V O L U M E 26, N O . 5, M A Y 1 9 5 4 than acet,ate and its subsequent partial hydrolysis during the titration. The method must be considered as unreliable for tertiary alcohols. The high values obtained for p-hydroxybiphenyl when the heating time was 40 minutes indicated incomplete reaction rather than subsequent hvdrolysis, since results in agreement with theory were obtained by longer heating times. I n the case of hydroquinone, however. heating for 105 minutes still resulted in a somewhat high T-alue. This may have been the result of some during the titration as suggested by Rrunner and Thoma.? ( 1J , or more probably it represents incomplete acetylation, eyen with the extended heating period. The results obtained with 2,3-butylene glycol indicated that heating periods of something more than 1 hour were required for complete acetylation of glj-cols. The precision of the method for samples of from 10 to 20 mg. is within &2%, arid the accuracy is of the same magnitude. This method thus allows a relatively accurate estimate of the weight of sample containing one equivalent of hydrosyl. presuming t,hat nitrogen-containing compounds have been shown to be ahsent,, when only ver.r- small amounts of the material under in-
927
vwtigation are available. If the physical properties of the material indicate that only monohydric alcohols can be present, an awurate determination of the amount of priniar:. and secondary alcohols can be obtained by using a 40-minute heating period: if phenolic compounds or glycols are present. a longer heating time will be necessary. The use of the longer heating period as a standard procedure when available information does not indicate the nature of the compounds under investigation does not introduce appreciable error iii the case of the more e:i.*ily nretylated alcoholi. LITERLTVRE CITED
(1) Brunner, H., and Thomas, €1.
R.,.I. 1 p p 1 .
Chetii,
(Londorl). 3,
49 (1953). (2) Kaufmann, H.P., and Funke. S . , Ber.. 70B, 2549 (1937). (3) Smith, D. AI.. and Bryant. TT-. It. L L .I. A m . c ' h r w , S i c . 57, 61 (19353. (4) Webb. A. D., Kepner, R . E.. nitd Ikecln. I:. lf.. As i r . , ~ H E K , 24, 1944 (1952). (5) Kilson, C. E..a n d L u c n s . 11. .J.. .J. 1 , t j . C ' h ? t t z . Soc.. 58, 2401 (1936). R E C E I Y Efor D review S o v e u i b e i 13, l!lS3,
.iccepterl Febri.ury 1, lQ24
Determination of Water in Easily Hydrolyzed Fluorides JOSEPH G.FEIBIG' and JAMES C . WARF2 Institute for Atomic Research, Iowa State College, Amer,
I
S T H E preparation of metals by the metallothermic reduc-
tion of the fluorides, it is of critical interest' t o know the water content of the salts employed. Analytical methods based on thermal evolution of the water are handicapped by the generation of hydrogen fluoride via hydrolysis, which leads to high results and thus precludes the use of siliceous apparatus. These difficulties were circumvented by Sipocz (8) in 18i7, during the course of analysis of fluorine-bearing silicate minerals, by admixture of sodium carbonare with the sample. The classical methods of Gooch (a), employing a platinum apparatus and sodium carbonate t o retain the fluorine, and of Penfield ( 6 ) , employing glass apparatus with calcium, bismuth, or lead oxide as a retainer of the fluorine, have been much used and modified. Jannasch and Xeingarten (3,4 ) employed boras t o retain the fluorine, and, in addit,ion, passed the exhaust gases over heated lead chromate or oxide before collecting. I n the related problem of anal>-tical dehydration of sulfates, Kuzirian (.5) used sodium paratungstate to retain sulfur trioxide. The determination of xater in uranium(I\') fluoride has been suniniarized by Rodden ( 7 ) , and a copper-tube method is recommended in which hydrogen fluoride is reconverted t o water by copper oxide a t 500" C . The present method ( 1 ) has been checked through total analysis of hydrated uranium(I\-) and thorium fluoride?, and has heen found to be satisfactory for routine analyi.is. AILTERIALS AND APP.4RhTUS
A4two-section apparatus was fabricated from standard nickel pipe, the first holding the specimen and the second a material to reconvert hydrogen fluoride to water. The rear section (0.5-inch pipe, 15 cm. long) was connected to the fore section b y tapered threads, fitting into a tapped hole in a disk welded to the end of the fore section (1-inch pipe, 19 cm. long). Each section was heated by a hinged-top resistance furnace equipped x i t h a variable transformer. Thermocouples and a potentiometer n-ere used to measure temperatures. Present address, 3Iartin Engineering Service. St. Louis 1. 1\10. Present address, Drpartment of Chemistry, Cniversity of So:itliern California, Lob Angeles i, Calif. 1
2
