sponding t o the neutralization of the alkali. This inflection is very small, and is not likely to be confused with the true end point that follows. A refluxing time of 1.5 hours has proved generally satisfactory for both the glyoxals and their bisulfite addition products. For individual compounds, it is possible that a shorter reaction time will suffice. LITERATURE CITED
(1) Ariyama, N.,J . Biol. Chem. 77, 359 (1928).
(2) Becke, F., Gross, O., Z. anal. Chem. 147,9 (1955). (3) Friedemann. T. E.. J . Biol. Chem. 73,331(ig27j. (4) Fritz, J. S., “Acid-Base Titrations in Nonaqueous Solvents,” p. 13, G.
J. J., LLElectrochemical Chemistry,” 2nd ed., p. 93,Interscience, Xem- York, 1958. (11) Moffet, R. B., Tiffany, B. D., Aspergren, B. D., Heinzelman, R. V., J . Am. Chem. Soc. 79,1687 (1957). (12) Pifer, C. W.,Wollish, E. G., ANAL. CHEM.24,519 (1952). (13) Salomaa, P.,Acta Chem. Scand. 10, (10)Lingane,
Frederick Smith Chemical Co., Columbus, Ohio, 1952. ( 5 ) Gabrielson, G., Samuelson, O., Svensk Kern. Tidskr. 62, 214 (1950). (6) Glasstone, S., Hickling, A., J . Chem.
306 (1956). (14) Soderbaum, H. G.,Ber. 24, 1386 (1891). (15) Wise, C. S., Rlehltretter, C. L., Van Cleve, J. K., A x . 4 ~ . CHEW 31, 1241 (1959).
SOC.1936. 824. (7) Goldyrev, L. N., Postovskii, I. Y., J. Gen. Chem. U.S.S.R. 10,39 (1940). (8)Hofreiter, B. T., Alexander, B. H., Wolff, I. A., ANAL. CHEM. 37. 1930 (1955j. (9) Xolthoff, I. M., Z. anorg. u. allgem. Chem. 109,69 (1919).
RECEIVED for review February 29, 1960. Accepted May 20, 1960.
Surface-Active AI kylene Oxide Condensation Products Determination of Polyethylene Glycols in Selected Cationic Ethylene Oxide Condensates J. V. KILHEFFER, Jr., and ERIC JUNGERMANN Armour lndustriul Chemical Co., 7 355 West 3 Jst St., Chicago 9, 111. ,A method has been developed for the determination of free polyglycols in a group of cation-active ethylene oxide condensates. The polyglycols are isolated as dilute aqueous solutions by batch treatment with ion exchange resins. Both the amount and the average molecular weight of the polyglycols are then obtained from the results of two rapid and simple determinations: a spectrophotometric analysis and a dichromate total oxidation.
of the present-day commercial surface-active agents have, as the hydrophilic part of their structures, one or more polyoxyethylene chains. These products are usually prepared by the base-catalyzed addition of ethylene oxide to a hydrophobic substrate containing one or more replaceable hydrogen atoms (14). The surfactants are not single chemical species, because the starting hydrophobe is often a mixture and the ethylene oxide addition invariably gives a range of chain lengths, distributed about a most frequent value. Flory (7) predicted that the distribution should be Poissonian, and several investigators have reported data (2, 6, 10, 11, 16) which show a t least qualitative agreement with Flory’s theory. Two classes of surfactants may be prepared by this process. Thus, nonANY
1178
ANALYTICAL CHEMISTRY
ionic materials are obtained from carboxylic acids (usually fat- or rosinbased), alcohols, phenols, polypropylene glycols, mercaptans, and amides; primary or secondary amines, on the other hand, yield cation-active products, which may, in turn, be quaternized to even more strongly cationic derivatives. During t,he addition of ethylene oxide, the presence of water leads to the formation of polyethylene glycols. The low molecular weight of water and its reactivity toward ethylene oxide imply that the percentages of these glycols may become large, especially in the products containing many moles of ethylene oxide. The similarity of the structure of the glycols to that of the desired adducts makes their removal usually impractical, so that commercial surfactants of this kind are mixtures. Because the presence of glycols in the products may affect their performance in some applications, i t is important to know the extent of their formation. CHEMISTRY A N D STRUCTURE
I n the addition of ethylene oxide to a primary amine, for example, the amino hydrogens are rapidly substituted, but in the absence of a catalyst, further condensation (ITith the hydroxy groups of the 2-mole adduct) is negligible (12). The production of higher condensates is catalyzed by strong alkalies, which
convert the hydroxy groups to the reactive conjugate anions (4, 20) the complexity of the product mixture increases rapidly with the oxide-amine ratio as the result of the presence of two reactive sites, not only in the original substrate, but in each successive adduct. The origin of the numerous constituents can Le seen in the schenie of Figure 1, in which the expression nz, n represents the substance H(OCH2CH2),KR(CH2CH20),,H, and the specific rate constant. k,,, corresponds to the reaction in which the eompound(m, n) is converted to the compound ( m f l , n). Lack of quantitative values of the rate constants precludes predictions of the exact distribution of the product mixture, although some qualitative inferences may be drawn from results appearing in the Literature. The order of reactivity appears to be 0 - > XH > OH, because amines add ethylene oxide in the absence of alkali, but the bis(hydroxyethy1)amines (12) or alcohols (4,80) do not, and because large amounts of caustic catalyze the formation of a product containing a considerable amount of primary and secondary amines, even after the addition of several moles of ethylene oside (12). With smaller amounts of alkali, the product is chiefly tertiary aminesi.e., both amino hydrogens react before appreciable chain lengthening
