Cation-Exchange Capacity vs. Phosphorus Content Phosphorus, Cation-Exchange, Compound Mg. AtomlG. Meq./G. 0.286 HCPA 0.274 0.222 0.232 HDPA 0.365 0.412 0.381 0.397 BCPA 0.194 0.199 0.252 0.254 Control ... 0.052
Table 111.
cation that the reaction with cellulose hydroxyls proceeds through the chlorine atoms to attach the phosphinic acid by ether linkages. literature Cited (1) Am. SOC. Testing Materials, Philadelphia, Pa., Committee
D-13, "ASTM Standards on Textile Materials," ASTM Designation D 1424-56T. (2) Ibid.. D 1295-53T. ( 3 j Zbid.; D 39-49. (4) Zbid., 629-59T. (5) . . Chance, L. H., Warren, J., Guthrie, J. D., U. S. Patent 2.743.232 (Ami1 24. 1956). (6) 'Drake, G. L., Jr.,' Reeves, W. A., Guthrie, J. D., Textile Res. .J. 29 ('3). \-,,270-5 - - - (1959). -\ - -
I -
(7).&garine, D. M., U. S. Patent 2,985,501 (May 23, 1961). (8) Grayson, M., Chem. Eng. News 40, 94 (Dec. 3, 1962). (9) Hoffpauir, C. L., Guthrie, J. D., Textile Res. J . 20, 617-20 (,I 1o ,.,",m. ) (10) Marie, C., Compt. Rend. 133, 219(1901); Ann. Chirn. 3, (8) 347-9 (1904); Beilstein I, 652 (1918). (11) Moedritzer, K., J . Am. Chem. Soc. 83, 4381-4 (1961). (12) Pons, W. A., Jr., Stansbury, M. F., Hoffpauir, C. L., J. Assoc. OFc. Agr. Chemists 36, 492-504 (1953). RECEIVED for review January 28, 1966 ACCEPTED April 18, 1966 Division of Cellulose, Wood, and Fiber Chemistry, 150th Meeting, ACS, Atlantic City, N. J., September 1965.
PRODUCTION OF LEVOPIMARIC ACID-FORMAL= DEHYDE ADDUCT AND HYDROXYMETHYLATED MATERIALS FROM RESIN ACIDS AND ROSIN B. A. P A R K I N , J R . ,
H. B . S U M M E R S , 1 R . L . S E T T I N E , * A N D G . W . H E D R I C K
Naval Stores Laboratory, Olustee, Fla.
Processes adaptable to large scale production of the levopimaric acid-formaldehyde adduct from pine gum or from purified levopimaric acid are described. This material was further converted to 6-hydroxymethylabietic acid and hydrogenated over palladium-carbon catalyst to a thermally stable hydrogenated hydroxy acid and over copper-chromite catalyst to 6-hydroxymethyltetrahydroabietinol. A hydroxymethyltetrahydroabietinol was prepared by copper-chromite reduction of oxonated rosin. The thermal reaction of abietic acid and rosin with formaldehyde gave mixtures of hydroxymethylated acids which were not separated but showed an average of 0.7 to 0.9 equivalent of hydroxyl per equivalent of resin acids. Partial hydrogenation of these materials stabilized the methylol content considerably over the nonhydrogenated material. Pimaric acid did not react with paraformaldehyde under the conditions used.
c
effort has been made by this laboratory and industry to make polyfunctional derivatives of rosin through the introduction of hydroxymethyl groups into the resin acid moiety by use of the oxo reaction (4, 5 ) and formaldehyde (8- 70). Investigations of the reactions of formaldehyde with resin acids led to the preparation of the levopimaric acid-formaldehyde adduct (11) (a), but the processes described are not well suited to large scale production of the material. Processes which can be scaled u p and require no specialized equipment have been worked out using purified levopimaric acid and crude pine gum. This product has been further converted to a hydrogenated 6-hydroxymethylabietic acid (IV) and 6hydroxymethyltetrahydroabietinol (V) via catalytic processes. Levering (4)reported the preparation of a glycol by the copper-chromite reduction of the ester function from oxonated rosin obtained from the reaction of the methyl ester of rosin, cobalt octacarbonyl, and hydrogen. Rosin is a mixture of resin acids. According to these authors, this reaction then resulted in a mixture of monohydric alcohols similar to abietyl alcohol and a mixture of glycols. They did not separate the glycol from the monohydroxy materials. T h e present work ONSIDERABLE
Ga.
Present address, Union Bag-Camp Paper Corp., Savannah,
Present address, Deqartment of Chemistry, University of Mississippi, University, Miss.
