rate can happen with a slag, consideriiig that its type as silicate is determined beforehand in tlie charges of a blast furnace, by furnisliing to the silica of ores, ston(J, a n d asli of coal, perfectly definite basic elements, in certain proportions primarily calci11at~?d, w 11i c 11 n ecessa r i I y d et e r m i 1 i e ii co i n po si ti 01i w 11 ic 11 t 11e u I ti mat e analysis of tlie s h g , as i - t i n from tlir. furnace, corroborates very closely in the great majorit1 of casr’s. \ V h r r i it does not hsppeti, it I M also to be attribute(1 t o clistur~)iiigfsctoi,s. :init tiley can be fountl, for instance, in this fact tiisit calciiitn, combined w i t h sulpliur, which the slags oftc>n wntitiii i i i very st’rious qaantities. has calculated :is oxide of ( ! ~ l ~ : i u i ioi r / i / w d , ‘L‘lic s:me ciin be said of magiiesiuni aiitl mangibtiesc. ci$ of sulpliur in a slag is by 110 metins :LII rsceptioiiiil circornstance ; S$ of inanganesc sulpliidt: is frequently inct with i i i Scotch slags. 111 otlier cases aluniiuii, owing t o certain condition of the furnace, or the proportions of flnses, w i l l play the par’t of a n acid, and, in these conditions, ouglit not to figure i n thc determinations of the type tis saturating silica. I x A e n
( T o De cvntinzcetl.)
T ) ~ ~ ’ l ’ E I ~ M I ~ A TOF IOK LI’l’HIA IS M I N E R A L WATERS. BY
E.
\vAI,I.EH.
1”.
1).
Practically, three methods are now available. 1. T h e phosphate method (Mayer’s modification) ( A t ~ i t C . h z . Q. Pharnz. 98, 193). 2 . T h c amylalcohol method (Gooch, Am. C‘hev). Jour., 9, 33). 3. T h e fluoride method ( C a ~ m tBur’/. , Soc. Chin). [3] 1, 280). Kammelsberg’s method (Poyg. Ann., 66, 79) somewhat similar in principle to that of (:oocli, in that it depends upon the comparatively greater solubility of lithium chloride in a n organic solvent, has been comparatively little used, on account of t h e difficulty and expense involved in obtaining the pure anhydrous alcohol and ether necessary for the process. Moreover t h e experiments of J. L. Smith ( A n , . Jour. Sci. [ 2 ] 16, 56), rearranged in convenient form for reference by Gooch (loc. c i f . ) do not indicate t h a t it is
DETERMIXATION O F LITHIA I N MINERAL WATERS.
215
very satisfactory in its application, even with the best of care. Some indirect processes, such as t h e weighing of mixed chlorides of sodiom, potassium a n d lithium, a n d then determining the chlorine a n d potassium (Bunsen, Aim. Cl~eni.u.Plmmz., 122, 348), have been also proposed, b u t they are troublesome in execution, and likely to be unsatisfactory in result. F o r all of these processes, it is necessary to obtain from some known quantity of the water, t h e alkalies as chlorides free from admixt u r e with other bases, and in most cases, a considerable proportion of the sodium a n d potassium salts, which usually rredominate largely over those of lithium, m u s t be removed. To accomplish this the usual method may be followed, acidification with hydrochloric acid, evaporation, treatment with barium hydrate solution, removal of t h e excess of baryta by ammonium a r b o n a t e , driving off tho ammonium salts, a n d extraction with alcohol or nlcohol and ether, to take o u t t h e lithium chloride which is inevitably i~cconipaniedby some sodium and potassium chlorides. Throughout this treatment, t h e spectroscope must constantly be used t o determine when t h e extraction or wasliing is complete and these preliminary operations often prove very tedious. Some suggestions in this connectioii may bo of value. T h e small platinum wires usc~lto test the precipitates, solutions, etc., need critical examination. A wire which has been once used with litliinm salts may perhaps be held in t h e flame until i t will give no trace of color to t h e flame, nor show the lithium line by the spectroscope, but on moistening wita hydrochloric acid a n d inserting in the flame, the line will sliow almost as brightly a n d distinctly as if no lithium had been removed from it. Repeated scouring, immersion in acid and insertion i n the flame, or long soaking in acid may be necessary to remove this trace of lithium from the wires. It has been found convenient to keep several wires dipping into a test tube partially filled with dilute hydrochloric acid, a n d to use them in successioii, so that each wire shall have a tolerably long immersion in t h e acid, before being tested again, as a preliminary to using it for tt test on a precipitate, etc. Barium precipitates (Ba CO, and Ba SO,) when formed in t h e presence of lithium compounds, carry clown and retain perceptible
D E T E R M I N A T I O N OF LITHIA I N M I F E R A L WATERS.
