified by filtration under suction, and cooled to room temperature. The trimethylolethane that separates is collected on a filter. T h e mother liquor is re-

INDUSTRIAL AND ENGINEERING CHEMISTRY

heated to boiling and used to extract a little additional product from the sodium formate. The combined product is washed with a little fresh, cold MIBK and dried in the air. The yield of crude

Table II. Cannizzaro Reaction Causes Loss of Formaldehyde and Sodium Hydroxide in Trimethylolethane Synthesis Reactants Tempeiature.

1 7 2 5 moles SOYc formaldehyde, 0 625 iiioles 10% sodiuin hydroside 25" C .

Time, Hr.

SaOH Reacted Mole

0.0 0.5 2.0 5.0 21.5

0.0 0.034 0.136 0.241 0.429

CH2O Reacted, Mole

... ... ...

0.505 0.875

trimethylolethane is 94 to 95% of theory based upon propionaldehyde. Typically, crude trimethylolethane obtained in this manner has a melting point of 183-5" C.; pure trimethylolethane melts at 199' C. On heating to the boiling point the crude product decomposes rapidly with evolution of acrid fumes; pure trimethylolethane boils without change at 283' C. I n estimating the yield of pure triol obtained in laboratory experiments, a weighed sample of the crude product was sublimed under a pressure of 0.1 mm. in an oil bath at 150' C. Weight yields of sublimed tri-

T R I M ETH Y LO L A M INE P R 0 CESS methylolethane melting at 196-7' C. averaged 95%. Accordingly, 89 to 90% yields of trimethylolethane based on propionaldehyde are possible. Calcium and sodium hydroxide were compared as bases in carefully conducted experiments. A higher yield and better purity were obtained with sodium hydroxide (Table I). Formaldehyde itself will undergo the Cannizzaro reaction, but this proceeds much more slowly than the "crossed" Cannizzaro reaction resulting in trimethylolethane. I n an additional experiment 30% formaldehyde and 10% sodium hydroxide were mixed and held at 25' C. This duplicated the situation before the addition of propionaldehyde in preparing trimethylolethane. Small samples were withdrafvn at intervals and the sodium hydroxide and formaldehyde estimated by titration with standard acid and sodium sulfite, respectively (8). Table I1 indicates that the Cannizzaro reaction cannot be ignored. Addition of propionaldehyde to the mixed solution of formaldehyde and sodium hydroxide should not be unnecessarily delayed, and an excess of the latter reactants is desirable to offset the loss to the Cannizzaro reaction. A 15y0 molar excess of formaldehyde and a 20 to 25% molar excess of sodium hydroxide over propionaldehyde were adequate. Formaldehyde will condense initially with propionaldehyde in the presence of a relatively small amount of base as catalyst, which suggests addition of the bulk of the sodium hydroxide subsequently, when its sole function will be to bring about the crossed Cannizzaro reaction. A possible saving in materials was foreseen. Unfortunately, this modification gave a lower yield of trimethylolethane. No evidence was obtained that aldol condensation of propionaldehyde occurred under the conditions of this work. The temperature of the trimethylolethane synthesis reaction appeared not to be very critical. The yield varied little at 25', 30°, and 35' C., but declined appreciably at 40' C. As the Cannizzaro reaction is reported to be a third-order reaction with a fairly high temperature coefficient (7, 5 ) it was concluded that the excess of formaldehyde and sodium hydroxide was insufficient to offset the competing Cannizzaro reaction at the higher temperature. Using simple calorimetric methods, it was found that the heat liberated on adding propionaldehyde to the mixed formaldehyde and sodium hydroxide solution was 38,600 calories per gram mole. Controlled addition of the propionaldehyde accordingly is necessary. After the addition, the reaction mixture can be allowed to stand without further cooling or heating. Trial experiments showed

Table 111. Methyl Isobutyl Ketone Is a Preferred Solvent for Trimethylolethane Temp., O

c.

35 55 75 95 110

Solubility, Grams/100 Grams Solvent 1.3 3.5 10.5 25.9 50.0

that a holding time of about 2 hours was adequate for completion of the reaction. Dilution is favorable to high yields and facilitates temperature control. I t was practical to dilute technical 50% caustic to 10% before mixing with the 30% formaldehyde solution.

for easy removal from the recovered product, and is stable and inexpensive. While the bulk of the water can be removed from the reaction product by simple evaporation, the concentrate is difficult to obtain completely dry except by azeotropic distillation. MIBK is satisfactory for this purpose. Trimethylolethane obtained after separation from the sodium formate using MIBK solvent contained traces of sodium formate and sodium propionate. Ignition always yielded an ash. I t was estimated from titration of a sample dissolved in water with standard acid that about 0.135 mole of sodium salt per kilogram was present in a typical product. The shape of the titration curve suggested the presence of a salt of a second weak acid in addition to sodium formate. When the crystals were wetted with cold phosphoric acid, the odor of propionic acid was unmistakable.

