A Fusion Method for the Production of Sodium Fluoride Citation Ranney, Kenneth Wyckoff (1925) A Fusion Method for the Production of Sodium Fluoride. Bachelor's thesis, California Institute of Technology. doi:10.7907/3vz0-2w58. https://resolver.caltech.edu/CaltechTHESIS:08072024-161746447 Abstract No abstract. Item Type: Thesis (Bachelor's thesis) Subject Keywords: (Chemical Engineering) Degree Grantor: California Institute of Technology Division: Chemistry and Chemical Engineering Major Option: Chemical Engineering Thesis Availability: Public (worldwide access) Research Advisor(s): Lacey, William Noble Thesis Committee: Unknown, Unknown Defense Date: 1 January 1925 Record Number: CaltechTHESIS:08072024-161746447 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:08072024-161746447 DOI: 10.7907/3vz0-2w58 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 16616 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 07 Aug 2024 23:54 Last Modified: 08 Aug 2024 00:03 Thesis Files PDF - Final Version See Usage Policy. 12MB Repository Staff Only: item control page

A FUSION MET"tlOD FOR THE PRODUCTION OF SODIUM FLUORIDE Theei• by Kenneth W. Ranney In partial fulfillment of the requirement• of the Degree of :Bachelor of Science in Chemical Engineering. C.ALIFORNI.A INSTITUTE OF TECHNOLOGY Pasadena, California 1925 A FUSION METHOD FOR THE PRODUCTION OF SODIUM FLUORIDE The ultimate object of this research is to develop a commercial method of converting calcium fluoride (f'luorapa.r) into sodium fluoride (NaF), which is very valuable as an insecticide and as a preserv- ative for wood, especially railroad ties 1 and telephone and telegraph poles. The chief competitors of sodium fluoride as a wood preserva.- tive today are zinc chloride (ZnCl 2 ) and creosote, but as creosote has the advantage over zinc chloride in that it resists being leached out, neither zinc chloride nor sodium fluoride can serve as a substitute for creosote in very moist sections of the country, although they a.re considerably chep.per 2 •. Zinc chloride ia, however, being used very extensively to treat ties in the drier sections of the country. For example, $2, ~0,000 worth of ZnC1 2 was produced in the United States in 1920 and. over 95% of this was used to treat railroad tie•. This shows that if NaF was able to compete with the ZnC1 2 there would be a large market for it. The following table from Chem. and Met. Eng. (~, 1124, 1920) ie good proof that NaF ie as good a wood preservative as ZnC1 2 , if not better. lRy. Maintenance Eng., 16 (142-4), 1920. ' ~Ry. Maintenance Eng., 16 (338-9), 1920. -2- COMPA.RISON OF ZINC CHLORIDE nm SODillM FLUORIDE AS W'OOD PRESERVATIVE NaF Solubility in water 00 Toxicity EnoUih to materially increase the life of the wood. Over twice as toxic as ZnC1 2 • Corrosive effect on steel Somewhat corrosive, but is conmonly used in steel apparatus. Much less corrosive Effect on , paint Cannot be successfully painted over. Toes not injure or Convenience of ehippir.g Shipped in 50% solution or solid; but ao11d form is deliquesent and must be sealed in air-tight Shipped in dry powder in slack cooperage. than ZnCl 2 . effect paint. containers. .15 cts. per lb • Cost, 1920 Cost , Apr. 1925 • OS½ ct8. per lb. As indicated in the above table the price of NaF is almost as cheap as ZnC1 2 and if tha price of NaF could be reduced by even one cent per lb, it could compete with ZnC1 2 at the present price,. T'.ais is assuming that NaF is as good . as ZnC1 2 and the table indicates that NaF is considerably better. -3- The present method of producing NaF by the action of concentrated H2 SO~ on nuorspar, and the dissolving of the resulting hydrofluoric acid in either a sodiwn hydroxide or a sodium carbonate solution, involves many dif'ficul ties due to the fa.ct that HF is very poisonous and corrosive, which rtiake it difficult and expensive to handle. The method which has been proposed to do away with hydro- fluoric aoid in the process 18 tha direct fusion of the calcium fluoride with a sodium salt and the leaching out of the Na.F produced. This fusion method was the basis for the research carried out by H. W. House and w. N. Lacey. House started out with a mixture of calci\lm fluoride, silica and sodi"Um carbonate, which gave a very poor yield on fusion and leaching. His low yield was probably due to too much and uneven heating which (as will be shown in this report) have a great effect on the yield. Later House tried charges of fluorspar and sodiuir1 oarbonate, without the silica, whi ch gave better yield.a, but they would have been still better if he had continued with the lilies.. Lacey obtained a. mu.ch better yield using silica. with the charge, so it was decided to continue with the same charge a.a that Lacey used. The silica -probably reacts with the calcium t o form the insoluble cal cium silicate and the fluorine is then in a pos ition to combine with t the sodium from the sodium carbonate, t o form NaF. The carbon diox- ide from the sodium carbonate that combines with the fluoride 1• driven off as a gas, while a large portion of the excess sodii.im car·oonate remains unchanged and is recovered in the l eaching solution. The equation that expresses the above reaction is as follows= 'Iha temperature at which tho reaction takes plEce is between 650°c and 775°F and as this ternperature is below the melting of the compounds involved, the reaction must go on at the surface of the :particles. This means that the reacting substances must be gfuund fine and thoroughly mixed to get the best results. Some of the other methods used for the laboratory production of NaF or have been suggested for its commercial production will be listed below. (l) "Sodi"Um fluoride is obtained by calcining a mixture of fluorspar, calcium carbonate 1 glauber's salt, and coal, followed by leaching with water and crystallizingn . 