Determination of Fluorine in Quantitative Organic Microanalysis T.
S. M A
Brooklyn College of New York City, Brooklyn 70, N. Y
b Determination of fluorine in organic microanalysis has become increasingly important because of the large number and variety of organic fluoro compounds being synthesized. Because of its peculiar properties, conventional micromethods for halogen determination are not applicable without modification. The principal steps in the determination of fluorine are: decomposition to convert organically bound fluorine to the fluoride ion, separation of the fluoride ion from ions interfering in the subsequent step, and measurement of the fluoride ion. Methods of decomposition include catalytic oxidation, alkali metal or carbonate fusion, reduction with sodium biphenyl, and heating with concentrated sulfuric acid. Fluoride ion has been separated b y the volatility of silicon tetrafluoride, steam distillation of fluorosilicic acid, or ion exchange resin. It has been determined gravimetrically as lead chlorofluoride, volumetrically with sodium hydroxide or thorium nitrate, colorimetrically with ammonium molybdate, and potentiometrically with calcium ions.
A
N UPSURGE in
the preparation of new organic fluorine compounds began during T o r l d War I1 in the search for volatile organometallic compounds, fireproof materials, and poison gas. Fluoro compounds find commercial application as aerosol. refrigerant, and insulator, and in agriculture, plastics, and the dye industry. The recent discovery that certain fluorinated uracils and steroids possess specific physiological activities has stimulated the preparation of a great variety of fluorine derivatives as potential drugs. Judging from the influx of fluorine-containing samples in microanalytical laboratories, probably more fluoro compounds will be submitted for microanalysis than all other halogenatd compounds in the near future. The reason is twofold. First, unlike chloro, bromo, or iodo compounds, most fluoro derix-atives of the multitude of organic substances are unknon n, thus necessitating quantitative analysis for the establishment of the forniula of the nen- compound. Secondly, a new class of organic fluorine compounds-the perfluoro compounds-
has attracted considerable interest in synthetic organic chemistry and has very fen- analogs among the other halogens. The presence of fluorine in the sample may affect the course of microanalysis in several ways. Oxidation of the sample invariably produces hydrogen fluoride, which attacks glass apparatus. Xost fluoro derivatives are more difficult to decompose than the parent compounds. Furthermore, when the organic molecule is fluorinated by three or more fluorine atoms, its boiling point decreases as the molecular weight increases, giving rise t o lowboiling substances. 4 s the conventional microanalytical equipment and methods for carbon, hydrogen, nitrogen, etc., are designed for samples of normal physical properties and chemical reactivity, the microanalyst has found it necessary, in many cases (10, 25, @), to modify the apparatus and technique in order to obtain satisfactory results when dealing with fluorine-containing samples. The most formidable task, however, has been the determination of the element fluorine in the organic compound. No method for fluorine determination is described in current laboratory manuals on quantitative organic microanalysis, and it is apparent that the methods for chlorine, bromine, or iodine are not applicable to the determination of fluorine. Consequently, synthetic organic chemists tended t o ignore the determination of fluorine in the proof of the structures of new fluoro compounds. A survey of the publications on organic fluorine compounds before 1956 reveals that other elements than fluorine FT-ere determined in most fluoro compounds. The situation has improved in the last two years, after reliable methods for the niicrodetermination of fluorine became available. This paper reviews and discusses these methods. METHODS OF DECOMPOSITION
Xierodetermination of organic fluorine compounds usually involves three steps: decomposition of the organic substance, separation of the resulting inorganic fluorine compound, and determination of fluorine by a direct or
