Glycerol

neighborhood of 900 to 1000 grains per gallon (CaC03 equiva- lent). This resin operated successfully over 200 cycles before tests were concluded. Chec...
0 downloads 0 Views 476KB Size
1070

INDUSTRIAL AND ENGINEERING CHEMISTRY

3

Rinse Service cycle of decationized glycerol 5. Sweetened off 4

By obtaining an automaticcycleevery 2 hours, work that otherwise would have extended over a much longer period was considerably shortened. Diluted crude glycerol (about 30% total solids) was decationized batchwise and stored for this work. The total acidity of this material after decationization was in the neighborhood of 900 to 1000 grains per gallon (CaC03 equivalent). This resin operated successfully over 200 cycles before tests were concluded. Check runs were made every 12 cycles, and the capacity of the resin n-as checked against the original capacity. After each twelfth cycle, the resin was given a special treatment to minimize capacity loss. Throughout the test there was no appreciable loss in capacity of the resin for the glycerol solution nor was there any material change in the pH curve during the service cycle. Consequently, when there were indications that the resin would give satisfactory service, the tests \$-erediscontinued. Subsequent tests of the cycled resin on simulated standard acid waters confirmed the low capacity loss. In addition to the resins involved in any ion exchange process, adequate equipment must be provided to house the resins and to provide for the various steps of the process. In general, ion exchange equipment used for glycerol purification is the same as that used for water deionization. A11 reactor tanks, piping, and valves must be of corrosion-resistant materials. The evaporation equipment ehould preferably be constructed of stainless steel. This equipment may be designed for manual or semiautomatic operation. Figure 8 shows a panel board for the auto-

Yol. 43, No. 5

matic control of a commercial drionizcr designed for g l y c e l ~ l purification. Conductivity, ~ € 1 and , specific gravity indica.tors and/or recorders are helpful in the operation of such ion eschaiige equipment,. Costs for complete ion exchange equipment t o purify glycerol solutions rvill vary because of the highly speciali nature of the equipment,. The coiicentrated glycerol produced b y ion exchange follo\vctl by evaporation is equal t o , and in most cases, superior t o that

produced by distillation. The ion exchanged glycerol genera,Ily contained less color, ash, and fatty acids and esters than distillctl glycerol. The glycerol purified by ion eschange has a greater stability to the effect of exposure t o sunlight. Samples of both materials exposed to sunlight s h o m d greater color devolopinent in the distilled glycerol. Probably the greatest advantage of employing ion exchange !'or glycerol purification is that substantially all the glycerol fed to the process is obtained as a C.P. material. I n distillation, only about 70% of the product is of C . P . grade, the rest of the material being dynamite grade and glycerol foots. Certain nonionized impurities in glycerol solutions are not removed by ion exchange. Among t,hese materials are esters, sugars, and polyglycerols. However, these constituents were riot present in amounts great enough in the vsriou,? glycerols treat,cvl tmoexceed qualifications for C . P . glycerol. LITERATURE CITED (1) (2) (3) (4) (5) (6)

Branilner, John D., IT. S. Patent 2,463,677 (Llarch 8, I ! M > British Patent 633,343 (Dec. 12, 1949). Hoyt, Howard E., U. S.Patent 2,381,055 (.lug. 7, 1943). Kahler, F., Chem. Eng., 57, S o . 7 , 109 (1950). Metzger, F.J., U. S.Patent 2,409,441 (Oc,t. 15, 1946). Schwarz, Theo., Ibid., 1,824,507 (Sept. 22, 1931).

RECEIVED Serjtember 9, 1950.

Glycerol S. E. ZACER' AND T. C. DOODY Purdue University, L a f a y e t t e , Ind.

HE physical separation of glycerol from water has long been accomplished by evaporation or more recently by solvent extraction (6-8, 12-14), However, with the advent of new anion exchange resins, having properties comparable to those of strong bases, a chemical method has been developed for this separation. It is well known that boric acid cannot be titrated satisfactorily with strong base, using phenolphthalein as an indicator, unless a suitable material such as glycerol or mannitol is added (10). Boeseken and coworkers (2-5) postulated that such behavior was caused by a complex formed b e b e e n boric acid and glycerol nimilar to the type