3 (2) nJ. J. Berzelius says that fluorspar is not decomposed when fused with alkaline hydroxide or carbonate, but P. Berthier showed that this is wrong, nuorspar is decomposed ,,men fused. with a mixture 4 of silica and sodi'UIIl carbonaten. (3) Eltperiments of Berthier: CaF2 + K2 C03 fused. Eq-qal molecu- lar proportions gave more fusible mass and greater yield of KF than 21½C03 + CaF2 • 5 (4) N. Roae found fluorapar completely decomposed when fused 6 with a mixture of silica and sodi'Um carbonate. (5) The manufacture of soluble fluoride by treatment of sodiilm fluosilicate with alkaliea. 7 3Jean, Techn. J.B. 1368, 199. 694. 5Ann. Chim. Phys. :;6, -253, 1S2S. 6Mellor, Comprehensive Treatise, p. 694. 1½Jie1 lor, p. 7uiman, Chem. Abet., 9, 2578. (6) Manufa.c ture of sod.ium fluosilicate. 8 (7) Sodium fluoride is obtained by dissol ving ammoni-um fluoride in sodiwn chloride . .Arorr.onium fluoride is produced by heating to- gether calcium fluoride and ammonium sulfate. The sodium fluoride Cl separates on cooling.~ (8) Same method as above, except that ammonium fluoride and amrconium sulfate a.re heated together, and the ammonium fluoride vapor condensed. lC (9) A suapension of sodium fluoeilicate is treated with sodium carbonate at such a rate that the mixture remains acid or neutral and sodi'\ml fluoride plus silicon dioxide are separated by fractional settling. ll As the fluorspa.r (CaF2 ) that was to be used in the research was impure and the anal ysis of it not known, a representative sample of the fluorspar was analyzed for CaF2 , thus furnishing a more accurate basis on which to calculate the percent conversion. The method of analysi•12 consisted essentially of converting the fluoride present into sulfate by heating the sample of fluorspar in a platin'Wn crucible with an excess of sulfuric acid (cone.), and then driving off the excess acid. From the increase in weight, due to the replacing of the fluoride 'by the comparatively heavy sulfate radical, the pet·c ent of CaF2 aan be calculated. The results of the different determinations made will be given in the following table. - Tchem. Abet. P. 2, 605. 9chem. Abst. P.ll, 1527 10 Chem. Abet. P. 12, 981. llchem. Abst. P. 13, 2112 12cairns , Quan. Anal . page 45. -6The first thing one will notice in the results obtained is that the method used is somewhat uncertain, but beca;,~se of the simplicity of the method indicated and because a high degree of accuracy was not necessary, a more complicated and exact method was not used. The rea1on for the low result in analysis No. 1 is probably due to the small amount of acid used, which in turn accounts for the fact that only 30 minutes was needed to drive off the excess acid and heat the crucible and. contents to a red heat. In the tt.ird determination the sulfuric acid was heated alone, in order to determine the amount of dissolved solid•, if any, in the acid. There was no residue aft_e r the acid was driven off, which shows that the re could be no error from this source. Determination 6-a is with the same sample aa 6, only the crucible is reheated, and a lower percentage is obtained, which shows that it was net heated long enough in 6. It was decided that the most reliable result for the percent of CaF2 in the tluorspar could be obtained by averaging the results ob' tained in the following determinationa: 2, 4, 5, 6-a, 7, S, and 9. Thie average gives 95-3% Ca.F2 whicm is probably correct to within 1.0%, and 95% will be used in calculating the amount of CaF2 con- verted to NaF. Methog pf Calculation Wt. of Residue Wt. of Sample (Ce.S04=l36.l) {CaF2 =87.07) x 100 = %CaF2 in s a.mple. None 40 2.0 40 45 2.0 2.0 2.0 2.0 2.0 l.0404 1.0404 1.0072 1.0174 1.0296 6 6-a. 7 g 9 40 45 35 96.S 93. 7 l.6802 94.l 97.2 97.6 96.7 93.7 95.0 91.5 sample. %C-aF2 in 1. 7194 1.6521 1.7611 1. 7705 l. 5g99 45 2.0 .9417 5 l. 7394 1.6152 50 2.0 1.0634 4 3 1.7797 45 2 .. 3 1.0695 2 30 1.1 1.oos5 CaSO'+, et 13. gr.1e" Weight of residue l Approximate time heated. Min. added. sample gms. sample Cc. H2SOi. Weight of No. of RESL'LTS OF THE All.AL 'rSIS OF FLUORSPAR FOR CaF2 I t' Analysis of "Tro~~" As a successful operation of the fu~ion method depends upon a cheap source of sodium carbonate, and as a pure carbonate is not necessary, the mixture of sodi'llm carbonate (Na2 C03) and bicarbonate . (NaHC03) known and sold under the name of "Trona." was selected. as the u:ost logical source of supply. In the first place "Trona." was the nea.res t supply, corning from Owena Lake ( Inyo Co. Calif. ) , and also it costs only about one-third as mu.ch as N92C03 per pound. The difference in purity offsets part of this difference in price, but taking the purity into consideration, the Na2 C03 from "Trona." costs_considerably leas per unit Na2 C03 than other comneroial N22C03. In order to be able to calculate the amount of NaeC03 or its equivalent in NaHC03 used, the "Trona" was analyzed for N22C03 and NaHC03. The method of analysis used was the ordinary two-end point titration, except that the new mixed indicators proposed by Stephen G. Simpson1 3 were used. That is: a one-gram sample is dissolved in 50 cc. distilled water and. it is than titrated with O. 500 N. H2 SO.,. using Indicator No. IV (see below) which gives the end-point for the conversion of the carbonate to the bicarbonate. Indicator N~. V is than added to the same solution and the solution again titrated with the same acid solution until the second end-point is reached. 