Nephelometric Determination of Fluorine - Analytical Chemistry (ACS

Nephelometric Determination of Fluorine. Rollin E. Stevens. Ind. Eng. Chem. Anal. Ed. , 1936, 8 (4), pp 248–252. DOI: 10.1021/ac50102a007. Publicati...
3 downloads 0 Views 686KB Size
Nephelometric Determination of Fluorine ROLLIN E. STEVENS, U. S. Geological Survey, Washington, D. C.

I

N THE analysis of rocks and Fluorine in minerals may be determined Preliminary Separation minerals fluorine has long with the nephelometer to about 1 per cent of Fluorine been regarded a of the fluorine. The determination is The preliminary decomposielement. Although in rocks the made on an Of the sodium chloride tion and removal of s i l i c a , determination of fluorine is fresolution of the fluorine, obtained by the alumina, etc., is effected by t h e quently omitted, it is an essent i a l c o n s t i t u e n t of m a n y Berzehs method of extraction. The flueBerzeliusmethod, essentially the rine is precipitated as colloidal calcium procedure given by Washington minerals. P h o s p h a t e r o c k s (17). Hydrochloric acid is used usually contain an appreciable fluoride in alcoholic solution, gelatin servthroughout the procedure inquantity and it is found in some micas, particularly biotite, in ing as a Protective colloid. Arsenates, S U I stead of nitric acid. hornblende, in tourmaline, and in fates, and phosphates, which interfere with About 2 grams of p o w d e r e d certain fluorine minerals. AIthe determination, must be removed. mineral (or a smaller quantity if though not an abundant element, the fluoride content is high) are f u s e d w i t h 10 grams of anhyfluorine must be frequently dedrous sodium carbonate until decomposition is complete. The termined in mineral analysis. Few methods of determining cake is thoroughly leached with hot water, filtered, and washed. fluorine are available, and most of them are time-consuming. If there is any reason to doubt the completeness of the extraction, the residue and paper are ignited a t a low heat and fused The evolution method, whereby fluorine is distilled as silicon with 2.5 grains of sodium carbonate, the cake is extracted as tetrafluoride, is described by Reynolds, Ross, and Jacob (10). before, and the second extract added to the first. The yield is stated by Hillebrand and Lundell ( 6 ) to be about To the combined filtrates, containing the fluorine, about 3 92 to 94 per cent of the fluorine, and a large correction factor is grams of powdered ammonium carbonate are added. The necessary. solution is digested for 1 hour at 40' C., allowed to cool, and 1.5 Steiger's method (15), which consists in measuring the bleachgrams more of ammonium carbonate are added. After standing ing effect of fluorine on oxidized titanium solutions, is useful in 12 hours the solution is filtered, and the beaker and precipitate the estimation of small percentages of fluorine in rocks but is not are washed with weak ammonium carbonate solution. The sufficiently accurate for the determination of large percentages. filtrate is boiled until the odor of ammonia can no longer be deThe volumetric method of Knoblock (8) was improved by tected. The burner is removed, a few drops of methyl red are Fairchild (4), who reported some satisfactory results. The added, and then hydrochloric acid is added until the red color present writer, however, obtained discordant results with the appears. If phosphates are present 1 ml. of 0.1 N ferric chloride method. It depends on the formation of un-ionized ferric fluois added, and then weak sodium hydroxide until the iron is preride when a standard solution of ferric chloride is added to the cipitated and the methyl red just becomes yellow. About 0.5 fluoride solution. The excess of ferric chloride sets free iodine gram of 20-mesh metallic zinc in the solution removes arsenic: from potassium iodide, and the iodine is determined by titration. and prevents bumping when the solution is boiled. Two milliMinute