Characterization of blank correction in determination of oxygen in

Davis and Co. Characterization of Blank Correction in Determination of Oxygen in Sodium by theAmalgamation Method. J. M. Scarborough and P. F. DeVries...
0 downloads 0 Views 458KB Size
~~~~

~~

Table 111. Analysis of Oils Av vol of 0.100M

Oil used Corn Olive Cotton seed Tung

Solvent for oil Isopropanol Isopropanol

Concn (grams/ml)

No. of

NaBHl s o h

replicates

0.1693

6

(ml) 1.78

0.1830

6

IsoDro-

0.1836

pano1 Ethyl 0.1495 acetate See footnote in Table I. * Hydrogenation iodine values up to 240.

Standard deviation (ml)

Iodine value found

Literature value

0.0063

F.S.D.“ 0.1065

1.21

0.013

0,0829

79.01 f 0 . 6 0

79-88

7

1.75

0.027

0.1031

111.8 f 1.60

99-1 13

5

2.85

0.012

0.1481

219.5 f 1.00

163-171”

123

f 0.60

111-128

0

range of =t 1.0-1.5 % was found in measurements, even on the smallest sample. Accuracy was about =t2% on the lower levels of sample size. These analyses are shown in Table 11. Table 111shows the analysis of several representative vegetable oils: corn, olive, cottonseed, and tung. The analyses were quite precise, within a range of =tl.Sz. A more extensive study of the application of this procedure for the determination of unsaturation in a large variety of oils has recently been reported (8). Finally, the difficulty-reduced A9~1O-octalin was analyzed. Successive replicate analyses required 6, 24, and 40 hours. Nonetheless, 0.975 mmole was found, an error only 2.5% low. The range of measurements was +0.5 %. (8) T. K. Miwa, W. F. Kwolek, and I. A. Wolff, Lipids, 1, 152 (1966).

It may be concluded that this procedure provides a rapid, convenient method for determining unsaturation in organic compounds. Even compounds which reduce with difficulty may be analyzed. The use of hydrogenation to determine iodine numbers eliminates the difficulties of incomplete reaction of IC1 with conjugated systems and the errors resulting from concurrent substitution. The hydroanalyzer eliminates the necessity for hydrogen burets with their thermostating problems and provides a simple, compact method of assuring pure hydrogen. It promises to be a valuable tool for the organic analytical chemist. RECEIVED for review February, 6, 1967. Accepted March 17, 1967. Study was assisted, in part, by a grant from Parke, Davis and Co.

Characterization of Blank Correction in Determination of Oxygen in Sodium by the Amalgamation Method J. M. Scarborough and P. F. DeVries Atomics International, A Division of North America Aviation, Inc., 8900 DeSoto Ave., Canoga Park, Calif. The presence, magnitude, and source of a blank associated with amalgamation method for the determination of oxygen in sodium have been investigated. The preparation and use of oxygen-free amalgams in lieu of oxygen-free sodium for establishing blank corrections are described. The principal source of the blank is attributed to adsorbed water. High temperature vacuum drying techniques are found effective in reducing and stabilizing the blank. I N RECENT YEARS the use of sodium as a heat transfer medium has created the need for improved methods for the chemical analysis of metallic sodium. Because oxygen (oxide) is believed to play a significant role in the corrosion (or mass transfer) of containment systems, reliable methods for the determination of oxygen are required. For several years, methods which would determine oxygen a t concentrations of 20-50 ppm in sodium were deemed adequate for most applications. More recently, fundamental studies of corrosion, mass transfer, solubility, and diffusion, especially with high purity

826

ANALYTICAL CHEMISTRY

materials, require analytical methods which will follow small changes in oxygen content of alkali metals in the range of 0-10 ppm. The amalgamation method, which is widely used for the determination of oxygen in sodium, is based upon the fact that sodium oxide (Na20) is insoluble in sodium amalgam and in pure mercury ( I ) . Sodium metal is dissolved in mercury and the oxide is separated by flotation. Free sodium is removed from the mixture by extraction with pure mercury, the oxide residue being retained by flotation after each extraction. When all of the free sodium has been separated, the oxide residue is dissolved in water or dilute acid. Oxygen content of the sample is based upon the alkalinity or the amount of sodium in the oxide residue. Current practice assumes the residue t o be NasO, but in some cases it is desirable to identify species in the residue. Sample weight is determined directly or by titration of the sodium in the extractant, ( 1 ) L. P. Pepkowitz and W. C . Judd, ANAL.CHEM., 22,1283 (1950).

