DEAmide of these samples as determined by the two methods represented. Needless to say, we believe the data based on the nonionic analysis are correct, and that the titration of the unsaponified sample to pH 4.8 does not adequately correct for all of the nonamide nitrogen in these commercial low-active amides. Because an accurate analysis of mixtures of low-active mono- and diethanolamide requires the isolation of the nonionic residue, it is feasible to run the differential saponification rate method on a portion of the nonionic residue. To evaluate this possible modification, four sample mixtures of these amides were prepared, and the analysis was carried out on the samples directly and on the nonionic residues of the samples. The results are compared in Table 111. Again, the direct analysis method appears to be accurate, although the reproducibility is not as good as we thought it should be. Analysis of the nonionic residue appears to introduce a small negative basis for diethanolamide, and we know of no reason why this should be so, for this procedure more closely approximates the calibration system than the commercial amide mixtures themselves. In spite of this small bias, however, the data of Tables I1 and I11 demonstrate the applicability of the differential kinetic technique to the analysis of fatty alkanolamide
mixtures by saponification when an appropriate method, such as the one which has been described here, for carrying out the saponification and following its course is employed. Of the various modifications of the basic differential saponification rate method which were evaluated, that employing a prior separation of the nonionic fraction of the sample followed by saponification and Kjeldahl N analysis of the nonionic residue would be the most generally applicable. Its use on finished products containing mixture of these amides seems feasible. The practice of obtaining the total millimoles of amide by a Kjeldahl N analysis on the sample itself and correcting for amine and amine soap would have utility for only very select systems such as the high active amides. The presence of bases other than amines or amine soaps (NH4C1 or sodium acetate for example) would interfere. The modification employing direct saponification of the sample and a determination of nonionic content of the sample and N content of the nonionic residue to give total millimoles of amide should be rather generally applicable because the method of measuring the extent of saponification is itself rather specific. RECEIVED for review April 16, 1968. Accepted November 14, 1968.
Determination of Nonionic Surfactants by Atomic Absorption Spectrophotometry J. C. Sheridan, E. P. K. Lau, and B. Z. Senkowski Analytical Research Laboratories, Hoffmann-La Roche, Inc., Nutley, N . J. 07110 A quantitative method for the rapid determination of nonionic surfactants is described. The surfactant is precipitated as a heteropoly phosphomolybdic acidbarium complex and the molybdenum in the supernatant solution is assayed by atomic absorption. Optimum conditions for the analysis of high levels of molybdenum in the presence of high levels of barium were determined. The effect of other cations on the solution was determined.
THECONCENTRATION of nonionic surfactant in multicomponent solutions has previously been determined by a method used for biological materials ( I , 2). The surfactants were precipitated from solution with phosphomolybdic acid and barium chloride and the precipitate weight was determined. The amount of surfactant in solution was related to the precipitate weight by an empirical factor determined for each surfactant batch and each multicomponent system, since the precipitate is nonstoichiometric. We have achieved a considerable saving in time by isolating the precipitate with centrifugation and measuring the residual molybdenum concentration of the supernatant solution by atomic absorption spectrophotometry, and relating this to the quantity of surfactant by an empirical factor determined for each batch and each multicomponent system. The conditions for measurement of molybdenum in the residual reaction mixture were determined.
EXPERIMENTAL
Instrumentation. A Perkin-Elmer Model 303 atomic absorption spectrophotometer equipped with a recorder readout and a Westinghouse WL 22937 molybdenum hollow cathode lamp was used to obtain the experimental data. A premix burner with the short path head attached (Perkin-Elmer, part number 290-1169), rotated 90" from normal to minimize the absorption path length, was used. This configuration was necessary because of the high concentration of molybdenum in the solutions. The fuel-oxidizer mixture used in the burner was highly reducing air-acetylene that gave a white flame on top of small blue base. When the gas regulator gages supplied with the instrument were used to meter the air and acetylene to the premix burner, a very unstable flame resulted. Introducing the solution into the aspirator with a constant injection pump (Harvard Apparatus Co., Model 901) improved the reproducibility of the absorption values found on replicate runs of the same solution. The pump did not improve the reproducibility of absorbance values sufficiently and was cumbersome in routine use, SO it was abandoned. The sample injection was controlled with the aspiration air flow rate in the conventional manner and varied from 2 to 4 ml of solution per minute. Sufficient reproducibility was obtained with a gas regulation system modified in the following way (3). Acetylene from a tank (Union Carbide, Linde Division, purified No. WC) passes through a single stage regulator existing at 15 psi (Matheson Gas Co., Model lL), into a precision rotameter (Brooks Instruments Co., Model 8900
(1) C. Boyd Shaffer and Francis H. Critchfield, ANAL.CHEM.,19, 32 (1947). (2) J. Oliver and C.Preston, Nature, 4162,242(1949).