quantities of fluorine in natural waters may be deterliters of a strong solution of zinc oxide in ammonium carbonate mined by the colorimetric method of Foster (5), which depends solution containing ammonia are added, and the solution is on the bleaching of ferric thiocyanate by fluorine solutions. boiled (covered) until free from ammonia and then filtered and Other methods for determining fluorine in water are described washed with water. by Armstrong ( I ) , Wilcox (18), Thompson and Taylor (16), Sanchis ( I I ) , and Barr and Thorogood (8). Most of the determinations on minerals, described later, Accurate results are reported by Hoffman and Lundell (7) were made using the above procedure for removing silica. by the use of a modification of the lead chlorofluoride method, Double precipitation with zinc oxide, as recommended by originated by Starck (13). The compound, PbClF, is precipitated and weighed as such or the chlorine is determined by the Hoffman and Lundell (Y), requires less time and was later Volhard method. The method is not satisfactory for the analyadopted. sis of phosphate rock or in the presence of much phosphoric acid. If sulfates or chromates are present in appreciable quantity An older method, still much used, is the gravimetric method of they are removed by adding barium chloride, avoiding a n Berzelius (S), consisting of precipitating and weighing the fluorine as calcium fluoride. If precipitated alone, calcium fluoride excess. The small quantities of sulfates usually present in forms a fine suspension which cannot be readily filtered. It is silicates may be ignored. therefore precipitated together with calcium carbonate, filtered, ignited, and weighed, after which the calcium carbonate is reThe solution, in a platinum dish, is evaporated on the steam moved by several treatments with dilute acetic acid and the puribath to about 100 ml., cooled, and filtered into a volumetric flask fied calcium fluoride is again ignited and weighed. The preof volume such that the concentration of sodium chloride will be cipitate may be contaminated by sulfates? phosphates, and chro55 grams per liter. Thus, if 10 grams of sodium carbonate were mates. Precipitation of calcium fluoride in conjunction with the used in the extraction of fluorine, a 200-ml. volumetric flask is oxalate is described by Starck and Thorin (14). used. The filtrate is made just acid to methyl red with weak hydrochloric acid, then just alkaJine with weak sodium hydroxide, diThe tendency of calcium fluoride to precipitate as a fine luted to the mark, and thoroughly shaken. suspension suggested the possibility that the precipitate might be measured nephelometrically. Preliminary results, Nephelometric Measurement of Fluorine although promising, showed the need of reducing the solubility by adding alcohol and of stabilizing the suspension with a REAGENTS.Two grams of gelatin in 40 ml. of distilled water are heated on the steam bath with occasional shaking until disprotective colloid. Gelatin was found satisfactory as a prosolved to make the 5 per cent gelatin solution. If the solution tective colloid, keeping the calcium fluoride in colloidal suscongeals on cooling, it is warmed slightly until fluid. A fresh pension almost indefinitely. solution should be made every few days. The procedure finally developed gives results which appear The precipitant is prepared by making 100 ml. of a solution containing 20 grams of calcium chloride just acid to methyl red to be accurate to about 1per cent of the fluorine. For samples with 0.1 N hydrochloric acid and adding 2 ml. excess of 0.1 N requiring a carbonate fusion the lower limit of the method hydrochloric acid. This solution is diluted with 100 ml. of disis about 0.3 per cent of fluorine, although for water-soluble tilled water and filtered through an S. & S. 589 or Whatman No. materials, such as salines, a smaller percentage may be 42 paper. Commercial 95 per cent alcohol. determined. 248

