V O L U M E 2 2 , NO. 1, J A N U A R Y 1 9 5 0

corresponding to the sum of the two masses is then SS - SI. The AS us. m curve is then plotted on rectangular coordinates. This method of plotting has the advantage that the origin gives an additional point through which the curve must pass and it becomes easier to detect flaws in the data.

is found and by reference to the standardized rurve, such as Figure 5 , the molecular weight is obtained.

A single determination of molecular weight can be made in a few minutes under ideal conditions. Where readily soluble com- pounds are being measured, and no cleaning difficulties are en- countered, the average time required for a complete determination is about 25 minutes.

From the curve thus constructed the value of (2) rn - 1w

DISCUSSION

The change in the position of the upper nieiiiscub 15 the only one followed in these measurements. This is done for purposes of simplification and is permissible, because the drop in the lower meniscus is a constant function fJf the rise in the upper meniscus. The method of plotting automatically takes care of the fact that A S is not Z A S . hctually, A 8 equals approximately 9iYc of Z A S in practically all measurements made on this apparatus.

The usual precaution of maintaining an adequate difference be- tween boiling point of solvent and solute must be observed, and in this laboratory no attempt is made to measure molecular weight by the ebullioscopic method of samples boiling below 210" C.- i.e., 150" C. above the boiling point of chloroform.

With careful attention to details, the accuracy attainable xi th this equipment is only slightly less than that obtained on the cryoscopic apparatus using benzene as a solvent and a Beckman t,hermometer for temperature measurement. A romparison of

175

results obtained by the two methods on duplicate samples ap- pears in Table I.

Certain types of compounds yield anomalous results, but this is common to the other widely used methods for determining molecular weight. I n particular, the molecular weights of stearic acid and azoxybenzene are far too low when measured in the ap- paratus described here.

A word of caution is in order relative to the grease used in the ground joint of the thermometer. The silicone greases cause ex- cessive foaming in the boiling solution and render the apparatue inoperable. A satisfactory grease is Celvacene Heavy, a product of Distillation Products, Inc., Rochester, K. Y .

Experience over the past 4 years has shown that this equipment meets all the requirements of speed, quality, and ruggedness in a laboratory where a large volume of molecular weight work is constantly being turned out.

ACKNOWLEDGMENT

It is a pleasure to acknowledge the careful workmanship and ideas contributed by W. R. Doty in the construction of the glass equipment.

LITERATURE CITED

(1) Kitson and Mitchell, - k i . k L . CHEM., 21, 401 (1949). (2) Kitson, Oemler, and Mitchell, Z b i d . , 21, 404 (1949). (3) Mensies, J . Am. Chem. Soc., 43, 2309 (1921). (4) Meneies and Wright, Ibid. , 43, 2314 (,1921). (5 ) Reilly, J., and Rae, W. N., "Physico-Chemical Methods," 3rd

ed., Vol. 1, p. 600, London, Methuen & Co., 1939.

RECEIVED September 11, 1947. Presented before the Division of lnalytical and Micro Chemistry at the 112th >feetine of t,he AMERICAX CHEMICAL SOCIETY, X e w York. PIT. Y.

Long-chain Alkyl Sulfates Colorimetric Determination of Dilute Solutions

FRED KARUSH' A R D MARTIN SONENBERG' iVew York TJniversity College of Medicine, New York, N . Y

A simple colorimetric method has been developed for the determination of long-chain alkyl sulfates. Solutions as dilute as 5 X M can be analyzed with an accuracy of about 2YG. The method depends on the formation of a complex between the deter- gent anion and the cationic dye rosaniline hydro- chloride. This complex is extracted into a mixed or- ganic solvent, 50% chloroform and 30% ethj-l ace-

N COSNECTIOS with a study of t,he interaction between I bovine serum albumin and sodium octyl, decyl, and dodecyl sulfates ( C,,H1,+ lSOaSa), the need arose for an accurate analyti- cal method for the determination of these alkyl sulfates in very dilute solutions, of the order of 10-j 111. h search of the litera- ture revealed that no adequate method had been published.

