content of slags containing various quantities of phos
tained in the manufacture of phosphoric acid b y furnace
pal source of error seems to lie, however, in the formation of a non-volatile oxyfluoride, probably SiOF2 (I), by the action of hydrofluoric acid on the silicic acid resulting from the action of sulfuric acid on silicates present in the slag.1 It seemed necessary, therefore, to resort to some method involving
1 R eceived M a y 14, 1931. Presented before the D ivision of Fertilizer Chem istry at the 80th M eeting o f the American Chemical Society, C in
cinnati, Ohio, September S to 12, 1930.
1 See the im m ediately following paper, "E ffe c t o f Certain Forms o f Silica on Determ ination o f Fluorine b y Volatilization M eth od ."
alkali fusion in order to de
termine fluorine in the slags.
A t the same time it was de
sired to obtain, if possible, a method that would be ap
plicable also to the determi
nation of fluorine in phos per cent of the fluorine pres
ent in this material (12). recovered. It seemed desirable, however, to make a further study of the method to determine whether it might be' modified to give satisfactory results on phosphatic materials.
After considerable experimental work a modification was de
veloped which seems to give fairly satisfactory results for fluorine in phosphate rock and phosphatic slags. The results of this study are given in the present paper.
T h e flu orine in h igh ly p h osp h a tic, calcareou s m a te
October 15, 1931 IN D U STRIAL AND ENGINEERING CHEMISTRY 367 P relim inary Experim ents
The method as outlined by Hoffman and Lundell is de
signed for the determination of fluorine and silica on the same sample. Briefly it is as follows:
The sample is fused with sodium or potassium carbqnate and the insoluble residue remaining from the disintegration of the melt with hot water is filtered off. Silica is removed from the filtrate by two precipitations, first with acid zinc nitrate solution, and second with ammoniacal zinc oxide solution. The filtrate from the second silica precipitation is treated with hydrochloric acid, lead nitrate, and sodium acetate to precipitate lead chloro- fluoride (PbCIF), and the chlorine in the latter is determined volumetrically by means of standard silver nitrate and potassium thiocyanate solutions.
Preliminary experiments showed that the low results ob
tained on phosphatic materials by the Hoffman and Lundell procedure are due to interference of the phosphate. It was thought that the trouble lay either in the lead chlorofluoride precipitation or in the alkali fusion.
In order to obtain information on the effect of phosphate on the lead chlorofluoride precipitation, various quantities of phosphoric acid in the form of potassium dihydrogen phos
phate were added to solutions containing 8 grams of sodium nitrate (equivalent to 5 grams of sodium carbonate) and 0.040 gram of fluorine as sodium fluoride. The Hoffman-Lundell procedure was followed in precipitating the lead chlorofluoride and determining the fluorine equivalent of the chlorine. In the absence of phosphoric acid, 0.0401 gram of fluorine was found, whereas in the presence of 0.005 to 0.06 gram of phos
phoric acid (P2O5), the quantities of fluorine found increased progressively from 0.0407 gram to 0.0467 gram. As pointed out by Hawley {4), the high results in the presence of phos
phates are probably due to the fact that when lead phosphate is precipitated in the presence of chlorides, it carries with it some lead chloride.
Since it is evidently necessary to remove phosphate prior to the lead chlorofluoride precipitation, experiments were carried out to determine whether the two zinc treatments used by Hoffman and Lundell for precipitating silica3 are effective also in removing the phosphate. A solution contain
ing 0.040 gram of fluorine as sodium fluoride, 0.35 gram of phosphoric acid (PjOo), and 5 grams of sodium carbonate was carried through the zinc precipitations and the subsequent operations outlined by Hoffman and Lundell, with a recovery of 0.0402 gram of fluorine. Only about 0.005 gram of phos
phoric acid remained in the solution after the zinc treatments.
In a second experiment, 1 gram of silica was fused with 10 grams of sodium carbonate and the melt was dissolved in a solution containing 0.040 gram of fluorine and 0.35 gram of phosphoric acid. About 0.008 gram of phosphoric acid re
mained in solution after the zinc treatments and 0.0398 gram of fluorine was recovered.
These results show that the zinc treatments correct the error arising from the presence of phosphates in solution.
They indicate also that the low results obtained on phosphatic materials are due to failure of the alkali fusion to convert all the fluorine into a water-soluble condition. This was shown to be the case in a number of experiments with various fusion mixtures.