(3) J. Johnson, Westinghouse Electric Corporation, Atomic Power Division, Madison, Pa., personal communication, 1967. VOL. 41, NO. 2, FEBRUARY 1969
247
flow controller and Sho-Rate 50 rotameter), and into the fuel inlet of the premix burner. Air from the laboratory bench supply, at 50 psi, is cleaned by a coarse filter (Commercial Filter Corp., Fulflo BAL x 6-4 14D) to remove moisture and large particles, a fine filter (Matheson Gas Co., purifier Model 450) to remove oil droplets and small particles, and a filter dripwell (as supplied by Perkin-Elmer). The air flow is shunted into two systems, each equipped with a precision regulator exiting at 28 psi (Moore Product Co., nullmatic Model 40A50) into a precision rotameter (Brooks Instrument Co., Model 8900 flow controller and Sho-Rate 50 rotameter). One air supply line leads to the auxiliary air inlet of the premix burner and the other to the aspirator. During use of the premix burner, the flow rates of both acetylene and air to the burner are adjusted with the rotameters. Prior to using the burner, the desired solution flow rate is obtained by appropriate adjustment of the rotameter in the aspirator air line and the atomizer capillary. Molybdenum Absorption Curves and Barium Suppression. The optimum operating conditions for the measurement of the molybdenum concentration in the residual mixture using the atomic absorption spectrophotometer were determined. The mixture contained high concentrations of both molybdenum and barium, so the previously reported conditions by David ( 4 ) and Mostyn and Cunningham (5)did not necessarily apply. Our starting reaction mixture contained 1.25% hydrochloric acid, 1.25% barium chloride (or 2.406 mmol Ba) and 1.25% phosphomolybdic acid (or 2.538 mmol Mo as acid phosphomolybdic Merck 20 M o o 3 2H3P04,48H20) in a total volume of 40 ml of water. After the reaction occurred the precipitate was a barium-molybdenum-surfactant complex ( I ) in which the inorganic fraction was approximately 1 mol of barium to 5 mols of molybdenum. Barium was removed from the reaction mixture less rapidly than was molybdenum with increasing additions of surfactant. Consequently, the concentration range of the molybdenum and barium in the residual reaction mixture would be from 2.406 mmol to approximately 1.9 mmol for Ba and from 2.538 mmol to almost 0.0 mmol for Mo. The criteria for selection of optimum operating conditions included linear response and use of a maximum portion of the instrument transmittance display meter over the expected molybdenum concentration range. Sensitivity was not a considered factor. Test solutions containing varying levels of molybdenum and barium were aspirated into the flame and the molybdenum absorption was measured. The absorption of these solutions was measured at two different flame positions and at the two molybdenum resonance lines 3133 A and 3158 A. The two flame positions used were with the source beam passing through the flame at 1.0-1.4 cm above the burner head slot (optimum) and 2.2-2.5 cm above the burner head slot (high). Each test solution contained HCI; 1.25% molybdenum as phosphomolybdic acid: 0.125%; 0.25%; 0.375%; 0.50%; 0.625%; 0.75%; 0.875%; 1.0% barium as barium chloride: 0% ; 0.25% ; 0.75% ; 1.00%. All possible combinations of the above solutions were prepared and analyzed. An aspiration rate that gave approximately 24.0% transmittance for a solution containing 0.25% Mo, 0.75% Ba, and 1.25% HCl was used for each flame position. The absorbance values obtained for each solution were divided into four categories for evaluation: optimum flame position, high flame position, 3133 A resonance line and 3158 A resonance line. The absorbance values ranged fr0m~0.745 high molybdenum, no barium, optimum flame, 3133 A line to less thanoO.O1 for low molybdenum, high barium, high flame 3158 A line. Inspection of the data showed that the (4) D. J. David, Analyst, 86,730 (1961). (5) R. A. Mostyn and A. F. Cunningham, ANAL. CHEM., 38, 121 (1966). 248
ANALYTICAL CHEMISTRY
3133 A line was superior to the 3158 A line, because the absorbance range was much greater for the expected molybdenum concentration Fange. Only the experimental values obtained for the 3133 A line were further evaluated. At this line, regression curves of the absorbance of each solution on the molybdenum concentration of the solution, at each added barium level, were calculated separately for optimum flame and high flame positions. In the latter case, absorbance values were transformed to the natural log of the values before regression curves were calculated. A linear relationship between absorbance and molybdenum concentration was found for all values obtained in the optimum flame position for all levels of added barium. The correlation coefficient was 0.99 and a chi-squared test between experimental and calculated values of absorbance showed