occurs. I n chain growth, i t seems likely t h a t the similarity of environment of all the hydroxy groups would ensure their similar reactivity. This similarity -which should extend over adducts of differing molecular w i g h t (degree of polymerization) as well as those of differing symmetry-would favor the more symmet'rical products (VI% n) over those hearing one very long and one very short chain ( m >> n), and would lead t o the Poisson-type molecular weight distribution derived by Flory ( 7 ) . Further speculation seems unwarranted in the absence of more quantitat'ive data. Several methods have been applied to the analysis of ethylene oxide condensates, including column chromatography (15, 18), continuous countercurrent distribution ( 6 ) , selective extraction (13, 1 6 ) , and precipitation or complexation by inorganic reagents (3, 17, 21, 2 3 ) . hIoreover, ion exchangers h a r e been used (1, 19, 24, 26) to remove both anionic and cationic surfactants from nonionics. None of the known procedures, however, offered two featurcs desirable for the present purpose-namely, rapidity and an estimate of the molccular weight of the glycol. The present paper describes a procedure n-hich iiicorporates these features and is applicable to fatty-derived surface-active amines and quaternary ammonium salts covering a wide range of molecular weights. The amines, including a n y unreacted starting material, can be separated from the glycols b y contact with the hydrogcii form of a cation exchange resin; the quaternaries require a mixture of cation and anion exchangers t o remove both the organic cation and the associated anion. I n both cases, the glycols are obtained in dilute aqueous solution. Evaporation of the x employed to isolate mater could I the glycols for 17-eighing, but the operation is slow and troul)lesome. Application of spcc4fic and more rapid teste yields satisfactory results. EXPERIMENTAL
Separation. ALIINGP. A sample (0.4 t o 0.6 gram) is dissolved or suspendcd in 25 nil. of distilled water and 5 grams of D o m x 50W-X4 (200 to 400 mesh) is addcd. T h e mixture is allowed t o stand n i t h occasional swirling for 16 minutes, a n d is then filtered n i t h suction through a coarse sintered-glass filter. T h e resin is n-ashed t h r w times n i t h IO-ml. portions of water, and combined filtrates a r e transferred t o a 100-m1. volunietric flask and diluted t o t h e m a r k with water. QUATERNARIES.The same procedure is used, but a mixture of 5 grams of Dowex 50W-X4 (200 to 400 mesh) and 5 grams of an anion exchanger
5,5
c.r
Figure 1. Competitive consecutive reactions in the addition of ethylene oxide to a primary amine
(such as Amberlite IRA-400) in the hydroxide form is used. Oxidation. OXIDIZIXGSOLUTION. Potassium dichromate (8.7 grams) is dissolved in 9 pounds of concentrated sulfuric acid (the amount contained in t h e common commercial package). Warming a n d agitation hasten t h e solution process. b u t must be conducted with due regard t o t h e hazards of handling such a mixture. An aliquot (10 ml. or PROCEDURE. less: The proper size gives a titration of about half the blank, and has to be found by trial) of the glycol solution (filtrate from the separation) is pipetted into a 300-ml. Erlenmeyer flask, and diluted to 10 ml. with mater. Cautiously, 10.0 ml. of the oxidizing solution is added, and the mixture is swirled and placed on a preheated hot plate. The solution is boiled vigorously 15 seconds, then removed and for 75 allowed to cool. The sides of the flask are rinsed with 100 ml. of water, the solution is again allowed to cool, 10 ml. of 10% potassium iodide is added, and the liberated iodine is titrated with standard thiosulfate. A blank is run on the oxidizing solution. Hydroxyl Determination. T h e procedure used is t h a t described b y Hillenbrand (9). The chain length mas shown to have no effect by checking a range of commercially available polyethylene glycols a t several concentrations. Calculations. Per cent glycol = (4.42 9y)/lOw Average number of ethylene oxide units in glycol = r/5y Average molecular weight of alvcol = 8.8$y 18 where 2 = millieauivalents of carbon present in glychl, calculated from results of the oxidation: 100 ( B - S ) S x = A
+
+
I
-
I
in which B = ml. of thiosulfate required for blank S = thiosulfate required for sample N = normality of thiosulfate A = ml. of aliquot oxidized y = milliequivalents of hydroxyl present in the glycol, cal-
culated from the results of the colorimetric analysis : 2/ =
100 C / D
in which
C
of hydroxyl present in the aliquot (read from calibration curve) D = ml. of aliquot used for colorimetric determination 20 = grams of original sample = milliequivalents
DISCUSSION