Alkali F u sion o f F lu orin e-B earin g Phosph atic Materials
Bureau of Standards standard sample No. 56, Tennessee brown-rock phosphate, was used in the majority of the ex
periments with different fusion mixtures. Previous analyses {1 1) of this material by the volatilization method, together with analyses during the present investigation, indicate that
1 A ccording to Hawley, silica itself does not interfere in the determina
tion o f fluorine b y the lead chlorofluoride method.
it contains between 3.45 and 3.55 per cent fluorine. For the purpose of the present paper a fluorine value of 3.45 per cent, is used. Several experiments were also made with synthetic mixtures of tricalcium phosphate and Bureau of Standards fluorspar, standard sample No. 79, containing 46.19 per cent single fusion with sodium carbonate, sodium carbonate plus potassium carbonate, or sodium hydroxide. The sodium per
oxide-carbon fusions also gave poor results. Addition of silica had a beneficial effect, but even then a maximum of only about 82 per cent of the fluorine was recovered. Similar results were obtained 011 mixtures of fluorspar find either
Fairchild (S) states, in a recent paper, that the fluorine in a mixture of dicalcium phosphate and fluorspar is completely converted into the water-soluble condition by fusion with a mixture of sodium carbonate and feldspar. The writers were unable, however, to confirm Fairchild’s statement. The figures given in Table I show that sodium carbonate-feld- spar fusions gave no better results on either phosphate rock or mixtures of dicalcium phosphate and fluorspar than those obtained by sodium carbonate-silica fusions. (The phosphate rock and feldspar were ground to 200 mesh.) Fairchild’s which are easily decomposed by sulfuric acid. The insoluble residue from the sodium carbonate-feldspar fusion of the mixture of dicalcium phosphate and fluorspar did not give a test for fluorine by the etching test, and the results by the hanging drop test were questionable, despite the fact that the residue actually contained approximately 0.008 gram of fluorine.
308 A N A L Y T IC A L EDITION Vol. 3, No. 4 The figures given in Table II show that, in general, three
successive fusions did not bring into the water-soluble condi
tion all the fluorine present in phosphate rock and in synthetic mixtures of tricalcium phosphate and fluorspar. It seems that, in general, less than 90 per cent of the fluorine can be recovered in this way, although a 99 per cent recovery was obtained in one case. The results give further evidence of the beneficial effect of silica in the fusion mixture.
T a b ic II— R e c o v e r y o f F lu o r in e b y T h r e e S u cce s s iv e F u s io n s
c Representative sample o f com m ercial material.
•i Approxim ate figure.
9 Fluorspar, Bureau o f Standards standard sample N o. 79.
R eaction s O ccu rrin g in A lkali F u sion o f P h osph ates When identical fusion mixtures were used, for example, 2 grams of sodium carbonate and 0.5 gram of silica, duplicate determinations on a particular sample of phosphate rock, or mixture of tricalcium phosphate and calcium fluoride, usually cheeked within 0.001 gram of fluorine, despite the fact that, in general, complete recovery of the fluorine was not obtained even with three successive fusions. Almost identical results were obtained when the determinations were repeated at a later date. Calcium fluoride alone is completely converted into the water-soluble condition by a single fusion with sodium carbonate and silica, but mixtures of calcium fluoride and calcium phosphate behave like phosphate rock.
These facts indicate that in the alkali fusion of fluorine- bearing phosphatic materials, failure to obtain all the fluorine in the water-soluble condition, even after repeated fusion, is due to equilibrium reactions in the melt involving a water- insoluble fluorine-phosphate compound. It seems to be definitely established that the fluorine in phosphate rock is present principally in the form of a complex calcium fluor
phosphate having the same empirical formula, 3Ca5(PO.t)2.- CnF», as crystalline fluorapatite. A number of investigators (.9) have prepared this compound b y heating mixtures of cal
cium fluoride and calcium phosphate in the presence of alkali
salts. •
A further study of the reaction showed that alkali fusion, with or without addition of silica, does not convert the same percentages of the total phosphoric acid and total fluorine into the water-soluble condition. In one experiment, fusion of 1 gram of phosphate rock with 10 grams of sodium carbonate converted 60 per cent of the fluorine and only 32 per cent of the phosphoric acid into the water-soluble condition. A second fusion solubilized 28 per cent of the remaining fluorine and 11 per cent of the remaining phosphoric acid. These inequalities may result from the formation of water-insoluble calcium sodium phosphate, CaNaPO, (3, 13), but further in
vestigation of the conditions under which this compound is formed are necessary before a definite statement to that effect can be made.
With these facts in mind, the principal reaction occurring when phosphate rock is fused with pure sodium carbonate, Similar reactions occur in the fusion of fluorine-bearing phosphatic slags and of synthetic mixtures of calcium phos
phates and calcium fluoride. complete extraction of the fluorine could be obtained by treat
ing the residue with nitric acid (1 to 10). The acid treatment introduces, however, several complications. In the first place, the calcium that is brought into solution simultaneously with the fluorine must be removed in order to prevent loss of fluorine as calcium fluoride during the zinc precipitations.