217
and water, and finally ignite and weigh as Li, PO,. A second or third portion of precipitate may be recovered, by evaporating the filtrate a n d washings, adding ammonia and allowing to stand as before, filtering, etc. The chief difficulty with the accuracy of the process, consists in the practical impossibility of obtaining all the lithium as phosphate, free from any other alkaline salts. Too much washing will cause appreciable amounts of lithium phosphate to go into solution. Indeed, in my experience the filtrate and washings have always showed a decided lithia line in the spectroscope, from the start. On the other hand, too little washing leaves some alkaline salt along with the lithium phosphate-shown by its tendency to cake on ignition-but whether it cakes or not, lithium phosphate separated by this method when tested by t h e flame, almost invariably gives so strong a sodium flame, as practically to obscure t h e red of the lithium to the naked eye. Consequently i t becomes to a considerable extent a matter of judgment, to decide when the washing is completed, and theit the amount of lithium phosphate obtained is a compromise between the precipitate dissolved off by washing, and alkaline salts left with it. However, by the aid of a little experience, the error can be usually brought within moderat,e limits, if the proportion of water which the chlorides represent is sufficiently large (ordinarily 10 t o 20 litres). The use of such large quantities of water is naturally attended with more or less labor, in evaporation, removal of bases, etc., and is in itself objectiontible aside from the sources of error inherent in the method of determinntion. This method has, however, until recently been practically the only one in general use. The method of Gooch (Zoc. cit.) used by him in the examination of the waters of the Yellowstone Park (BzdZ. li. S. Qeol. #zit-vey, NO. 4'7, also Chem. ..Ve?us, 59, 113, e t may be described as follows : The concentrated solution of the alkaline chlorides should contain only about 0.3 grm. of salts in all. To this solution, in a casserole or dish; YS added 30 to 50 c. c. of pure nnhyd7'07LS amyl alcohol. The vessel is then heated on a sand bath over a low flame, so as to boil off the water through the amyl alcohol, leaving t h e undissolved salts adhering to the sides of the dish. The heat is kept u p until the volume of amyl alcohol has
a$.),
218
DETER~XIlIhTIOSOF LITHIA I N MINER11; IVATERS.
been reduce1 to about 15 or 18 e. e. after cooling. A few drops of hydrocliloric acid are added to restore to the form of c,hloride any lithium oxide or 1iydr;ite wliicli may have been formed, and the lieiiting is rcpe:ited for ii sliort time. The amy1 nlcohol is then filtered through pa1)er, or through a Gooch crucible. into it measuring syliiider, and its volume notcd (ustially 10 to 1.j e. e.). 111 case t h e proportion of lithium is large t h e undissolved salts sliould be taken up with a little water, and the tre;ttment repeated in tlie sitme W:LSas jiist desc*ribed, the amount of amj1 nlwliol whirl1 has been heated with the caliloridcs beiiig measured as before. The salts are then wasliecl with cold amyl alcoIiu1 until 110 trace of lithiiini is perceiitible i n them by tlie spectroscope ; the filtrate and wasliiugs are crupor:~teil in it weighed p1:ttinnm dish, m d the chlorides ronvcrtcil i n t o salphates, ignited and weighed. From this weight, for el-ery 10 i. L'. of :tm!l alcohol which remained in contact with tlie chlorides after IieiLtiiig, the following dcductiun is made : \Vlien only sodiurii and lithiuni chlorides were present, 0.00050 grni. " 0.00059 " potnssiuni " Wlien both sodium a n d potassium, ns well as lithiuni 0.00109 i i chlorides were present, * (
'(
The cold amyl alcohol used for mashing dissolves SO little t l n t it is iiwxlless to take i t into account. T h e relative solubilities of K a C1, KCl and Li @1 in amy1 alcoliol, its determined by (;ooch, are essentially : . ....... . ...- - l in 30,000, or 0.0041 grm. i n 100 c. c. K Cl _ . . _ . . . . _ _ .1. _ - '_24,000. _ _ . or 0.0051 " 100 " Li C1 ... . _ _ _ _ _... ._ . . -1 _ or 10 c.c.) and pure ammonium fluoride :tnd ammonia added ; after tliorough mixing tlie d i s h is set wide oyer night for tlic Li F to piwipitatc,. The solution is t h e n clccmtetl through a small tiltvi'. a i i d t h e pi'ecipitute is ~v~tslictl by decantation three or foiir tiniw j 2 t o ;' V.C. :it it time), with a. solutiou consisting of tlie w a g e t i t niiscd w i t l i to 10 times its bulk of nmmonia ; betweell decaritntioiis tlit: solution must be allo~vctlto stand some littlr' tiiiic with stirring. Tlie bulk of 61tr:ttc:tncl wishiiig (30 to 5 0 2 . c . ) is iiotetl, the filter paper :~iidcontents placed i i i tlic dish, sulpl~iii*ir;acid itddeil, a i ~ dlieat itpplietl nntil t h e paper has been iiiciiieratcd, and t h e lithium converted to sulphatc, iii which form it is weighed. To this weight is added 0.'0040 g r m . for cvory ; C.C. of filtrate and wasliiiigs, and the result estimate$ as Li, SO, is calculated to Li, Li HC03, etc., according to the requiremeii ts of t h e case. C'n1.e i, necessary i n preparing tlic reagent, and wash liquor. Carnot seems to have found that the ammonium fiuoride ordinarily supplied for laboratory iise, is the only member of the combination liable to contain impurities which would interfere (chiefly fluosilicic acid, which migb t' precipitate alkaline fluosilicates), but
DETERMINATION O F LITHIA I N MINERAL WATERS.