Quantitative removal of the cationactive amines from the neutral glycols could be conveniently effected by cation exchange resins in the hydrogen form. The oxyethylated amine molecules, although often extremely water-soluble, are apparently too bulky to penetrate even the more porous resins; thus, adsorption occurs only at the surface of the resin bead. Accordingly, use of finely divided resins, with their much greater surface-volume ratios, was found preferable. I n the separation of the quaternary salts, the inorganic anion (usually chloride) is removed by using an anion exchanger in the hydroxide form in combination n i t h the cation exchange resin. Both column and batch procedures were euamined. The column technique was generally slower, and n-as ineffective for the oxyethylated materials of poor water solubility; in cases where the column method removed the amine, regeneration of the resin often proved troublesome. The batch treatment, on the other hand, proved effective for the entire range of materials examined in this study. The dilute aqueous solutions remaining after the ion exchange contain the polyethylene glycols HO(C2H40),H, in which n may vary from 1 to 50 or more. The total weight and the average chain length of the glycol mixture may be found by utilizing the method of end group analysis-well known in the polymer field-in combination with B VOL. 32, NO. 9, AUGUST 1960
1179
total carbon analysis. The wet combustion (essentially a determination of the total carbon content of the solution) alone would suffice for the weight determination if the percentage of carbon in polyethylene glycols were independent of molecular weight, but in the range of chain lengths encountered in this work, the proportion of carbon in the glycol varies too strongly with the degree of polymerization to permit this simplification. The dichromate oxidation of mono-, di-, and triethylene glycols to carbon dioxide and water has been described (8). The general equation for the reaction is HO(CZHIO),H *2
+
+ + 5n C ~ * ( S O+~ ) ~ 3 26n + 3 3KPSOa + KzCrzOi
H ~ S O ~2nco2
3
-f
-
3
Table I. Typical Analysis of Commercial Oxyethylated Fatty Amines and Quaternaries
Glycol Found Av. chain Weight Samplea 5% length Ethomeen C/l2 2 9 2 6 Ethomeen 5/15 4 6 6 8 Ethomeen S/25 13 3 13.6 Ethomeen 18/60 22 7 78 Ethoquad C/12 2 4 2 6 Ethoquad C/25 12 1 13 2 Ethomeen C/12 is the product made by adding 2 moles of ethylene oxide to the primary amine derived from coconut fatty acids; Ethomeens S/l5 and 5/25, by adding 5 and 15 moles, respectively, to the soybean-derived amine; Ethomeen 18/60 by adding 50 moles to commercial octadecylamine. Ethoquad C/12 and C / 2 5 are monomethyl quaternary ammonium salts containing 2 and 15 moles, respectively, of ethylene oxide.
from which i t follows that each mole (44 n 18 grams) of polyglycol re5n quires formula weights (10n equivalents) of dichromate. The concentration of the organic hydroxy group in the dilute aqueous solution can be determined b y a spectrophotometric procedure which depends on the formation of a red complex with hexanitratocerate ion (28). This method was applied to the first three members of the series by Hillenbrand (9) ; in this work i t was extended to polyglycols of average molecular weights ranging up to 1000. It was found that log T is linear with concentration of hydroxyl and independent of chain length. Establishment of a calibration curve for the molecular weight range makes possible the desired end group analysis. Combination of this result with that of the dichromate oxidation, as shown in the calculations section, gives both the amount and the average chain length of the polyethylene glycols isolated. A large number of ethoxylated fatty amines and quaternaries containing as few7 as two and as many as 50 units of ethylene oxide have been examined. Some typical results are shown in Table
+
I. An interesting point is noted: The average number of ethylene oxide units attached to the cationic nitrogen and those making up the polyglycol formed as a by-product, agree within close limits. To study the efficiency of recovering polyglycols from the cationic materials, known amounts of glycols were added and determined by this procedure. Some of the results are summarized in Table 11; they show t h a t the recovery is essentially quantitative. Extension of this general method to propylene oxide adducts of various
Recovery of Polyethylene Glycols Weight, Calculated Found Compound G. hieq. C Meq. OH Meq. C Meq. OH Ethomeena C/12 0.8284 4.70 0.36 PEG 106 0.1062 19.97 2.00 Total 0.9346 24.67 2.36 24.7 2.37 Ethomeen 5/15 0.3981 3.93 0.116 PEG 200 0.0395 8.31 0.368 Total 0.4376 12.24 0.484 12.2 0,493 Ethomeen S/25 0.7344 21.5 0.315 PEG 600 0.0732 16.2 0.248 Total 0.8076 37.7 0.563 37.7 0.553 Ethomeen 18/60 0.4774 24.5 0.063 PEG 1000 0.0472 10.5 0.090 Total 0,5246 35.0 0.153 35.0 0.160 a See footnote to Table I. PEG 106 is diethylene glycol; PEG 200, 600, and 1000 are commercial polyethylene glycols with average molecular weights of 200, 600, and 1000, respectively. The carbon and hydroxyl contents of the glycols were calculated from hydroxyl numbers, corrected for moisture (Karl Fischer titration). The carbon and hydroxyl contents of the Ethomeens were calculated from the data of Table I. Table 11.
