STUDY OF FLUOROAPATITE REACTIONS
THE EFFECT OF METAPHOSPHATE MELTS ON
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
UNIVERSITY OF FLORIDA
3 1262 08552 2877
TABLE OF CONTENTS
LIBT OF TABLTS *.
The Occurrence and Use of Apatites
The Fluorine Problem . .
The Objectives of Further Studies
SYNTHESIS OF FLUOROAPATITE . . *
* 0 6 0 0
* 0 0 0 0
* 0 0 0
0 0 0
Synthesis Methods * * * .
Recommendations of the TVA Group .
Synthesis of Chemically Pure Fluoroapatit& .
A Simplified Procedure for Fluoroapatite Synthesis
ANALYSIS OF FLUOROAPATITE
0 0 0 0 0 0 0 0 0
Preparation of Analysis Standards
Calcium Determination . .
Phosphorus Determination . .
Fluorine Determination . .
* 0 0
* 0 0 0 0
* 0 0 0 0
* 0 0 0 0
THE MOLECULAR STRUCTURE OF FLUOROAPATITE .* .
SSOLUBILITY OF FLUOROAPATITE IN AQUEOUS SOLUTIONS
General Behavior . . * *
The Effect of Sodium Carbonate Solutions on
Fluoroapatite .* * . . .
REACTIONS OF FLUOROAPATITE 'ITH A GASEOUS PHASE ,
Formation of Fluozoapatite and Hydroxyapatite
Reactions with Hydrogen-Containing Gases .*
Experimental Tests with Gases 0 * *
Reduction of Fluoroapatite by a Gas Phase *
Conclusions * * *
REACTIONS THE SOLID STATE . 0 0 0 .. 38
Types of Solid State Reactions . . * 38
Thermodynamic Considerations . . . . 39
Exchange Reactions Between a Pair of Salts . . 40
REACTIONS OF FLUOROAPATITE If MELTS .* . .* * 44
Melts of Neutral Salts * *. 44
Molts of Basic Salts * * * * 5
Melts of Acidic Salts . . . . . 5
rTTERACTION OF SODIUM METAPHCSPHATE WITH FLUORID1]S . 46
Preliminary Observations . . . 46
Literature Survey . *. . * * 46
Objectives of Study of the Metaphosphate Melts . 47
Selected Metaphosphate Systems . * . 49
Characteristics of the Evolved Gases * * 50
Construction of the Vacuum Furnaces .* * * 54
Method of Measurement with the Vacuum Furnaces * 57
Experimental Results of the Vacuum Measurements . 58
Experimental Procedure at Atmospheric Pressure * 60
Standard Mixtures and Conditions * . * 62
Experimental Results of the Measurements at 700C with
the Standard Compositions * . 0 64
Variation in Composition at 7000C. . . . 65
Temperature Dependence * * * 72
Reactions Involving Other Cations . * 72
INTERACTION OF METAPHOSPHATE WITH FLUOROAPATITE . . 75
Experimental Conditions * * * * * 75
General Results * * . * * 75
Temperature Dependence . . . * 77
Effect of Supplementary Materials , . 79
EXPER TS IN THI CRUCIBLE FURNACE . . . . 81
Solubility of Salts in Molten Sodium Motaphosphate 81
Weight Loss of Standard Mixtures and the Component
Salts * * * e 8 82
Microscopic Observations . * 83
SUMMARY OF THE EXPERIMENTAL RESULT 'ITH MTAPHiSPHATE MELTS 84
Reaction Formnllas * * *. 84
The Major Factors and Their Influence Upon the
Reactions * * . . . 84
PHYSICAL ITERPRETATIONS . . . . . 87
Metaphosphate Melts . * . * *. 87
Effect of the Type of Cation * * 88
Extension of the Reaction Concept * * * 89
CONCLUSION. * . 90
BIBLIOGRAPHY . . . . . . 93
ACKOcTI DGET:IFJS * . * 0 * 97
BIOGC;A'IC/L NOTE . . . . 98
LIST OF TABLES
1 Composition of Pure Fluoroapatite Samples . 14
2 Composition of Fluoroapatite Samples Prepared by
the Simplified Procedure . . . 16
3 Free Energies of Formation of Fluoroapatite and
Hydroxyapatite at Four Temperatures .* * 31
4 Standard Free Energy Changes of Reactions of
Fluoroapatite with a Gaseous Phaso .* * 34
5 Rate of Gas Evolution from Metaphosphate Melts in
the Vacuum Furnace * * * * 59
6 Effect of Experimental Conditions at 7000C. on the
Evolution of Fluorine-Containing Gases from
Metaphosphate Melts . . . . 66
7 Compositions of the Experimental Metaphosphate-
Fluoride Mixtures . . 69
8 Evolution of Fluorine-Containing Gases from
Metaphosphate Melts at Various Compositions 73
9 Evolution of Fluorine-Containing Gases from
Metaphosphate Melts at Various Temperatures 73
10 Evolution of Fluorine-Containinp Gases from
Metaphosphate Melts in the Presence of Other
Cations * . . 74
11 Compositions of Metaphosphate Melts Containing
Fluoroapatite *. . * . 76
12 Evolution of Fluorine-Containing Gases from Meta-
phosphate Melts Containing Fluoroapatite 78
The Occurrence and Use of Apatites
One of man's most useful minerals is phosphate rock. Its major
use as the source for phosphate fertilizers is well known. Phosphate
rock is primarily composed of a group of minerals called apatites, the
most typical representative of which is fluoroapatitel with the formula
Ca10F2 (PO)6. Natural apatites vary greatly not only in kind and variety
of impurities present, but also in the composition of the apatite itself.
