Table IV. Determination of N-Alkyl Acetamide In Amine Acetates
N-Octadecyl Acetamide, % DeterAdded mined6
Sample Tallowamine acetate, lab. prep. 0.6 0.6 Tallowamine acetate, lab. prep. 3.9 3.8 Tallowamine acetate, lab. prep. 9.1 9.1 Tallowamine acetate, commercial-1 . . . 0.4 Tsllowamine acetate, commercial-1 1.1 1.6 Tallowamine acetate, commercial-2 ... 1 . 0 Tallowamine acetate, commercial-3 .., 1 . 4 Tallowamine acetate, commercial.-4 1.7 e Values for samples containing added N-octadecyl acetamide corrected for amide originally present.
...
dodecylamide) Absorptivities for solutions containing 2 grams per liter of dodecylamide were 2.81 liter per gramom. a t 5.92 microns and 0.990 liter per gram cm. a t 6.01 microns in a nominal 1-mm. cell. The same concentration
of pure N-dodecyl dodecylamide gave absorptivities of 0.520 liter per gramcm. a t 5.92 microns and 1.24 liter per gram-cm. a t 6.01 microns. Simultaneous equations were employed in the usual manner to calculate the percentage of each type of amide. Table 111 lists recovery data for synthetic mixtures of dodecylamine, dodecylamide, and N-dodecyl dodecylamide. Primary amine acetate, another commercially available fatty nitrogen derivative, is prepared by neutralizing primary fatty amine with acetic acid. A possible by-product of this neutralization is N-alkylacetamide. The ion exchange-infrared method also was applied to the determination of small amounts of this acetamide in tallowamine acetates. For this determination the nonamines from the ion exchange column were dissolved in 10 ml. of chloroform and the solution was scanned from 5.4 to 6.2 microns in a nominal 1-mm. cell against chloroform. The Beckman IR3 was employed a t settings of 5% gain, 0.3 micron per minute speed and w Csecond period. The absorptivity a t 5.97 microns was determined and the percentage of Nalkylacetamide was calculated from the absorptivity of purified N-octadecyl
Alk
olyrner Distri Condensates
tion in
Determination o
onomer through Te
acetamide, which was 1.741iterper gramcm. Appropriate correction was applied for any unsubstituted amide and N-alkyl alkylamide in the amine used for preparing the acetate. This was accomplished by carrying a sample of the amine through the same procedure and correcting for the 5.97-micron absorption of its nonamines. Table IV lists results by this method on a series of amine acetate samples. ACKNOWLEDGMENT
The authors gratefully acknowledge the work of Barbara Polister, Wesley Tolberg, and Dolores Bell, who contributed to various phases of the work reported herein. LITERATURE CITED
(1) Barber, A., Chinnick, C. C. T., Lincoln, P. A., Analyst 81, 18 (1956). (2) Killheffer, J. V., Jungermann, Eric, ANAL.CHEM.32, 1178 (1960). (3) Myers, R. T., Ohio J . Sci. 58, 34 ( 1958). (4) Watkins, S. R., Walton, H. F., Anal. Chzm. Acta 24,334 (1961).
RECEIVED for review July 3, 1961. Accepted October 5 1961. Paper No. 275 Journal Series, bentral Research Laboratories, General Mills, Inc.