Sensitivity of the amalgamation method in the earlier work of Pepkowitz and Judd was limited by a precision of A0.005 (absolute). With impIovements (2), a precision of *0.001% was achieved. Other investigators (3-5) have extended the r m g e of the method downward and improved its reliability. A variety of other techniques such as neutron activation, dissolution in liquid ammonia, and vacuum distillation have been investigated (6-8). Although the amalgamation method is the most widely used of the techniques for determining oxygen in sodium (or potassium), several unsolved problems are encountered in the application of the technique. One of these problems which becomes quite serious in the analysis of samples containing less than 10 ppm oxygen, is the determination of a valid blank correction. From the inception of this method ( I ) until the last 2-3 years, no serious concixn has been evinced for the need to establish the existence cmf or to characterize a blank which may be associated with the mnalgamation procedure. Until now, no systematic investigation of the question has been reported. Poor precision and s:ensitivity of the method, which have limited its effectiveness and discouraged many investigators, are probably due in 3art to the failure to recognize the problem. The lack of satisfactory standard samples has sometimes been given as a n excuse for bypassing a study of blank characterization. A blank correction is often ascertained by analyzing several portions of the same sample, such that subsample weights cover a range of about 1-10 grams in increments of 1--2 grams. The total oxygen content of the subsamples is plotted as a function of weigh,:; the zero weight intercept provides a n approximate blank correction. Such a procedure is not always possible due to limitation in the amount of sample available. Compiicaticlns may also occur because large samples are often not homogeneous with respect to oxygen. Consequently, another means of evaluating and determining a procedural blank is desirable. Oxygen-free sodium would be a n ideal material with which to study the blank problem; however, such material is not readily available. An alternative material is suggested. Because oxides can be separated from sodium by the amalgamation technique described above, the same technique can be used to prepare oxygen-free amalgam. Such amalgams can then be treated a:; oxygen-free sodium samples. The use of these pure amalgam samples to investigate the blank problem is the principal subject of this discussion. The procedures and twhniques used to obtain data reported here are the outgrowth of several years of practical experience with the technique in this laboratory. Details will be reported elsewhere (5). No attempt has been made to cite all of the literature on the amalgamation method, nor will a n attempt be made to describe or justify all aspects of detailed procedures or to compare relative merits of various experimental techniques.

z

(2)

L.P. Pepkowitz, W. C . Judd, and R. J. Downer, ANAL.CHEM.,

26, 246 (1954).