This is true even for two mineral samples from the same deposit.
The phosphate rock deposits of varying composition can be found
in many places throughout the world. The North American continent, how-
ever, is particularly rich in this mineral, which in this case consists
essentially of submicrocrystalline fluoroapatite containing some excess
of fluorine (27), and is usually found in an admixture of iron, aluminum,
and silica compounds.
The structure of crystalline apatites has been studied, and it
has proved to be remarkably stable, permitting a variety of unusual types
of substitutions involving a considerable number of ions. The most im-
portant of these changes occur where the fluoride ion in fluoroapatite is
In this work the prefix "fluoro-" is applied in the term
"fluoroapatite" in conformity with the general usage of this prefix to
designate compounds containing fluorine (e.g., fluoroaluminates). The
Tennessee Valley Authority investigators use "fluorapatite", and some
substituted partly or fully by hydroxyl, chlorine or carbonate Croups,
forming the respective apatites. The occurrence of extensive series of
solid solutions among these apatite type structures are also very common.
The apatites are readily formed in solid state reactions as voll as
in aqueous solutions in which they are quite stable. For example, a
suspension of hydrated tricalcium phosphate may be used to induce the
formation of fluoroapatite by the precipitation of solute fluoride ions,
thus effecting complete removal of fluorides from ground waters (49).
Ever since the phosphate fertilizer industry existed, numerous in-
vestigators have been seeking better industrial procedures for processing
phosphate rock, but in recent decades more attention is being paid to
these materials as chemical systems. Basic studies have been started,
and are being carried out by several independent groups, each interested
in their omw particular applications. Considering the complexity of
phosphate systems, the research done until now has been only a modest
beginning in disentangling the variety of problems related to these
In this country the Tennessee Valley Authority has done extensive
and excellent work in the field of phosphates in developing new industrial
procedures as well as conducting basic studies of the phosphatic compounds.
The major purpose of their work has been the increase of the phosphorus
availability in the processed materials. The results of their findings
have been made public in a series of reports and articles (18b, 68).
The study of the structure of apatites has deserved considerable
other authors prefer "fluor-apatite" or "fluoro-apatite".
attention on the part of medical people and crystallographers particu-
larly since the discovery that the mineral constituent of bone and of
the enamel and dentine of teeth is essentially hydroxyapatite, as
established by X-ray crystal analysis (h).
In recent years a number of new practical applications have been
found for synthetical apatite type materials. For example, the alkaline
earth halophosphates possess very favorable characteristics as fluore-
scent materials, and are already in wide-apread use in fluorescent lamps
in England (55).
The Fluorine Problem
In the near future ride attention will be directed to the fact that
the North-American phosphate rock constitutes a practically inexhaustible
source for fluorine-containing compounds. Presently this element is won
from fluorspar (CaF2), but since the domestic resources of this mineral
in highgrade purity are being rapidly depleted, it has been placed on the
list of critical materials by the Strategic Materials Committee of the
U. S. Government (21), and rccomended for stockpiling.
At this time no efficient and economical industrial process is
known for the extraction of fluorine from phosphate rock.
The fertilizer industry has long been concerned about this problem
for its own reasons, since the presence of fluoride exerts a deleterious
effect on the availability of phosphorus in the soil and acts as a
poison if fluorine-containing phosphates are used as cattle-feed
The known and practiced methods for the extraction of fluorine-
containing components from phosphate rock are:
1. dissolution of the mineral in strong aqueous acids, and the
precipitation of fluorine as fluoride, or silicofluoride (18b);
2. treatment of phosphate rock with sulfuric acid and heating (23);
3. defluorination of the mineral at high temperatures, above
1300C., in the presence of silica and water vapor (17, 67).
Part of the extracted fluorine, in the form of hydrogen fluoride
and silicon tetrafluoride, can be recovered from the flue gases by the
adsorption of hydrogen fluoride in (18a) lime-packed towers, and by
scrubbing silicon tetrafluoride with alkaline solution (52). In the
latter case fluorosilicates will be formed.