The development of a colorimetric method w-as therefore undertaken, following the previous work of Brodie, Udenfriend, and Levj- (1). As the basis of methods for the analyses of certain strong organic acids, these investigators utilized the formation of organic soluble complexes between the acids and the cationic dye rosaniline hydrochloride. The method described here is based on the formation by this dyestuff and pararosaniline hyclro-

York. x. T. I Present addrws. Sloan-Kettering Insti tute for Cancer Research. S e w

tate, and its spectral absorption is read with a Klett- Summerson colorimeter. Pararosaniline hydro- chloride is also suitable as a dye reagent. The molar sensitivity of the method increases from octyl to decyl to dodecyl sulfate, because of increasing ef- ficiency of extraction of the dye-detergent complex with increase in chain length. The molar ratio of d?-e to detergent in the extracted complex is 1.

chloride of complexes with alkyl sulfates and their extraction into an organic phase. It has been applied to the determination of the alkyl sulfates noted above and is sufficiently sensitive to permit the analysis of solutions as dilute as 5 X ill (involv- ing 2 X Its ujefulness is not limited to alkyl sulfates and it can undoubtedly be applied to other strong organic acids which contain relatively large non- po1:tr groups.

mole) 1% ith an accuracy of about 27,.

MATERIALS

The rosaniline hydrochloride was obtained from the National .hiline Division, Allied Chemical and Dye Corporation, and was of 82YG strength according to the manufacturer, the impurity being largely salt. A solution of the dye a t a concentration of 4 x 10-4 31 was prepared in 0.025 M phosphate buffer, pH 6.1 This solution was extracted several times with the same solvent

176

usrd in the analysis until the intensity of color in the organic phase remained unchanged on consecutive washings. The dye wai then stored in a Pyrex bottle in the dark. The pararosaniline hydrochloride used in this study nas a specially purified sample supplied by this manufacturer, and specified to be of 96.5yo .strength and to contain 2.5% moisture.

After some trials with a number of organic solvents, a mised \olvent consisting of 50% (by volume) chloroform (c.P.) and .io% ethyl acetate (c.P.) was selected for the extraction of the colored complex. The treatment of this solvent with aqueous alkali to remove organic acid was found to be an unnecessary precaution.

The spectral absorption of the complex was measured m ith the Iilett-Summerson photoelectric colorimeter using a green (540) filter. However, it was found that the soft-glass absorption tubes ordinarily usrd with this instrument were undesirable br- wuse of the relatively high blanks they gave. Instead, spwiall] made Pyrex tubes of outside diameter 14.0 * 0.1 mm. \+ere ob- tained and were found to give lower blanks than the Klett tubec I\ hen used in accordance with the procedure described below.

To minimize errors introduced by differences in the optica I properties and dimensions of the Pyrex absorption tubes, zero and thickness corrections were established for each tube. Only tube5 n-hose thickness correction was not more than 1% (of the Klett leading) were selected for use, thereby rendering this corrrctiori negligible.

For purposes of efficiency, aimplicity, and the avoidance or (.ontamination, it nas desirable to carry out the extraction, cen- trifugation, and absorption measurement in the same tubeb This was possible with Pyrex tubes if they were treated before uze nith a boiling solution of approximately 1 S hydrochlorir .ic-id for about 1 hour. .4fter this acid treatment the tubes \\ere rinsed well nith distilled water and dried in an oven for 10 to 15 minutes. Glass-stoppered Pyrex tubes were employed initiallT , but inverted serum caps ( S o . 2) were found to be as suitahlr .is the glass stoppers. These caps were cleaned after use b! waking in 0.1 11' hydrorhloric acid solution, folloaed bv thorough rinsing.