221
experiments have shown that ammonia, which has been standing for some time in contact with glass, will give a cloud (presumably ammonium fluosilicate) with a mixture of solutions of amnionium fluoride and ammonia after boiling and filtering clear. This solution, so long as it contains a fair amount of free ammonia, appears to be without action upon glass. It has been found advisalile therefore to make u p (and cork up) the reagent and washing solution some time beforehand, and to filter off such poi*tioiis as may be required at the time of using. Saturally, it is necessary to use for the final filtration, R filter paper which has been extracted with hydrofluoric acid. Sclileicher tk Schull’s papers mere found satisfactory in this connection. Carnot also recoinmends that the resulting Li, SO, should be dissolved i n 40 to 50 C.C. of water, and a test made for the presence of magnesium, which may have remained with the alkaline chlorides. If any IS found to be present, i t must be determined as phosphate and a correction made accordingly. I n connection with this process it was observed t h a t lithium sulphate ignited in contact with the carbon of the filter paper, is especially prone to reduce t o sulphide and especial caution is necessary a t this stage of t h e operation. T h e sulphide, when heated in contact wit11 the platinum, attacks i t in a very marked manner. T h e process seems to be very good, althougli not rapid. I t s tendency is to yield results a little high, apparently because t h e allowance for solubility is u s n d l y larger than the actual amounts of precipitate dissolved. Test analysis tended to show also t h a t unlcss the amounts of potassium and sodium chlorides present with the lithia are kept within narrow limits, the results mill be high. Unfortunately a number of the tests and comparison experiments with these methods have not yet been completed, and will have to be deferred to a sccond communication. I order to test these methods iipon water containing lithia, samples of several of the best liriowii and widely advertised waters were purchased and submitted to examination. T h e results were somewhat surprising, and showed unquestionably t h a t either the original an~lyses,on t h e strength of which thobe waters are now sold, were erroneous, on account of imper-
fcctioii in tke metlio~lsused, or: what is.niore pi~obal)1etthat tlie proportions of lithiuni i r i tliose witvrs art? liable to gretit finctuation>.. 'l'lic 1-esults gircii were cliiefly oi~taincclIJJ ('iiriiot's fluoride mctliod, but were i l l several citscs coiifirincd by tile nsc' ilt' other m c t i i d s . T h e m o s t scru1)iilons ciirc' w i i ~exer obt;~iiiingall of tlie litliiorn iri t l i c t ivibtk'm iiiider ~!sitniiii:itioii,tlic specti,oscopic indiciitions liavii~gbceu nscd a t cvcry stag14 of tlie pl'occss. 111 the FarmvilIc. Litlii~t \\'atclr, l~iircliased: I t t h e 0ffic.e i J f the cornijiLiiy, n o lithium could be cletecteil by t h e sl)ectroscope on moilerate amuuiits of the water. On evaporating eight litrcs of tlie water, and treating in the manner described for the c:oiice~itriitivii of the lithia into R sulutioii of smtLll biilli, ii 1itlii:i line x i s obtitiiird in the spcctroscopc, b u t t h e amount wiis foiiiicl to he too small to permit of il quantitative estimation. The esp(~riiiic.iit\vas repwtcd with ten litrcs of tlie wntei., with csselitially the same result. \\-it11 tlie Uuffaio Ilitliia IVater the reaction f