indirect method. The cleavage of the organically bound fluorine can be effected by oxidation, reduction, alkaline fusion, or hydrolysis, as shown in Table I. Hubbard and Henne (19) in 1934 discussed the decomposition of the sample over heated silica in a special apparatus. Because the procedure \?as designed for gas samples and required more than 1 mg. of fluorine for a determination, it received little attention in the microanalytical laboratory. The application of catalytic oxidation using the regular quartz microcombustion tube and platinum foil was first reported by Clark in 1951 (8). He recommended a furnace temperature of 900' C., which is satisfactory for mono and difluoro compounds. Later Clark and Rees (10) reported that some polyfluoro ethers, biphenols, benzoic acid, and substances which on oxidation may go through the benzoic acid stage, such as acetophenones, required 1150' C. Freier and coworkers (18) found it necessary to use this temperature for all highly fluorinated substances and to introduce water vapor into the oxygen stream. Belcher ( 3 ) constructed a platinum microcombustion tube which can stand very high temperature but has the drawback of being opaque and pliable. The chief advantage of catalytic oxidation in a microcombustion tube is that the hydrogen fluoride produced may be determined by direct titration with a 0.01S sodium hydroxide solution. HolT-ever, this method is not applicable if the sample also gives rise to other strong acidic oxides, unless these can be removed before the gasps reach the absorbent for hydrogen fluoride. Clark and Rees (10) described methods of retaining the other halogen acids and sulfur oxides in the combustion tube by incorporating a section of silver; but nitrogen oxides cannot be so removed. Hence the procedure described by Kojima, Sagase, and Moramatsu (si?),which entails introduction of ammonia into the combustion train, seems to defeat its purpose. Oxidation by means of the conventional sodium peroxide-sucrose mi\ture was reported by Rush, Cruikshank, and Rhodes (40). The addition of potassium nitrate to the miVOL. 30, NO. 9, SEPTEMBER 1958
* 1557
Table 1.
Methods of Decomposition
Method I. Oxidation methods 1. Combustion in oxygen a. In quartz combustion tube b. In platinum combustion tube 2. Sodium peroxide fusion 11. Reduction methods 1. Reduction in solution a. W'ith sodium in liquid ammonia b. With sodium biphenyl in glycol ether 2. Fusion with alkali metal a. In sealed glass tube b. In metal bomb 111. Alkaline fusion methods 1. Fusion with sodium carbonate 2. Fusion with calcium oxide IV. Hydrolytic methods 1. Heating with sulfuric acid 2. Heating with silica and acid
Resulting Fluorine Compounds
HF, SiFd, HtSiFa HF NaF
KaF NaF
NaF or KF NaF or KF
NaF CaF2
HF SiF4
dizing mixture was advocated by Nichols and Olsen (35),while Eger and Yarden (14) suggested potassium perchlorate. I n the writer's experience, sodium peroxide fusion is the most difficult microdecomposition procedure. The proportion of the sample and reagents has to be carefully controlled; otherwise incomplete oxidation or explosion results. Cleavage of the organically bound fluorine by reduction in solution, if effective, would be a convenient procedure. The use of liquid ammonia as the solvent and sodium as the reducing agent was suggested by Vaughan and Nieuwland (47) in 1931 for semimicrodetermination of fluorine in organic compounds. Attempts to adapt this method t o the micro scale have not been successful. Recently, Benett and Debbrecht ( 7 ) reported that reduction by sodium biphenyl in ethylene glycol dimethyl ether is generally applicable to fluoro compounds, and that the decomposition reaction is complete in a few minutes. The detailed procedure has not been published. Reduction by fusion with an alkali metal is a reliable method for the quantitative decomposition of organic fluorine compounds. It is also the most
1558
ANALYTICAL CHEMISTRY