(3) W. A. Dupraw, J. V i . Graab, and R. Gahn, Ibid., 36, 430 ( 1964). (4) E. W. Hobart, 1 F_'CRepr., TIM 900,1965. 15) J. M Scarbolough a i d P. F. DeVries, NAA-SR-12250, 1967. (6) E L. Stee!e and R. Lukens, N A S A Rept. GA-4855,1964. (71 R V. Jaworowski, J . R. Potts, and E. W. Hobart, ANAL. ':riEM, 35, .2?5 (1963). :$) ii; 8. Bergstresser, CI. R Waterbury, and 'I H M e ' z , ;EC Yrpa '.A ,7543. iY55

9.0

I

I

I

I

I

I

I 4.0

I 6.0

I 0.0

I 10.0

12.0

I

2.01 t

01

0

I

2.0

I

14.0

GRAMS OF SODIUM

Figure 1. Estimation of blank correction by intercept method EXPERIMENTAL

All analyses were performed in accordance with detailed procedures to be presented in reference 5. The only specialized equipment used in the work was a n inert atmosphere box and purification system capable of maintaining water and oxygen near the 1-ppm level. Standard 125-ml cylindrical separatory funnels were used as amalgamation/extraction vessels. All glassware, unless otherwise indicated, was dried overnight at 150" C in a vacuum oven. Sodium content of the oxide residues was determined by flame photometry. Oxygen-free amalgam was prepared by mixing sodium and mercury in the ratio of 1 gram of Na per 10 ml of Hg and separating the oxide impurities by flotation. RESULTS AND DISCUSSION

Several samples from a source of high purity sodium prepared a t Atomics International were analyzed to demonstrate the intercept technique for establishing a blank correction and to provide data for comparison with that obtained using oxygen-free amalgam. The results of these analyses are shown in Figure 1. The zero intercept of the least squares line shows that the blank correction for this set of seven analyses, using the laboratory's routine procedure, was 4.5 f 0.7 pg of oxygen. Allowing for this blank correction, the sodium contained less than 0.8 ppm oxygen. Batches, each of several hundred milliters of amalgam containing 1 gram of N a per 10 ml of H g were prepared as needed. Ten samples of amalgam ranging in size from 7.5 to 75 ml were analyzed for oxygen t o determine whether all of the oxygen had been removed. These analyses showed that the apparent oxygen content (4.5 rt 1 .O p g ) was independent of the amount of amalgam analyzed. It is therefore reasonable to conclude that the apparent oxygen obser\e:l constitutes a blank and :hat the amalgam is indeed free of oxygen. It should be Jbserved that this blank miw is in excellent agreement with that rkterriiined by !he intercept technique.

Table I. Oxygen Content as a Function of Surface Area and Material Approximate surface area, Apparent oxygen Em2 Type of material equivalent F g 45 Borosilicate glass" 3.3 f 0 . 2 75 Borosilicate glassn 4 . 5 f 1.3 110 Borosilicate glassa 7 . 6 f 1.9 200 Bxosilicate glass" 32.6 f 3.5 3.5c 40 Quartz" Quartza

70 100 195

8.7c

9.& Quartz" Polyethyleneh 4.3 f 1 . 0 0 Dried overnight at 120" C in air. * Dried several days at room temperature in air. c Single runs.

Vessel

No.

Table 11. Variation of Blank Value with Amalgamation/Extraction Vessel of oxygen -~~ _ Micrograms _ Blank 1 Blank 2 Blank 3 Average across 4.4 5.2 4. I 4.4 2.6 3.5

1 2 3 4 5 6

Average down

6.1

5.1 4.4 4.4 3.0 4.4

4.6 5.3 4.2 4.2 3.8 4.2

5.0 5.4 4.2 4.3 3.3 4.0

f 0.9 f 0.3 f0.2 f0.2 f 0.6 z!=

0.5

4 . 0 f 0.9 4.7 f 1 . 1 4 . 4 f 0.5

Because prior work had eliminated environmental and reagent contamination as the probable source of oxygen, it was considered plausible that a surface phenomenon could account for the blank. Experiments were performed in which the surface area was varied while keeping the amount of amalgam constant. The effects of contacting borosilicate glass, quartz, and polyethylene with amalgam were compared. Volumetric flasks of borosilicate glass and quartz and polyethylene bottles were used in the study, Oxygen-free amalgam (10 ml in each case) was added to the containers and shaken vigorously to ensure contact of the amalgam with the entire surface. The amalgam was then drained out, including any oxide residue, if present, leaving only surface contamination if any. The containers were then rinsed vigorously eight times with mercury. Each time the mercury was completely drained. Finally the flasks were removed from the inert atmosphere box, rinsed with water, and the sodium content of the wash water was determined. The sodium value was converted to its oxygen equivalent. Results in Table I show that apparent