&OH
n
I
m HzlPd-G
n
H
CH,-OH -
V
H
'\coon IV
describes the preparation of the glycol by the copper-chromite reduction of the polyester resulting from the thermal polymerization of oxonated rosin and separation by distillation of the glycols from the monohydric type materials. T h e thermal reaction of processed rosin and formaldehyde has been investigated with the objective of producing di- or polymethylolated materials for use in industrial poly01 applications. The reaction of rosin and abietic acid in the presence VOL. 5
NO. 3 S E P T E M B E R 1 9 6 6 257
of acid catalysis has been investigated by St. Clair (70) and Royals and Greene (9). Thermal addition of formaldehyde to rosin and abietic acid without added acid catalyst results i~ a product containing 0.7 to 0.9 equivalent of hydroxyl per equivalent of resin acid. The addition of one mole of hydrogen to the material from rosin and subsequent reduction of the carboxylic acid functions gave materials having 1.7 to 1.9 equivalents of hydroxyl per equivalent of original resin acid. Both the methylolated rosin and adduct I1 from levopimaric acid are thermally unstable and lose formaldehyde on heating. Partial hydrogenation stabilizes the products considerably. Pimaric acid does not appear to react thermally with formaldehyde. Work is in progress to find uses for a number of the products resulting from these reactions. Experimental
Adduct Preparation (11). The earlier preparations of adduct I1 from levopimaric acid (I) were carried out by heating the acid and paraformaldehyde together until the mixture resolidified and became immobile. Addition of an inert solvent such as mineral spirits kept the mixture in a viscous but mobile state after crystallization. T h e following example illustrates the procedures developed. FROM LEVOPIMARIC ACID (I). Resin acids [lo0 grams, neutral equivalent (n.e.) 3051 containing 89 grams (0.295 mole) of levopimaric acid (I) were mixed with powdered paraformaldehyde (24.6 grams, 0.82 mole) and mineral spirits (30 ml.) in a 300-ml. three-necked flask equipped with a paddle stirrer and a reflux condenser. T h e flask was immersed in an oil bath maintained a t 130' C. and the mixture was stirred vigorously. T h e solids dispersed, giving a mobile liquid and then, after about 15 minutes, solids precipitated and the mixture became viscous. Stirring and heating were continued for another 20 minutes. The flask was removed from the oil bath and 100 ml. of ether were added. The ether-insoluble product was removed by filtration and washed with ether (2 X 50 ml.). T h e dried product contained some paraformaldehyde (5.5%), weighed 89.6 grams, and accounted for 86.4% of the original levopimaric acid. Analyses. Calcd. for CzlH3208: n.e. 332.5. Found: n.e. 352. Tlic filtrate contained 3.5 grams ( 4 7 3 of the original levopimaric acid unreacted. The 9.6% unaccounted for was probably converted to 6-hydroxymethylabietic acid (111). The presence of I11 was demonstrated by lithium aluminum hydride reduction of the materials in the filtrate. I n a run, in addition to 66.8 grams of adduct [me. 352; 94.5% pure; 64.65; yield (0.19 mole)] obtained initially as a crystalline precipitate, there were 29.2 grams of V (0.09 mole) isolated from the filtrate. The combined yield in moles of I1 and V was 0.28 mole or 95.4%. T h e details for lithium aluminum hydride reduction are given below for the reduction for the product resulting from the reaction of formaldehyde and abietic acid. LABORATORY PREPARATION FROM PINE GUM. Adduct I1 was also prepared directly from crude pine gum. The following examples will illustrate the process. Crude pine gum [400 grams which contained 88 grams (0.291 mole) of levopimaric acid and 10% water] was mixed with flake paraformaldehyde (35 grams, 1.16 moles) in a 1liter three-necked flask equipped with a heating mantle, reflux condenser, and paddle stirrer. T h e mixture was heated to reflux a t 100' C. in 15 minutes and stirred vigorously a t reflux for 45 minutes without removal of water. Heptane (500 ml.) was added, and the solution was filtered and allowed to stand 72 hours a t room temperature for crystallization. The product was isolated by filtration, the cake was washed 258
I B E C PRODUCT RESEARCH A N D D E V E L O P M E N T