generally used method reported in the literature. Fusion in a glass tube on the macro scale was described by Elving and Ligett using sodium metal (16) and by Silvey and Cady using potassium ( 4 2 ) . Kainz and Scholler ($1) adapted the procedure to microanalysis. The drawback of the sealed glass tube technique is the requirement of evacuation and glass blowing facility. Therefore the use of the metal bomb is preferable. Belcher and coworkers (4) constructed a nickel bomb for this purpose and heated the bomb containing the sample and sodium or potassium a t 600" C. for 1 hour. Savchenko (41) reported that heating a t this temperature gave low results and recommended 900" C. instead. A steel bomb was described by Korshun, Klimova, and Chumachenko (23),who suggested heating at 800" C. for 10 minutes. Working with the regular Parr microbomb, hIa and Gwirtsman (66) found that it served well for the alkali metal fusion of fluorine compounds, if the usual rubber or lead gasket was replaced by one made of copper, which is noF commercially available. Metallic sodium is an effective reagent for the decomposition of mono and difluoro compounds, while potassium is recommended for highly fluorinated substances. Because potassium is much more dangerous to handle than sodium when exposed to air and moisture, the use of a glass capillary containing potassium is recommended (25). The fusion reaction may be carried out by means of the Bunsen burner and it is not necessary t o heat the microbomb for longer than 15 minutes. Alkaline fusion methods employing sodium carbonate, calcium oxide, or calcium hydroxide (37) do not always give quantitative yields of the corresponding inorganic fluoride even for monofluoro compounds. Hydrolytic methods are of limited application, being useful only for compounds such as tertiary fluorides which liberate hydrogen fluoride on heating in an acid solution. REMOVAL OF INTERFERING SUBSTANCES
Unlike the microdetermination of the other members of the halogen family, quantitative organic microdetermination of fluorine seldom can proceed to the final stage of determination immediately after the decomposition process. This is because of the interference due to the presence of inorganic compounds of nitrogen, sulfur, chlorine, bromine, iodine, phosphorus, and arsenic, which are produced from the organic sample containing these elements upon oxidation or reduction. The separation of substances which interfere with fluorine determination can be approached from two directions:
On the one hand, the interfering ions may be removed through the formation of thermally stable compounds or by precipitation with a suitable reagent or expelled by volatilization, keeping the fluoride ions in the solution. On the other hand, the separation may be based on the steam distillation of fluosilicic acid, leaving the foreign matter behind in the reaction mixture. Clark and Rees (10) investigated the removal of sulfur and other halogens by retaining them as silver sulfate and halides, respectively, in the microcombustion tube and recommended the following procedure. If other halogens but no sulfur are present, the silver should be located in the auxiliary furnace section maintained at 400' to 425" C. If there are no other halogens but there is sulfur, the silver should be in the same position but maintained between 500" and 550" C. If other halogens and sulfur both are present, the silver must be in the principal furnace section, replacing the platinum contact, and the furnace must not be heated to more than 900" C. Belcher and coworkers (4) reported that attempts t o remove cyanide and sulfide ions from the fluoride solution by the precipitation of silver cyanide and sulfide, respectively, were not satisfactory because the precipitates were too fine to filter. These workers claimed that hydrogen cyanide and hydrogen sulfide could be expelled by controlled heating of the acidified solution containing fluoride ions without affecting the latter. Sulfide could be separated by precipitation using zinc oxide, according to Belcher and Macdonald (6), while phosphate and arsenate ions were removed by the addition of solid zinc carbonate ( 5 ) . Eger and Yarden (14) applied the ion exchange technique for fluorine determination. The solution obtained after sodium peroxide fusion was passed through a column containing Amberlite IR-112, and the hydrofluoric acid was recovered in the effluent. Separation of fluoride ions by steam distillation of fluosilicic acid is the preferred micromethod, because this technique permits the removal of practically all interfering substances and is relatively easy to control. The distillation procedure was originally suggested for the macrodetermination of fluoride by Willard and Winter (49), who established the optimum condition for the volatilization of fluosilicic acid a t a temperature of about 135O C. in the solution. To prepare an aqueous solution which boils a t this temperature, the general practice is to add perchloric acid to the solution containing the fluoride ions, though sulfuric acid has also been employed (44). Murty, Viswanathan, and Ramakrishna (54) claimed that recovery of fluoride