The above-mentioned procedures have been used only to a minor
extent, and it is improbable that any of them could be adopted to a
wide-scale economic industrial process for the purpose of fluorine
The Objectives of Further Studies
More knowledge about the apatite systems is necessary for the
development of now approaches uhich would enable more efficient bene-
ficiation of all the materials present in phosphate rock. This can be
accomplished only by thorough fundamental studies, which should be
extended to a large number of phosphate systems under a variety of
Almost all previous work has been done on natural minerals, and
although much of a general nature has been published very little precise
scientific knowledge is available on phosphate rock. A reproducible
fluoroapatite has only been prepared within the past several years and
the rates of reactions of this pure material with other chemicals have
not yet been studied.
The contribution of a small group of investigators must of necessity
remain rather limited considering the demands of the field. It was,
however, hoped that this work would help to emphasize the need to look
at the problem as a whole, and add to the basic knowledge concerning
these important materials.
The general objective was set to initiate a broad program of study
of apatites and other related materials for the purpose of finding new
basic facts about the physical properties of well defined phosphatic
systems, their reactivities and rates of reactions. The coordination of
the available information from various sources would form a part of this
This dissertation summarizes parts of the work which was carried
out with the purpose of studying the possibilities of novel fluoroapatite
reactions below 1000C,. with special attention to the behavior of fluorine.
One of the favorable decomposition reactions was then selected for a
Furthermore, there was the preliminary task of choosing a feasible
method for the production of chemically pure fluoroapatite, and the
comparison of suitable analysis methods for these materials.
The subject matter in this dissertation is composed of three in-
In the following three chapters the synthesis, the analysis and the
crystallographic properties of fluoroapatite are described.
The chapters in the middle section discuss the general possibilities
of fluoroapatite reactions with chemicals in various phases. The refer-
ences to the known reactions have been included, and the results of a
variety of experiments carried out in this laboratory have been
presented, he standard free energy changes vere calculated for a
number of reactions and the results have been given.
The third and major part of the dissertation describes and dis-
cusses the interaction of metaphosphate melts with fluorine-containing
substances. The experimental procedures for the study of the special
reactions were developed in this laboratory.
SYNTIHESI OF FLUOROAPATITE
Tricalcium phosphate combines readily with pulverulent calcium
fluoride forming mixtures which register characteristic properties of
fluoroapatite. Numerous investigators have observed this reaction under
The process may take place in aqueous solutions, although in this
case the precipitate is always contaminated with other phosphates (6).
Stoichiometric amounts of tricalcium phosphate and calcium fluor-
ide, intimately mixed, produce fluoroapatite after gradual heating of the
mixture (5) to 9500C.
Fluoroapatite has been prepared by the reaction of hydroxyapatite
with calcium fluoride in the solid state (8) above 6000C.
Wallaeys has synthesized mixed fluoro-, chloro-, and hydroxyapa-
tites. The structure and isomorphism of these compounds are described
The phase diagram of the system calcium fluoride tricalcium
phosphate has been studied by R. Nacken. According to him fluoroapatite
will be formed in a molten mixture of calcium fluoride with an excess of
tricalcium phosphate (50, 5S4).
In the foregoing preparations the identity of fluoroapatite was
usually established by X-ray analysis, and little was mentioned about the
chemical purity of the products or the presence of other phases. In
this respect the latest and the most reliable procedure has been given
by the TVA scientists under the direction of Kelly L. Elmore, Chief of
Research Branch, Division of Chemical Development. It was decided to
adopt their method for the synthesis of fluoroapatite, hydroxyapatite,
and tricalcium phosphate as described by Dr. K. L. Eliore (14).
Recommendations of the TVA Group
The folloirting directions are an excerpt from a personal letter by
Dr. K. L. Elmore, April 1, 1953.
Prepare a stock of raw materials as folloust
Calcium Carbonate: Recrystallize reagent-grade calcium nitrate
and precipitate calcium carbonate with ammonium carbonate.
Calcium Fluoride: Distill reagent-grade hydrofluoric acid in
a platinum still and precipitate calcium fluoride with precipitated
Calcium Metaphosphate: Recrystallize reagent-grade monocalcium
phosphate monohydrate. Heat the monocalcium phosphate carefully
to 6000C. in a platinum boat to form crystalline calcium
Tricalcium Phosphate: Mix calcium metaphosphate and calcium
carbonate in stoichiometric proportions to form tricalcium
phosphate. Heat the mixture in a platinum boat at 1100C, for
five hours. It may be necessary to make adjustments in composition.