with heptane, and the solids were dried to give 61.8 grams of material (n.e. 337.5; 98.2y0 purity, 62.8% yield of pure I1 based on I in the gum charged). There were 4.02 grams (5.7%) of unchanged acid (I) in the filtrate, leaving 22.4 grams (31.3%) unaccounted for, which was probably hydroxy acid (111). Recovery of the by-products, solvent, etc., was not attempted on the laboratory runs. PILOTPLANTPRODUCTION FROM PINEGUM. Addiict I1 was prepared in the pilot plant in a 35-gallon jacketed kettle heated by circulating Dowtherm. A typical run was as follows: Crude longleaf pine gum [81 pounds which contained 17 pounds (0.056 mole) of levopimaric acid] was mixed with powdered paraformaldehyde [7.5 pounds (0.25 mole)] and water (6.5 pounds). T h e charge was stirred and heated to a reflux in 55 minutes. The heater was shut off but the Dowtherm was circulated for an additional 40 minutes to maintain refluxing. The temperature of the charge a t reflux was normally 98.5-99.5'. Heptane (12 gallons) was added and the batch was immediately filtered through a pressure filter and allowed to stand for 48 hours. The adduct was isolated by filtration using a vacuum filter and washed with heptane. Air drying gave 10.1 pounds of material (n.e. 353, 94.2% purity, 51% yield of pure I1 on the available levopimaric acid originally charged). I n a typical recrystallization, 23.6 pounds of the crude product (94.2% purity) were dissolved in 28.4 gallons of hot benzene, filtered, and allowed to crystallize a t 10'. Filtration and drying gave 20.2 pounds of product (11) (91% yield, n.e. 334). The loss in yield was due to conversion of adduct to hydroxy acid (111). Distillation of the original filtrate allowed recovery of the solvent, normal turpentine, in 94 to 97% of that obtained directly from the gum and a iosin, which after steam sparging had a color grade W G and softening point (ring and ball) 80' [n.e. 350 and hydroxyl equivalent (OHE) 13411. This is equivalent to a rosin having a hydroxyl value of 1.27,. The hydroxyl value for a commercial rosin is about 1.0% (5). Thermal Stability of Adduct (11). Adduct (11) (30 grams, 0.1 mole) was placed in a 500-ml. flask with mineral spirits (200 ml.) and heated at 130' for 6.5 hours with periodic sampling. The samples were allowed to cool and stand overnight. The supernatant liquid from which virtually all of the adduct had crystallized was titrated to determine the resin acids present. The acid content of the liquid increased steadily (0.07 meq. per hour per ml. of solution) during the first 5 hours and then leveled off. The final solution was analyzed for levopimaric acid by Lloyd's procedure ( 6 ) . The solution contained 31% levopimaric acid. Gas-liquid chromatography of a sample of methyl ester prepared from the solution on a 5-foot SE 30 Chromosorb W column a t 200' indicated the presence of high boiling materials, presumably the methyl ester of the hydroxy acid as well as esters of resin acids formed by isomerization of the levopimaric acid (7). Derivatives from Adduct (11). The adduct (11) was further converted to 6-hydroxymethylabietic acid (111) by treatment with acid, reduced to a hydrogenated 6-hydroxymethylabietic acid (IV) over palladium-carbon catalyst, polymerized by heating to a polyester, and hydrogenated to 6-hydroxymethyltetrahydroabietinol (V) over copper-chromite catalyst. The latter product was also prepared directly from I1 by hydrogenation over copper-chromite catalyst in dioxane. CONVERSION OF 11 TO 6-HYDROXYMETHYLABIETIC ACID (111) AND
HYDROGENATION TO 6-HYDROXYMETHYLDIHYDROABIETIC
ACID(IV). Crystallized adduct (11) (525 grams. 11.56 moles, n.e. 336) was mixed with 1500 ml. of 95% ethanol and 150 ml. of 3-47 hydrochloric acid. The mixture was stirred until solution was completed and then was maintained a t 5' overnight. After neutralization with 150 ml. of 3N sodium hydroxide the mixture was charged to a pressure reactor with 7.9 grams of 5% palladium on carbon catalyst (Girdler G 81-C) and reduced with hydrogen a t 1000 pounds' precsure for 3 hours. The mixture was then heated to 100' and held there for 1 hour. It was removed from the reactor, 5 grams of charcoal were added, and it was filtered hot through a pressure filter. The charcoal and catalyst were extracted in a continuous extractor with ethanol. The combined filtrate and extract were diluted with water to reduce the alcohol concentration to 50% and heated to 80' to give complete solution. T h e mass was agitated while cooling to room temperature and then was maintained a t 5' in a refrigerator overnight. The
product was isolated by filtration and dried on a steam bath (n.e. 339). The acid was recrystallized by dissolving in 1475 ml. of 95y0 ethanol, diluted with 1325 ml. of water, cooled, and filtered to isolate product (IV) (505 grams, 96%, m.p. 189-91'). T h e N M K spectrum of a purified sample showed one proton in the vinyl region. Analyses. Calcd. for C21H3403: n.e. 334.27; C, 75.41; H, 10.22. Found: n.e. 336; C, 75.15; H, 10.28. Hydroxyl equivalent calcd. 334.57. Found: 329.0. Analysis by gas chromatography of the dimethylsilyl ether of the methyl ester (of the unpurified material) gave three barely distinguishable peaks using a 12-foot column packed with silicone SE-52 on Chromosorb G (2). CONVERSION OF ADDUCT (11) T O 6-HYDROXYMETHYLTETRAHYDROABIETINOL (V). Adduct (11) (60 grams, 0.178 mole, n e . 