oxygen content is a function of surface area as well as surface type. Once the relationship between the blank value and surface area was established, it became desirable to determine the average blank correction associated with all of the amalgamation/extraction vessels in use for routine analysis of sodium samples a t that time. Forty 10-ml samples of oxygen-free amalgam each containing 1 gram of sodium were analyzed by the routine procedure. Twenty amalgamation/extraction vessels were used for these forty determinations. The apparent oxygen content of these amalgam samples was found to be 3.9 + 1.0 pg of oxygen. Six of these vessels were then selected a t random and labeled to maintain identity. Three additional analyses were then performed in each of the vessels. The results are shown in Table 11. The values obtained are comparable to those reported above. The data also show that each piece of glassware probably has a characteristic blank associated with it. As it was shown that pure amalgam interacts in some way with the surface of glassware to produce a reasonably precise blank value, it was desirable also to establish the source of the blank. The most obvious source would be the failure to wash all traces of amalgam from the glassware in the extraction steps of the procedure. I n general, failure to d o so would give higher more erratic blanks than reported here. Mercury should be a satisfactory solvent for metallic sodium; therefore, it is not likely that metallic sodium remains on the surface. It is conceivable that the amalgam could interact with the surfaces of the glass to produce a water-soluble layer containing sodium which would contribute to the sodium content of the oxide residue. It was considered more probable that a thin film of moisture present o n the surface of the glassware hydrolyzes sodium, from the amalgam, at the interface to produce a film of hydroxide or oxide which is not readily removed by repeated washing with mercury. Two radioactive tracer experiments were carried out to determine which of the above hypotheses could best account for the sodium in the blank-Le., whether the sodium originated from the sample or from the glassware. In one experiment a typical amalgamation/extraction vessel was irradiated with thermal neutrons to produce *"a activity in the borosilicate glass matrix. After irradiation, the vessel was washed thoroughly and rinsed until the final wash solution contained no sodium detectable by flame photometry or gamma counting. After washing, the vessel was vacuum dried (approximately 2 hours a t 150" C ) and transferred to the inert atmosphere box. The vessel was then used to analyze 10 ml of oxygen-free amalgam as described in previous experiments. The final aqueous solution resulting from the hydrolysis of the oxide residue contained a total of 21.8 ~~

Sample No.

Sample wt., gram

1 2 3 4 5 6

0.463 0.554

b

b b

b b

Table 111. Blank Determination Using Activated Sodium Tota Na Activity Maximum in oxide Oxygen equivalent of original theoretical residue, pg of total Na, pg Na, dps/pg activity," dps 50.5 82.3 23.8 20.4 18.8 25.0

17.6 28.6 8.3 7.1 6.5 8.7

b

...

128 128 128 110 110

10,530

Activity expected if all of the sodium in the blank originates in the sample. Not determined.

828

e

ANALYTICAL CHEMISTRY

3,046 2,611 2,068 2,750

Activity found, dps 5,454 10,Ooo 2,241 1,958 1,607 2.013

z

Of

theoretical

... 95 74 75 78 73

pg of sodium, equivalent to 7.5 pg of oxygen. This quantity is of the same magnitude as a typical blank and is in satisfactory agreement considering the fact that the standard overnight vacuum drying technique could not be used because of the short half-life of 24Na. N o 24Na was detected in the oxide rizsidue. However, because of limitations imposed by aliquotting and by the detection limit for gamma-counting, as much as 15% of the sodium present could have been 24Na and not have been detected. The sensitivity of this experiment was limited by the impracticability of irradiating the glassware t o a higher specific activity. To have done so would have required nontypical handling techniques which might have invalidated the experiment. I n the second tracer experiment, several pieces of sodium were sealed in polyethylene vials (inside the inert atmosphere box) and irradiated w.th thermal neutrons. Two samples of the irradiated sodium from one vial were analyzed for oxygen using the laboratory’s routine analysis procedure (see samples 1 and 2 in Table 111). The oxygen-free amalgam from sample number 2 was recovered and analyzed twice in succession as sodium samples (3 and 4 in Table IV) to determine a so-called ‘‘piggy back” blank value. The sodium from the other vial was used solely to prepare oxygen-free amalgam. Two samples of this amalgam were analyzed for comparison