Separation of Trimethylolethane from Sodium Formate Purification of Trimethylolethane

Recovery of trimethylolethane from the aqueous reaction product by extraction with solvents was explored and rejected as unattractive. Butanol and 2-butanol were promising solvents but experiments showed that the volume requirement would be objectionably large. An appreciable amount of sodium formate passed into the extract and appeared as a contaminant in the trimethylolethane subsequently isolated. Attention was then turned to complete removal of water and extraction of trimethylolethane from sodium formate with an organic solvent. Investigation of possible solvents led to the selection of methyl isobutyl ketone (MIBK). Trimethylolethane is very soluble in hot MIBK and its solubility at room temperature fortunately is low (Table 111). The solvent boils low enough (116' C.)

Table IV. Extraction with Small Amount of Water Removes Residual Sodium Salts from Trimethylolethane in MIBK Solution Single Two Extrao- Extraction tions MIBK solution of trimethylolethane, g. 230 Estimated trimethylolethane content, g.O 30 Water used, g. 8 Trimethylolethane recovered, g. 21.8 M.P., ' C. 193 B.P., O C. 267 Ash on ignition, % 0.13 a

230 30 2 x 8 12.5 197.5 283 0.01

Sample isolated had m.p. 186O C., b.p.

258' (deo.).

Recrystallization of trimethylolethane containing residual sodium formate from fresh MIBK solvent resulted in little improvement in purity. Vacuum sublimation, employed in laboratory experiments for estimating yields, did not appear practical. Trimethylolethane could be melted and distilled at 110-mm. pressure, but not without prior addition of a trace of phosphoric acid, which apparently improved its heat stability. Without this precaution, decomposition was extensive. The distilled product had an appreciably lower melting point than sublimed product. I t was found by trial that residual sodium salts could be removed from the hot MIBK solution of trimethylolethane by extraction with a small volume of water. Exploratory experiments (Table IV) indicate the extent of purification possible in single and double extractions. Unfortunately, while a relatively small amount of water efficiently extracts the sodium salts, an excessive amount of trimethylolethane is lost to the extract. This trimethylolethane is recoverable. The water extract was returned to the ketone mother liquor and the so1u;tion dried by azeotropic distillation. This hot solution was again extracted with fresh water in the same amount as before. O n cooling the separated organic phase, a second crop of pure trimethylolethane was obtained. After this sequence of operations had been repeated twice more, the total trimethylolethane recovered was 87% by weight of that originally present. Extraction as described can be accomplished in a practical manner with an extraction column in which water enters at the top, hot feed solution at midpoint, and fresh MIBK solvent at the VOL. 52, NO. 1

JANUARY 1960

35

Table V. Countercurrent Extraction Data Show High Purity Trimethylolethane Obtained

Feed solution used per cycle, 237 g. Estimated trimethylolethane content, 34.5 g. Sample isolated had m.p. 184-6" C., b.p. 258' C . (dec.). Sample Re-

covered Cycle 1 2 3

4 5 6

4

7

Wt., G. 27 31 35.5 35.1

M.P., "C. 196 198 196 196

B.P., "C. 283 283 283 283

diagram. Only three flasks, A , B, and C, were needed to carry out the steps in the order shown. Where ketone phase is shown as retained for recovery of trimethylolethane, this involved distilling part of the ketone (200 grams) and cooling the remainder to room temperature for crystallization. Four samples of trimethylolethane were recovered by extending the steps of the diagram and recovering fifth stage raffinates. Table V shows that recovery of trimethylolethane was very high and the successive crops of crystals were about equally pure. Pilot Plant Process Tests of Continuous Process