13 Ind. and Eng. Chem. Vol. 16, No. 17, pp. 709, 1924. -9- I. -0.0l " + O. 58 ml. O. 05 N. • NaOH, dil. to ·25 ml. III. Eromphenol bl~~ -0.02 " + 0.66 ml. 0.05 N. NaOH, dil. to 25 ml. Indicator 1st ~nd Poin!, IV. 1 vol. Cresol red to 2 vol. th;y-mcl blue ( use three drops) 2nd ~d Point , V. Use four drops of bromphenol blue. The sulfuric acid solution wa.s standardized by using a sample of Na2 C0.3 (special anal;;rtict-u quality), which was kept in the even for three hours at 115°c, ancl titrating the N.:::.2 00:; with the acid us ing the above indicators. After several trial analyses to get fa"Yliliar with the color changes, tl1e two follo wir.g were taken to calculate the normality of - - 1.0000 gms. 37.99 cc. HeSOi;. sol. to neutralize 2 x lOCQ __ __ 37.99 X 106 = .497 N. of 1½S04- sol. (2) Wt. of N~C03 2 X 1000 38.00 X 106 - - - - - .. = .496 N. of I-I2S04- 1.0000 gms. -10- ' (l) Wt. of se.mple (not dried) -= l.00CP gms., 2nd ti " (2) Wt. of sample " 1.0000 gms. " 25. 86 cc. 2nd " lOO 8. 77 ~ _ ! _ 4965 x 106 1000 II = 46. 2% Na2 co3 • l No. II 100 x (g5~86 - 2 _x _8•IILx .4965 x 8~ _ 6 1000 30. %NaHCOz 46.5 + 46.2 2 = II 65. 42% "N8',2C0:, 11 in the Trena. Note: T'ne above indicators are probably all right to use where trie operator has a chance to become used to the color chat'.ges at the end-points• but for a limited number of determinations even with a "comparison color" it is very hard to get the results to check. This is r.r.ainly du.e to the great number of color changes and the fact that the end-points are not v er~· sharp. -11- ~ysis of Charg~ after Fu.sion The method u sed by House and Lacey, which consisted of precip14 itating calcium oxalate with the c e1cium fh::.oride, was given up after a. nur.aber of trials, because .of the difficulties encountered in filtering and weighing the precip itate. The method finally worked ou t and used was a direct fil traticn of the precipitate using "fil tercel". The procedure for this method will be given in some detail. A portion or all of the charge recovered aft e r fusion is ground so that about 90% will go through an 80 ruesh screen. It is than weighed and poured in to abo-ut 700 cc. of be il ing distil 1 ed water which is allowed to cool with frequent stirring. (This process has been found to lea.ch out prac~ically all of t he s oluble material from the sample) . After it has cooled the s olution and rasidu.e 1is poured into a liter volumetric flask and diluted to one liter. This is then allowed to stand over night to 1 et the fine particles settle (or a portion of the s 0lution can be filt~red) . .As two samples of the solution were t aken in each case, care was taken in getting the first portion of the soiution without distu.rbing the fine residue. The volume of the solution taken to analyse for NaF was chosen so that the weight of CaF2 prec i pitate would not exceed 1 gm. as it was much more difficult to filter a large amount of CaF2 than a small amount. If a high degree of accuracy is not des i red a sample that will produce between 0.25 and 0. 50 gme. of CaF2 is much ::aore convenient to filt e r. il+ Zeit. f. .Anal. Cham., 51 , 1912, p. 14. The 5al14.)l el -12- cf solution are then acidified with acetic acid and brou ght to boil ing to decompose the car1mate present . .An excess of CaC1 2 solution (100 gma. per liter) is then added to each sample and the solution al lowed to cool and the precipi tats of CaF2 to settle. The filt e ring layer in the Gooch cruci·o1e is the most important part of this method of analysis ae a. precip itate of CaF2 is always very finely divided and will go through any ordinary filter. The filtering layer comists of about 1/3 of an inch of asbestos packed. down very firmly and on this is added a layer of filtercel about 1/16 to 1/8 inches thick. The filter is then washed thoroughly and dried at 210°0 (ite iitagra) for 2 hours. Complete new filters were not made every time as the CaF2 preci:pi tate could be scraped off the top and a small au:ount of new fil tercel ado.ad. The sample is t r:.en fi 1 tered by O\J.~tion through a small weighed Gooch crucible (1 inch in diam.). As much of the liquid is decanted off a$ possible, to hasten the filtration. The precipitate is then washed with about 25 cc. of distilled water. The filter and precipitate are then dried at 21O°C in the air bath and then allowed to cool in a desiccator, after which it is weighed. From the weight of the CaF2 precipitate, a.~d the ratio of the total weight of the charge after fusion and the amount used in the analysis, the percent of conversion can be calculated. The above ratio is found for each charge by experiment as follows: the crucible plus the charge is weighed before fusion and again after fusion, and from this can be found the loss in weight of crucible and charge during -13- fusion. Then by subtracting from this los s the loss in weight of the crucible, the loss in weight of the charge alone is found. Then the weig..lit of the charge after fusion can 'be found by subtracting the loss in weight on fusion from the original weight of the charge. The amount of charge actually recovered after grinding in each case is found by wei.ghing. The loss in weight of the cr..i.cible alone was found by heating a similar crucible under the same condition& and finding the loss. Thie loss is approximately 1,5 gms. for the average crucible and average heating. Some of the advantages of this method of analysis over the similar one using the oxalate are: (1) The proce~ire is simplified due to the elimination of the sodium oxalate solution and the running of the blank on the oxalate solution. (2) The filtration is easier because of the nr~ch smaller a.ao'l:?lt of precipi ta.ta to hand.le, (3) The method used is more accurate, because the error in the blank detennination is eliminated, and also the oxalate precipitate absorb& moisture rapidly which makes it very hard to weigh aceu.rataly. (4) The calculations a.re also simplified somewhat by eliminating the oxalate. The data for a sample run will be given with the method of oal cul a t ion: -14- Mix no._2±5 ':P gme. Wt. of Trona (N22C03) . " " Sand ( Si02) " " Fl'UOrspar (CaF2 ) 20 " Total weight of charge . . . . . . 