IOWJ The entrance to the fore section was closed by a rubber stopper through which the carrier gas (tank nitrogen) was admitted. Thermal protection of the stopper was assured by a Teflon radiation shield held against its inner face, and a water-cooled coil of copper tubing silver-soldered to the end portion of the pipe. Similarly, a 2.5-em. condenserlike cooler, made of copper, -way silver-soldered to the free end of the rear section, to prevent overheating the rubber tubing connection to the collection tower. -1 slow stream of water from a constant-head reservoir flo7ved in series through the coil of the fore tube to the cooler of the rear tube, this arrangement assuring that the latter was never cooled below the dew point of the gas stream within. The rear section was unscren-ed from the fore part and packed with sodium carbonate over most of the length heated by the furnace. It was found necessary to use fused sodiuni carbonate, broken into 1- t o 2-mm. pieces, as the powdered material offered too much resistance to the gas stream. dlt,ernatively, pressed pills of the pori-der, broken into granules of suitable size, were found satisfactory. Llarble chips were also used to convert hydrogen fluoride t o water, but blanks were erratic, and the results were more variable than with sodium carbonate. Copper oxide was satisfactory, except that it \vas effective in removing hydrogen fluoride for a rather short time conipnred with sodium carhonate. PROCEDURE
h slow stream of water is directed through the coolers on the ends of the nickel sections, and the svstem flushed with A4nhydrone-dried nitrogen. The fore sectioi is brought to 500" C. and the sodium carbonate-packed rear section brought to 300" C'. by regulating the variable transformers. The water stream through the coolers is reduced until the temperature of the inlet end of t'he fore section is but a little above room temperature, while the outlet tube leading to the Anhydrone-packed collection tower (Nesbitt type) is at 60" t o 70' C. The nitrogen stream, measured by a flon-meter, is regulated to 40 to 50 ml. per minute and the system flushed an hour or more. Finally a half-hour blank is run, which generally amounts to 0.1 t o 0.2 mg. h sample of the fluoride to be analyzed for water, generally between l and 10 grams, depending on the water content, is weighed out into a nickel boat. The general procedures and precautions of 'combustion analysis are followed-i:e., weighing the collection tower against its counterbalance, inserting the boat with the weighed specimen into the fore tube and closing quickly, allowing adequate time for the reaction and flushing, reweighing the collection tower, and subtracting the blank. Half an hour WLS found to be sufficient time :ifter inserting the hn:it into the tul)e.
ANALYTICAL CHEMISTRY
928 Table I.
Water in Precipitated Thorium and Uranium Fluorides
Fluoride ThF4,nHiO
Sarnple
F,
7
H2O Calcd.,
I
Tli, c; 69.46
22 i 2
ThF, nHxO
B
67 0.5
21 98
10 97
VF, nHtO
-4
70 28
22 39
i.33
I-F,.?iH?O
B
. .
..
CiO
7.82
H20
Found, cO
- -/ . , I
i 79
,..
IO 90 10 95 6.67 68 I 66 7 6.1
6
Difference 0 04 0.05
each prepared by hydrofluorination of the oxide. By the use of large (10-gram) samples in these analyses, satisfactory precision was obtained. The results for the uranium(1V) fluoride samples are believed t o be within a few hundredths of 1% of the true value, while an absolute standard for beryllium fluoride is lacking.
0 66
...
Table IT. Water Contents o€ Conimercial Fluorides Fluoride
Sample
Water Content, % ’
The rear section must, be unscre\\-ed and recharged with fresh sodium carbonate after each 10 t o 50 analrses, depending 011 the moisture content of t,he material analyzed. The first result obtained using fresh sodium carbonate was generally low, and was discarded. LITERATURE CITED
RESULTS
The accuracy of the method was determined hy application to thorium and uraniuni(IT-) fluoride samples whose metal and fluorine contents had already been reliably established through pyrohydrolytic anal~-sie ( 9 ) . The thorium and uranium(1V) fluorides were prepared by precipitation from aqueous solution with h>-drofluoric acid, centrifuging, washing with water, air drying, and grinding. Their water contents were calculated by eubtracting the total metal and fluorine percentages from 100, a procedure satisfactory for thorium fluoride but giving too high a water content for uraniuni(IT7j fluoride, prohahly atti.ilmtahle to partial oxidation during drying. Talde I s h o w that the water contents calculated and found for thorium fluoride agrce within a few hundredths of I%, but the difference in the case of nraniuni(IT-) fluoride is more than 0.5%. Table I1 shows the water contents of various batches of commercial anhydrous uraniuni(IT-) fluoride and beryllium fluoride,
Feibig, J. G., and Warf, J. C., Manhattan Pmject Rept., CC-2939 (June 29, 1946). Gooch, F. .4.,A m . Chem. J . , 2 , 247 (1880). Jannasch, P., and Weingarten, P., Z . anorg. Chem., 8 , 362 (1895). Jannasch. P.. and Veinearten. P.. Ibid.. 11. 37 (1896). Kusirian, S.B., Am. J.-Sci.. 4th Ser., 36, 401 (1913); 2. anorg. Chwn., 85, 127 (1914).
