I New Uses f o r .
WESLEY L. ARCHER, ROBERT S. MONTGOMERY, KEITH B. BOZER, and JAMES B. LOUCH The Dow Chemical Co., Midland, Mich.
..
Coal Acids Controlled oxidation should be looked at more closely, as a process for producing inexpensive chemicals directly from bituminous coal OxmAmm O F ALKALINE slurries of bituminous coal yields a mixture of watersoluble polycarboxylic acids which have three principal aromatic ring systemsbenzene, napthalene, and diphenyl ( 2 , 3 ) . These acids could be an inexpensive raw material if suitable commercial applications could be developed. The water solubility of these acids, their resin-forming properties, and their film-forming nature suggest certain applications. Water-soluble resins for coating foundry sands looked promising as also did the use of a coal acid, water-plasticized film for a warp size for continuous multifilament yarn. A novel feature of the resins is that a mere physical mixture of the coal acids and other reactant in water may be applied directly to the substrate which is to be bonded. This eliminates the hazard and expense of organic solvents. The resin adduct in a concentration range of 50 to 80% solids gives a water solution of workable viscosity and good film-forming propertics. Thermosetting Resins
Coal acids can react with an alkanolamine, alkylene oxide, polyhydroxy1 compound, or polyamine to give a wide spectrum of resins with varying properties. The following equations illustrate some of the reactions.
+
R(C0OH)T HN(CHzCHz0H)z -+ R(COOCHz-)z(CONCH2-) (1) R(COOH)* -I- HOCHzCH20H + R(C0OCHp)z (2) R(C0OH)z
+ HzN(CHz),NH2
-*
R(CONCHz-)2 ( 3 ) 1 eq. R ( C O 0 H )
+
1 mole HN(CH2CHZOH)z + R[CON(CHgCHzOH)2] (4)
R(COOH)2 f CHgCHzO
+
i-_I
Preparation and Chemistry. Resins were prepared from coal acids obtained from the coal research laboratory of the Carnegie Institute of Technology ( I ) , or prepared in these laboratories. Their average molecular and equivalent weights are about 270 and 82, respectively, resulting in a n average carboxylic acid functionality of 3.3. I t was advantageous to prepare a partially cured but still water-soluble resin. Heating equivalent amounts of the coal acids and monoethanolamine a t 170' C. for 3 hours gave 85y0esterification, but very little amide formation. Another partially cured but water-soluble resin can be prepared from the coal acids and pentaerythritol. A suspension of a n equivalent of pentaerythritol in a 70% solids solution of 1 equivalent of coal acids in water was stirred and heated at reflux for 3 hours. The resultant solution showed 27% esterification of the pentaerythritol. After 24 hours of reflux the esterification approaches the equilibrium value of 40%. The resin is a viscous solution which does not exhibit any precipitation on standing for prolonged periods. Ethylene, propylene, and butylene oxide have been allowed to react with the coal acids in dioxane to produce partially reacted esters containing carboxylic acid and hydroxyl groups. These
products, which are water soluble in concentrated solutions, can be thermally cured to give resins. The most interesting of these resins was that produced by the reaction of 1 mole of ethylene oxide with 2 equivalents of coal acids (Equation 5). The alkylene oxide addition is generally allowed to proceed until one half of the carboxylic acid groups have reacted, giving a resin intermediate which on final cure has no reactive groups. Wide deviation from the one to one acid-hydroxyl group ratio may give a thermoplastic resin on final cure. Ethylene oxide adducts with equivalent weights rangingfrom 162 to 328 havebeen prepared as have propylene oxide adducts with equivalent weights of 170 to 440. Reaction with butylene oxide was slower and required the addition of a small amount of sulfuric acid and additional heating a t 60' C. to give a n adduct with an equivalent weight of 223. A reaction can occur between coal acids and the phenol-formaldehyde resins, Various amounts of coal acids were mixed with both A and B stage phenol-formaldehyde resins and cured. The A stage resin could combine with approximately 43y0 of its weight of coal acids, but the B stage resin with only 7%. This result occurred because the A stage resin contains a larger proportion of free methylol groups. ~
Coal acids can be a cheaper source of raw materials for foundrl'es and also textile and plastics industries. They have an outstanding advantage: The hazard and expense of using organic solvents can be eliminated. More specifically, they can give b Solutions of workable viscosity and good film-forming properties for warp sizing b A wide spectrum of themosetting resins with varying properties
b Outstanding mold-release compounds for foundries
R( COOH)( COOCHzCHgOH) ( 5 ) VOL. 52,
NO.