100 II . 528. 5 gms. Wt. of charge plus c~~cible before fusion " " " fl n after ~ n Loss in weight of charge and crucible Average loss in weight of crucible . 29. 7 " Loss in weight of charge alone . . Wt. of charge after Musion = (100 - 29.7) = 70.3 graa. Wt. of charge recovered from crucible= 53.5. (This takes into ac~o-unt the loss on srinding and handling , therefore it equals the amount of sample leached.) Leaching solution diluted to 1000 cc. Two 50 cc. portions taken for a:nalyai11::... Anal;,rsia (1) Wt. of Gooch cru.cible (no. 1) fl n n " ti " pl us CaF2 ppt. . .16. s632 grns. .17-3818 " • JU.86 _, 11 Analysis (2) Wt. of Gooch crucible (no. 4) . . " " " "CaF2 ppt. " . . 16.0310 grns. " "plus CaF2 ppt. 16. 5470 II • 5160 " . • Average of analysis (1) and (2) = .5186 + .5160 = .5173 gma. 2 = 71.7%CaF2 converted to Na.F and recovered. (Note: .95 is the fraction of CaF2 in the fluorspar; see anal ysis of fluorspar.) (100%) f2ooI = 29-7%loss in weight cf charge on fusion. Fusion ~~st of the fusions were carried on in a 5" by 5u. (inaide dimensions, app rox.) cylindrical fire clay furnace, having a. l" hole at the bottom into which fitted the end of a gas burner. T"nie was covered ,vith a fire clay top withal½" hole in it to let out the gases. The charges fused in this furnace were placed in a 2½" by 4" fire clay crucible, a new one being used each time. The fusions which are marked as "stirred" (see table of results which follows) were stirred nearly all the time of fusion \vi.th a wrought iron wire. Rune nos. 49, 50, 53 and on were not fused in the fire clay crucibles as indicated. above, but were spread on a flat tray a.nd this heated in a small muffle furnace. Charges 49 and 50 (see t able) were fused in an iron tray \vi th an area of 9 sq. inches , w1:ile charge• 53 and on were fused in nickel trays with an area of 30 sq. inches. -16- The following table gives a record of each indi,,idual mix and the results obtained with some of the conclusions to be drawn. The more general conclusions will be given later. -17i c;h t in grmns. ,-.. .... ·o >o ~ -ri a 'H . 0 i •rl (\J (l) ....... :g. zaJ 'ii:,'?._ =al ..::t • s: ..::t 01..0 J.. 8 0 = 8 ('harge - ar:r ,,......_ C\I 0 ...... rf.l '-' l'tlO 'O 0 ; ti) .... .,-'?_ .p ~ 10 2 50 16 20 3 50 16 20 (!) S::-rl bl) s.. d 8 g fl) ~ ".....,. 25 .p :;: s: :::$ I.!"\ 1 ~ J..i 'd Cl) 'M ~~ .µ o, 'd e G) •rl 0 (;..i 0 ~ ...., ..., 0 tt-. 0 ti}+> .cf! ~'Eli i:: '<i;a__ ~ m 0 H 0 C) s... 0 .... ..-, +> {I) 0 a, 'M lll r-1"'1 if 0 "d 0l.. § ,-.... (l) CD l:0 s...> rl 'O Cl> bD s.. <tl ,s::: 0 f+-4 • 0 Cl) ....,~ s 0 r - ....S'--"'b-3 C\J ...., ~ ~ G> bO .p ~ +> C\l 0 © 8~ Purpose of Run and Remarks s;: s:• 'O Condition of ....,~.!! G) product afte r f'usion. s +> (~ .,-j 8 Cl) 13.6 31.6 43.1 no 22 Glassy; light grayish green. To check results of Lacey. Lew yieli, due to carrying it too far and too high temperature. 86 28.6 33.5 9.s no 23 Glassy; green .Another tri al to check. 86 25. 5 29. 7 53.0 " 12 Part glassy ;part granular; part An at'Gempt Impossible to leach if present in glassy form. to eliminate glassy form. Impossible 'co heat evenly in ·furnace and conditions used. not fused. 4 50 16 20 86 24.3 2s.3 50. 4 " 1i::: _, Same as 3. Part glassy; part granular;not mixed. Not homogeneous; shows need of stirring. ' .,I 50 16 20 86 19.2 22.2 45.1 II 6 Part granular; large part not fused. Better furnace used with better ·heat control. The part not fused was not taken as part of sample leached. 6 50 16 20 86 26.6 31.0 44.5 " 9 GlQssy;light grq Too glqssy to get good yield. 7 50 16 20 S6 20.6 24.o 23. 6 l m. 17 Semi-glassy;gray Tried stirring. g 50 16 20 86 25. 5 29. 7' · 27.1 yes 11 9 ~ 161 20 86 27.3 31.s 65.9 r::; " 14 Poor yield due to lack of experience. More granular thS127 · A. 2 grq; granular No. 9 and 10 to find effect of fine sand; go mesh .. Stopped heating just after 2nd Solid stage. · Homogeneoue; light 10 50 16 20 86 26.7 ti 12 a Same as 9: easy to grind. Coarse sand used again. Heating conditions same as no. 9- ll 50 16 20 86 30. 7 35. 7 50.3 " 18 62 Gloaey; hard to grind;light green To study the physical changa- in charge on fusion. This was carried too far to get good yield. 12 50 25 20 95 26.9 27. 3 74.1 n 15 ½ Same as 9. Nos. 12, 13 and 14 to find effect of large excess Si0 2 (14 gm1. theor. amt. SiO~). 13 50 14 20 84 24.8 29.6 61.6 II 13 .! .a Sa."lla as 9. 14 50 10 20 &) 23.2 29.0 51.6 " 14 ·2 1 Same as 9. Less than required Si02·• 15 46 16 20 82 23.2 28. 3 64.. 9 " 13 Sarne as 9. To check value obtained in nos. 9 and 10 and to show what effect decreasing the carbonate has (116. 5% theor. used.) 15-a. 46 16 20 82 23.2 2s.3 0.131 " 13 ½ ½ 1 fine 2 semi-solid 31.1 65.0 l No. 15 reground ahcl releached. To check the leaching used. Seems to be nearly complete. = 'U .,.. rl 0 g:i ......C\I .,..H Purpose of Rim and Remarks . 0 a ori Condi tion of p r oduct a f t er fusion . r.r:i ........ 'i-1 0 ~ttJ rn 16 37 16 20 73 17 46 16 20 82 22.2 27.0 63.6 " 11 18 46 16 20 66 8.9 13.5 61.6 ff 9 Same as 9, An ..equiv. am't. of Na::C03. 19 301 16 20 83. 6 17. 6 21.0 62. 4 Same as 9. An equiv. am 1t. of NaHC03 instead of Trona. 20 47.6 216 20 S2 24.4 29.8 72.6 " 9 ½ " 23 19.5 Same as 9 To find the effect of holding the charse in 2nd solid form. 21 46 16 20 82 24.9 30.4 64.9 " 21 Same as 9 Held. charge in semi-solid formj hatter than no. 20. 22 46 16 20 82 24.4 29.8 70.7 " 15 Same as 9. Held at s~e temp. as no. 21 but only 4 min. 46 16 20 82 24.6 30.. 0 45.1 II 13 Same as 9 No. 23 and 24 to study effect of rapid cooling by pouring red hot charge into water. 46 16 20 82 24. 6 51.1 II 27 12 Same as 9 Same as 23, but was held in solid form longer than 23 and at higher temp. 2}--a 46 16 20 82 24.6 30.0 6.4.2 " 13 5 24 1. Same as 9. 