Penfield, S. L., A m . J . Sei., 3rd Sei., 48, 31 (1894); 2. nfzorg. Chem., 7 , 23 (1894).
Rodden, C . J., “Analytical Chemistry,” Natl. Nuclear Energy Series, Di\-. VIII, yo]. I, p. 250, Sew Tork, McGraw-Hill Book Co.. 1960. Sigocs, L., .4nz. A k a d . Wisa. Kim. Math. naturw. KZ.,14, 135 (18iT). Warf, J. C . , Cline, T T . D.. and Tei-ehaugh, R. D., -4w.r.. CIHEY., 26,342 (1954). RECEIVED f o r review Norember 23. 1953. Accepted February 12. 1954. Contribution S o . 249, Institute for Atomic Research and Department of Chemistry, Iowa State College, Ames, Iowa. Work performed under the Manhattan Di-trict of the U. S. Corps of Engineers.
Effect of Side Chain on the Chromatographic Adsorption of Some Ketones on Carbon EDGAR D. SMITH’and ARTHUR L. LEROSEN~ Coates Chemical Laboratory, Louisiana State University, Baton Rouge, La.
I
S C O S S E C T I O S with a fundamental study of the forces involved in chromatographic adsorption ( S ) , it was desirable that a simple method be developed for determining ratio of distance moved by zone to that moved by developing solvent ( I ) of various adeorptives (compound adsorbed) ( 5 ) on colored adsorbent materials. The present work describes a method which has been found to give good results, and presents tentative conclusions reached in a study of the adsorption affinities of some simple ketones on carbon. If a chromatographic tube iq packed so that a short coluiiin of adsorbent A is placed above a longer column of adsorbent B, it should be povible to calculate the R value of a chromatographic zone on .i by noting the apparent change in R value of this zone on 13. Obviously, if the R value determined on adsorbent B in this way is the same as that found by direct measurement, then adsorbent .1 has not slowed the movement of the zone relative to the developing solvent and the R value of the zone on A is 1.00. Any apparent decrease in the rate of zone movement in the second adsorbent, however, would be an indication of adsorption of the adsorptive by adsorbent A. It remained to be proved whether this “sloning” of the ad-orptive could be made the basis for 1 2
Present address, The Chemstrand Corp., Decatur, -41s. Deceased.
calculating I? value^ 011 the upper adsorbent with any reasoliable degree of p i ~ i e i o n . PROCEDURE
Preliminary work was carried out using silicic acid as the upper adsorbent and Florid as the lower one, u.qing a number of typical adsorptives having R values from 1.00 to 0.033. It was soon learned that a certain minimum length of upper adsorbent column was necessary in order to obtain connktent, values for this adsorbent. The minimum lerigt,h increased with increasing R value on the upper adsorbent SO that for R values of about 0.1 to 0.3, a rolunin 10 to 15 mm. in length could be used, while for R values from 0.5 to 1.0, the length of the upper adsorbent column should hc 30 to 40 mm. (The upper adsorbent should be a t least ll/z times as loiig as the adsorbate zone width.) The method of making the requisite calculations can best be made clear hy wing a specific example and carrying out t’he calculations i i i three steps as follows: EXPERIMESTAL DATA
1,ength of upper adsorbent’Ai,20 mni. Length of lower adsorbent B, 60 mm. Distance leading edge of adsorptive has moved into 13, 10 1n1n. R value of adsorptive on adsorbent B, 0.50. Rat,io between the distance a given volume of solvent travels in adsorbent -4to that traveled in B, 1.12. Total distnncc developing solvent haq moved, 80 nun.