10
OCTOBER 1960
849
Reaction Rates. Reaction rate studies of the alkanolamine-coal acid resin adducts have afforded information concerning the relative rates of ester formation as compared with amide-imide formation. T h e usual methods for kinetic studies were useless in this case because of the highly cross-linked and insoluble nature of the coal acid resins. The most satisfactory solution to the problem was to cure weighed samples of the resin adduct on squares of aluminum foil which were floated on the surface of a Wood's metal bath controlled at a fixed temperature. The resin samples were prepared from a mixture of 1 to 1 equivalent of alkanolamine-coal acids containing 109;., water. The sample after cure was digested in a methanolwater solution and titrated with I.&' sodium hydroxide. A plot of p H us. milliequivalents of alkali revealed two points of inflection. one occurring a t approximately pH 7 and indicating free carboxylic acid groups, and the second infleaion point at p H 9.5 to 10.5 which corresponded to the amine salt component. Samples were run for various time intervals and the milliequivalents of unreacted carboxylic acid groups and/or amine salt groups plotted us. the cure time in minutes. The semilogarithmic plot approximated a straight line, suggesting a first-order reaction. This behavior is reasonable, because both reactants are combined into one molecular species by means of the amine salt formation. -4 typical plot is shoivn in Figure 1. In this case, samples of a one-to-one
10 0
80 60 $50
%40 30
5 22.0
. (L
z 4 4 10 SOB
-
06 05
0
05
1
15
10
-
20
TIME, IC N
Figure 1. Rates determined at 265" C. indicate first-order kinetics
equivalent formulation of the diethanolamine-coal acid resin were cured at 265' C. The reaction rate constants and half lives of the mono-: di-, and triethanolamine coal acid resin reactions are shown in Table I. The slower rates exhibited by the di- and triethanolamine resins probably result from steric hindrance. Use of an acidic catalyst gave the expected reaction rate incrcase. Reaction rates of the coal acidglycol resins can also be determined provided a lower temperature, such as 150' C., is used to reduce the evaporation losses of the glycol. ,4 plot of the reciprocal of the square of the carboxylic acid concentration us. time as obtained with a coal acid-diethylene glycol resin system gives a straight line plot and thus a third-order reaction is indicated. This is in accord with the mechanism of an acid catalyzed esterification. The catalyzing acid in this reaction is the coal acids becauseof their strong acidic nature. A rate study of the coal acid-pentaerythritol resin adduct indicates a firstorder reaction similar to the alkanolamine resins. The reactants were heated in water to give a partially esterified resin w-hich on titration showed 18% advancement. Samples of the 757c solution were then cured in Ihe normal manner at 230' C. A first-order reaction is reasonable if one considers that a large portion of the partially cured resin adduct is in the form of a molecular species having both acid and hydroxyl groups. Thus the reaction in this study would be primarily an intramolecular polycondensation and first order. Calculation of the first-order reaction rate constant for this reaction gives a value of 0.175 minutes-' and a half life of 3.93 minutes. Saponification rates of some of the cured coal acid resins are shown in Figure 2. The completely cured resin (225' C. for 2 hours) ground to less than 250 mesh was suspended in O . l # sodium hydroxide and stirred a t room temperature. The rate and extent of degradation was proportional to the alkali concentration. The lower saponification rates of the diethanolamine and pentaerythritol resins can be explained by steric hindrance. Physical Properties of the Resins. Sand briquets were used as the test media in determining the physical strength of the coal acid resins. T h e 1-inch thick briquets were of a figure-eight shape
60t-
'e1 ETHYLENE GLYCOL
Figure 2. The ethylene glycol-coal acid resin i s subject to alkaline degradation
identical to the form adopted by the American Foundry Society for evaluating foundry molding resins. The resins in order of decreasing strength are the ethylene oxide (830 p.s.i.), pentaerythrito1 (740 p,s,i.), ethylene glycol (460 to 690 p.s.i.), and ethanolamine adducts (130 to 460 p.s.i,). The alkanolamine-type resins can be considerably overcured without impairing their sirzngth, but glycol resinbonded briquets overcured for 5 minute3 at 260' C . suffer a strengrh loss of some 20%. Investigation of the coal acid resins as glass-fiber binders prompted a study of their thermal stability. Briefly, the method of test consisted of suspending a powdered sample of the cured resin in a wire screen holder placed in a vertical furnace. A thermocouple placed in the sample continuously measured the temperature of the sample as the furnace temperature was gradually increased. The slope of the combustion temperature plot between 700' to 110O0 F. was used to express the rate of resin combustibility. If no resin combustion took place under the test conditions, the resin sample temperature did not exceed 700° F. Therefore, the temperature line between 700' and 1100' F. results from only the resin's heat of combustion. The critical temperature at which the resin would completely burn without additional furnace heat was de-
- 500 &
E5 400 c 300
z Y _1 )
Table I.