ions was more rapid and complete when sulfuric or phosphoric acid was used instead of perchloric acid. Several distillation apparatus for the microdetermination of fluoride have been reported (1, 90). The apparatus described by M a and Gwirtsman (26) appears to have the most compact design, and it gives blank values of less than 2 y of fluorine. A study was made to ascertain the minimum volume of distillate to be collected for accurate microdetermination of fluoride (97). The results, shown in Table 11, indicate that it is not necessary to prolong the distillation beyond 150 ml. of distillate, when the determination is for the purpose of identifying a new organic fluorine compound. The steam distillation of fluosilicic acid from a perchloric acid solution eliminates the interference of sulfate, phosphate, and arsenate ions because the corresponding acids are not volatile under this condition. If chloride, bromide, iodide, and sulfide ions are present, addition of silver perchlorate to the solution containing the fluoride sample removes the interfering substances by the precipitation of the corresponding silver salts. The writer also obtained satisfactory results for fluorine in nitrogen-containing compounds. However, according to Steyermark (46), cyanide ions are not completely removed by silver perchlorate, in spite of the extremely low solubility of silver cyanide in aqueous solutions. Steyermark claimed that hydrogen cyanide was found in the distillate when the organic sample contained nitrcgen, and he recommended the removal of cyanide ions by heating the solution with hydrogen peroxide prior to distillation. Hydrogen peroxide was tested by Belcher and coworkers (4) as a reagent to decompose cyanide ions, and they reported that it also oxidized the ethanol which was used to destroy the excess alkali metal after fusion, and the acetic acid produced affected the determination of fluoride ions. METHODS OF DETERMINATION
After the conversion of the organically bound fluorine to hydrogen fluoride, sodium or potassium fluoride, or silicon tetrafluoride, followed by the separation of interfering substances if necessary, the final step of fluorine determination involves the quantitative analysis of the resulting compounds in amounts corresponding to organic microanalysis, using approximately a 1to 10-mg. sample. Xumerous methods for determining fluorine and fluoride have been published, and comprehensive reviews were written by McKenna (89) and Elving, Horton, and Willard (16). These methods practically cover all techniques used in quantitative
Table 11.
Volume of Distillate and Recovery of Fluoride Volume
FIuoof ride DistilAdded, late, y
M1.
500
250 150
250
250 150
100
250 150 100
100
100
Fluorine Found 7 0 recovery 493 98.6 482.5 96.5 475 95.0 248 99.2 244.5 97.8 240 96.0 100.5 100.5 99.0 99.0 97.5 97.5 y
analysis, but most of them are not suitable for the microdetermination of fluorine in organic compounds. The term “micro” appears to have different connotations in the minds of authors of papers on fluorine determinationseveral micrograms, parts per million, or over 5 mg. of fluorine in the sample taken for a determination. For organic microanalysis, the procedure for determining fluorine should be concerned with about 150 y to 2.5 mg. of the element in a volume of 20 to 200 ml. of the final solution. Methods which deal with very small amounts of fluorine entail difficulty in weighing out the sample, or introduce errors if aliquots are taken. Procedures which require more than 2.5 mg. of fluorine may lead to low results due to incomplete decomposition of the organic compound in the microapparatus. Thorium Nitrate Titration of Fluoride. This is the most common procedure used for determination of fluorine in organic compounds. It is based on the reaction between thorium nitrate and fluoride ions under controlled p H conditions in the presence of sodium alizarin sulfonate. Thorium fluoride is formed, which prevents production of the pink colored lake containing thorium alizarin sulfonate, unless excess thorium ions are present. The optimum condition for this reaction is a t a p H of 3.0 and the acidity of the fluoride solution is usually adjusted by means of hydrochloric acid (18), but acetic acid (SO), monochloroacetic acid, and other organic acids (38) have been used. Other indicators besides the commonly accepted sodium alizarin sulfonate were extensively investigated by Willard and Horton (48). The reaction is not stoichiometric. If one can control the amount of fluoride ions to be determined !Tithin narrow limits, it is possible to employ the regular technique of titrimetry by running the titrant into the fluoride solution and watching the end point, after the experimental condition has been standardized. Hom-ever, in quan-