cationic materials is now being investigated. Furthermore, the methods and calculations used to analyze the dilute aqueous solution of glycols can probably be used in the analysis of other ethoxylated materials, such as acids, phenols, alcohols, or amides, if suitable separation techniques can be developed. On the other hand, care must be taken that the material being analyzed does not contain substances, such as alcohols, which would interfere in this analytical scheme. ACKNOWLEDGMENT
The assistance of Clemens Roter and John Hinson in performing much of the experimental work is gratefully acknowledged. LITERATURE CITED
(1) Barber, A., Chinnick, C. C . T., Lincoln, P. A., Analyst 81, 18 (1956).
(2) Birkmeier, R. L., Brandner, J. D., J . Agr. Food Chem. 6,471 (1958). (3) Brown, E. G., Hayes, T. J., Analyst 80, 755 (1955). (4) Chitwood, H. C., Freure, B. T., J.A m . Chem. Soc. 68, 680 (1946). (5) Drew, H. F., Schaeffer, J. R., Ind. Eng. Chem. 50,1253 (1958). ( 6 ) Fineman, N. M., Brown, G. L., Myers, F. J., J. Phys. Chem. 56, 963 (19.52). ( 7 ) Flory, P. J., J . Am. Chem. Soc. 62, 1561 (1940). (8) Hillenbrand, E. J., in ”Glycols,” G. 0. Curme and F. Johnston, eds., DD. 340-2. Reinhold. New York. 1952. (9j i b i d . , PI;. 343-4. ’ (10) Karabinoe, J. V., Quinn, E. J., J. Am. Oil Chemists’ SOC.33, 223 (1956). (11) Kelly, J., Greenwald, H. L., 6.Phys. Chem. 62, 1096 (1958). (12) Komori, S., Sakakibara, S., Fujiwara. A.. Technol. Rewts. Osaka L’niv. 6,387 (1956). (13) Kortland, C., Dammers, H. F., J. Am. Oil Chemists’ SOC.32, 58 (1955). (14) hialkemus, J. D., Ibid., 33, 571 (1956). (15) Malkemus, J. D., Swan, J. D., Ibid., 34,342 (1957). (16) hlayhew, R. L., Hyatt, R. C., Ibid., 29,357 (1952). (17) Oliver, J., Preston, C., Nature 164, 252 (1949). (18) Papariello, G., Higuchi, T., Martin, J. E., Kuceski, V. P., Abstract 25, 49th Annual Meeting, American Oil Chemist,s’ Society, Memphis, Tenn., April 1958. (19) Rosen, J. J., ANAL. CHEE~I. 29, 1675 \ - - - - I .
i1957 i.
(207-Satkowski, W.B., Hsu, C. G., Znd. E n . Chem. 49, 1875 (1957). (21) chonfeldt, N., J . Am. Oil Chemisls’ SOC.32, 77 (1955).(( (22) Smith, G. F., Cerate Oxidimetry,” 51-6, G. Frederick Smith Chemical Columbus, Ohio, 1942. (23) Stevenson, D. G., Analyst 79, 504 (1954). (24) Weeks, L. E., Ginn, M. E., Baker, C. E., Soap Chem. Specialties 33, 47 (1957). (25) Weiss, F. T., O’Donnell, A. E , Shrew, R. J., Peters, E. D., Ax.& CHEM.27,198 (1955). RECEIVED for review August 27 1959. Accepted April 6, 1960. Divihon of Analytical Chemistry, 136th Meeting, ACS, Atlantic City, N. J., September
k
81,
1959.
1 180
ANALYTICAL CHEMISTRY