This may be accomplished by precipitation as calcium oxa
late under carefully controlled conditions. It is very diffi
cult to obtain fluorine-free precipitates of calcium oxalate if more than about 20 mg. of fluorine are present.
It is then necessary to remove the excess of oxalic acid re
maining in the filtrate from the calcium oxalate precipitate.
If this is not done, the oxalate is carried into the lead chloro- fluoride precipitate as lead oxalate and makes it impossible to obtain a satisfactory end point in the thiocyanate titration.
This is effected by oxidation with potassium permanganate in slightly acid solution, and the manganese is precipitated as a mixture of the phosphate, carbonate, and dioxide by treatment with sodium carbonate. The filtrate from the manganese precipitate is then added to the solution obtained by water extraction of the melt resulting from the alkali fusion of the original material. The method used for the final deter
mination of fluorine is essentially the same as that described by Hoffman and Lundell. Direct treatment of phosphate rock with dilute nitric acid was quite effective in bringing the fluorine into solution, but the final results were erratic and usually much lower than those obtained by the com
bination of alkali fusion and acid extraction.
R ecovery o f F lu orine in Presence o f Pyrite and C a lciu m Su lfate
According to Hawley (4), the presence of 0.1 gram of iron sulfide (FeS2) causes a serious reduction in the recovery of fluorine from fluorspar by fusion methods. Hawley’s results indicate that this loss is reduced but is not entirely prevented by oxidizing the sulfide with a small quantity of sodium per
oxide or potassium nitrate.4 Hoffman and Lundell (7), on the other hand, did not find a significant loss of fluorine when 10 per cent of iron sulfide (FeS2) was added to an opal glass containing 5.75 per cent fluorpic.
Some types of phosphate rock, particularly Tennessee blue- rock, contain appreciable quantities of iron sulfide, princi
pally pyrite, and acid-soluble sulfate, principally gypsum.
Inasmuch as the fusion-acid extraction method in the ab
sence of oxidizing agents gave results appreciably lower than those obtained by the volatilization method on several sam
ples of Tennessee blue-rock phosphate (Table IV) experiments were made to determine whether better recovery could be obtained in the presence of oxidizing agents such as potas
sium nitrate and sodium peroxide. In making these experi
ments the 200-mesh materials were thoroughly mixed with
4 Hawley does not state whether the oxidation of the sulfide was carried out prior to or simultaneously with the fusion of the sample.
October 15, 1931 IN D U STRIAL AND ENGINEERING CHEMISTRY the fusion mixture and oxidizing agent, and the oxidation
and fusion were carried out simultaneously. The fusion-acid extraction method, as outlined later, was used.
The results given in Table III show that when no oxidizing agent was used, addition of 0.1 gram of pyrite or of gypsum had no pronounced effect on the recovery of fluorine from mixtures of tricalcium phosphate and fluorspar. The addi
tion of 0.1 gram of potassium nitrate had no effect on the re
covery of fluorine from mixtures containing only tricalcium phosphate and fluorspar, but in the presence of 0.1 gram of pyrite, the recovery from synthetic mixtures was seriously reduced by the use of either potassium nitrate or sodium per
oxide as oxidizing agents. In the case of the Tennessee blue-rock phosphate No. 930, however, addition of 0.1 gram of oxidizing agent gave, in general, a significant increase in the recovery of fluorine, but the results were still lower than those obtained by the volatilization method. This particular sample contained 5.18 per cent of SOj in the form of acid- insoluble sulfide. The addition of 0.1 gram of potassium nitrate did not improve the recovery of fluorine from Tennessee brown-rock phosphate, Bureau of Standards sample No. 56, which contained 1.18 per cent of SO3 as acid-insoluble sulfide.
T a b le II I — R e co v e r y o f F lu o r in e In P resen ce o f P yrite and Tennessee brown-rock phosphate,* Bureau o f None 33.4 Standards N o. 56, 1 g. (34.5 mg. F) N one 32.9 tricalcium phosphate, calcium fluoride, pyrite, and gypsum.
In the absence of an oxidizing agent, low and erratic results were obtained. When 0.1 gram of potassium nitrate was added to the fusion mixture, the results were more consistent but were still lower than those obtained by the volatilization method. This indicates that the low figures obtained on Tennessee blue-rock phosphate are caused by a reaction, or reactions, involving the gypsum, pyrite, and calcium fluoride, or calcium fluorphosphate, present in the rock. Roasting at temperatures up to 800° C. prior to fusion with sodium car
bonate did not improve the recovery of fluorine from Tennessee blue-rock phosphate. The results, as a whole, indicate that.
with the possible exception of Tennessee blue-rock phosphate, the recover}- of fluorine from phosphatic materials is not carbonate in 100 ml. of water and 10 ml. of ammonium hydroxide (sp. gr. 0.90). Add 5 grams of zinc oxide and heat on the steam bath until a clear solution is obtained, adding more ammonia if necessary.