titative organic microanalysis, samples with a wide range of fluorine contentsmostly unknown-are encountered, and it is difficult to estimate accurately the amount of fluoride produced. Therefore the titration is preferably carried out in two Nessler tubes (12, 96), one containing the sample and the other a blank solution a t the same pH. Thorium nitrate solution is added to the sample until a pink color is observed. Exactly the same volume of thorium nitrate solution is added to the other Nessler tube, follom-ed by the standard sodium fluoride solution to match the color. Thus the quantity of fluorine in the sample is given by the amount of standard sodium fluoride solution delivered to the blank. When 0.05N thorium nitrate is used-for a solution containing about 2 mg. of fluorine-the titration may be performed under ordinary light. However, a fluorescent lamp is recommended, and it is a necessity when working with the lower limits of fluorine determination. A permanent color standard for the titration was proposed by Smith and Gardner (43). Mavrodineanu and Gwirtsman (31) described a photoelectric filter photometer which permits the direct titration of fluoride ions with standard thorium nitrate solution. The end point is detected by the disturbance of a beam of light when thorium nitrate is added in excess to the fluoride solution, producing the thorium lake of alizarin sulfonate. The amount of fluorine present is read from a calibration curve prepared by titrating known quantities of fluoride with the same thorium nitrate solution. This curve is linear up to 300 y of fluorine, and hence serves well for routine microdetermination of organic fluorine compounds. A differential spectrophotometric procedure, described by Lothe (94),involves comparison of the color of the thorium lake, bleached by fluoride ions, against a set of three reference standards containing 50, 100, and 200 y of fluorine per 50 ml. A precision and accuracy to 1% were reported, and the p H of the solution was found to be critical. Other Titrimetric Methods. Savchenko (41) titrated t h e fluoride solution with zirconium chloride, while Kojima and coworkers (22) used aluminum chloride. These reagents do not appear to have advantages over thorium nitrate. Debal, Levy, and Moureu ( I S ) proposed an alkalimetric method for the microdetermination of fluoride, based on the observation that the acidity of the solution is reduced when potassium fluoride reacts with silicon dioxide to produce fluosilicate ions. However, the reaction is not stoichiometric and a correction factor is necessary. Titration of fluoride ions with standard calcium solution using the VOL. 30, NO. 9, SEPTEMBER 1958
1559
bismuth-bismuth fluoride electrode system in acetone-water solvent was reported by Benett and Debbrecht (7). The use of zirconium electrode for fluoride determination was suggested by Megregian (38). It was found that the current generated by the spontaneous electrolysis of the zirconium-platinum cell is proportional to the fluoride ion concentration between 0.25 and 2.0 mg. of fluorine per 100 ml. Alkalimetric Determination of Hydrogen Fluoride. When fluorine is converted to hydrogen fluoride, a simple procedure is t o absorb the gas in solution and titrate the latter with 0.01N sodium hydroxide, if no other strong acids are produced and pass into the absorbent a t the same time. The standard alkali solution should be stored in a paraffin-coated bottle or polyethylene container. According to Clark (9), when barium chloride is incorporated in the standard alkali solution, the titration gives a sharp end point if phenolphthalein is used as the indicator, because it involves the neutralization of hydrochloric acid produced by the precipitation of barium fluoride from barium