336) was dissolved in 100 ml. of dioxane and charged to a pressure reactor with 12 grams of Girdler G-13 copper-chromite powdered catalyst. Hydrogen (2500 p s i ) was added and the batch was heated to 275' and the pressure maintained a t 4000 to 5000 p.s.i. for 4 hours. The mixture was cooled, removed from the bomb, treated with 5 grams of charcoal, heated to obtain complete solution, and filtered through a pressure filter. The filtrate was cooled and the product, in the form of colorless crystals, was isolated by filtration. The filtrate was found, by titration, to contain only 0.56 gram of acid calculated as I V . The catalyst in the filter cake was extracted in a continuous extractor to remove occluded products. Forty grams (69.6Y0 of crystalline V) was obtained. Evaporation of the solvent from the filtrate gave 15.3 grams of resinous material. This represents a recovery of 96y0. I n most runs the filtrate was set up for removal of the water formed in the reaction as a result of interesterification by azeotropic distillation. T h e water-free filtrate then was fortified with more adduct, catalyst added, and reduction repeated. There was no noticeable drop in yield on recycling residual materials four times. Crude glycol was purified by distillation in vacuo (b.p. 192', 0.2 mm.) and crystallization from dioxane (m.p. 179-80'). Direct crystallization from dioxane or acetone gave m.p. 179.5-81 '. T h e glycol was undoubtedly 6-hydroxymethyltetrahydroabietinol (V), since it did not absorb ozone and there were no vinyl hydrogens as evidenced by NMR spectroscopic examination. Analyses. Calcd. for C21H3@2: C, 78.05; H, 11.86. Found: C, 78.05. H, 11.86. Hydroxyl equivalent: Calcd. 161.15. Found: 165.6. Scaling up the process to 360 grams of I1 gave 1290 grams of glycol (V) (4.0 moles) from the initial reduction of 2520 grams of adduct (11) (7.54 moles). Drying the filtrate as above and reworking resulted in a n additional 669 grams of V (2.08 moles) from a second reduction, giving a total of 1959 grams of V, (80,8y0yield). An attempt to obtain more glycol from a third reduction gave a n oil which was slow to crystallize. This was distilled through a 12-inch column packed with stainless steel helices. T h e distillate, (392 grams of solid, 16.170) was almost colorless (b.p. constant a t 170', 0.15 m m ; O H E 260.7). By comparison, tetrahydroabietol (b.p. 154', 0.1 mm.) was easily separated from V using the same equipment. From this evidence this second material is not a mixture of V and tetrahydroabietol, which might be expected since the elimination of formaldehyde from the adduct during the reaction would have resulted in formation of this alcohol. Crystallization of the distillate from ethyl acetate gave a crystalline solid (m.p. 100-101.5") which contained one hydroxyl and had no ultraviolet absorbance. I t has not been characterized further, however, for use in calculating elemental analyses; it is presumed to be hydrogenated 6-methylabietinol (C21H80) which could have been formed by hydrogenolysis of the 6methylol derivative. Analyses. Calcd. for C ~ l H 3 ~ 0C,: 82.27; H, 12.5; O H E , 306.3. Found: C, 81.73; H, 12.24; O H E , 298. POLYMERIZATION OF 6-HYDROXYMETHYLDIHYDROABIETIC ACID (IV) A N D HYDROGENATION TO 6-HYDROXYMETHYLTETRAHYDROABIETINOL (V). A polymer (polyester) of the hydroxy
acid (IV) was prepared by heating 60 grams (0.18 mole) of acid to 275' a t 0.2-mm. pressure for 2 hours. Molecular weight by end group titration was 1625 (2.4 ml. of 0.1N sodium hydroxide per 0.39 gram of resin in dioxane). The residue, a clear colorless resin (m.p. 227-32') was dissolved in 120 ml. of dioxane and charged to a pressure reactor with 12 grams of Girdler G-13 copper-chromite powdered catalyst. Hydrogen was added (2500 p s i ) and the batch was heated to 275' for 4 hours a t a hydrogen pressure of 3000 to 5000 pounds. The batch was filtered and product isolated as above (50.7 grams, 88.570 of crystalline (V) (m.p. 179-80'). By titration the filtrate contained 0.67 gram of acid (IV). Reaction of Abietic Acid with Paraformaldehyde. Abietic acid (30.2 grams, 0.1 mole) was placed in a pressure reactor with 12.4 grams (0.4 mole) of paraformaldehyde. The reactor was sealed and placed in an oil bath a t 135' for 8 hours. During the reaction period the material liquefied and was agitated by frequent shaking of the reactor. After cooling, the amber resin formed was chipped from the reactor. Since no material was separated from the crude, the calculated neutralization equivalent should be 426 (found 448). There was 14.6y0 unreacted formaldehyde. The methyl ester was prepared by reaction of the crude product with diazomethane in ether solution. Solvent removal from the resinous or near resinous material was somewhat difficult. Evaporation on a rotary evaporator a t 130' to 150' and 25 mm. gave a n essentially solvent-free product. Infrared analysis of the methyl ester showed strong bonded hydroxyl absorption a t 3.95 microns and a strong band between 9.5 and 10 microns where the methylol hydroxyl group absorbed. A sample (3 grams) of the crude product was reduced by addition, with stirring, of the material in dry ether solution (25 ml.) to a clear solution of a large excess of lithium aluminum hydride (2 grams, 125 ml.) in ether. The clear to slightly cloudy solution which resulted was allowed to stand 12 hours a t room temperature. The reduction mixture was then hydrolyzed by pouring it into a rapidly stirred mixture of 6 N HCI (50 ml.) and crushed ice (150 grams). Ether (100 ml.) was added and stirring was continued until all the solid had dissolved. T h e ether solution was separated, washed with water until a silver nitrate test showed no cloudiness in the wash, and placed over anhydrous sodium sulfate. After standing overnight the product solution was filtered and evaporated (135' a t 25 mm.) yielding a yellow resin. Analyses. Calcd. for C21H3402: active hydrogen equivalent 159.24. Found: 155.2. Acetylation of the reduction product in pyridine-benzene solution using acetyl chloride (7) gave an amber, near-resinous material. Analyses. Calcd. for C2SH3804 : saponification equivalent 201.4. Found: 207. Neutralization equivalents in this series were determined from the p H titration curve of the sample run in alkaline alcohol solution on a n automatic recording titrator using 1.O.V HC1. The free formaldehyde left after reaction was determined on the same sample by adjustment of the p H to the midpoint of the first break, and addition of excess Na2S03 solution, followed by titration of the liberated alkali. Active hydrogen equivalents were determined by the gasometric technique of Zerewitinoff using lithium aluminum hydride in diglyme. REACTION OF PIMARICACID WITH PARAFORMALDEHYDE, Ten grams (0.033 mole) of pimaric acid and 4.1 grams (0.13 mole) of paraformaldehyde were placed in a pressure reactor and heated a t 135' in an oil bath for 4 hours. The crystalline mixture was then heated at 180' to 185' for 4 more hours. O n cooling the still crystalline mass was recrystallized from ethanol to recover exclusively pimaric acid (9.4 grams). Preparation a n d Hydrogenation of Hydroxymethylated Rosin. The addition of formaldehyde to rosin was effected readily a t 120' to 150' C . This material can be hydrogenated to relatively stable dihydro material, which can be further reduced to a partially methylolated rosin alcohol. W C gum rosin (1000 grams, n.e. 346) was placed in a rocking autoclave with powdered paraformaldehyde (100 grams) and heated a t 150' with rocking for 4 to 5 hours. The bomb was VOL. 5
NO. 3
SEPTEMBER 1966
259
opened when the temperature had dropped to 100' and the molten resin was poured to cool on a flat plate. The product was a friable resin containing free formaldehyde (n.e. 390; O H E 434, 3.9% hydroxyl; ring and ball softening point 86.5' F.). T o test the stability, a portion of the product (100 grams) was heated under vacuum (4 to 5 mm., 275') for 4 hours. T h e material darkened considerably (n.e. 488). Reduction with lithium aluminum hydride gave a product having a hydroxyl equivalent to 278.7; 6.1% hydroxyl. (Calculated for alcohol from rosin n.e. 346, O H E 332; 5.1% hydroxyl). A second portion of the product was hydrogenated over palladium catalyst (0.15% of 5y0 palladium on carbon, 1000 p.s.i. hydrogen). The hydrogenated material showed little absorption in the ultraviolet spectrum (no abietic type acids) and on heating at 275' for 4 hours, 4- to 5-mm. vacuum, gave a material having n.e. 626 which upon lithium aluminum hydride reduction gave a very viscous almost resinous material, O H E 199, 8.6% hydroxyl. (Calculated for diol from rosin, n.e. 346, with 3.9% added hydroxyl, O H E 179.5, 9.0% hydroxyl). Copper-chromite reduction of the hydrogenated material resulted in some tyPe ,. of degradation as evidenced bv a high O H E , 287. Reduction of Oxonated Rosin. Oxonated rosin Dreuared from rosin and cobalt octacarbonyl (4, 5 ) (140 grimsj was converted to a polyester, dissolved in dioxane, and reduced by the procedure used for the polyester prepared from IV. After removing from the reactor, stripping solvent, then distilling the volatiles, a glassy residue (polyester) was left as still residue. This was returned to the bomb for reduction. After two additional reductions there were 126.4 grams of distillate and residue, which when fractionated using a short column gave 37.8 grams of monohydroxy compound (b.p. 154', 0.1 mm., 3OyG), 11.1 grams (9.8%) of intermediate fraction, 39.5 grams of diol (b.p. 189', 0.1 mm., 31.5%), and 38 grams of polyester residue, 30.1%. Analyses. Calcd. for C 2 0 H 3 6 0 :C, 82.11; H, 12.41; O H E , 292.29. Found: C, 82.52; H, 11.78; O H E , 296.5. Calcd. for C21H3802:C, 78.05; H, 11.86; OHE, 161.15. Found: C,78.34; H, 11.71; O H E , 167.6. Reduction of 50 grams of oxonated rosin with lithium aluminum hydride resulted in 44.7 grams (96%) of resin, which when fractionated gave 15 grams (b.p. 154', 0.1 mm., 33.67,), 5.6 grams of intermediate fraction (12.5'%), and 14.7 grams (b.p. 189', 0.1 mm., 32.9%). Elemental analyses and hydroxyl equivalent for the first and last fractions were in agreement with calculated values for mono- and dihydroxy compounds.