yethylene
MARY ELLEN PUTHOFF and J. H. BENEDICT Procter and Gamble Co., lvorydale Technical Cenfer, Cincinnati, Ohio
b A new method that separates and quantitatively determines components containing one to four ethylene oxide units in surface active alkyl ether polyoxyethylene polymers is presented. The samples are reacted with p phenylazobenzoyl chloride and the resulting esters are chromatographed on alumina. The individual components are eluted with solvents of increasing polarity and the amount is determined gravimetrically. The combined standard deviation for the method is 1 .270b.Samples examined contained a wide range of components, including 10 to 40% of unreacted fatty alcohol. The oxyethylene distribution in the samples examined does not follow the expected Poisson distribution. 1884
ANALYTICU. CHEMISTRY
A
ether polyoxyethylene polymers produced by the condensation of fatty alcohols and 1 to 100 moles of ethylene oxide are widely used as surface active agents. The low molecular weight polymers that contain an average of 1 to 5 moles of oxyethylene exhibit good emulsifying and wetting properties. Since these condensation products consist of a broad range of molecular weights, ' knowledge of the polymer distribution is needed to correlate performance with composition. The reported procedures for determining the total polymer content (S, 8-10) do not show the distribution of the species. Only molecular distillation (6, 7) has been used to separate and to determine the various oxyethylene species present in conLKYL
densation products. However, this technique requires considerable time and is not suitable for routine analysis. Kelly and Greenwald (6) used chromatography on silica gel to separate the species containing two to 13 oxyethylene units in an octylphenol polyoxyethylene product, Silica gel chromatography of alkyl ether polyoxyethylene polymers did not give complete resolution of the low molecular weight components. Also, the Kelly and Greenwald separation requires several liters of solvent and long elution time for each sample; therefore, a new procedure was developed which involves chromatography of esterified polymers. This paper presents a method for the rapid, routine determination 'of the oxyethylene distribution in alkyl poly-
oxyethylene polymers containing an average of one to four osyethylene units. The condensates are esterified with p-phenylazobenzoyl (PPAB) chloride, and the esters are separated chromatographically on an alumina column. The individual components are eluted with solvents of increasing polarity, and the amount of each is determined gravimetrically. EXPERIMENTAL
High purity p-phenylazobenzoyl (PPAB) chloride is used as received. Anhydrous pyridine is required for the esterification. All solvents used in the esterification and evaporation steps must be alcohol-free, because i t interferes in the subsequent chromatography. Woelm acid-quality alumina (Alupharm Chemicals, New Orleans, La.) is adjusted to 3% moisture. The chromatographic tubes, 1.9 x 28 om., are equipped with separatory funnels as solvent reservoirs. Each column is filled with 45 grams of alumina, and a glass wool plug is placed on top of the alumina. Procedure. Reflux for 1 hour 0.1 gram of sample, 0.18 gram of PPAB chloride, and 1 ml. of pyridine in a PO-ml. Erlenmeyer flask with an air condenser. Add several milliliters of benzene and dichloromethane (1 to 1) and evaporate the mixture to dryness. No odor of pyridine should be detectable. Transfer the esters to the chromatographic column using 10 ml. of hexane and then pass 150 ml. of benzene through the column. The benzene effects a separation of orange-yellow bands; each band is a separate species and is collected in a separate flask. Benzene elutes the ester of the unreacted fatty alcohol first and then the monomer (ester of the alkyl ether containing one mole of ethylene oxide). Collect each in a separate flask. Elute the dimer with 100 ml. of 95 to 5 benzene rind ethyl ether, the trimer with 100 ml. of 90 to 10 benzene and ethyl ether, and the tetramer with 100 ml. of 75 to 25 benzene and ethyl ether. Evaporate the solvent from each fraction, coo1, and weigh. Calculate the material remaining on the column by difference. Reagents and Equipment.
% species = F x l! x 100 P X W
where F = fraction weight M = molecular weight of species P = molecular weight of PPAB ester of species W = original sample weight DlSCUSSlON
Alumina with a y crystalline form is employed in preference to the more
widely used a monohydrate crystalline structure. Previous studies a t Procter and Gamble and by others have shown that different production batches can have significantly different adsorptive properties. Recent publications of standard methods, such as that of the Association of Official Agricultural Chemists (I), specify the use of the y form for more uniform performance. Because of its more consistent properties, the y form was selected for this work. Also, the pH of the a h minum oxide is important in obtaining a good separation. Acidic alumina with a pH of about 3.5 gave better results than the neutral or alkaline types. Although the order of elution was as expected, a verification of the fraction identities was deemed advisable. Therefore, a polyoxyethylene polymer containing an average of 3 moles of ethylene oxide and made from coconut fatty alcohol. was chrornatographed. The chemical analyses obtained on the individual PPAB ester fractions are given in Table 1. The per cent oxygen was determined by the direct oxygencombustion method. The saponification value, which was determined by a semimicro modification of the standard procedure, was used as an indication of the molecular weight of the fraction. I n the fraction coding, the E represents an alkyl polyoxyethylene polymer and the numbers indicate the moles of ethylene oxide in the molecule. Thus, E-0 is unreacted alcohol and E-1 is the monomer. The experimental oxygen and saponification values are in good agreement with the theoretical values, and the variations are within the standard deviation of the methods. Thus, the expected identities of these fractions are confirmed. The accuracy of the analysis was checked by comparing results with those obtained by molecular distillation and gas chromatography. The oxyethylene distributions as determined by molecular distillation and by the new method are shown in Table 11. The sample analyzed was made here from dodecyl alcohol and has an average of three ethylene oxide units per mole of alcohol as determined by hydroxyl value. The data show that the percentages as determined by the two methods are in good agreement and that the proposed method gives an accurate analysis of the condensation product. Further evidence of the accuracy of the new method is shown in the comparison of the results obtained by it and by gas chromatography (Table 111). The samples examined were commercial products made from coconut fatty alcohol (thus the 6!in the sample identification) and were reported to contain
Chemical Analyses of PPAB Ester Fractions
Table 1.