2 30.0 To check no. 15 No. 23 r eground and ral eached. 24-a 46 25 50 26 16 25 20 82 24 .. 6 30.0 71. 0 fl 27 94% theor. am't. Na2 C03 used in form of Trona as in above ca.sea. 12 No. 24 r egro·.md To prove value of Trena as source of Na2 0. Nos. 23-a and 24-a are nos. 23 and 24 ground and rel eached. and rel eacp.ed. Same as no. 24-a. 20 95 27.5 29.0 68.3 11 .5 Dark gray gr~nular To check no. 12 in which large excess Si02 used. 50 20 95 28, 6 30.1 73. 6 " 16 7 Light gray granul ar To check no. 12; good check. 27 50 20 100 30.0 30.0 76. 6 fl 15 5 Granular i o find the effect of still larger am't. Si02 • 28 50 20 28. 5 25.9 43.6 " 22 12 Light gray; tendency t o be glassy. A further increase of Si02 . 19 g 29 35 110 20 105 28.6 fl Very light gray; somew:hat glassy . 30 50 30 20 100 26.0 28,.0 49.9 fl 31 5-~ 40 20 110 27. l 24.6 72.2 " l 2 Naz ·C03 NaHC03 3Pasty and solid . 4r>asty. LQw yield due to too high temp. II fl II Somewhat lower yield than 12 (74.1%) Increased the yield. Pasty when stopped heating. " II ff ff 17 8 Somewhat glassy. To check no. 27. Poor yield becuase of too high temp. 17 5 Light gray; very granul ar ; easily crumbled. Repeating mix. no. 28. Higher yield due to lower temp. and better temp. regulation. Poor heat regulator. - ~- -r Wei_gh.t__ i.n._g_r ~ . __ . "' >o ..-to ....a.c t+-, 0 ::i = 'O .... Char,re o•:z G)....., "'?fa--- =al ..:::t• S::..:::t . s.. 0\.0 'O - li ,.., '( \I ....0 U) ........ 11>0 § ii, .... Cl) G> ~ a: ~ ~ rl ;:i tn .p fit.?) [ -! 0 ~ ,;,~ '(j .p ~ ,q t> • . ,-f (I) .p .p iii .... ~ ~ = I> .µ aG> ......a •r! s:: s:: ....s::~:::$ f!) Q> GI '°m ft.i s:: s:: '11~ 0 0 0 11) .p N ,....~ iii ..... - I ft.t--; OS:: I> ~ 0 rn fD f.4 OJ G> ~ s:: td 0 0 l:r., J.t (\1 0 'd ~Q.__ § .....r.. al bO ~ d .... I .p b.!) s:: .-! .... {lj ~ .p t\l ,,::j C\l Purpose of Run and Remarks • 11 ....s:: ......... l1) (I) ..µ 8 1:/i s d ..-1 .µ Condition of pro:duct after fusion. ~ E-t 0 = E-t 32 50 35 20 105 28.4 27.0 78.3 ye, lS 5 Same as 31, 33 50 30 20 100 28. 3 79. l " 14 34 :;61 y:; 20 86 21.3 24.s 54.7 " 16 5 6 " " Probably very ciose to the best conditions of temp. control. To det armine the effect of decreased Trena with increased Si02, Decreased yield tiue mainly to low Trona - as temp. regulation good. 35 41 2 ':I) 20 91 24.4 26. s 69. s " 14 5 n 36 463 30 Z) 96 23, 9 24, 9 63.0 n 16 5 Dark grav (Jg.9s. 34, 35, 36) effect of decreased Trona. tow yield due to too low temp, ae indicated by low %loss in wt. of cha.rge on fusion. 37 45 -;o 'it) 95 26.7 2s.1 67.2 " 19 5 Gray fine granular To check better yield. 3g 46 '30 20 96 27.3 28.4 69.2 " 16 5 Gray; granular To. check no. 37; good cheok. 39 60 30 20 110 34. 3 31. 2 68. 3 " 18 6 Light gray; fine gramua.r Effect of large excess Trona; heating conditions good. Seems to get hotter without changing to 2nd solid stage. 40 'fJ 30 20 Very light gray To check no~ 33 which was highest yield so far. No apparent reaeon for lower yield than no. 33 except loss in wt. higher. m 0 'H 0 0 H 0 ~' ~ 28, 3 .s:! 0 CD 0 8~ 5 granular 41 50 30 20 5 Tendency to be glossy 42 50 30 20 " 12 3 Light gray; fine gram1lar 43 50 77. 7 20 100 28.0 41.i-a 50 304 20 100 30.6 30.6 64.o 44-b " " 10 50 y:; 20 46 47 30 50 50 1 91% th. 2103% th. 30 4 20 High yield due to lower temp. and better temp. regulation. To check no. 33. Low yield due to too hi gh temp. on fusion. •ro check no. 33; low yield. Might be due to too low temp. c:ir too ., short time of holding charge in 2nd solid stage. To check no. 33. Heated quickly by hot fire; somewhat pasty but this was offset by short time of holding. Short, high temp. seems desirable. To check no. 33. (Not stirred enough to be homogeneous, so was divided into two parts to analyse no. 44-a..) To check effect of fine Si02•. 10 l Gray , granular " 16 5 Coarse , slightly glassy, part of 44. 74,l " " " Fine gray granular part of 44. no . 44-b (see above) lC'O 29.7 29.7 71.7 tr 11 4 Gray grarr12.lar To chaok no. 33;. very slightly pasty\ vhen ta.ken out of crucible. Coarse Si02- used to compare with fine Si02· in no. 44. 100 28. l " 13 4 Gray granular To check no. 33 and use coarse SiO~. 21 8 " 100 3116%th. 480 mesh 5coarse 26.0 II 28, l 73, l ff " size of peas Gray granular Another trial with fine 5102 ; not carried far enough so wae reheated before it cooled materi a 1ly. -20- = -d il"° ~~ . ~-- -~ .... l rl 0 qj ti) = fl) ~ Purpose of Run and Remarks. s.. ....t ft.I -4,> 0 m ro _.,. Q) fl) QI) O ,-i <> ..r:b(l ~ -.➔ 6 II) ~-- s: 48 50 30 20 100 49 5 3 2 10 5 2 28. 5 28.. 5 . 1 Condition of product after fusion. 65. 6 yes 16 ·3.0 30.0 50.6 no 1 1.5 30.0 19.4 " 5 6 Gray, coarse gran. Dark gray due to 4 Fine Si02 . Slightly pa.'3ty on last end; acts the same way as an excess Si02 in that solid form is over a more limited range. iron. To find the effect of thin layers on metal surface without stirring. Iron vessel used. Nearly black due iron Sarne as no. 49; low yield due to poor temp. regulation. 30 20 100 _3~·01 30.0 70.0 yes 15 4 Light gray: gran. To study effect of fine (SiOi) dehydrated silicic acid. 50 30 20 100 27.01 27.0 4 Fine, granular 53 30 13 12 60 10.4 17.4 24.4 no 7 To compare with no , 51. Layer of charge spread on Ni tray and put in hot muffle furnace. Lo.-yield due to too low temp. 54 30 18 12 6o 10.4 17.4 17.5 " g 55 30 1$ 12 60 15.4 25.7 79.5 " 3 56 30 lS 12 60 17.4 29.0 75.s " 4 57 YJ 18 12 60 14. 3 23. S 74.9 " 58 50 30 20 100 23.2 23.2 79.3 11 30 20 83 16.6 20.0 61.9 30 20 83 12.2 14~7 61-a. 50 10* 20* 80 19.0 61-b 50 l.j()* 20* 110 19.0 62-a 50 62-'b 50 51 50 52 33 59 60 l 33 3 n 13 Dark gray Same as no. 53; still not enough heat. Light grq Much hotter furnace than used in no- 1. 5 Very light grav Higher temp. than used in no. 4 0.5 Light gray To check no. 55; not cpite as hot aa no, 55. 