JULY 15, 1936

ANALYTICAL EDITION

249 MILLIGRAMS

2 262

2.036

FLUORINE

OF 1.810

IN

25ML.

IS84

1.311

1.131

.?O

.60

.so

w

v)

n as 5 W

.Z 30 W Y

4I I 6 W I

a

z 20

Y

I6

1.0

.eo

.80

RATIO

OF

SOLUTIONS

CURVES OBTAISEDON DIFFEREST FIGURE 1. NEPHELOMETRIC DAYS

FIGURE 2. NEPHELOMETRIC CURVEFOR DILUTESOLUTIONS

Sodium chloride solution, 55 grams per liter, filtered through a fine paper. To make the standard sodium fluoride solution, c . P. sodium fluoride is dissolved in water, filtered, and twice recrystallized. The pure crystals are dried, crushed, then heated for 2 hours at 500' C. The pure sodium fluoride is kept in a wide-mouth glass-stoppered bottle, the stopper being coated with petrolatum to exclude moisture. Of this 0.2000 gram is weighed accurately and made up to 1 liter with the sodium chloride solution (55 grams per liter). The flask is shaken until solution is complete; then the standard is stored in a bottle coated inside with araffin. APPARATUS.The nephelometer used was a Bausch Lomb Kober constant ugper-end type, with nephelometer cups of 5ml. capacity and lamp-housing equipment. Volumetric flasks and pipets of different capacities. PROCIDURE.A 25-ml. aliquot of the unknown solution is taken with a pipet and placed in a wide-mouth glass-stoppered bottle. A number of other unknown solutions may be determined at the same time. Twenty-five milliliters of the standard sodium fluoride solution are measured into another bottle, 1 ml. of 5 per cent gelatin solution is added with a pipet to each solution, 10 ml. of 95 per cent alcohol are run into each solution, and the bottles are swirled. Five milliliters of the precipitant are then added, the solution being swirled during the addition of precipitant. The bottles are then stoppered and shaken vigorously, and the suspensions allowed to stand 15 minutes. The nephelometer cups are then rinsed with the standard suspension, and the standard placed on both sides of the instrument. The left side is set a t a definite reading, 10 or 20, the lamp turned on, and the right side adjusted to equal intensity of light. A series of six or more readings of the right side is taken, and the right side set at the average of these readings. This setting of the right side is retained for the whole series of comparisons with the standard, being potentially equal to the setting of 10 or 20 on the left side of the instrument. The standard solution in the left cup is discarded, and the cup is rinsed with unknown suspension, filled with the unknown suspension, and replaced in the instrument. The left side is adjusted l,o equality of light with the right, four or more readings being taken.

with a standard suspension prepared at the same time. In the final comparison the unknown should be within about 5 per cent of the standard. Usually two comparisons suffice. If the original unknown solution is weaker in flhoride than the standard, the standard must be diluted to the proper concentration for the final comparison; however, a dilution of the standard to less than one-third strength is not practicable. The fluoride concentration of the final diluted unknown solution is read from the nephelometric curve and, knowing the dilution, the quantity of fluorine in the sample is calculated.

The average of these readings is the comparison of the unknown with the definite setting of the standard (10 or 20). The approximate concentration of the unknown is read from the determined curves (Figures 2 and 3) which give the readings for different concentrations of fluoride compared with a setting of 10 for the standard in Figure 2 and of 20 for the standard in Figure 3. This preliminary figure having been obtained, a part of the original unknown solution is diluted with sodium chloride solution (55 grams per liter) to about the same strength in fluoride as the standard. An aliquot of this solution is precipitated and compared nephelometrically, as described above,

Determination of the Nephelometric Curves In the determination of the nephelometric curves the standard was diluted with sodium chloride solution (55 grams per liter) to 90, 80, 70, 60, and 50 per cent and comparisons were made with a definite setting of 10 for the undiluted standard. Knox gelatin was used as a protective colloid and the precipitant was added with a pipet draining in 17 seconds. The values obtained are given in Table I. TABLEI. COMPARISON OF MOREDILUTESOLUTIONS TO STANDARD (Standard = 2.262 m g . of fluorine per 25 ml. Standard diluted to 90% 80% 70% Deo. 14 . ., 17.99 23.28 ... 18.23 23.42 Dec. 17 14.20 ... 24.78 13,88 ... 24.79 Deo. 18 14.00 22.12

...

Standard set a t 10.0)

60% 30.39 30.29

... ...

...

50% 46.75 46.82 49.47 49.00 43.09

These values are plotted in Figure 1. They fall precisely on the curves, but the three curves show a slight difference in slope. This difference is in part due to a slight error in comparison of standard to standard, by which the right side of the instrument is set a t a value equal to 10 on the left side. This error would become magnified in comparing the less concentrated solutions. For accurate work the unknown and standard should be within 5 per cent of equality in fluorine concentration, the curve serving as a means of approximating the proper dilutions to be made. In Figure 2 these values and some others are averaged to give an average determined curve. The determined curve is much above the hypothetical, in which readings are assumed inversely proportional to concentrations. This is not objec-

VOL. 8, NO. 4

INDUSTRIAL AND ENGINEERING CHEMISTRY

250

attained. This time interval was arbitrarily taken and a shorter period may be found sufficient. The results were obtained at a room temperature of about 22' C. If the standard and unknown are a t the same temperature, variations in room temperature would presumably have little effect.