PROCEDURE

In a t ypiral analysis eight or nine samples IT ere simultaneousl) rB\amined in triplicate with threr blank controls. To each tube itas added 1 ml. of the rosaniline hydrochloride solution (4 x

This was followed by the addition of the appropriate volume of alkyl sulfate solution I I I the same buffer, not exceeding 4 ml. The final volume of the aqueous phase was adjusted to 5 ml. and contained 0.025 .I/ phosphate, pH 6.1. For the extraction 5 ml. of the mixed sol- vent were added and the tubes were stoppered and shaken liv hand about 50 times.

Centrifugation for a fen minutes resulted in the complete sepa- rlition of the t M o phases with the organic phase, containing the volored complex, a t the bottom. Its spectral absorption was then measured with the colorimeter against the reference tube filled Kith solvent. After the necessary corrections, including that tor the blank, the amount of alkyl sulfate was read from the *+ppropriate calibration curve.

Considerable variations in the conditions of the procetiur e C M I ~

be tolerated. The use of pararosaniline hydrochloride does not lead to an increase of the blank nor does the use of 0.05 Jf phos- phate buffer, pH 7.0. Furthermore, Mith dodecyl sulfate at least, the same sensitivity is observed with both dyes and under either set of buffer conditions.

Typical calibration curves for octyl, decyl, and dodecyl sulfate. based on the procedure described above are shown in Figure 1. The upper scale refers to octyl sulfate and the lower one to the other sulfates. These curves mere obtained with specially puri- fied samples of alkyl sulfates supplied by the Fine Chemicals Division of E. I. du Pont de Semours & Company. The cali- bration data are plotted in terms of the amount of alkyl sulfate used in the analysis.

M in 0.025 M phosphate, pH 6.1).

DISCUSSION

Though the authors have not investigated in detail the nature of the dye-anion complex, it is fairly evident that its formation and extraction depend on the electrostatic interaction between opposite charges and the presence of a large nonpolar group in the organic anion. The significance of the latter is apparent from a comparison of the calibration curves in Figure 1. With

A N A L Y T I C A L C H E M I S T R Y

MOLES OF ALKYL SULFATE x 10-8

Figure 1. Calibration Curves

increase in chain length the molar sensitivity of the method in- creases and the curves deviate less from linearity. This is, of coursr, precisely the behavior onr would anticipatr if the frac- tion of organic anion extracted into thr organic phase as a colored complex increased with chain length. This conclusion has been verified experimentally by making a second extraction of the aqueous phase. The usual procedure was carried out with 3 X

mole of dodecyl sulfate and 10 X lo-* mole of octyl sulfate to give net Klett readings of 309 and 326, respectively. A second extraction with 5 nil. of solvent was performed on 4-ml. aliquots of the aqueous phases including blank controls. The net Klett values obtained, 1 for thr dodecyl compound and 175 for the octyl compound, clearly demonstrated the dependence of the efficiency of extraction on the chain length of the detergent.

Inasmuch as only electrostatically neutral complexes would be expected to be extractable into the organic phase, it was inferred that the molecular ratio of dye to detergent in the complex is 1 to 1. This conclusion was subjected to experimental test by extract- ing 5 ml. of a buffered solution containing 4 X mole of para- rosaniline hydrochloride and 100 X 10-8 mole of dodecyl sul- fate nith 5 ml. of solvent. Because of the large excess of deter- gent, the dye was completely transferred to the organic phase and gave a net Klett reading of 326. On the other hand, with 4 X 10-8 mole of detergent a net reading of 30; was found by the usual procedure (involving dye excess). The ratio of the readings is 0.94, a value that is sufficiently close to 1 to establish the 1 to 1 stoichiometry of the dye-detergent interaction.

Whereas the drop in molar sensitivity is slight between do- decyl and decll sulfates, the sensitivity for octyl sulfate is less than one half that for the other homologs. The concentrations of the alkyl sulfates used for analysis were far below the critical micelle concentrations (2 , 3 ) .