chloride. Potentiometric titration with 0.01N sodium hydroxide using a p H meter having a glass-calomel electrode system was reported by Clark and Rees (IO). Gravimetric Determination of Fluoride. T h e gravimetric method has some advantages for occasional microdeterminations. Precipitation of calcium fluoride is t h e conventional gravimetric method for determining fluorine in the macro scale, and was used by Ruff (39) for the analysis of organic fluorine compounds. This procedure cannot be adapted t o microanalysis because calcium fluoride is slightly soluble in water and the precipitate is difficult to filter. Triphenyl stannous chloride was recently suggested by Ballczo and Schiffner ( 2 ) as a new reagent. However, as the procedure prescribes the addition of a chloroform solution of the reagent to the fluoride ions in aqueous medium, the precipitation takes place in a two-phase mixture, which is troublesome. I n the writer’s experience (28) 0.5 to 1 mg. of fluoride can be accurately precipitated and weighed as lead chlorofluoridc, using a mixture containing sodium chloride and lead nitrate as the precipitant. This method was also used by liazor (32). Because the composition of the precipitate is somewhat varied, it is necessary t o adhere to the established experimental conditions and it is advisable to perform a parallel determination with a solution containing a known amount of sodium fluoride. The presence of sulfur in the organic sample interferes in this method, because lead sulfide or sulfate
1560
ANALYTICAL CHEMISTRY
precipitates m-ith the chlorofluoride. Steam distillation to separate the interfering ions is not recommended in this case, as it entails evaporation of the distillate to a small volume suitable for the quantitative precipitation of lead chlorofluoride. A more convenient way is to convert all sulfur present t o sulfate ions and precipitate lead sulfate and chlorofluoride simultaneously. The supernatant liquid is removed by inverted filtration. The lead chlorofluoride is then dissolved in nitric acid and chlorine is determined by means of the Volhard titration. Determination of Silicon Tetrafluoride. Fichera (17) used a n alkali solution t o collect the silicon tetrafluoride obtained b y heating t h e fluoride with silicon dioxide in sulfuric acid. T h e excess alkali was then titrated with t h e standard acid solution t o a phenolphthalein end point. Sulfuric acid t h a t was carried over was removed by precipitating it with barium chloride. Peregud and Boikina (36) determined silicon, instead of fluorine, after the silicon tetrafluoride was absorbed in mater. Ammonium molybdate in sulfuric acid was added to the resulting solution, followed by tartaric acid and ascorbic acid, and the blue color produced was measured against a series of standards prepared from sodium silicate. Curry and Xellon (11) recentlv proposed a method which involves hydrolysis of silicon tetrafluoride in a sodium borate-boric acid buffer and spectrophotometric determination of the silicon through molybdosilicic acid and a heteropoly-blue. LITERATURE CITED
(1) Ballczo, H., Kaufmann, O., X i k r o chemie uer. Mikrochim. Acta 38, 237 (1951). (2) Ballczo, H., Schiffner, H., 2. anal. Chem. 152, 1 (1956). (3) Belcher, R., private communication. (4) Belcher, R., Caldee, E. F., Clark, S. J., hlacdonald, A. 11.G., Mikrochim. Acta 1953, 283. (5) Belcher, R., hlacdonald, A. hi. G., Ibid.. 1956.899.
(6) Ibid., 1957, 510. (7) Benett, C: E., Debbrecht,, E. J., 131st AIeeting, ACS, Miami, Fla., April 1957, Abstracts, p. 24B. (8) Clark, H. S., AKAL. CHEU. 23, 659 (1951). (9) Clark, H. S.,private communication. (10) Clark, H. S., Rees, 0. K., Illinois State Geol. Surveir, Reat. Invest. 169 (1954). (11) Curry, R. P., Mellon, LI. G., -4x.4~. CHEX 29, 1632 (1957). i 121 Dahle, D., Bonner, R. V., Wichmann, \--, H I., J. Assoc. Ofic. A y r . Cheniists zi, 459 (1938). (13) Debal, E., Levy, R., lfoureu, H., Mtkrochani. Acta 1957, 396. (14) Eger, C., Tarden, A,, AS.& CHEX 28, 512 (1956). (15) Elving, P. J., Horton, C. A,, R’illard, H. H., in “Fluorine Chemistr; J. ,I’
H.. Simons ed.. Vol. 2. D. 51. Academic Priss, New Yoik, 1954.’ ’ (16) Elving, P. J., Ligett, FV. B., MD. ESG. CHEM.,-43.4~.ED. 14,449 (1942). (17). Fichera,, G., Boll. sedute accad. uioenza sci. nat. Catania [41 1, Ei99 (1952). (18) Freier, H. E., Sippoldt, B. K., Olsen. P. B.. Veiblen. D. G.. -4s.~~.