Effect of Rate of Addition of Precipitant In Figure 4 is shown the result of slower addition of the precipitant. With a 5-ml. pipet that drains in 47 seconds the curve is definitely lower than that obtained when the precipitant is added with a pipet draining in 17 seconds. If the precipitant is added dropwise or the fluoride solution is added to the precipitant, results showed that the weaker fluoride solution may have greater light-reflecting power than the stronger.

Stability of Suspensions

FIGURE3. NEPHELOMETRIC CURVEFOR CONCENTRATED SOLUTIONS tionable, provided the readings are reproducible, as they are shown to be. Indeed, the steepness of the curve makes it possible to obtain a much more accurate figure for fluorine with the average observer's sensitivity to light. Curve Y is obtained by dividing the difference between the observed reading and the setting of the standard by three and adding this figure to the setting of the standard. It follows the hypothetical curve closely and was found useful in calculating results. More concentrated solutions are compared to the standard in Table 11,and the values are plotted in Figure 3. The curve is useful in determining the dilution necessary to make the unknown and the standard of approximately the same strength for a second comparison. Some uncertainty was experienced in determining these values because of the difference in appearance of the two solutions in the nephelometer. Solutions twice standard and stronger showed a clear sky-blue in the nephelometer, whereas the standard in comparison was a creamy yellow.

A suspension was prepared from the standard solution a t 9:30 A. M. and kept in a glass-stoppered bottle. At 3:30 P. M. a second suspension was prepared. Comparison in the nephelometer showed that the first suspension was 97.8 per cent of the second, a comparatively small change for a 6-hour period. Duplication of Results in Absence of Interfering Elements In order to determine the accuracy of the method in the absence of interfering elements (sulfates, phosphates, and arsenates) eleven suspensions were prepared from the standard solution and compared in turn with one selected for reference. Ten readings were taken of the reference standard, and four of each succeeding suspension. (See Tahle 111.) The results show a high precision for nephelometric work ( l a , 19). Comparisons with preparation 1 were made continuously in the order left to right across the table, and the poorest results, those on preparations 6, 7, and 8, seem to be caused by a temporary eye strain rather than by an error in the preparation of the suspensions.

Errors Due to Maladjustment of Sodium Chloride Concentration TABLE 11. COMPARISON OF MORECONCENTRATED SOLUTIONS TO

STANDARD

(Standard = 2.262mg. of fluorine per 25 ml. Standard set at 20.0) Strength oompared to standard 1.25 1.5 1.75 1.875 2 2.5 5 7.5 10 Re adi ngs

14.6 14.5 15.7 18.1 .. 14.5 . . ,

.

..

..

.. ..

19.5 16.4 11.9 6.6 4.5 18.8 16.3 11.9 6.7 4.6 17.2 ...

..

.. .. .

PROTECTIVE COLLOID. The curves shown were obtained by using Knox gelatin as a protective colloid. With Difco gelatin the curves were nearer the hypothetical. It is advisable for each observer to determine the nephelometric curves before proceeding t o unknown samples. TIMEAND TEMPERATURE. Fifteen minutes were allowed after the addition of the precipitate for equilibrium to be

Mistakes in adjusting the sodium chloride content of the solutions result in small errors; if the sodium chloride concentration is too high the results are slightly low, and vice versa. The standard for the tests contained 1.697 mg. of fluorine in 25 ml. and the concentration of 55 grams sodium chloride per liter. A solution containing the same quantity of fluorine but with a sodium chloride concentration of 69 grams per liter gave 98.2 per cent recovery. Another having 41 grams sodium chloride per liter gave 103.3 per cent recovery. By recording the approximate quantity of sodium carbonate used in the extraction, the proper dilution is made to give a concentration of 55 grams of sodium chloride per liter, small differences in sodium chloride content having little effect.