5 l .. O Light gray '.fo check no. 55 with thicker layer on trey; hotter than no. 57 • " 6 1.0 Light gray 62.1 " 5 1.0 Light gray 23. S 85.2 " 12 7• 5 Gray 17.3 74.1 " 12 7.5 Gray 20** 20** 90 19.3 21.5 82.5 " 6.5 1.5 Gra¥ 40** 20** 110 19.3 17.6 76.2 11 6.5 Gray 4 4 .Assiml8d 2 salicic acid 3so mesh 70.9 l.5 4Na2C03 *Also 30 ~s. leached out rer.idue from no. 57 (which contains 2. 86 gms. CaF2 dn) and 10 gma. Si0 2 . **.Also 20 gme. leached out residue from no. 58 (which contains l. 57 gms. CaF2 &lu) and 20 gme. Si02• 53 and 54. 55. T_o compare value of N¾C0 3 with Trona. Too hot at end of fusion. Sarne as no. 59. Temp. control seemed all right. To study the effect of using s ome of leached out residue in charge with a decrease of Si02 • Calculated on basis of 20 gms. CaF2 • Same as no. 61-a, only calculated on basis of total CaF2 present, which includes that from residue. Leached out residue from no. 58; calculated on 20 gms. CaF2, Same a, no. 62-a. Calculated on basis of total CaF2 • 20 gme. and am't in residue added. -21- The physical changes in the cbarge du.ring the fusion are quite 1::.arked and the effects of stopping the fusion at a given time were studied in the different mixes. The changes that take place and the approximate ten~erature of the fusing charge will be given. The temperatures are very rough and were obtained by comparing the color with a color temperature chart. The temperatures given are probably within 100°C and the relative temperatures somewhat closer than 50°c. (1) 20°c - 500cc. Drying action only. (2) 5000c - 6oo 0 c. Some fusion and. bubbling ( CO2 given off). (3) 600°0 - 675°c. Completion of fusion. (4) 675°0 - 750°c. A solid formation again. (5) 750°0 - 800°0. A second fusion begins (pasty). (6) 800°C . and up. A viscous liquid (glassy on cooling). (See conclusions for best condition of terr~erature control.) TAe temperatures as given vary for different arnounts of Na2COG and Si02, - I I . -2Z- I i.--- ....- ~ /;J() - I ~ ~ , . I #IJ *~ i ,_ l•A .... . ,.. ... _.., .2-,4 t ,_ . L . 60 6D r Zl 20 I /_6 +o .. . 17 15 •.Z3q• • I ' I 7'1 afo 'Conversion. I . •~ r.i); 19 1 /30 - ... ,'Zt . . I + = :-3ct. ~ lh.l' I • \) ~ -' <S> . ~.,;. -lo -16 \'.._;.I ~ , _., ~ " , .. I - ~ \.} /fb rI '. t l t . "'13 ' 31 • I I I ! ,· - ; 11 :1/. -F 38, I I I Cvi -= l~'O"fa rheo I /~ I F (f<:c z; of am•t pt' 71 ooa·: t/s~ I L .90 l I ~- .::" . cJO I I I 7i ' -rlie_o_. CqrS :: I oo,fi Theo . f'9 I '.J Sill. =- ~ /0 ,. rt - I .. .. ~ -~ #;O I I rrler tr "'' am·t or 'Tronn" ~~£J.JjU~~d I '-- ,-- I I I . 3,S • I I ' . 3~ ~() ,-. &: i oD 40 60 1'0 ~Convers,~ >n . tI ~ 80 90 1 --- I : : ' ' - - . . ' -~ TT( E.(f1!cl of 4 mao.nt ' - ~ ,_ ,.._ .__ l -- '' ' :1 /)f - I . . --,__ ,__ ~ ' \l r-, ~ f -- t ,u S6 ,fJ ~ 41> f - . 1· L (I'.) -- ,_., ~ f j ' I ~' 2s ...... I L... 27 .sr3.J 4S t cy -- I . . . .s1....... 4-Z. ~ I~ ' ,- - ' .. ,'£,b - 3,Z. . i- ' ~Ob . - I - --- l i --- ' ,_ L , I I I r -- ' .,. I .JI ----,., ' I I -- I + _( o fz+L O i/4.TMa 2,-61) ' I - . . . ' -- ..... .__ + I t .. -- ' t ' + ·• - ' t I -- ,-~..., ~..... 10 -- s: I~ t l /,f ;;o I [9 -- -- - 1 ' ! I I~ ,- . ~;a2 Us.er/ ~~co.,~ 126,f. rhtd. --- -;--- I I l ' t . "'., j d/() - "'40 60 ~l) ,_ Cor, ve r, $ion I 90 8t> 7i°O -· t -' -- -- .... 1-,- .... -24- T'ne preceding plots are an attarA:pt to indicate graphically the main effect on the yield due to variations in the mix used. Due t o the fact '(;hat there are many other factors that affect the yield besidae the ratio of Na,zC0J, Si02-, and CaF2 in the mix, such as rate of heating, time of heating, stirring, time of holding in-second solid stage, etc., it is very difficult to make curves that show one effect onlJ. The points taken were from the charges in which the outside effects were nearly equal or did not enter. That is they represent charges in which there were no obvious reasons for low yields, and not necessaril? the charges that gave high yields. In plot I and II the points represent values in which the Si02 and CaF2 used were held constant while the N8.;?C03 was changed. The percent theoretical Na2 C03 is plotted against the percent convereion. The difference between plot I and II is that in I the basis is 100% theoretical Si0 2 and 100% CaF2 , while in II the basis is 188% s10 2 and 100% CaF2 . In plot III the percent theoretical Si0 2 is plotted against the percent CaF2 converted to NaF, while the amount of CaF2 and N~C0:, a.re held constant at 100% and 126q-~ theoretical raspect1.vely. The Reco~y of NaF from the :Bus ion Product Analysis of solution from mix no, l.~6 for aolids in solution (NaF, Na-?C03 and ot her solids) is given in deta il, as it was considered a fairly representative run. was as f ollows. The method of anal ysie used Four 5()-cc. sa~mles of sc l~tion we re tak3n. One was acidified , a c evi. ~ • c ac 1• d an.... ~ s.n a ~ ·" .i.or ~ 'IIT "' r-. se ,9 p r oc e(J.ure _,_ ;;,. ·..vi 'v.'l .J.:'.)'Zeu. .. a.J .1 or anal;:,rsis of• Another 50-cc. seimple was no t acidi fied to destroy the rni =c for NaF). N~CC):; but t he CaCl 2 was added d.irectl;,1 und the CaF2 and CaC0 3 were precipitated toget::er 2.n d the amo~ nt of precipi ta.ta found by filtering and weighing as in the an al7rs is fer N.,,,F, The third 50-cc. sample was ev~.pora t ed. to dryness and. tl~e total amo,mt o f solid in the s olution found by weighing the residue• .A. fo ·,1rth s :.:U1,-ple was anal:,rzed to find out if the r e was any Na.: Si ') 3 ll:. the le£:.Ching soh:tion which wJuld 'oe precipitated with the CaF2 The 50-cc. saz;:;ple was eve,:po rated nearly to dryness in a beEJ<.er and t h en it vvas transferred. t o a ple.tinum crucible. Eight, cc . of ccir.Cel, tra ted H2 SOi. was added and t11e s ol u tion evapo r atad tc- about 3 cc . The so l ut ion was then diluted with dis tilled. wat e r and pou.re d into a ·::ieak er . No p reci -p i tate (Si0 2 ) f orme d on dih.1.tio:n which indica ted. that there was no soluble silicate i n the scl·..i.t ion. S:he calculations and re sults o f t he a.na.l 7si&: v;t. of crucible (Gooch) I! II !I plus CaF2 ppt. 18. 