Br o m o p i i e n o l Bl u e In d i c a t o r— Prepare a 0.4 per cent solu
tion by grinding the dry powder with sodium hydroxide solution (1.5 ml. of 0.1 N sodium hydroxide to 0.1 gram of powder) and diluting to the proper volume.
Le a d Ni t r a t e— It is the writers’ experience that the use of the so-called c. r. grades of lead nitrate may lead to low results for fluorine, particularly if the salt has been dried at too high a temperature or for too long a time at a lower temperature. It is advisable to test the salt by using it to precipitate lead chloro- fluoride from a solution of sodium fluoride of known purity.
If the results for fluorine are too low, the lead nitrate should be recrystallized from a solution containing 2 ml. of concentrated nitric acid to 100 ml. of water. Dry the recrystallized salt at a temperature below 100° C. and determine its behavior towards a solution of sodium fluoride of known purity.
Le a d Ci i l o r o f l u o r i d e Wa s h So l u t i o n— (a) Dissolve 10 grams of lead nitrate in 200 ml. of water. (b) Dissolve 1 gram of sodium fluoride in 100 ml. of water and add 2 ml. of con
centrated hydrochloric acid. Mix solutions a and b, allow the precipitate to settle, decant the supernatant liquid, and wash the precipitate 4 or 5 times with 200 ml. of water by dccantation.
Transfer the precipitate to a large beaker or flask, add about 1 liter of cold water, and allow to stand for 1 hour or longer with occasional stirring. Filter and keep the clear filtrate for use as a wash solution. Save the remainder of the precipitate for the preparation of further quantities of wash solution as needed.
0.1 N Si l v e r Ni t r a t e— It is preferable to standardize this solution by precipitating and weighing the chlorine as silver chloride. One milliliter of 0.1 N solution is equivalent to 0.00190 gram of fluorine.
ml. beaker, wash the residue 5 or 6 times with hot water, and evaporate the combined filtrates to a volume of 50 to 75 ml.
This solution, A, which contains the greater portion of the fluorine, is reserved for further treatment.
By means of a jet of warm water return the residue to the same beaker in which the water and carbonate digestions were made, using a total volume of about 50 ml., add 3 ml. of con
centrated nitric acid, and allow the mixture to stand for 0.5
« In the case o f Tennessee blue-rock phosphate somewhat higher results are obtained b y the addition of 0.1 gTatn of potassium nitrate or sodium peroxide to the fusion mixtures. The nitrate is preferable because it is easier to handle.
370 A N A L Y T IC A L EDITION Vol. 3, No. 4 to 1.0 hour with frequent stirring. This treatment will
usually bring all the fluorine into solution, but in case con
siderable insoluble material remains, it is advisable to filter it off, fuse with a small amount of sodium carbonate, and add the water extract of the melt to solution A . Add 50 ml.
of a 5 per cent oxalic acid solution to the cold nitric acid solution and precipitate calcium oxalate by adding 10 per cent sodium carbonate solution drop by drop until the solution is solution drop by drop until the solution becomes permanently colored, or a brown precipitate forms. Neutralize the excess acid by adding solid sodium carbonate in small quantities until frothing ceases, and then add about 2 grams of car
bonate in excess. If the precipitate is light colored, add per
manganate solution drop by drop until it becomes dark brown, and bring the mixture to boiling, stirring continuously. Filter the hot solution through a rapid-filtering 12.5- or 15-cm. paper and wash the precipitate four or five times with a hot 1 per cent sodium carbonate solution. Catch the filtrate in the beaker containing solution A and adjust the volume to about 250 ml.
The procedure from this point is essentially the same as that recommended by Hoffman and Lundell.
Heat the solution to boiling, add 25 ml. of acid zinc nitrate solution, and bring to a boil, stirring continuously to prevent bumping. The precipitate is very heavy and loss of the deter
mination is almost certain to occur unless bumping is pre
vented. Filter through a rapid-filtering 15-cm. paper into a 600-ml. beaker, wash to a volume of about 400 ml., and discard the precipitate. Neutralize the solution with nitric
vented. Filter through a rapid-filtering 15-cm. paper into a 600-ml. beaker, wash to a volume of about 400 ml., and discard the precipitate. Neutralize the solution with nitric