to blank values obtained by the oxygen-free amalgam technique described earlier (samples 5 and 6). Agreement was good. Through experimental oversight, the specific activity of sample number 1 was not determined; however, its specific activity lies somewhere between samples 2 and 6. Although data for this sample probably fit the pattern established here, the assignment of a yield value would not be valid. These data show quite clearly that the sodium constituting the blanks obtained in i:his set of experiments is made u p of a t least 75% labeled sodium. This observation is not in disagreement with the: first tracer experiment, described above, in which it was concluded that less than 15% of the sodium in the blank originated in the glassware. The fact that the oxide residue :Yam sample 2 contained 95% of the maximum theoretical a1:tivity is consistent with the observation that at least 75 of the blank originates in the sample. The preceding experiments indicate that the blank is related t o a surface phenomenon and that the source of the sodium in the blank is primarily the sample. It is therefore probable that the surfacemoisture hypothesis offers the best explanation for the origin of the blank. Several additional experiments t o compare the efficacy of improved drying techniques were performed. In one series of tests, the amalgamation/extraction vessels were dried in a vacuum oven overnight a t 190” C. The vessels were then transferred in air to the inert atmosphere box and used for the analyses of oxygen-:kee amalgam as previously described. An average blank of 3.0 i. 0.1 pg of oxygen was obtained. The same experiment was repeated with the difference that the vessels were dried in air at 300” C overnight. A n average blank of 2.7 f 0.1 pg of oxygen was observed. Once again the experiments were repeated but with vessels that had been dried overnight a t 300” C in a vacuum. The vessels were transferred under vacuiim t o the inert atmosphere box. An

Table IV.

Effect of Drying on Blank Value

Method of dryng Normal “routine” technique; 150” C in vacuum, overnight; transferred in air Vacuum at 190” C , overnight; transferred in air Air at 300” C, overnight; transferred in air Vacuum at 300” C, overnight; transferred under vacuum

Blank value of oxygen

pg

3.9 f 1.0 3.0 f 0 . 1 2.7 f 0 . 1

1 . 3 f.0 . 3

average blank of 1.3 =t0.3 pg of oxygen was obtained. From these experiments, the results of which are summarized in Table IV, it is clear that high temperature vacuum drying was effective in reducing and stabilizing the value of the blank. The presence of moisture o n the surface o f the glassware appears to account for most of the blank along with a smaller contribution from the glass. CONCLUSION

This work has shown that oxygen-free sodium amalgams can be prepared using flotation t o separate and remove sodium oxide. There amalgams can then be treated as oxygen-free sodium samples t o establish a total blank correction for the amalgamation method which includes all systematic errors for the entire procedure. The observed consistency of the large quantity of data presented in this investigation, especially that for the analysis of high purity sodium, should dispel any misgivings regarding the reliability of the amalgamation method for determining oxygen in sodium. Problems encountered in the past are those of sampling, contamination, and environmental control, and especially the failure t o recognize the tenacity with which water is retained o n the surface of glass. I n earlier applications of this method, analyses were performed in a vacuum. The assumption was made that a vacuum system Torr a t was “clean” and dry when a pressure of IOp5to room temperature was maintained. The acceptance of this assumption, which is valid only if the glassware is outgassed for long periods of time a t elevated temperatures, caused the problem of adsorbed moisture t o be overlooked. This work has shown that if environmental conditions are adequately controlled, the major source of the blank is a surface phenomenon involving moisture. I n addition, a small amount of sodium is removed from the surface of the glassware which also contributes t o the blank. High temperature vacuum drying was shown t o be effective in reducing the blank correction. The techniques described herein can be applied t o reduce and stabilize the blank correction, making the amalgamation method useful for determining oxygen in sodium a t the 1-ppm level. RECEIVED for review September 12, 1966. Accepted April 7, 1967. Work sponsored by the U.S. Atomic Energy Commission under Contract AT(1l-1)-Gen-8.