E Frac-
tion 0 1 2 3 4
Savonification - value Oxygen, % Theory Found Theory Found 8.1 11.0 13.2 15.2 16.8
142 128 116 107 98
8.7 10.8 13.0 14.8 16.3
146 131 114 112 101
Table II. Comparison of Results on 12-E-3 Polyoxyethylene Polymer
Found, % E 4
E-1
E-2
E-3
E 4 E-54-
Molecular Distillation 15.0
14.0 9.0 12.5 10.5 39,O PPAB Ester Separation
17.5
1 3 . 1 1 1 3 . 8 12.3 10.0 33.3
Table ill. Cornparison of Results Polyoxyethylene Polymers
Sample DE-1" C-E-1* 6-E-2" C-E-2b c-E-3~ C-E-3* a
b
E 4 43.1 42.3 21.4 23.2 11.0 11.4
OR
Found, % ' E-1 E-2 11.0 13.2 1.1.2 15.5 11.5 12.9
20.8 20.3 14.7 13.2 9.7 7.9
Analyzed by gas chromatography. Analyzed by PPAB ester separation.
an average of 1, 2, or 3 moles of ethylene oxide, respectively. The gas chromatographic method, using a 6-foot column of 15% Apiezon-L on Chromosorb VV a t 240" e., separates the sample according to the alkyl chain length and the number of ethylene oxide units. Only the unreacted alcohols and the monomeric and dimeric components are determined quantitatively by an internal standard. Although other higher polymeric materials are eluted from the column, they can only be estimated, because their long retention times and incomplete elution from the column give low results. The gas chromatographic results for total unreacted alcohol, monomer, and dimer are essentially the same as those obtained by the PPAB ester method. The ester procedure is preferred, a t this time, because it provides more information about the higher polymers. The reproducibility of the method was determined by replicate analyses on two samples (Table IV). Sample l@E-3 was made from C-10 fatty alcohol and sample 12-E-3 from a C-12 alcohol; both contained an average of three ethylene oxide units. A standard deviation was calculated for each fraction, using the values obtained from VOL. 33, NO. 13, DECEMBER 1961
a
1
le IV.
~ e ~ r o d u e i b iof ~ ~ ~Py Procedure
15-E3 12-E-3 Av. Std. Av. Std. value, devv., value, dev., % % % %
E
Fractiom 0
8.8 6.7 10.1 10.3 10.0
L
2 3 4
1.3
17.5 13.1 13.8 12.3 10.0
0.8 0.9 1.0 1.7
0.2 0.9 1.1 1.6 2.3
Combined standard deviation 1.2%
five determinations on 10-E-3 and six runs on 12-E-3. The combined standard deviation is 1.2%. The total elapsed time required for one analyst to run six samples is 8 hours; thus the method is suitable for routine examination of a large number of samples. Although this procedure was designed for alkyl ether condensates, it probably would be applicable to alkyl ester, alkyl amide, and alkyl aryl condensates with some minor changes in the elution scheme. Five commercial polyoxyethylene polymers from three manufacturers made with varying initial ethylene oxide contents were examined (Table V). E-5+ is the total of E-5 and higher polymers. These data show that, contrary to expectations, these polymers contain a aide range of species and in no case is the designated material the major component. Another surprising fact is the appreciable quantity of unreacted alcohol present in all samples. These variations in composition are not evident from determinations run on the total sample. For example, sample C-E-3 has an average of 3 moles of ethylene oxide per mole of alcohol as determined by hydroxyl and ethoxyl values on the total sample. The PPAB ester separation data show that the
Table V.