TABLE111. DUPLICATION IN PREPARINQ AND READINQ STANDARD SUSPENSIONS (Preparation 1 taken as standard and set at 20.0) 2 3 4 5 6 7 Preparation No. Reading 20.25 20.03 20.05 20.08 20.42 19.45 Calculated hypothetical reading 20.08 20.01 20.02 20.03 20.14 19.82 99.6 100.0 99.9 99.9 99.3 100.9 Per cent recovery Average per cent recovery, 99.8 Greatest deviation from 100 per cent recovery, 1.3 per cent Average deviation from 100 per cent recovery, 0.4 per cent

8

9

10

11

20.78 20.26 9s.7

19.90 19.97 100.2

20.22 20.08 99.6

19.92 19.97 100.2

JULY 15, 1936

ANALYTICAL EDITION

Effects of Interfering Substances Some effects of sulfates, phosphates, arsenates, and ammonium salts on the results are shown in Table IV. Sulfate was added as anhydrous NazSO1, phosphate as KHZPOa, and arsenate as Ka2HAs04.12Hz0. TABLE IV. EFFECTOF INTERFERING SUBSTANCES Fluorine on 2-Gram Fluorine Sample Gam

0.01629 0.01086 0.01086 0.01358 0.01810 0.01358

Interfering Substance

Interfering Substance on 2-Gram Sample

Gram

%

0.81 0.54 0.54

0 , 0 2 1 so8 0 . 0 2 1 SOs 0 . 0 2 1 SO8

0.91 0.68

0.0005 As205 0 . 0 2 NHiCl

0.008 PzOs

0.68

Recovery Shown by Nephelometer Fluorine Reading Found

%

1 . 0 so8 1 . 0 so8 1.0 sos 0 . 4 PzOs 0 . 0 3 As206

...

99.3 103.3 103.0 121.5 106.5 100.4

251

results in the middle section of Table V, where barium chloride was used in the absence of sulfate. Arsenic may be satisfactorily removed by adding about 0.5 gram of 20-mesh metallic zinc before the addition of zinc oxide. During boiling to remove ammonia, the metallic zinc reduces arsenates and a t the same time prevents bumping. Results obtained after using these treatments to remove. interfering elements are given in Table V. TABLE v.

REMOVAL OF INTERFERING

0.80 0.56 0.56 0.83 0.97 . 0.68

Treatment of S o h tion

Fluorine Wt. on2of Gram Interfering Fluorine Sample Substances Gram

The results in Table IV show that the quantity of sulfate taken causes a slight plus error when the fluoride content of the sample is low, but its effect is negligible when as much as 0.80 per cent of fluorine is present in a 2-gram sample. This is due to the steep nephelometric curve for fluorine, the slight turbidity caused by co-precipitation of sulfate being noticeable only in weakly turbid solutions. A large positive error results in weak fluoride solutions when a little phosphate is present, but this likewise would be less for high percentages of fluorine. A small quantity of arsenic causes an appreciable error. Failure to remove ammonium salts completely seems to have little or no effect. The precipitation with zinc to remove the last traces of silica, in the Beraelius method of extraction, does not remove the phosphate sufficiently. A solution containing 0.0152 gram of phosphate and 0.01358 gram of fluorine in 200 ml. gave 108.8 per cent recovery after the zinc treatment; and a solution having the same quantity of phosphate and 0.00905 gram of fluorine gave 128 per cent recovery. A fairly tmtisfactory method of removing this quantity of phosphate was to add 1 ml. of 0.1 M ferric chloride before precipitation with zinc oxide, but no method was found for removing phosphate so completely as to cause no interference in the fluorine determination. Sulfates may be removed with barium chloride. An excess must be carefully avoided, however, as it leads to high results for wea,k fluoride solutions and removes fluorine when the fluoride content is high, as illustrated by the high and low MILLIGRAMS 2.262

0.016

OF 1.810

FLUORINE 1.18+

IN

ELEMENTS

%

25ML. 1.357

41

1.111

%

Interfering Substance on 2-Gram Sample

%

%

%

101.2 100.6 101.9 100.4 100.0

0.92 0.68 0.46 0.45 0.68

None None 1 . 1 903 None

100.9 100.9 151.0 171.5 100.1 96.8 97.8

1.1 908 0 . 2 CrzOa

106.2

0.92 0.92 0.68 0.77 0.45 2.19 2.21 0.48

1 . 1 803 0 . 2 CrzOa 0 . 4 PiOs 1.1 803 0 . 42 Crz08 PzOs

102.0

0 69

100.3

9.08

1.1 80s

100.1

4.53

97.8

8.85

Gram

Metallic 0,01810 0 . 9 1 0.0027 As206 zinc (ZnO) 0,01358 0 . 6 8 0,0027 AszOs 0.00905 0 . 4 5 0,0027 AszOs 0.00905 0 . 4 5 0.0027 Ass06 0.01358 0 . 6 8 None 5 m l . of 2% 0.01810 0 . 9 1 None BaCh 0.01810 0 . 9 1 0 021808 0.00905 0 . 4 5 None 0.00905 0 . 4 5 None 0.00905 0.04525 0.04525