.,
Discussion
T h e reaction between levopimaric acid and formaldehyde is time-temperature sensitive as a result of the competition of the reaction of levopimaric acid to form abietic acid (7) and its reaction with formaldehyde as well as the further conversion of the adduct (11) to 6-hydroxymethylabietic acid (111) (8). The yield of adduct is also affected by the concentration of levopimaric acid in the starting resin acids, falling off rapidly with the concentration of levopimaric acid. This loss is due partially to a solubilizing effect on the product but probably largely due to increased isomerization as a result of higher carboxylic acid concentration. The effect of varying the ratio of formaldehyde to levopimaric acid is shown in Figure 1. The limiting ratio has not been found. Although aqueous and alcoholic solutions and flake paraformaldehyde were tried, the powdered form gave the best results. The effect of varying the reaction time a t 130' (oil bath) is shown in Figure 2. Reduction in yield was noted a t a temperature other than 130'. The actual temperature of the reaction mass rose to 120-24' a t the time when the mixture became viscous and fell to 110-14' during the remainder of the reaction time. Lower concentrations of levopimaric acid in the feed stock resulted in marked reductions in yields of products. The use of other diluting solvents such as mineral spirits, hexane, and xylene instead of ether gave lower yields. 260
I & E C PRODUCT RESEARCH A N D DEVELOPMENT
" ;c 90
n
0
l
"1
0
0
70
0
I
I 2
0.8
I 1.6
I 2.0
I
2.4
I
I
2.8
3.2
MOLES OF PARAFORMALDEHYM/MOLE OF LEVWIMARIC ACID Figure 1 . Variation in adduct yield with formaldehyde-levopimaric acid ratio
' O T
n
I 0
71-
601
I 0.2
0
1 0.4
I 0.6
1 0.8
I
1.0
TIME (HRS.) Figure 2. Variation in adduct with time under fixed conditions
yield
While maximum yields of crystalline adduct were obtained when 1 to 570 of the levopimaric acid was left unreacted, the reaction can be carried to completion by using higher temperatures or longer reaction times. Under these conditions part of the product is converted to the 6-hydroxymethylabietic acid (111). If this, the hydrogenated acid, or the diol is the desired end product, higher yields may be obtained by carrying the reaction to completion. 6-Hydroxymethylabietic acid does not crystallize well and its purification is difficult. The diols or the hydrogenated hydroxy acid, however, present no such problems. The adduct (11) can be prepared in about 60% over-all yield from longleaf pine gum. As a result of lower levopimaric acid content, the yield from slash gum is about half this amount. With the observation that the presence of water in moderate amounts produced no ill effect on the reaction, temperature control became simply a matter of adjusting the water content of the pine gum to about 10% and maintaining reflux conditions. In the laboratory there appeared to be a dependence of reaction time on batch size. I t was necessary to give 2000gram batches about 60 minutes a t reflux for maximum yield, but pilot plant batches (81 pounds) required only 40 minutes. From this it appears that mixing and movement of the material over the heated surface are the major variation, being best in the small laboratory runs, poorest in large laboratory runs, and intermediate in pilot plant runs. The yield rises with increased formaldehyde concentration, as shown in Figure 3. These data were obtained using 400-gram batches of gum containing 23.7% of I. The trend would be expected to remain the same but with lower yields from gum having a lower
70r
Table 1.
""t E $
0
T
0
40
'"t 20
I
0
o
;
I
I
I 3
2
4
MOLES OF PARAFORMALDEHYDE/MOLE LEVOPIMARIC ACID IN GUM Figure 3. Variation in adduct formaldehyde-pine gum ratio
yield
with
OF
increasing
concentration of I. At maximum yield appreciable levopimaric acid remained as such. Running the reaction to the point of total disappearance of this acid caused a decrease in yield of product. The choice of heptane as the diluting solvent for the pine gum was the result of a series of 400-gram runs (Table I) which were identical in all respects except the diluting solvent. Distillation of the filtrates obtained from formaldehyde reaction with pine gum gave a 94 to 97y0 yield of a commercial grade turpentine having a n cy- and 0-pinene content within the range normally obtained by conventional steam distillation processing techniques. These major terpenes d o not react appreciably with formaldehyde under the conditions used. I t was expected that the residual rosin would have some methylol content, but after steam sparging, the residue and the rosin from the original pine gum had hydroxyl contents in the same range as those found for commercial rosins. Any methylolated material in the residual rosin apparently breaks down under the conditions of isolation with loss of formaldehyde. T h e acid content and softening point are also within the range of commercial gum rosin. Heating the adduct a t temperatures above 130' resulted in its partial conversion to 6-hydroxymethylabietic acid (8) and some loss of formaldehyde, With this evidence it was surprising that hydrogenation of the adduct over copperchromite catalyst a t 2'75' took place without much apparent 1 0 s of formaldehyde. This may be accounted for by opening of the cyclic ether and subsequent hydrogenation of the double bond system a t a temperature below the point of decomposition. Hydrogenation of 6-hydroxymethylabietic acid with palladiumcarbon resulted in a mixture of hydrogenated hydroxy acids, one of which was a dihydro acid. This mixture may be similar to that obrained by Herz et al. (3) by the catalytic reduction of 6-hydroxyabietic acid. Hydrogenated hydroxy acid was converted to a polyester by heating without loss of formaldehyde. T h e preferred process for reduction would probably be continuous, direct hydrogenation of the adduct over copperchromite. Although the yield per pass would not be as high as with the polyester, the ease of separation of the crystalline diol should allow recycle of unreacted material without any difficulty.