9572 graa. 17.t~51_ t1 . 5321 II Ca.F2· from 50 cc. leaching s olution. (2) (. 5321) 84/78 = 1.148 grns. NaF per 100 cc . l eaching soluti on. Wt . of beal:er plu!3 so lid from 50 cc. sol u tion . !I " !I ,, ti solid frcrr. 50 c,~. s ol u t ion 57.9156 gms. 51_._995..Q__ " • 820 6 - 26- (. 8208) (2) = l. 6412 grns. to t al sclids frow. 100 cc. of leaching solution. Wt. of crucible (Gooch) . ti II 17. 4908 gms. plus CaF2 and CaC0.3 ppt. II !_§_~2J.3!f_ 11 • ,.,,. 74GO " leaching solution. Solids in leac'!:ing $Olution (gms. per 100 ca.) NaF 1.148 gms. Na2Si0.3 0.000 ti Other sol ids.Q_1±1_ " Total l. 641 " 'Ihe bciling point rise of the leaching solution and the purity of the crystals produced by evaporaticin were next studied. .Abo,it 6. 5 liters of the dilute leaching solution t aken from some of the runs with good yields was evaporated to dryness. The volu.@e of solution left and the te1::.per ature of the liquid was noted at in- t ervals during the evaporation. Also the crystal!! that ta.<l formed. were filtered out of' the solution w:tile hot (to keep the Na2 C0:3 frorn. crystallizing) 2eve ral times du ring the evapo ration. The cli fferent cr~rstals we:!'.'e washecl with a srJall ai.iount of water, dried, and then analyzed. for 1TaF. -27The boiling poi nts, arnoi.;nt of solution left, weig.,.1.t of' cr;)'stals formecl during the period between filtra tions, and. the percent of lfaF in thes e crys tals are given in the following ta'ble. The reason the boiling points do not have a S'My rise is probably mainly due to the tendency of the solutiori to 'become superheated, 'but since only the app roximate boilir1g point rise wt1.s wanted, the temperatures taken are close eno~gh. It is seen from the table that the la.st crystals contain very little NaF and that they are nea,rly pure Na2 C03 which can be used a.gain. By calcula.tion from the a.>no-1.:.nt of NaeC03 recovered in the lec"lChing solution it is founcl that 14% of the Na2 C03 in the forrA cf Trena is recovered or that 58% of the excess Na2 C03 is recovered. The table also showe that the first two crops of crystals contain more than 99% NaF and that t11ey contain most of the NaF in the solution. By calculation it can be shown that 90% of the NaF is recovered in these first two crops of crystals. If the third and fourth crops are included they contain 97% of the 1!a.F in the form of crystals that are over 98% NaF. -28- Table Cc. of solution Boiling po int left. of the solution °C. Gros. crystals forroed since la.st fil tration. 6500 1400 99°c 600 100.5 l.!OO 100.5 250 100.0 16.0 l~ 101.3 3.0 120 102.5 95 102.0 50 104.2 o.o 36.0 gms. %NaF in crystal 1. The Corrosive Action of NaF Solution Since a l a r ge q_uanti ty of the NaF leaching s olution v~1;ld have co be handled in :pipes, vats, evapo r a tors, e!ic. if the deacri bed pro cess was to be put on a cornrnercial scale, it was t h01.: ght advis- able to set some corrosive samples. The corrosion tests r.-a.de were as follows. Wrought iron wire in dilute leaching solution from mix 46. I. This sample was kept at rcom terrrperat1.1re al 1 the time. Wro~ght iron wire in dilute leaching solution from mix II. 46. This sample was kept hot on a hot plate part of the time, tha,t is when it was convenient. III.* Strips of copper and steel were placed in tap water which was kept at re orr, temperature. IV.* Strips of copper and steel were placed in a concentrated leaching solution (approx. 4.2% NaF) .vhich was kept at room temperature. 1 V.* Strips of copper and steel were placed in concentrated leaching solution which was kept hot :part of the time ( see II). Each piece of metal was polished bright and weighed before it was put in the solution and after it had set the time indicated in the table. The metal was then cleaned and polished carefully again taking care not to wear away any of the uncorroded metal in the polishing. The metal y1as weighed again and the less in weight found. of the pieces of metal was fo1md by measurement. The area The main data. ob- tained and the results are given in the following table. *Strip1 were in same bottle and in contact with each other, which would tend to increase the rate of corrosion offhe steel and decrease that of the copper. Kind of .0019 2.18 71 V (hot-cone.) Copper 0.9 0.1 .0002 2. 20 71 IV (cold- cone. ) Copper 3.0 .0064 2.10 71 III ( t ap water) Copper 1.7 .0040 2.41 71 V (hot-cone.) Steel 0.5 .0008 l. 71 71 IV ( cold-cone.) Steel 38.l .0648 1. 70 71 III ( tap water) Steel 9.1 77 II (hot-dilute) Wrought iron .Ol 72 77 1. 89 sq. in. Loss mg. per 4.0 1. 86 (gms.) Loss in weight .0074 - Area of metal (sg_. in.) (days) Time set I ( cold-dilute) ~elution Wrought iron metal. Table I 'SI Conclusions The folbv.d. ng conclusions are based. partly on direct observation and partly on expe rimental data which has already been given. In some cases the conclusions are only ro ugh apprc,:d.1nations and in other cases they should be fairly accurate. Th~ Mix From a stucy of the runs made it is believed that the 'best mix to use ia 20 parts Ca.F2 , r::P parts "Trena", and 30 parts Si02 • Taking CaF2 as a basis, 50 part• "Trona" (source of Na2 C03) corresponds to 126% theoreticsil and 30 parts s102 correspcnds to about 188% theoretical. This is assuming that the follovnng reaction takes place: CaF2 + Na2 C03 + Si02 --.S,- 2NaF + CaSiO3 + CO 2 • As already indicated under the ~~rk on the leaching solution58% of excess Na 2 C03 is recovered and can be used again. .Also, if some of the leached