no significant difference. As the barium concentration increased, the slope of the regression curves decreased, which indicated that barium suppresses molybdenum absorbance. But in the expected concentration range of barium in the residual reaction mixture, no significant change in slope occurred. Hence in the measurement of Mo in the residual reaction mixture, no adjustment for variable Ba concentration was necessary when the standards contained a Ba level within this expected concentration range. ' A nonlinear exponential relationship between absorbance and molybdenum concentration was found for all values obtained in the high flame position for all levels of added barium. The correlation coefficient was from 0.83 to 0.88 and chi squared test between experimental and calculated values of absorbance showed them to be not significantly different. As the barium concentration increases, the slope of the exponential curve decreases indicating barium suppresses molybdenum absorbance. Because of the shape of the response curve, the optimum position in the flame was used for further work. When the burner was rotated through 90°, the optical system of the Perkin-Elmer Model 303 was such that the source beam had a cross-section diameter of 0.3 cm as it passed through the flame. The optimum position gives the maximum absorption value for a given aspiration rate and may correspond with the hottest site in the flame as described by Rann and Hambly (6). The high position was considerably less sensitive, required a higher aspiration rate to achieve 24% transmission, and may correspond to a zone where the concentration of ground state molybdenum atoms is considerably less than in the optimum position but where the concentration of ground state barium atoms is not much less than in the optimum position. The shape of the molybdenum response curve depends on the site in the flame through which absorption measurements are being made and not on the concentration of barium. Thus where the concentration of ground state molybdenum atoms is highest, the response is linear over a wide absorption range. At this flame site, addition of barium results in fewer ground state atoms, but does not alter the linearity of response. Where the concentration of ground state molybdenum atoms is considerably less, the response curve becomes nonlinear, independent of the presence of barium. The maximum level of suppression of molybdenum by barium is obtained with a lower concentration of barium for the high flame position than for the optimum flame position. These results indicate that absorption should be measured in only the hottest part of the flame to obtain linear analytical curves for those elements that give relatively small hot sites. When the concentration range at this site is unsuitable, altering the flame path length or using a different resonance line is to be preferred to using an area out of the hot site. Effects of Sample Composition on Molybdenum Absorption in the Flame. In most instances, the multicomponent solution in which the nonionic surfactant was being mea(6) C. S. Rann and A. N. Hambly, ANAL. CHEM., 37, 897 (1965).
sured (at 0.5 to 10% concentration) contained organic and inorganic materials which varied from trace amounts to more than 10%. These latter materials could affect the surfactant analysis by influencing the precipitation step or molybdenum absorption in the flame. The mechanism of precipitation was not considered in this study. Extraneous organic material in the sample is unlikely to affect the flame, in view of the small amount of sample added to the reaction mixture. However inorganic ions present in the sample might affect the flame (4, despite the high barium content. Calcium, magnesium, sodium, aluminum, and phosphate were added to a high barium test solution and the flame absorption of the molybdenum was measured. It was found that sodium and phosphate do not alter the absorbance of molybdenum, calcium and magnesium suppress it, and aluminum enhances it. The experimental results are shown in Table I. Although the amount of barium present was sufficient to attain maximum barium suppression of molybdenum absorbance, this barium level did not inhibit suppression or enhancement by other inorganic ions. When surfactant samples contain inorganic ions that will influence molybdenum flame absorption, these ions must be added to the standards. RESULTS AND DISCUSSION Analysis of Surfactant Solutions. Aqueous solutions containing approximately 25 mg of surfactant per milliliter of water were prepared from six batches of Emulphor EL 620 (five commercial batches and one prepared in our laboratories), five batches of Tween 80, and one batch each of Carbowax 400, Carbowax 600, and Carbowax 6000. These solutions were used for the following investigations : Solutions of one lot of Emulphor EL 620 and one lot of Tween 