Sample C-E-1
A
C
D E F
Av. C-E-3
1886
e
actual trimer or E-3 fraction is not the majority of the sample but only lOyo Gf the total material. This does not mean that the hydroxyl and ethoxyl values are in error; rather that the distribution of unreacted alcohol, the low molecular weight species, and high polymer material is such that these values average that of a polymer containing three ethylene oxide units. This same wide range of species was evident in all other commercial samples examined. A second very interesting point shown by these data is the similarity in composition of all these polymers, esciuding perhaps the 1643-6 sample. The major difference between, for esample, the C-E-1 and the 16-E4 condensates is in the amount of unreacted fatty alcoho? and percentage of materiai having five or more ethylene oxide units. These samples are typical of many other condensates examined and show the usual oxyethylene distribution encountered. This is evident from the analyses cited for the six C-E-3 samples in Table V. Sample il was made in the laboratory from dodecyl alcohol,
Oxyethylcne Distribution in Commercial Samples
E-0
Found, yo E-1 E-2 E-3 Various Ethylene Oxide Contents
42.1
B
Figure 1. Comparison of theoretical and experimental oxyethylene distribution in C-E-3 condensates
17.5 11.0 14.8 17.4 11.6 12.9 14.2
19.6 13.2
14.8 15.5
8.5 3.4 Various C - E 3 Samples 13.1 13.8 8.3 14.0 11.9 11.1 12.4 12.7 10.6 14.8 9.7 14.4 11.0 13.5
ANALYTICAL CHEMISTRY
E-4
6.9 10.4 10.2
5.8 9.2 9.0
4.8
2.8
12.3 13.0 9.7 9.3 8.4 10.2 10.5
10.0 10.9 8.8 9.0
9.2 9.0 9.5
E-5+ 10.8 28.6 43.8
33.3 42.7 43.7 39.2 45.5 43.8 41.4
and the others were commercially prepared from coconut fatty alcohol. Each sample was made by a different manufacturer. Although there are slight differences in composition from one manufacturer to another, the condensates are essentially the same and show the same distribution pattern. Flory (4)in studies on the theoretical composition of polyoxyethylene polymers deduced that the oxyethylene distribution should follow a Poissontype distribution. These relationships are true only when all the reactions are kinetically identical. Other workers (2, 6-79 have examined various alkyl aryl, alcohol, amine, and fatty acid polyoxyethylme condensates by molecular distillation and have found that these products do give the typical Poisson distribution. In all cases the products contained an average molar ratio of greater than six ethylene oxides. Knrabinos and Quinn (6) obtained distribution patterns for 12-E-7 and 1Z-E10 alkyl ether condensates with maxima a t E-6 and E-9, respectively. These values show slightly more of the lower polymer and less of the higher polymers than that calculated from the theoretical distribution. A Poisson distribution was derived from Flory’s equation (4)for 12-E-3 condensates. This is compared in Figure 1 with the average determined values for C-E-3. These low molecular weight alkyl polyoxgethylene polymers do not follow the theoretical distribution. Thus the various reaction rates occurring during the production of these samples must not be equal for all species. Since appreciable quantities of unreacted alcohol are present, the initial reaction of alcohol and ethylene oxide is probably the slowest one in the sequence and is a first-order reaction dependent upon ethylene oxide concentration. When the initial ethylene oxide concentration exceeds a molar ratio of 6, then the reaction rates for all steps in
the sequence should be essentially the same and the distribution should approach that of Poisson. This would explain why those condensates containing an average of 1 mole and those with 3 moles of ethylene oxide are so similar in composition. If a catalyst for the initial reaction could be developed, then these low molecular weight condensates would also be expected to yield a Poisson-type distribution.
(7) Mayhew, R. E., Wyatt, R. C., J . Am. Oil Chemists’ Soc. 29,357 (1952). ( 8 ) Schonfeldt, N., Ibid., 32,77 (1955). (9) Stevenson, D. G., AnaEyst 79, 504 (1954). (10) Weeks, L. E., Lewis, J. T., Ginn, M. E., J . Am. Qil Chemists’ Soe. 35, 149 (1958). RECEIVED for review March 27, 1961. Accepted September 18, 1961. Pittrjburgh Conference on Analytical Chemistry and Applied Spectroscopy, February 27, 1961.
LITERATURE CITED
(1) Aseoc. Offic. Agr. Chemists, J. Asaoc. O.@. Agr: Chemists 42, 61 (1959). (2) Birkmeier, R. E., Brandner, J. D., J. Agr. Food Chem. 6,471 (1958). (3) Brown, E. G., Hayes, T. J., Andyst 80, 755 (1955). (4) Flory, P. J., J. Am. Chem. Soc. 62, 1561 (1940). (5) Karabinos, J. V., Quinn, E. J., J . Am. Oil Chemists’ Soe. 33, 223 (1956). (6) Kelly, J., Greenwald, H. L., J. Phys. Chem. 62, 1096 (1958).