Cure Rates of Ethanolamine-Coal Acid Resins a t
Resin Monoethanolamine Diethanolamine Triethanolamine Diethanolamine with 5 % polyphosphoric acid
850
Total k mh-1 t ~ , zmin. 1.22 0.57 0.68 1.02 0.41 1.68 0.87
INDUSTRIAL AND ENGINEERING
Ester k mh-1 f 1 , ~min. 1.67 0.42 0.73 0.95 0.46 1.50
0.80
CHEMISTRY
0.96
0.72
265" C.
Amide /Imide k mh-1 t i , z min. 0.82 0.85 0.62 1.12 0.28 2.44 0.72
0.98
200 IO0
4 VASSER BANK AFS 100
4 % REACTED RESIN SOLIDS
COAL ACID5 termined by means of a second heat stability test. The over-all heat stability of a resin was expressed by dividing the resin combustibility rate by the critical temperature as measured in degrees Fahrenheit. The resins in order of increasing heat stability and respective critical temperatures are as follows: B stage phenol-formaldehyde (400' F.) ; ethylene glycol-coal acids (625' F.) ; diethanolamine-coal acids (700' F.) ; and diethylenetriamine-coal acids (690' F.).
Foundry Resins T h r monoethanolamine, diethanolamine, ethylenediamine, diethylenetriamine, and pentaerythritol adducts of the coal acids were investigated as shell molding resin binders, and the preadvanced pentaerythritol resin was outstanding in over-all performance.
Preparing Coal AcidPentaerythritol Resin The reaction of a 75ojo solids solution of 1 equivalent coal acids and 1 equivalent pentaerythritol during the first 2 hours of reflux is rapid. An equilibrium esterification of 3870 was obtained after 8 hours. The effect of solids concentration on the esterification rate is shown by the fact that a 90% solids solution gives 25% esterification a t the end of the first 20 minutes of reflux. The partially advanced resin is a viscous solution which does not exhibit any precipitation on standing if the advancement is greater than 1270. The resin solution a t 60 to 65y0 solids (reacted) gave a very workable viscosit)-.
Physical Properties of the Coal Acid-Pentaerythritol Resin Preparation of I/d-inch briquets from the dried, resin-coated sand allowed a study of the physical properties of the coated sands as they were related to the resin composition and pretreatment. The sand coating operation was done in a Kitchen Aid mixer or in large batches in a Simpson Muller. A source of hot air directed onto the sand mix shortens the time required for the mulling operation. After screening the coated sand through a 42-mesh screen, it was dropped into a l/4-inch deep briquet mold preheated to 425' F. The coated sand was allowed to stand for 10 to 15 seconds and then the excess unbonded sand scraped off with a thin metal strip. Curing of the briquet a t 650' F. for 2 minutes gave the finished test specimen. These were then tested on a motor-driven Dietert tensile tester, Tensile strength determinations have shown that the pentaerythritol conceii-
tration can be dropped to 0.9 equivalent per equivalent of coal acids without reducing the resin's bonding strength. Use of a 1 to 1 ratio of hydroxyl to carboxylic acid groups seems unnecessary because steric hindrance undoubtedly prevents the reaction of some of the carboxylic acid groups. The strength of the resin-bonded briquets varied with size distribution of the sand, its clay content, the resin composition, and the coating techniques. Suitable strengths were achieved a t the usual resin levels. Four per cent of the coal acid-pentaerythritol resin with a preadvancement of 17.5yG on a coarse Ottawa sand and cured for 2 minutes at 650' F. gave a strength of 360 p.s.i. The tensile strength of the resin coated sand is a function of the resin preadvancement. The preadvancement necessary for optimum tensile strength appears to Figure be in the range of 13 to 18'%, 3 shows the effects of preadvancement on the tensile strengths of two different resin coated sands. This particular data was obtained on small 11/2-pound batches of sand coated in a Kitchen Aid mixer. The one tensile strength value in Figure 3 marked "muller" demonstrates that higher tensile strengths may be expected from the more effective muller coating. The presence of clay and other nonsilica impurities in the Vasser bank sand explains the differences in tensile strength of the Vasser bank AFS 100 sand as compared with the high silica content Wedron AFS 116 sand. The clay and silt impurities decrease the flow of the resin before the thermcset. Thus, flow of the resin coating on a sand grain to the surrounding sand grains and formation of a complete bond would be decreased. The most important propertv controlled by resin advancement is the rate and temperature at which the resin flows. The improved flow behavior of the resin with less preadvancement is directly reflected in the improved tensile strengths. The temperature at which the resin film on a sand grain will start to flow can be determined and is referred to as the stick point. A lower preadvancement of the resin gives a lower stick point and a resultant higher tensile strength. Preliminary observations have shown that the amount of moisture pickup by the resin coated sand is another factor controlled partially bv the degree of resin advancement. Thus, adequate resin advancement is necessary so that the resin coated sand will remain in a free flowing condition before actual use. Sand coated in a Simpson Muller with a resin of at least 15% preadvancement will give a stable free flowing sand. This range of resin preadvancement may
also be expected to give nearly optimum tensile strengths.