80 at concentrations of 0 to 225 mg in 25-mg increments were analyzed in quadruplicate. Aliquots of the sample solutions were added to centrifuge tubes together with sufficient water to bring the volume to 25 ml. After the addition of 5 ml of 10% HC1 and 5 ml of 10% BaC12 the mixture was shaken for one minute. Five milliliters of 10% phosphomolybdic acid were added to the tube and the mixture was shaken again. The tube and contents were heated for one hour at 85 "C, cooled, and centrifuged. The molybdenum absorbance of the clear supernatant was then determined in the atomic absorption spectrophotometer using the instrument operating conditions previously described. The relationship between the amount of Mo precipitated and the amount of surfactant added was determined. The relationship was linear from zero to 125 mg of added Tween 80 and zero to 175 mg of added Ernulphor EL 620. The regression coefficientsare shown in Table 11. When more than 125 mg of Tween 80 or 175 mg of Ernulphor EL 620 was used, the relationship became nonlinear, in that further small additions of surfactant caused proportionately larger Mo precipitation. Inversely this means that as more surfactant was added, proportionately less Mo remained in the residual reaction mixture. Since the standard molybdenum-barium concentration cs. molybdenum absorbance curves did not show nonlinearity, the nonlinearity is possibly due to the nonstoichiornetric nature of the complex formation ( I ) . A suitable volume of water and amount of reactants must be used to avoid such high surfactant to reactant ratios. Three concentrations of each batch of surfactant were added to reaction mixtures, as described in the previous paragraph, and the molybdenum absorbance in the flame was measured. The concentrations, chosen to be within the linear range were 75, 125, and 175 mg of Emulphor EL 620 and 25, 75, and 125 mg of Tween 80. The regression coefficients for each batch were calculated and are shown in Table 111.
Table I. Inorganic Interference with Molybdenum Absorbance Solution Concentration
Added inorganic salt concentration 0.25 %
Molybdenum absorbance 0.256 0.244 0,248 0,256 0.260 0.328
0
CaClz MgCL NaCl NazHPOa AlCls (0.75 %)
Table 11. Regression Coefficients for Absorbance VS. Surfactant Concentration Assume y = ax
+xb =where y = mmols molybdenum precipitated: surfactant concentration Correlation coefficient R2 0.99 0.99
Regression coefficients Surfactant Tween 80 Emulphor EL 620
a
b
+0.0165 +0.0124
-0.0171 -0.029
Table 111. Regression Coefficients of Batches of Surfactants Assume y
= ax $ b where y = flame absorbance; x = wt of surfactant in mg
x 108
b
x
$0.59 +0.60 $0.61
6a
-2.93 -2.98 -3.13 -2.95 -3.01 -2.96
+0.59 $ 0 . 59
0.70 0.79 0.63 0.44 0.55 0.63
Tween 80 Batch #I 2 3 4 5
-3.38 -3.40 -3.28 -3.17 -3.30
+O. 53 0.52 0.51 0.50 0.52
0.26 0.67 0.96 0.97 0.52
Sample
a
Emulphor EL 620 Batch #I 2 3 4 5
$0.58
5
Batch prepared in laboratories of Hoffmann-La Roche.
b
Z a is the standard error of the estimated slope.
104b
Table IV. Precision of a Single Analysis 10 replicates of 125 mg of surfactant
Batch
Mean Std dev Relstddev
Tween 80 1 2 127.0 mg 124.6 mg & l . 15 10.96 0.93% 0.76%
Emulphor EL 620 1 2 127.6 mg 127.5 mg 12.27 f1.19 1.78% 0.93%
Table V. Comparison of Surfactant Types Absorbance from
Nonionic surfactant
125-mg sample
Emulphor EL 620 Tween 80 Carbowax 400 Carbowax 600 Carbowax 6000
0.218 0.125 0.026 0.029 0.145
VOL. 41, NO. 2, FEBRUARY 1969
249
The precision of a single analysis was determined for each of two lots of Tween 80 and Emulphor EL 620. Ten replicates of 125 mg of surfactant were added to the reaction mixture, as described in the previous paragraph, and the molybdenum absorbance in the flame was measured. The mean standard deviation and relative standard deviation of the surfactant found was calculated (7). The amount of surfactant found was calculated from the regression coefficients for the batch (Table 111). The precision of a single analysis was calculated from experimental values obtained under virtually uniform conditions in two consecutive working days (Table IV). The precipitation procedure was carried out on equal weights (125 mg) of each of the different surfactants and the molybdenum absorbance values of each of the residual reaction mixtures measured. The values, shown in Table V, show that the quantity of molybdenum reacted is quite different for each surfactant. Optimum ratios of reaction mixture to surfactant must be determined for each surfactant type, to ensure a linear relationship between molybdenum absorbance and weight of surfactant used. These results show that the concentration of nonionic surfactant in solution can be measured by determining the molybdenum concentration in the residual reaction mixture (7) C. Mack. “Essentials of Statistics for Scientists and Technologists,” Plenum Press, New York, 1967.