’ Acta 32, 145 (1944). 121) Kainz. G.. Scholler. F., Mikrochim. ‘ Acta 1956. 843. (22j- Kojima, R., Nagase, S., Moramatsu, H., J a p a n Analyst 4, 518 (1955). (23) Korshun, 11. O., Klimova, V. A., Chumachenko, 111. S . , Zhur. -4nal. Khzrn. 10, 358 (1955). ’ (24) Lothe, J. J., ANAL. CHERI. 28,
949 (1956). (25) Ma, T. S. Xicrochem. J., in press. (26) Ma, T. d., Gwirtsman, J., ANAL. CHEX.29, 140 (1957). (27) hla, T. S., Gwirtsman, J., unpublished work. (28) Ma, T. S., Mangravite, R., unpublished work. (29) McKenna, F. E., 12’ucleonics 8, 24, 9, 40 (1951). (30) hfatuszak, hf. P., Brown, D. R., IND.ENG.CHEM.,ANAL. ED. 17, 100 (1945). (31) Mavrodineanu, R., Gwirtsman, J., Contrib. Boyce 181 (1955).
Thompson Inst.
18,
(32) Mazor, L., Mzkrochim. Acta 1957, 114. (33) Megregian, S., AXAL. CHEM. 29, 1063 (1957). (34) Murty, G. V. L. N., Viswanathan, T. S., Ramakrishna, V., AnaLChim. Acta 16, 213 (1957). (35) Kichols, M. L., Olsen, J. S., I K D ~ ENG.CHEY.,ANAL.ED. 15, 342 (1943). (36) Peregud, E. A,, Boikina, B. S., Zavodskaya Lab. 22, 287 (1956). (37) Roth, H., in Houben-PVeyl-?v.Iuller, “hlethoden der organischen Chemie,” 4th ed., Vol. 2, “Analytischen Methoden,” p. 137, Thieme, Stuttgart, 1953. (38) Rowley, R. J., Churchill, H. V., ISD. ESG. CHEM.,ANAL. ED. 9, 551 (1937). (39) Ruff, O., Ber. deut. chem. Ges. 69, 299 (1936). (40) Rush, C. A., Cruikshank, S. S., Rhodes, E. H. J., Mikrochim. Acta 1956, 856. (41) Savchenko, A. Y., Zhur. Anal. Khzrrc. 10, 355 (1955). (42) Silvey, G. A., Cady, G. H., J . Am. Chem. SOC.72, 3624 (1950). (43) Smith, F. -4.,Gardner, D.E., A m . lnd. Hug. dssoc. Quart. 16, 215 (1955). (44) Smith, 0. D., Parks, T. D., A N A L , CHEJI. 27, 998 (1955). (45) Steyermark, A., private communication. (46) Throckmorton, \V. H., Hutton, G. H., AXAL.CHEJI.24, 2003 (1952). (47) Vaughan, T. H., Nieuwland, J. A., ISD. EKG.CHERI.,A K ~ L .ED. 3, 274 (1931). (48) Killard, H. H., Horton, C. A,, A4S.41..CHEX. 22, 1190 (1950). (49) Killard, H. H., Winter, 0 . B., Ibid., 5, 7 (1933). RECEIVED for review February 7, 1958. Sccepted June 9, 1958. Division of Analytical Chemistrv, Symposium on Microchemistry, l32”nd Meeting, ACS, N e w Tork, 5 , T., September 1957.