Shell Mold and Core Fabrication The better coal acid-pentaerythritol resins were then examined in the actual fabrication of shell molds and cores. The test shell mold was a small three prong cover plate. Hollow shell cores have also been prepared on a commercial Shalco core blowing machine. A Nugent AFS 75, coarse Ottawa or Wedron AFS 116 sand coated with 4yc resin gave good cover plate shell molds at a pattern temperature of 450° to 500' F., investment tim? of 15 seconds, and cure time of 40 to 60 seconds. The resin preadvancement of approximately 150/, is preferred for maximum tensile strength and pattern performance at a low pattern temperature of 450' F. Moisture stability of the coated sand appears to be good since no build-up of coated sand on the back side of the shell and/or peel back of a portion of the shell occurs. The differences in tensile strengths of a high silica sand and bank sand shell mold are again illustrated by the following values. The tensile strength of the 4Oj, resin (advanced 16y0) coated Nugent sand averaged 315 p.s.i. as compared to the value of 795 p.s.i. for the resin on Wedron sand. Hardness as measured by a Dietert hardness tester was 80 for the Nugent shells and 90 for the Wedron sand shells. Later work showed that the addition of a little nonionic surfactant such as 0.25y0 Triton X-100 to the resin coated Nugent or Vasser sand would give a 250/, strength increase. Thus the tensile strength of a 4Oj, resin coated Nugent sand can be increased to 400 to 450 p.s.i. The addition of 5y0 (based on resin weight) powdered ammonium chloride to the sand during the coating operation decreases the thermosetting time during shell fabrication. Acceptable commercial hollow shell cores (23 inches in length and 2 to 5 inches in width) have been fabricated on a Shalco core blowing machine using a 3.5y0 coal acid-pentaerythritol resin coated Wedron sand. The resin had been advanced to 28.5% esterification. Shell cores with excellent hardness (90 on Dietert tester), detail, and strength were prepared a t 600' F. with an investment time of 5 to 15 seconds and cure time of 1.5 to 2.0 minutes. These conditions correspond to cure conditions used with a commercial phenol-formaldehyde resin. Several of the coal acid type resins may find utility in foundry cores. Specifically. cores with tensile strengths of 570 p.s.i. have been prepared from an Ottawa VOL. 52, NO. 10
OCTOBER 1960
851
AFS 60 sand bonded with 1.0% coal acid-pentaerythritol resin, 1.O% Mogul B211 flour, and 5 y 0 water. Core mixes of this type have been successfully blown on commercial machines. The tensile data of other coal acid resins given in the preceding section also suggest possible use in core binder applications. Preparation of briquets from a moist resin coated sand by allowing the briquets to stand at room temperature leads to the development of considerable green strength. This green strength is the result of the air drying of the resin film to give a "dry" water-plasticized resin film surrounding and bonding each sand particle. The diethanolamine resin adduct, for example, yields an optimum strength of 325 p.s.i. at a 6% resin concentration while 6y0 of the diethylenetriamine adduct gives a green strength of 240 p.s.i. after only 20 hours a t room temperature. This green strength property could also be of interest in sand core fabrication. One of the earlier coal acid-pentaerythritol resins (preadvancement of 28.5Yc) was field tested as a shell molding binder in a malleable iron foundry. Shell molds 20 X 30 inches in size were fabricated at a pattern temperature of 500' F.: investment time of 34 seconds, and cure time of 50 to 60 seconds. Twenty-five castings weighing 11 pounds each were successfully cast without any major metal breakouts. This resin? however, seemed to lack sufficient tensile strength. Tensile bars prepared from the coated sand had an average strength of only 170 p.s.i. This problem has been obviated by using the proper resin preadvancenient and significantly stronger resins have been prepared. Warp Size Uses Aqueous solutions of the coal acids \vi11 drv to form a rather tough watersoluble film. One possible use for such a material is as a warp size in the textile industry. A warp size is a water-soluble or dispersible coating applied to the warp yarns to increase loom efficiency during weaving. The size must have good adhesion to the yarn and be sufficiently pliable to protect the warp yarns during weaving. The size must also be easily removed by dilute detergent solutions after the weaving operation. The coal acids were first evaluated on continuous multifilament nylon yarn. A polyacrylic acid size (Acrysol A-1) manufactured by Rohm 8r. Haas was chosen as the control. Use of a laboratory single-end slasher indicated that the coal acid coating could be easily applied to the yarn. Two methods were used to determine whethcr the yellow colored coal acid coating could be satisfactorily removed