0.45 2.26 2.26

0.021S03

FeCla-ZnO, 0.00905 5 ml. of 2% BaClz

0.45

0.021 803 0.004 CrzO3 0.008 Pros 0.021 803 0.004 CrzOa 0 , 0 0 8 PzOs 0 . 0 2 1 so8 0 . 0 0 4 Crz03 0 , o o s PlOS 0 . 0 2 1 so8 0.004 CrzOa 0.008 PzOs

0.01358

0.68

0.1810

9.05

0.0905

4.53

0.1810

9.05

None

0.021 SO8

None

Recovery Shown by Nephelometer FluoRead- rine ing Found:

0.14 0.14 0.14 0.14

AszOs AszOs AszOs ASZOS

None None

1.1908

1 . 1 SOs

0.4

PZ06

0 . 2 CrzOs 0 . 4 PzOs None

Failure of Method for Phosphate Minerals The method was not found satisfactory for materials of high phosphate content, owing to incomplete extraction of fluorine (9) and to failure to remove all the phosphate. For known solutions of high phosphate content, two precipitations with a large excess of zinc oxide gave slightly high results for fluorine. Two precipitations with 1 ml. of ferric chloride in combination with zinc oxide were better, but caused the loss of a small portion of the fluorine when present in large quantity. A higher concentration of hydrochloric acid in the precipitant might prevent the eo-precipitation of phosphate, but would probably change the nephelometric relationships. The results obtained on some phosphates, after removal of phosphate by two treatments with ferric chloride and sine, oxide, are given in Table VI. The fluoride solution obtained from triplite failed to follow the nephelometric curve on dilution, showing that the phosphate was not completely removed. The figure given for the fluorine in triplite, there-fore, is probably high. The tests on phosphate rock seem tc, show incomplete removal of fluorine. TABLE VI. DISCREPANCIES WITH PHOSPHATE MINERALS

a

FIGURE 4. EFFECT O F RATEO F ADDINGPRECIPITAVT

b C

Fluorine by Other Methods

Locality

Fluorine Found

%

%

Triplite

Amelia, Va.

9.76

Phosphate rock Phosphate rock Phosphate rock

Wyoming Tennessee Florida

2.83 2.93 3.14

9,580 8.93; 3.58 3.7bC 3.89.

Sample

3. G. Fairchild by volumetric. J . G. Fairchild by Berzelius. Distillation, Reynolds, Ross, and Jacob

INDUSTRIAL AND ENGINEERING CHEMISTRY

252

Results with the Method I n Table VI1 are given determinations on silicates and fluorine-containing minerals. The two results on phlogopite check closely; no theoretical figure for fluorine is given, as the fluorine content of phlogopite varies. Results on gearksutite and creedite are much closer to the calculated theory than previous determinations. Bureau of Standards standard sample 91 gave a figure in fair agreement with the result obtained by Hoffman with the lead chlorofluoride method. In topaz hydroxyl replaces fluorine, so that no theoretical figure for fluorine is given. The topaz listed is unusual in having a high water content (2.67 per cent of combined water) WITH TABLE VII. RESULTS

Sample

Locality

THE

METHOD

Fluorine Found Theory

Results by Other Methods

% 1. Phlogopite

Kin& Mountain,

2. Fluorite 3. Gearksutite

Rosiclare, Ill. Wagon Wheel Gap, Colo. 42.30 42.9

N. c.

4 31 ... 4.34 4 8 . 6 5 48.68

...... . . .... Average of results bv three

4. Creedite difference, 30.35 5. Bureau of Standards standard sample 91, opal glass

... Jefferson, S. C.

6. Topaz

13.23

...

....

VOL. 8, NO. 4

and hence a low fluorine content (13.23 per cent). The complete analysis gave a summation of 100.13, after subtracting the oxygen equivalent of the fluorine.