One of the most persistent difficulties encountered in this study was the high and variable blanks. This was not due to the solubility of the dye in the organic solvent, but rather to con- tamination and absorption of the dye on the glass surface. By the use of Pyrex tubes and the procedure described above it was found possible to obtain satisfactorily low and consistent blanks. Over a period of vrek.i the average Klett readings for the blanks

V O L U M E 2 2 , N O . 1, J A N U A R Y 1 9 5 0 177

varied from 20 to 30, though in an) particular series the individual readings generally agreed to within 2 units.

The dependence of the amount of colored complex extracted on the aqueous dye concentration was considerably different among the several homologs, but was consistent with the other differences noted above. A twofold increase in the dye concen- tration gave a 45c7, increase in the net Klett reading for octyl sulfate, about 8 7 , for decyl sulfate, and practically none for tiodecvl sulfate. Associated with this higher dye concentration f as a slight increase in the blank reading. Increase of phosphate concentration from 0.025 M to 0.05 &' yielded a 7% increase in the reading for octyl sulfate. S o change in blank reading was observed. However, the presence of higher salt concentrations- e.g., 0.1 .lf sodium chloride-reaulted in large increases in the blank.

The use of a mixed solvent instead of pure chloroform was dictated by the observation that the presence of ethyl acetate increased the effic-irncy of rutraction of the colored complex.

However, because its inclusion did give rise to higher blanks. a compromise figure of 50y0 was selected.

A few attempts were made to apply this method to long-chain fatty acids, but without success. I ts failure in this case was UII- doubtedly associated with the weakly acidic character of' the carboxyl group.

ACKNOWLEDGMEhT

The authors are grateful to R. Keith Canrian for the generous hospitality afforded them in his laboratory during this investiga- tion.

( l j Brodie, B. B., Udenfriend, S., and Levy. B., personal COIILIIIIIIII -

(3) Corrin, Sf. L., and Harkins, W. D., J . dm. ('hem. Soc., 69, 68:3

(3) Hoffner, F. D., Piccione, G. A. and Roaenbluni , ('., J . f ' hys . C'hem.,

RECEIVED August 8, 1949. Investigation wnducted during tenure of a fellowship in cancer research of the dmerican Cancer Society, recommended hy the Carrimittee on Growth, Xational Rewarch Council.

LITERATURE CITED

cation.

(1947) .

46, 662 (1942) .

Determination of Water in Glycols and Glycerol CHARLES B. JORDAS AND VIRGIL 0. HATCH

Paint and Chemical Laboratory, Aberdeen Proving Ground, .Wd.

A method has been dekeloped for determining the percentage of water, by volume, in glycol-water or glycerol-water solutions. The procedure is a reflux distillation and salting-out process. 1-Butanol was selected as a refluxing medium. This solvent forms an azeotrope with water, but not with glycol or glycerol, and is capable of breaking the hydrates which are formed in aqueous glycol solutions. Forty samples of known compositinn, varying in water

N PHASE rule studies of aqueous glycol or glycerol solutions, I it is essential that the exact percentage of water be known. Boiling points, freezing points, and corrosive action on metals depend upon the amount of water in the solution. An industrial application of this type of problem exists in the investigation of permanent-type antifreeze solutions.

Several methods of determining water in difl'erent substances have been perfected, but no method could be found which would accurately and rapidly determine large percentages of water in glycol or glycerol samples. The accuracy of the Karl Fischer

content from 1 to 95%, were analyzed. The method was found to be reliable and accurate. It is also applicable to mixed glycol-water solutions and mix- tures of glycols with glycol-ethers and/or 1-butanol and alcohols containing more than four carbons. The method is especially applicable to permanent- type antifreeze solutions. It works equally well for new or used antifreeze solutions, with no in- terference from inhibitors or inorganic impurities.

method of determining watei ( 3 ) drcreuhes rapidly as the water content exceeds 2%. The presence of dyes and certain inorganic salts affects the Karl Fischer method 13 giving false end points during titration. Therefore the method is not accurate for all mtifreeze mixtures. Stevens and Nickels (6) determined small amounts of water in glycols and glycerol by means of a cloud point; hon-ever, this procedure is ricit applicable TA here large percentages of water are present.