.In
itrirnetric Method for Continu f Carbon Dioxide and Its mino Acid Chemistry
n
ABRAHAM PATCHORNIK and YECHIEL SHALlTlN Department of Biophysics, The Weizmann Institute of Science, Rehovoth, Israel )c A method is presented whereby carbon dioxide, pure or contaminated with strong volatile acids, may be determined continuously. The carbon dioxide is absorbed in a benzylamine solution and titrated directly with standard sodium methoxide. Milligram quantities may b e determined with an accuracy of 2y0. The method has proved valuable in the determination of N-carboxyamino acid anhydrides and in following the kinetics of their polymerization. It has also been used in quantitative determination of amino acids by measuring the amount of COz evolved upon their decasboxylation, and in following the removal of carbobenzyloxy groups by catalytic hydrogenation.
N RESEARCH on the kinetics of polymerization of N-carboxyamino acid anhydrides (NCA’s) (14, it was necessary to develop a method for the continuous quantitative determination of 601. The method described in this article was also useful for the analysis of N-carboxyamino acid anhydrides (1) as well as for quantitative measurements of the decarboxylation of amino acids with ninhydrin (10) or with N-bromoeuccinimide (NBS) (4)and for following the catalytic hydrogenolysis of N-carbobenzyloxyamino acids over palladium-carbon (9). The methods in use for the determination of COz have been briefly surveyed by Blom and Edelhausen (8). These authors describe a method for the titrimetric determination of COS after its absorption in pyridine. Their method has two main disadvantages: If during the reaction the absorption
solution is not kept near neutrality by continuous addition of sodium niethoxide, the carbon dioxide is not quantitatively retained by the pyridine. The carbon dioxide should be free of volatile acidic substances, the presence of which interferes with the determination. I n the procedure outlined here the carbon dioxide which is released from the sample is removed by a stream of pure nitrogen and is quantitatively absorbed by a mixture of benzylamine, ethanol, and dioxane (Figure 1). A stable benzylamine salt of benzylcarbamic acid is formed as follows (6, 16): 2CsHsCHzNHz COz --* - + CaHaCHzNWCOONHsCHzCeHs
given in Table I. The accuracy of those determinations was within 1 to 2%. Any volatile acidic compound will. interfere with the determination of carbon dioxide, as it will. also be absorbed by benzylamine and be titrated with sodium methoxide. However, carbon dioxide dissolved in pyridine can be quantitatively removed by a stream of nitrogen, while strong acidic compounds are quantitatively retained. This property makee it possible t o utilize an alcoholic pyridine solution for the removal of acidic contaminants accompanying carbon dioxide.
This salt reacts with one equivalent of sodium methoxide :
Reagents. Dioxane, reagent grade, was purified according to Vogel (16). Pyridine, benzylamine, and apliline, reagent grades, were distilled a t atmospheric pressure. Potassium carbonate, analytical grade, was used as a standard. N-Carboxy-a-ammo acid anhydrides were purified by recrystallization from ethyl acetate and petroleum ether. In each case the equivalent weights were
+
EXPERIMENTAL
C ~ H ~ C H ~ N H C O O ~ H ~ C+H ~ C ~ H ~
+
NaOCHs -+ C6H&HzNHC06ha C&CH1NHz CHsOH
+
The absorbed carbon dioxide can thus be titrated continuously as the reaction proceeds, or at the end of the reaction, with sodium methoxide to the blue end point of the indicator thymol blue. The benzylamine mixture absorbs the COZ very efficiently, because of the formation of the stable salt. No carbon dioxide is lost during the absorption, and there is no need to add sodium methoxide as in Blom’s method. The accuracy of the method described was tested by titrating the carbon dioxide released on acidifying aliquots of a standard solution of potassium carbonate. This determination was carried out in the apparatus shown in Figure 1. The results obtained are
Table 1.
Micro- and S e ~ ~ ~ j ~ ~ a ~
Present,
Found,
Mg. 2.51 3.65 8.45 9.60 10.20 11 .o
Mg. 2.53 3.67 8.42 9.72 10.35 10.9 19.9 43.8
20.3
44.15 53.0 76.9
123.0
Recovery, %
62.4 76.2 121.7
VOL. 33, NO. 13, DECEMBER 1961
100.8 100.6 99.5 101.3 101.6 99.0 98,l 99.2 98.8 99.0 98.9
* 1887