Yield of Adduct II from Pine Gum Using Various Diluting Solvents Solvent Yield, 70 Mineral spirits 59.5 Acetone 39.7 50.4 Turpentine 12.7 Benzene Heptane 63.0 Isooctane 55.4
T h e small amount of by-product obtained from the large copper-chromite reductions was believed to be either tetrahydroabietol resulting from loss of formaldehyde from the adduct or a monohydroxy compound wherein one of the hydroxymethyl groups was converted to a methyl group by hydrogenolysis. The analytical data were not sufficient to differentiate between these two possibilities. The reaction of rosin with cobalt octacarbonyl and hydrogen is a practical one. Reduction of the carboxyl moiety by copper-chromite is a n alternative route to a hydroxymethyltetrahydroabietinol, which would lack the optical purity possessed by 6-hydroxymethyltetrahydroabietinol(8). The reaction of processed gum rosin and formaldehyde with acid catalysis ( 9 , 70) appears to be some\+hat different from thermal addition. The thermal reaction, even in the presence of a 2- to 3-mole excess of paraformaldehyde, gives products having 0.75 to 0.9 equivalent of hydroxyl per mole of resin acids. I n the absence of excess formaldehyde, the reaction appears readily reversible, giving rise to ambiguous results in many cases where materials were heated excessively in isolation or further reactions. An instance is seen in the attempt to form a polyester from the original methylolated product. Partial hydrogenation of the methylolated rosin allowed the formation of an interester without loss of formaldehyde, as attested by the fact that the hydroxyl content (8.6%) of the lithium aluminum hydride reduction product is in agreement with the value 9.0% calculated from the original hydroxyl content and neutralization equivalent of the methylolated rosin. The product obtained from abietic acid could not be crystallized, and attempts to separate a product by gas-liquid and thin-layer chromatography indicated the material to be a complex mixture of nonsubstituted and polysubstituted materials. From this, the fact that pimaric acid does not react and the probability that isopimaric, dihydroabietic, and dehydroabietic acids \vi11 not react, it would appear that the abietic acid in rosin reacts, on the average, with more than one mole of formaldehyde. In fact, since gum rosin contains 60% resin acid of the abietic type, most of the abietic acid present must have reacted with 2 moles of formaldehyde to get products having a n average of an equivalent hydroxyl per mole of resin acids. \Yhy abietic acid in rosin reacts with 2 moles while pure abietic reacts with only 1 mole is not clear. Perhaps it is the higher acid function and solubilizing effect of the crude rosin. Conclusions
Convenient procedures involving the reaction of levopimaric acid and paraformaldehyde are reported for the levopimaric acid-formaldehyde adduct in a yield of about 85%. Under optimum conditions (time and temperature) 4 to 5y0 of levopimaric acid remained unreacted and some of the adduct was converted to the thermally rearranged 6-hydroxymethylabietic acid. T h e combined yield of adduct and hydroxy VOL. 5
NO. 3
S E P T E M B E R 1 9 6 6 261
acid was about 95%. O n a laboratory scale, levopimaric acid in pine gum was made to react with formaldehyde, giving a crystalline product easily removed from the gum by filtration in a yield of about 60% based on levopimaric acid in the gum. O n a larger scale involving 81 pounds of gum, 9.5 pounds of adduct was obtained, 51% based on available levopimaric acid. I n addition to being time- and temperature-dependent, the yield of adduct varied greatly (30 and 65%) when the levopimaric acid content of the gum was 18 and 23%. Although the yield of product per unit of pine gum is low, gum is cheap and the process is simple. In 1965, some 290,000,000 pounds of gum were produced, which was approximately a fifth levopimaric acid (50,000,000 to 60,000,000 pounds). With a 50% yield this would have made 25,000,000 to 30,000,000 pounds of adduct. Distillation of the filtrate after removal of adduct resulted in a commercial grade turpentine and rosin. Most of the formaldehyde combined or otherwise vaporized during the processing of the rosin (sparging with steam a t 150' pot temperature). Although the by-product rosin is of acceptable quality, the principal deterrent toward making the adduct from pine gum is the necessity of having to dispose of 6 pounds of rosin for every pound of product. This situation, however, is similar to the turpentine-rosin situation the naval stores industry has lived with through the years. Conversion of the adduct in 95% yield to a thermally stable mixture of hydrogenated 6-hydroxymethylabietic acids resulted from acid isomerization and catalytic reduction. A glycol, 6-hydroxymethyltetrahydroabietinol,was obtained in 80y0 yield by copper-chromite reduction of the adduct. The hydrogenated hydroxyacid and glycol can be prepared in a high degree of purity and should find many industrial uses where a difunctional molecule having a rigid structure is desired. Oxonation of rosin has been used as a means of increasing the functionality of rosin and resulted in a hydrogenated hydroxymethylabietic acid. Copper-chromite reduction of the carboxylic acid function and distillation gave a rosin glycol similar to but less pure than the glycol above. The reaction of rosin with 1 to 2 moles of formaldehyde is a low cost operation. T h e cost of hydrogenation to stabilize the material will add appreciably to the cost of the product.