A method of determining witer 111 pt,troleum products ( 1 ,. using the Dean and Stark appdratns ( 2 ) with toluene as the re-

Table I. Determination of Composition of Glycol, GI) cerol, and Antifreeze Solutions Water Known Butanol Known Nonvolatile Constitueny'

F . 241. MI. 11.11. MZ. M1. M 1. ,111. M1. .w. Run Base Temp. Received Water Error Received Butanol k h o r Received Known 6rrol

4 5 6 7 8 9

10 1 1 12 13 14 15 16 17 18

Ethylene glycol Ethylene glycol Ethylene glycol Ethylene glycol Ethylene glycol Antifreeze A Antifreeze A Antifreeze B Propylene glycol Propylene glycol Diethylene glycol Diethylene glycol Antifreeze C Antifreeze C Glycol mixture Glycerol Glycerol Glycerol

385 385 385 385 385 385 385 385 370 370 450 470 385 385 370 450 450 450

0 . 5 3 0 . 0

7 0 . 5 9 5 . 0

5 . 0 5 2 . 5 2 0 . 0

1 . 0 5 0 . 5 2 . 0

5 1 . 0

50.;

4 . 0 5 1 . 5 1 0 . 5 3 . 0

4 1 . 5 7 0 . 5

0 . 6 3 0 . 3

9 4 . 0

5 0 . 3 7 0 . 2

5 2 . 8 1 9 . 5

5 . 5

1 . 0 5 0 . 5

2 . 0 5 1 . 0

4 . 0 5 2 . 0 1 1 . 0 3 . 0

4 1 . 8 7 0 . 9

0 0 . 3 0 . 3 0 . 3 0 0 . 5 0 . 3 0 . 5 0 0 0 0 0 0 . 5 0 . 5 0 0 . 3 0 . 4

2 0 . 0 2 0 . 0 20 .5 1 9 . 5 20 .0 2 0 . 5 2 0 . 5 5 4 . 5 2 0 . 5 20 .0 1 9 . 5 20.0 2 0 . 0 2 0 . 0 2 0 . 0 2 0 . 0 1 9 . 5 20 0

20 .0 20 .0 2 0 . 0 20 .0 2 0 . 0 20 .0 2 0 . 0 5 4 . 0 2 0 . 0 2 0 . 0 2 0 . 0 2 0 . 0 2 0 . 0 2 0 . 0 2 0 . 0 2 0 . 0 20 .0 2 0 . 0

0 0 0 . 5 0 . 5 0 0 . 5 0 . 5 0 . 5 0 . 5 0

0 0 0 0 0 . 5 0

9 9 . 0 6 9 . 5 4 9 . 5 2 9 . 5 5 . 0

9 4 . 5 47.0 4 6 . 0 9 8 . 5 4 9 . 5 9 8 . 5 4 9 . 0 9 6 . 0 4 8 . 5 8 9 . 5 9 7 . 0 5 8 . 5 2 9 . 0

9 9 . 5 6 9 . 7 4 9 , 7 2 9 . 8

5 . 0 9 4 . 5 4 7 . 2 4 6 . 5 9 9 . 0 4 9 . 5 9 8 . 0 9 9 . 0 9 6 . 0 4 8 . 0 8 9 . 0 9 7 . 0 5 8 . 2 2 9 . 1

0 . 5 0 . 2 0 . 2 0.3 0 0 0.2 0 . 6 O b 0 0 A 0 0 0 . 5 0 r? 0 0.3 0 1

Figure obtained by measuring volume of liquid remaining in distilling flask after distillation. .- ~ .__ __