bottom. Sodium salts are extracted by the water phase in the upper section of the column and trimethylolethane extracted from the water phase by fresh MIBK in the lower section. Calculations based on necessary simplifying assumptions indicated that two theoretical stages above the stage where the feed is introduced should suffice and that loss of trimethylolethane to the aqueous extract could be held to less than 5% by using two stages below the feed point. T h e use of fresh MIBK as suggested permits adjusting the water feed to a reasonable volume. T h e volume of dispersed water phase to organic phase must be kept relatively small and use of an extractor having mechanically agitated mixing zones is desirable. The proposed scheme for trimethylolethane purification was tested in the laboratory in a batch pseudocountercurrent extraction experiment. A large sample of hot MIBK product solution as obtained after separation of sodium formate was collected and recoverable pure trimethylolethane estimated by work-up of an aliquot. I n following the steps indicated in the extraction scheme, 327gram portions of hot ketone solution were used in the middle stage of each cycle. T h e water phase was created by adding 16 grams where shown. Where fresh ketone is indicated, 200 grams was added. Equilibrium was attained by stirring the mixtures in 1-liter flasks while maintaining the temperature a t 90" C. T h e phases were then allowed to separate and the upper ketone phase was drawn off into the next flask shown in the

36

The process flow diagram embodies steps in the preparation of trimethylolethane worked out in the laboratory investigation. Several unit operations were checked on the pilot plant scale in available or readily constructed equipment. Trial batch runs in a 100-gallon stirred vessel revealed the importance of good mixing in the reaction step. Crude trimethylolethane yield was only 627, when propionaldehyde was added to the stirred reaction mixture in a slow stream as in the laboratory preparation, but was increased to 94 to 95% by circulating the reaction mixture through an external loop and introducing the propionaldehyde a t a mixing point in the loop. Equally good yields were obtained with a continuous reactor. This consisted of a jacket-cooled, baffled reaction zone of 3.1 -gallon capacity, through which the reaction mixture was circulated by a 10-g.p.m. turbine pump connected in the external loop, and a 9-gallon holding tank. The average holding time in the reactor was about 40 minutes, with 2 hours more in the holding tank. For deMatering the reaction product, a Turbo-Film evaporator (Rodney Hunt Machine Corp.) was very efficient. When operated at atmospheric pressure with heat from steam a t 330' F. and a holding time of 9 minutes, the sirup had a water content of only 127,. This concentrate remained fluid and free of solids indefinitely if kept at 70' to 80" C. For testing the proposed extraction scheme a jacketed extractor of the Scheibel type (7) was constructed from 4-inch stainless steel pipe. T h e extrac-

INDUSTRIAL AND ENGINEERING CHEMISTRY

tion zone, 20 inches long, consisted of nine stirred zones and eight settling zones. T h e settling zones were filled to a I-inch depth with'/A-inch Alundum spheres. A motor-driven shaft with attached paddle blades positioned in the open zones provided agitation. T h e feed solution was introduced into the middle stirred zone. Both the feed solution and fresh MIBK solvent were preheated to 90' C. and the temperature within the apparatus was held as constant as possible. I n a typical run feed solution containing 8.27, by weight crude trimethylolethane was fed at 1.56 pounds per minute against water and fresh MIBK rates of 0.16 and 1.35 pounds per minute, respectively. Recovery of pure trimethylolethane was 90%. As shown in the process f l o ~ vdiagram, the water extract can be returned to the drying kettle and loss of prodnuct minimized. T h e MIBK-product phase from the extraction step contains only about 47, dissolved trimethylolethane and provision is made for its concentration by evaporation prior to crystallization. Acknowledgment

T h e assistance of J. P. Hickman in supplying the results of pilot plant tests is gratefully acknowledged. literature Cited (1) Abel, E., Z. physik. Chem. (N.F.) 1, 201 (1954). (2) Derfer, J. M., Greenlee, K. W., Boord, C. E., J . Am. Chem. SOC.71, 179 (1949). ( 3 ) Gottesman, R. T., Cake, W. R. (to Heyden Newport Chemical Corp.), U. S. Patent 2,806,891 (Sept. 17, 1957). (4) I. G. Farbenindustrie, PB Rept. 74978, Frames 925-6, "Inspection of Trimethylolethane Plant in Leuna." ( 5 ) Pajunen, V., Suomen Kemislilehti 21B, 21 (1948). ( 6 ) Poitras, H. A , , Snow, J. E., De Lorenzo, S. A. (to Heyden Newport Chemical Corp.), U. S. Patent 2,420,496 (May 13, 1947). (7) Scheibel, E. G., Karr, A . E., IND.ENG. CHEX 42, 1048 (1950). (8) IValker, J. F., "Formaldehyde," ACS Monograph 120, 2nd ed., p. 383, Reinhold, New York, 1953. (9) Wyler, 6 . A. (to Trojan Powder Co.), U. S. Patent 2,468,718 (April 26, 1949).

RECEIVED for review May 7, 1959 ACCEPTED Spptember 24, 1959