852
By detcrmining the fray point of samples of yarn having varying percentages of size pickup we were able to compare the efficiencies of the comSample Rd a b mercial and coal acid sizes. The number Porcelain 77.0 - 0 . 2 +lag of cycles necessary to fray sized nylon Unsized yarn 52.8 - 3 . 0 +3.0 multifilament yarn is shown in Figure 4 Washed yarn (Acrysol as a function of the size pickup. There is A-1) 54.9 - 2 . 7 +3.8 Washed yarn (coal little difference between the abrasion acids) 52.6 - 3 . 0 4-3.1 protection afforded by Acrysol A-1 and that afforded by the coal acid size. Evaluations carried out on Dacron polyester fiber, Orlon acrylic fiber, and from the yarn. First, sized samples of cellulosic-type yarns gave results similar yarn were rinsed \vith a stream of water to those obtained on nylon. The one for 10 minutes and after drying titrated noteworthy difference was that the for residual acid. This test indicated coal acids afforded much greater prothat the amount of coal acids left on tection to Orlon multifilament yarn the yarn was less than o ~ 0 5 y 0of the size than did Stymer-S size. An Acrysol pickup. The second and more sensiP-4 size was used as the control on tive test was based on residual color as Dacron while Stymer-S size served as determined on a Hunter color and color the control on the acetate multifilament difference meter. Samples of the sized yarn. yarn lvhich had been washed in an ,4 small scale weaving trial was SLICaqueous solution of 0.25% Triton X-100 cessfully conducted at Dow and afterand 0 . 2 5 7 . tetrasodium pyrophosphate ward a larger and more thorough were wound on white plastic chips and trial was carried out at North Carolina the residual color determined. State College, School of Textiles, Raleigh, Table I1 shows the results of this N. C . A continuous multifilament nylon evaluation with coal acids and Acrysol yarn sized with Acrysol A-I was used '4-1. as the control in these trials. While Fl'he Kd values indicate the per cent duration of the Meaving trials was not reflectance while the negative numbers sufficient to calculate the efficiency of in column a show the degree of blueness. the sizes, there were fewer yarn breaks The degree of yellowness is indicated attributable to "size failure" in the coal by the positive values in column 6 . acid sized \varp than in chr Acrysol This data demonstrates that the coal acid sized warp. color can be completely removed from 'The only serious problem \vas enthe )-am. countered in the scouring step. During The abrasion resistance of the sized this step the sized fabric was inadvertently yarn was used in the laboratory to exposed to live steam before the scourmeasure the size efficiency. Strands of ing was starled. The resultant temperasized yarn were abraded with the osciltures were sufficient to "set" or cause lating blade of a Duplan cohesion tester chemical combination of the coal acids and the "fray point" determined. This lvith the yarn. fray-point indicates when the size film Further testing in the laboratory is destroyed. indicated that temperatures of 180' F. or greater would cause varying degrees of discoloration in proportion to the I -exposure time. Size-drying temperatures are normally in this range, but 50 the high fluidity of the coal acids should allow drying at lower temperatures. I n addition, this discoloration 40 of the yarn a t higher drying temperatures may not be a problem with yarns other than nylon. Table
3
II.
Hunter Color Difference
and
Color
30
0
literature Cited
0
(1) Franke, N. W., Kiebler, M. W., Ruof, C. H.? Sovich, T. R., Howard, H. C . , IND. ENG. CHEM.44, 2784-92 (1952). (2) Montgomery, R. S., Holly, E. D.? Fuel 36, 63 (1957). (3) Montgomery, R. S., Holly, E. D., Gohlke, R. S., Ibid., 35, 60 (1956).
20
10
6
for review June 22, 1959 RECEIVED ACCEPTED May 31, 1960
Figure 4. The coal acids are equal to Acrysol A-1 as a warp size for nylon
Division of Gas and Fuel Chemistry, 136th Meeting, ACS, Atlantic City, N. J., September 1959.
0
0
1
2
3
4
5
6
7
PICKUP, %
INDUSTRIAL AND ENGINEERING CHEMISTRY