T h e removal of glycerol from an aqueous solution by a continuous adsorption process promised to be more economical than the distillation of large volumes of dilute solutions. Exploratory tests show-ed possibilities for the removal of glycerol on an anion exchange resin as a glycerol borate complex anion. Boric acid and sodium tetraborate both gave complex anions which were firm13 attached to anion exchange resins. The best runs showed 93 to 9Sqo removal of glycerol from the solutions up to the breakthrough point of the resins. Desorptions by aqueous solutions and organic solvents have not yet given practical methods for recovery of the glycerol. Analysis of aqueous glj cero1 solutions by refractive index gave excellent agreement with chemical methods of analysis. The refractive index, y , is related to the percentage of glycerol by weight, 2, by the equation y = 0.001247~3- 1.33321

where glycerol and certain other polyhydric alcohols may be irpresented by R. Schafer (9) reported that this type structure predominates Kith mannitol, fructose, erythritol, and glycerol. Tuzuki (11) confirmed the existence of such a complex by polarimetric methods. Hc also reported that a complex can be formed hetween glycerol and sodium tetraborate under conditions which 1

Present address, Johns-Manville, Ino. hlanville, N. J.

render complex formation between glycerol and boric acid impossible. As glycerol forms a complex anion n i t h borate compounds, strongly basic anion exchange resins ought to remove the complex from solution. The resuking investigation was directed toward the removal of the glycerol by this technique. Preliminary investigation showed that more glycerol could be removed if boric acid or sodium t,etraborate were held on the

INDUSTRIAL AND ENGINEERING CHEMISTRY

May 1951

1071

1.34400

1.34300

/20mm.

17mm. 1.0. O.D.

1.34200

I .34 1 0 0

I .34000

1.33900 4

w

E

1.33700

-

1,33600

-

I .33500

-

1.33400

-

0

Figure 1.

Experimental Column

resin, rather than in solution with the glycerol. Apparently the formation of the complex directly on the resin was more effective for the removal of glycerol than complex formation followed by ion exchange. Furthermore, the analysis of the aqueous glycerol solution could more easily be determined by refractive index. A diagram of the experimental column which was used in the investigation is shown in Figure 1. The column was filled with resin to a depth of 33.5 cm., which is approximately equal t o a volume of 75 ml. An aqueous glycerol solution under constant head was percolated through the resin bed at a rate of approximately 15 ml. of effluent per minute. The maximum flow suggested by the manufacturers for the 75-ml. volume of resin was 17 ml. per minute. The resins tested in the investigation were Permutit S (Permutit Co.) and Amberlite IRA-400 (Rohm & Haas Co.). Both resins were strongly basic anion exchangers developed to remove very weak acids such a carbonic, boric, and phenol from solution. The glycerol was of analytical reagent quality from the Mallinckrodt Chemical Works. Distilled water was used throughout. Analyses of the effluent to determine the amount of glycerol remaining in it were carried out by chemical as well as physical methods. A Zeiss dipping refractometer reading t o five places was used for the latter method; the periodic acid method of Allen, Charbonnier, and Coleman ( 1 ) was used for the former. By this method, glycerol may be oxidized by periodic acid at room temperature to two equivalents of formaldehyde and one equivalent of formic acid. The latter may be titrated acidimetrically with sodium hydroxide t o a methyl red end point. The agreement between the two methods is shown in Figure 2. The equation derived from these data by the method of least squares is y = 0.001247s

+ 1.33321

where y = refractive index and z = per cent glycerol as determined by chemical analysis. Because of the excellent agreement between the two methods of analysis, all the percentages of glycerol in the effluent for the 80 runs were determined at 20" C. b y refractive index. The effect of the variables of glycerol concentration, type resin, borate compound attached to the resin, and p H of the solution on the amount .of glycerol removed by the reRins was de-

a a LL

1.33300

0 2 4 B PER CENT GLYCEROL BY CHEMICAL

Figure 2.

0 ANALYSIS

Correlation of Analytical Data

termined as follows: A sample of each of the resins was treated with boric acid and another sample was treated with sodium tetraborate. Each sample was investigated at glycerol concentrations which varied from 1 to 10% and a t two p H values-an unadjusted one of approximately 5.5 and an adjusted one of 7.0. A total of 80 runs was made t o obtain complete d a t a on these variables. PROCEDURE

The procedure for making a run was as follows: The resins were regenerated with a 10% sodium hydroxide solution, were allowed to stand 30 minutes with occasional agitation, and were rinsed with distilled water until the supernatant liquid was neutral to litmus. After being in contact with a saturated solution of either boric acid or sodium tetraborate for 12 hours, the resins were rinsed with distilled water and washed into the column. The resin bed was agitated by a slow backwash rate before i t was allowed to settle. This procedure tended to minimize channeling and to remove air pockets. The distilled water was drained from the column until t h e bottom of the meniscus touched the top of the resin, The glycerol solution, under constant head, was attached to the column, and the effluent rate was adjusted t o approximately 15 ml. per minute by a pinch clamp. Samples were taken a t 2minute intervals except when glycerol concentrations were greater than 6%, in which case 1-minute intervals were employed. The runs were terminated after approximately 300 ml. of material had percolated through t h e resin. T h e refractive indexes of the samples a t 20' C. were determined by a dipping refractometer.