Literature Cited Armstrong, IND.EXG.CHEM.,Anal. Ed., 5 , 300-2 (1933). Barr and Thorogood, Analyst, 59, 378-80 (1934). Berzelius, Schweigg. J., 16, 426 (1816). Fairchild, J . Washington Acad. Sci., 20, 141 (1930). Foster, IND.EXG.CHEM.,Anal. Ed., 5 , 234 (1933). Hillebrand and Lundell, “Applied Inorganic Analysis,” p. 601, New York, John Wiley & Sons, 1929. Hoffman and Lundell, Bur. Standards J . Research, 3, 581 (1929). Knoblock, Pharm. Ztg., 39, 558 (1894). Reynolds and Jacob, IND.ENQ.CHEM.,Anal. Ed., 3,366 (1931). Reynolds, Ross, and Jacob, J . Assoc. Oficial Agr. Chem., 11, 225 (1928). Sanchis, IND.ENG.CHEM.,Anal. Ed., 6, 134-5 (1934). Snell and Snell, “Colorimetric Methods of Analysis,” New York, D. Van Nostrand Co., 1936. Starck, Z . anorg. Chem., 70,173 (1911). Starck and Thorin, 2.anal. Chem., 51, 14 (1912). Steiger, J . A m . Chem. Soc., 30,219 (1908). Thompson and Taylor, IKD. ESG. CHEM.,Anal. Ed., 5, 87-9 (1933). Washington, “Chemical Analysis of Rocks,” p. 264, New York, John Wiley & Sons, 1930. Wilcox, IND.ENG.CHEM.,Anal. Ed., 6, 167-9 (1934). Yoe, “Photometric Chemical ,4nalysis,” New York, John Wiley & Sons, 1929. RECEIVED March 11, 1936. Presented before the Division of Physical and Inorganic Chemistry at the 91st Meeting of the American Chemical Society, Kansas City, Mo.. April 13 to 17, 1936. Published by permission of the Director, U. S. Geological Survey.

Microvaporirnetric Determination of Molecular Weight JOSEPH B. NIEDERL, OTTO R. TRAUTZ, AND ALBERT A. PLENTL Washington Square College, New York University, New York, N. Y.

I

N 1929 Niederl (26) described a simple microvaporimetric

method for the determination of the vapor density or the molecular weight of low-boiling liquids. This method was improved (29) so that the results were within *2 per cent of the theoretical (%$,a?); consequently, its application to higher boiling substances for temperatures up to 200” C. was investigated (28). But the difficulty of obtaining temperature constancy increased proportionally with the rise in temperat u r e . 80 m u c h so that such

i m p r o v e m e n t s as

fused-in thermometer and other s u i t able c h a n g e s i n a p p a r a t u s did not appreciably i n c r e a s e t h e a c c u i a c y of the results. F u r t h e r experimental studies led the ment of a method for temperatures up to 320” C., retaining all the f a v o r a b l e feat u r e s of the low-temperature m e t h o d , s u c h a s v i s u a l observation Of the and c o n d e n s i n g point, as well as r e p e t i t i o n , and t o the cons t r u c t i o n of the improved,

’’

yet extremely simple and inexpensive apparatus as set forth in this communication.

Principle of the Method A few milligrams of a substance, either liquid or solid, are vaporized in a closed system in such a way as to permit an

accurate indirect volume determination with suitable provisions for pressure and temperature control and constancy a t any point for a temperature range of 300” C. The previously described micromethod The m e t h o d is n o t merely

for determining Of lowboiling liquids has been suitably modified to include high-boiling liquid and solid substances of various types. The obtained compare favorably with any of the existing macro- Or micromethods based upon similar principles. The method, in addition to the determination of the moleclllar weight, permits repetition as well as simultaneous observation of the boiling and condensing Point O n a single Sample of a few milligrams in weight.

a miniature m a c r o m e t h o d , but may be c o m p a r e d w i t h t h e c l a s s i c a l vapor density m a c r o m e t h o d s of Hofmann (16) and Meyer (22). Mercury, t h e s e a l i n g fluid, is d i s p l a c e d i n s t e a d of air as in the older m e t h o d . Both methods (9, 11, 32) and to a lesser degree their many modif i c a t i o n s (3, 6, 6-8, 10, 15, 18, 20, 21, 23, 30) a s w e l l a s the vapor d e n s i t y method of Dumas (12) and others (31)