by atomic absorption spectrophotometry. However, as in the classical method of measurement ( I , 2), the nonstoichipmetric nature of the reaction between the surfactant, barium chloride, and phosphomolybdic acid means that the empirical relationship between the amount of phosphomolybdic acid reacted and the amount of surfactant present can only be determined when the type of surfactant and the composition of the solution containing it is known or a placebo is available. Table I11 gives some indication of variation in empirical factor to be expected between batches of the same surfactant, Table V the variation between types of surfactants, and Table I the variation due to changes in the inorganic composition of a multicomponent solution which contains the surfactant. Valid quantitation can be achieved only by a comparison of samples and reference standards prepared from the same batch of surfactant and analyzed in an identical multicomponent system. ACKNOWLEDGMENT We thank A. G. Whittaker, systems analyst of HoffmannLa Roche, Nutley N. J., for the statistical analyses except those reported in Tables 111 and IV. We also appreciate the assistance of E. H. Waysek in carrying out atomic absorption measurements.
RECEIVED for review July 22, 1968. Accepted October 16, 1968.
Determination by Atomic Absorption of Molybdenum, Ruthenium, Palladium, and Rhodium in Uranium Alloys J. M. Scarborough Atomic International, Canoga Park, Gal$ 91304 The feasibility of determining Mo, Ru, Pd, and Rh in uranium alloys by atomic absorption spectrometry was investigated. The mutual interference of the four elements which is normally a complicating factor in the determination of these elements by atomic absorption was eliminated by the presence of uranium. A procedure for the analysis of uranium fissium is presented.
MANYCOMPLEX URANIUM alloys have been compounded for use as nuclear fuel. Uranium fissium, made up of uranium (95 molybdenum (2.46 ruthenium (1.96 %), rhodium (0.28 Z), palladium (0.19 zirconium (0.10 Z), and niobium (0.01 is one of these alloys, the properties of which are reported elsewhere ( I ) . The analytical chemistry of the alloy, involving traditional techniques, has been thoroughly investigated (2, 3). Recently, the need arose to provide methods of analysis for this alloy which would be faster, cheaper, and, if possible, more reliable than the existing methods for determining Mo, Ru, Pd, and Rh. Atomic absorption spectrophotometry, which
x),
x),
x), x),
(1) S. T. Zegler and M. V. Nevitt, “Structures and Properties of Uranium-Fissium Alloys,” A E C Report, ANL-6116, (1961). (2) J. 0. Karttunen, ANAL.CHEM., 35, 1044-69 (1963). (3) C. J. Roddin, Ed., “Analysis of Essential Nuclear Reactor
Materials,” Division of Technical Information, USAEC (1964), pp 456--60. 250
ANALYTICAL CHEMISTRY
had not been considered for the analysis of fissium previously, offered a promising approach to the problem. A significant amount of information is available on the determination of Mo by atomic absorption techniques; however, very little new information has been forthcoming since the work of David reported in 1961 ( 4 ) . Mostyn and Cunningham (5) later reported on interferences associated with the determination of Mo in ferrous alloys. References to the determination of Ru, Pd, and Rh are somewhat more limited. The works of Lockyer and Hames (6) and Menzies (7) on the determination of the noble metals, discuss the determination of Pd and Rh. That of Strasheim (8) treats interferences in some detail. Peter Heneage ( 9 ) and Deily (IO) reported on the determination of Rh. Erinc and Magee (11) discussed the determination of Pd using propane flames. (4) D. J. David, Analyst, 86,730 (1961). ( 5 ) R. A. Mostyn and A. F. Cunningham, ANAL.CHEM., 38, 121
(1966). (6) R. Lockyer and G. E. Hames, Nature, 189,830 (1961). (7) A. C. Menzies, ANAL.CHEW, 32, 898 (1960). (8) A. Strasheim, Appl. Spectry, 17, 65 (1963). (9) Peter Heneage, Atomic Absorptiorz Newsletter, 5, NO. 3, MayJune 1966. (10) James R. Deily, ibid., 6 , No. 3, May-June 1967. (11) G. Erinc and R. J. Magee, A m l . Chim. Acta, 31, 197 (1964).