I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY

1072

Vol. 43, No. 5

DISCUSSION OF RESULTS IO

4

8

sr w

3 U

f J5 0

a W

0

54 0 I-

z

w

s3 W

a. 2

I

I

I

50

0

Figure 3.

I

I

100 150 200 VOLUME I N m l .

1

250

I

300

3

Adsorption of Glycerol by Permutit S with H8B08(pH = 7.0)

Both resins gave S-shaped curves, typical of ion exchange materials, when the glycerol concentrationfi were plotted against the volume of effluent (Figures 3, 4,and 5 ) . Figure 3 illustrates one of the less effective methods for the removal of glycerolnamely, Permutit S with boric acid a t a pW of 7.0. Figures 4 and 5 illustrate the best methods for the removal of glycerolnamely, Amberlite IRA-400 with sodium tetraborate. The p H of the solution apparently had only slight effect on the amount of glycerol removed. In all the methods investigated, a family of curves was o b t a i n d similar to those shown in Figures 3, 4, and 5 Thc curve initially was flat, indicating complete or almost complrte removal of the glycerol (93 t o 98%) with the slope gradually increasing until suddenly the resin decreased in adsorptive capacity. At this point or breakthrough the curve sloped shaiply upward, followed by a s l o r tapering off as small amounts of material were still rrhloved. The final value on the curve should be approximately equal to the initial concentration of the glycerol solution. iimberlite IRA-400 with sodium tetraborate appeared to be the most effective in the removal of glycerol, since the breakthrough appeared after approximately 75 ml. of effluent. For all of the other samples, the breakthrough occurred a t approximately 50 ml. of emuent. A better idea of the superiority of Amberlite IRA-400 with sodium tetraborate compared to the other methods is best illustrated by Table I and Figure 6 . Table I shows the total grams of glycerol per gram of resin for all 80 runs. The total weight of glycerol removed for each run was obtained by a material balance. The glycerol removed was obtained by subtracting the summation of the glycerol in each sample through the column from the total weight of glycerol fed to the column. The weights of solution through the column were corrected for

e

rz r

W

3 U 5 A5

0

K

w $4

E ow3 K

W

a

e I

0

fio

I00

150

200

260

300

3!

VOLUME IN m l

VOLUME IN m l .

Figure 4.

Adsorption of Glycerol by Amberlite IRA-400with NasB40,

Figure 5.

Adsorption of Glycerol by Amberlite

IRA-400 with NasB40r (pH = 7.0)

INDUSTRIAL AND ENGINEERING CHEMISTRY

May 1951 OJ 8

0.16

F4

-

1073

bility of influencing the answer. All other interactions were small, and the probability of their influencing the results would occur in 5% or less of the runs. Although removal of the complex from the resin can be accomplished by the usual methods which employ acids, bases, or salts, it was desired to recover the glycerol in some nonaqueous solvent instead. Exploratory work in a Soxhlet extractor with methanol indicated that from 15.5 to 34.6% of the glycerol held by the resin could be removed. Here Permutit S appeared t o be more easily extracted than the Amberlite IRA-400.

P:.”

E

v)

w

Table I. Grams of Glycerol Removed per Gram of Resin for Operating Variables

Glycerol

Concn., % 1

I O IRAr400, HIBO1, pH.7.0 2.. IRA-400, HaBOa 3AlRA-400. Na284O7, P H m

2 3 4 5 6 7 8

9 10

Amberlite IRA-400 Permutit S HsBOa NdhO7 %Boa NazB4Oi _____ pH 7.0 p H 5.5 p H 7.0 p H 5.5 pH 7.0 p H 5.5 pH 7.0 pH 5.5 0.009 0 009 0.048 0.046 0.003 0 000 0 022 0 018 0.025 0:028 0.075 0.087 0.015 0:020 0:037 0:034 0.047 0.042 0 097 0.078 0.039 0.022 0.057 0 053 0.062 0.054 0:114 0.101 0.054 0.038 0.070 0:066 0.067 0.072 0.109 0.126 0,053 0.059 0.075 0.086 0.085 0.086 0.121 0.126 0.073 0.069 0.094 0.095 0.083 0.079 0.160 0.139 0,080 0.076 0.099 0.102 0.099 0.106 0.151 0.159 0.092 0.105 0.119 0.123 0.099 0.113 0.156 0.167 0.093 0.107 0.114 0.122 0.102 0.124 0.150 0.167 0.094 0.117 0.126 0.144

CONCLUSIONS

0

,e,

2 4 6 8 10 GLYCEROL CONCENTRATION IN PER CENT

Figure 6.