The thus stabilized reaction product will be a crude hydroxymethylated rosin in which about half the molecules contain a t least one methylol group. Conversion of the carboxyl group to a methylol with lithium aluminum anhydride gave a crude glycol which could have value as an industrial chemical if the carboxyl group was reduced in a practical mannere.g., copper-chromite reduction. When this was tried, degradation occurred, resulting in a resin having a hydroxyl equivalent similar to monohydric materials rather than a glycol. Acknowledgment
The authors thank A. J. Green and J. B. Lewis for many of the analyses necessary in this work; J. C. Braun for assistance in the initial studies on reduction of the acid to hydroxy acid and glycol; R. A. Tyre for making pilot plant preparations of some of the materials; and G. J. Bordreaux, Sylvia H. Miles, and R. T. O'Connor for running the nuclear magnetic resonance and infrared spectra analyses. Literature Cited
(1) Cohen, S. G., Schneider, A., J . Am. Chem. Soc. 63, 3386 (1941). ( 2 ) Hedrick, G. W., private communication, November 1965. (3) Herz, Werner, Wahlborg, H. J., Lloyd, W. D., Schuller, W. H., Hedrick, G. W., J . Org. Chem. 30, 3190 (1965). ( 4 ) Levering, D. R., U. S. Patent 2,906,745 (Sept. 29, 1959). (5) Levering, D. R., Glasebrook, A. L., Ind. Eng. Chem. 50, 317 (1958).
(6) Lloyd, W. D., Hedrick, G. W., J . Org. Chem. 26,2029 (1961). ( 7 ) Loeblich, V. M., Baldwin, D. E., O'Connor, R. T., Lawrence, R.V.,J . Am. Chem. Soc. 77, 6311 (1955). ( 8 ) Parkin, B. A,, Jr., Hedrick, G. W., J . Org. Chem. 30, 2356 (1965). ( 9 ) Royals, E. E., Green, I. T., Jr., Zbid.,23, 1437 (1 958). (10) St. Clair, W. E., master's thesis, Tulane University, New Orleans, La. RECEIVED for review January 7, 1966 ACCEPTEDJune 3, 1966 The Naval Stores Laboratory is one of the laboratories of the Southern Utilization Research and Development Division, Agricultural Research Service, U. S. Department of Agriculture. Mention of commercial products by name is for purposes of identification only and does not constitute their endorsement by the department over others which might be applicable.
SYNTHESIS OF HIGH VISCOSITY POLYOLS H . F. L E D E R L E A N D J . E . M A S T R O I A N N I Olin Mathigson Chemical Cor$., New Haven, Conn.
A number of high viscosity polyols for potential use as lubricants for automotive central hydraulic system fluids were synthesized. To achieve the desired viscosity range of 1000 to 4000 cs. at 100" F., small amounts of diepoxides (butadiene dioxide, 3-vinylcyclohexene dioxide, or resorcinol diglycidyl ether) were copolymerized with ethylene and propylene oxides during the synthesis of the polyols. Sodium oxide or potassium hydroxide may b e used as catalyst, but the former requires a higher reaction temperature.
RISK (4) has recently proposed a n automotive central hydraulic system which would include not only all currently independent hydraulic systems, but also some functions now operated electrically. Such a system must be operable over a temperature range of about 500' F., and consequently a hydraulic fluid of high quality was needed. Specifications for 262
I&EC PRODUCT RESEARCH A N D DEVELOPMENT
central system fluids have been published recently (5). O u r approach to an SAE 71R2 type of hydraulic fluid (5) required a n alkylene oxide-based lubricant of high viscosity and excellent ASTM viscosity slope. T o this end, small amounts of diepoxides were copolymerized during the poly01 synthesis, and a new catalyst, sodium oxide, was evaluated.