out residue is used with the mix (see runs 61 and 62) the amount cf Si02 used can be greatl y decreased and t};e yield, considering only the ne·,v CaF2 used, is materially increased. Condi t_ions of fusion The fusion sho~ld take place in such a place that the taraperarture of the fusion mixture never exceeds 750°c (approx.), but the 'heat shou ld be so gov8rned that it rea ches this temperature within a short time (2 to 10 minutes). GU'ii The charge sho~ld be either taken of the furnace or the temperature not allowed to exceed 750°c soon after it s t ops bubbling and reaches the second solid stage. A short time of holding at this higher tenperature, fer instance one minutet will tend to increase the yield, but it seems to reach an equ.ilibriwn -32- seem after one minute and the yield will not increase further a.nd tcere is a tendency of getting the charge too hot if held too long, which decreases the yield very rapidly. The rate of cooling does not seem to have any appreciable effect on the yield. (See mixes 23, 23-a, 24, and 24-a.) The charge must be hendlad during the fusion so that it is nearly homogeneous all the time, or when it is taken out of the furnace some of the charge will have gone toe far, while some of the rest may not have been heated enriugh. The best we.y fo1;.nd to do thh was on a flat surface with the charge not more than one cm. thick, if it is heated from both top and bottom. &rinding and Leaching .Al thou gh very little experiment.al data was obtained on the a- mount of grinding and the time and conditions of leaching, some observations should not be out cf place here. It was found that the fusion product must be grc~nd in ordar to get a thorough leaching. (See runs 23, 2}-a, 24, 24-a.) By grinding the fusion procu.ct so that 90% goes through an 00-mesh screen over 99% of the NaF ie leached out, (See runs 15 1 15-a..) The time of leaching was not studied but the leaching is complete after the solution (not more than 1 %Na.F) has been in contact with the grm:md. f\1sion product for two ho-:..:rs with frequent stirring. How much this time c an be cut down or the concentration increased is not known. By the use cf a counter current system of leaching a c on centrated solution (4.2~ NaF) can be obtained. -33- .Q.ITstallization Since the solubility of NaF does not materially increase with the te?I;)arature*, and the solubility is only about 4.2%,a large quantity of water will have to be evaporated. is in rt!Ultiple-effect evaporators. Probably the beat way to do this There will be crystals forming in each effect so that crystal traps will have to be provided for each effect. As the boiling point rise of the solution during evaporation was small (about 5°0) this will not mat Arially decrease the possible number of effects. If the quantity of solution was large enough five or even six effects would not be impracticable and would mean very cheap evaporation. The ordinary steel plate and tubes can be used in building the evaporators as the solution was found to have a mu.ch less oorrosive action on steel than ordinary tap water. *The only data th~t co,1ld be found. on the solubility of NaF: T~. oc -5.6 Solubility % Ice and NaF eutectic Reference " Landolt-Bornstein +15. 3. 85 n +lg 4.22 " +?.l 4.00 n +25 4.03 n +100 4.2~ z. Ph. Chem.1912, 82,632. +100 4. 56 Mellor, Yol. II, p. 515. Po~sibility of Eco~.omic Production Al though the pro~e$s ha.a not been carried out on any but a pure1.y labora:tory scale, the economic side of the pr oces e should be oon- 3idered, because the first consideration in contemplating semi-commercial production:, would depend upon a rough estimate of the cost of production. Assum,ing 80-85% yield, the Cotlt of raw materh.ls would be approximately as follows for one lb. NaF produced. CaF2 - 1.1~ ots. - Basis $25.00* per ton :or fluorspar . U~C03 1, 6 ch. " ing Trona cost 70%as much f or the same amount c f Na2 C03. Total Bas is $18. OC * per ton. 3.6 cts. cost of ru.w r..-:at erial to produce 1 lb. NaF. The val·u.e of the NaF was S-3/4 cts. * vmich leaves a margin of 5.15 cts. within which the cost of p roduction must come. This mariin , is qu ite large, and if the acale of.pr oduction was suffi ciently large 1 it probably would lno t be dif ficu.l t to keep within these limits. Suggesti ons for Fi~ture Work (l) A development of a continuous method of fusion wi tho·11t stir-ring and the kind of material tha t would best stand the fused substance and the high temperature. Nickel which wa.1 used for soma of the later runs soon becomes 'brittle. Monel metal was suggested but not tried. (2) A continuation of the study 'of the tempera ture, time of holding, etc. on the charge. *Prices , April 1925. -35- (3) Further invesUgation of the best proportion of the different -;;ubeta:nces tc use. In thie connection a very importa."lt :point to ·zheck and v.'o rk on is the use of the leached out residue. Runs 61 and 62 were a start alone this line, and they show interesting possibilities. (4) After the above points are cleared up, a.Yid if they are favorable, the process should be ready to be put on a semi-commercial scale. The writer at this time wishes to express his thanks to W. N. Lacey for his gt1iding suggestions in the performance of the experimental work and for the helpful advice and cri tidsrns in the :p reparation of t his report. Also credit is due W. N. Lacey and Harvey W. House for the use · of the reaul ts of their work on this fusion process which preceded this work. Sumnary Sodium fluoride is produ.ced "by the fusion of a mixture of CaF2 , Na2 C03 and Si02 at about 750°C according to the following e·q_~a.tion CaF2 + Na2 C03 + Si02- ~ 2Na.F + CaSi03 + CO2. The NaF is recovered by leaching, evaporation: and cr:tstallization in nearly pure form. Yields of 00 - S5% are possible.