Glycerol Removed per Gram of Resin

20 ml. of distilled water which was the approximate amount drained from the resin bed a t the start of each run. It was necessary to keep the resin bed under water to prevent the formation of air pockets and several blank runs were made to determine the approximate amount of water contained therein. The curves for each method of glycerol removal are plotted in Figure 6. These curves show that Amberlite IRA-400 with sodium tetraborate wa8 the most effective method for the removal of glycerol. Although somewhat less effective, Permutit S with sodium tetraborate removed more glycerol than the resins treated with boric acid. According t o data supplied by the manufacturer, Amberlite IRA-400 with sodium tetraborate was operating at its maximum capacity. The maximum amount of glycerol removed by this technique was 0.167 gram of glycerol per gram of resin. Calculations based on the manufacturer’s data indicated that 0.149 gram of glycerol per gram of resin should be removed. This value was based on the capacity of the resin for sodium chloride and hydrogen chloride and will vary according to the roncentration and type material being removed from solution. I n order to determine which variable or variables wore of major importance, a four-factor analysis of variance was applied to the results for initial glycerol concentrations of 5 and 8%. The results indicated that Concentration was the most important variable-that is, the greater the concentration the more glycerol will be removed. The type resin or the type borate attached to the resin were approximately equal in magnitude; next wa8 the interaction between the resin and the borate. The last major variable, pH, Bras small in magnitude compared t o the others. Statistically speaking, the magnitude of the variable is related to the mobability t h a t it will affect the result. In other words, the larger the vaiue of the variable, the greater will be its proba-

In the resin columns used in this investigation, it was possible to remove 93 to 98% of the glycerol from dilute aqueous solutions, based on the volume of solution through the column to the break-point. The best results were obtained with Amberlite IRA-400 and Permutit S, two strongly basic anion exchange resins, which had been treated with sodium tetraborate prior to the percolation of the glycerol solution through them. The Amberlite IRA-400 appeared to have R higher capacity than the Permutit S. The variables of concentration, resin, and borate attached to the resin were found by variance analysis to be the most important. Of slightly smaller magnitude was the interaction between resin and borate attached to it. The p H was found to be a much lese important variable than these. This new method for the removal of glycerol from dilute aqueous solutions makes possible a n approach to the problem which is entirely different from any of the existing techniques. ACKNOWLEDGMENT

The authors wish to thank The Permutit Co. for the Permutit ion exchange resins used in these studiefi. LITERATURE CITED

(1) Allen, N., Charbonnier, H. Y . , and Coleman, R. M., IND. ENG.CHEM.,ANAL.ED., 12, 384 (1940). (2) Boeseken, J., Rec. Trav. chim., 49, 711 (1930). (3) Boeseken, J., and Coops, Ibid., 45, 407 (1925). (4) Boeseken, J., and Vermaas, N., J . Phys. Chem., 35, 1477 (1935). (5) Boeseken, J., Vermaas, N., Gayer, W. H., and Leefers, J. L., Rec. Trav. chim., 54, 853 (1925). (6) Britain, E. H., Joshua, W. P., Whitmarsh, J. M., U. S. Patent 2,377,306 (June 5, 1949). (7) Elgin, J. C., U. S. Patent 2,436,209 (Feb. 17, 1948). (8) Natl. Maize Products, Ltd., French Patent 843,074 (June 26, 1939). (9) Schafer, H., 2.anorg. allgem. Chem., 247, 96 (1941). (10) Scott, W. W., “Standard Methods of Chemical Analysis,” 5th ed., p. 168, D. Van Nostrand Co., New York (1939). (11) Tueuki, Y . , Bull. Chem. SOC.Japan, 16, 23 (1941). (12) Walmesley, R. A., British Patent 515,831 (Dec. 15, 1939). (13) Walmesley, R. A., U. S. Patent 2,235,056 (May 18, 1941). (14) Ibid., 2,235,057. RBCEIVED August 30, 1950.