Hydroxyapatite: Precipitate a composition approximating
hydroxyapatite by the method of Rathje (57). Dry the precipitate
at 105C. and then heat in a platinum boat at 9500C. in an
atmosphere of 20 per cent steam and 80 per cent nitrogen for at
least 72 hours to promote crystal growth.
Fluoroapatitoe Prepare a stoichiometric mixture of tricalcium
phosphate and calcium fluoride. Heat the mixture in a platinum
boat at 13500C. for at least 30 minutes in a stream of dry
nitrogen. Again, adjustments of composition may be necessary.
Complete reaction requires grinding and intimate mixing. The
resulting products have small crystal size. In a separate
publication (13) the conditions are described by which large,
well defined crystals of fluoroapatik can be produced from
sodium fluoride flux in a vacuum furnace heated by induction.
Difficulties were encountered in keeping the crystals completely
free of flux. In none of the preparations was it advisable to
prepare more than fifty grams in one batch.
The first task of the present study was to prepare various samples
of fine crystalline fluoroapatite according to the TVA recommendations.
The directions given above are quite generally worded, and it was found
that an extensive review of the literature on the subject along with
actual experiments had to be carried out in order to determine the in-
fluencing factors in the preparation of the materials needed.
The procedure for fluoroapatite synthesis can be divided into seven
1. Recrystallization of monocalcium phosphate monohydrate.
2. Conversion of monocalcium phosphate monohydrate into metaphosph-
3. Recrystallization of calcium nitrate tetrahydrate.
4. Distillation of an-monium carbonate and precipitation of
5. Ignition for formation of tricalcium phosphate.
6. Precipitation of calcium fluoride,
7. Ignition for formation of fluoroapatite.
In the following section some of the essential details of each step
will be pointed out along with major references for background infor-
Synthesis of Chemically Pure Fluoroapatite
Recrystallization of Monocalcium Phosphate Monohydrate.-The mono-
calcium phosphate obtainable on the market usually contains free phos-
phoric acid, sometimes up to $ per cent, although the salt may appear
quite dry. If the commercial material contains only monocalcium
phosphate as a solid phase, a satisfactory acid-free product can be
obtained by washing out the free acid with a suitable organic solvent.
Frequently, however, the C.P, salt carries a small. quantity of dical-
cium phosphate, which can be removed only by recrystallization from a
solution of the proper composition (9).
Monocalcium phosphate monohydrate, Fisher Certified Reagent,
was used in this laboratory, and recrystallized according to the direc-
tions of V. L. Hill and S. B. Hendricks (32).
It is difficult to remove every trace of excess phosphoric acid
from the crystals. Washing with alcohol and ether has been used for
removing (64) the excess acid, but acetone (31, 60) is to be preferred.
Acetone was dried over anhydrous potassium carbonate and filtered before
Conversion of Monocalcium Phosphate Monohydrate into
Metaphosphate.-Heating of monocalcium phosphate produces varying results,
depending upon whether charges are heated directly or in a step-wise
fashion to the selected temperature. Step-wise heating gives somewhat
more reproducible results, particularly when hydrated monocalcium phos-
phate is used as a starting material. This salt, when it is heated
rapidly, melts incongruently near 150C., and the accompanying rapid
loss of water vapor causes frothing; frequently with loss of material
from the container, and formation of a tough mass. Fusion can be
avoided (33) by first heating the charge at 1250C. to expel water of
crystallization, but further heating in the range 2000 to 6000C. yields
an unpredictable mixture of phases consisting of glasslike amorphous
material, and one or more of at least three crystalline phases. This
mixture, however, can be converted to stable (-calcium metaphosphate by
heating between 600 and 70000C,
In this laboratory, about a fifty gram lot of pure, acid-free
monocalcium phosphate monohydrate was heated, first at 1250C. for two
hours to expel water of crystallization, and then at 4000C. for two
hours; after reheating between 600 and 70000., the substance was cooled
in air. The product obtained by this method (34) had a sandy crystalline
appearance and could be easily pulverized.
Recrystallization of Calcium Nitrate Tetrahydrate.-Calcium
nitrate, Mallinckrodt Analytical Reagent, contains as major impurity a
total of 0.20 per cent of magnesium and alkali salts. Richards and
Honigschmid proved (50) that such impurities can be removed by the re-
crystallization of the nitrate.
Calcium nitrate is a deliquescent and very soluble salt with a
great tendency to form supersaturated solutions. Owing to the ready
solubility of calcium nitrate in organic solvents, the crystals could not
be freed from solution by washing with alcohol, acetone, pyridine, etc.
In this laboratory, about 203 grams of calcium nitrate tetra-
hydrate were dissolved in 25 ml. of water in a platinum dish and heated
to about 1000C. The solution was cooled overnight in a refrigerator and
seeded with a small crystal of calcium nitrate tetrahydrate. The crystal
formation progressed very rapidly in cool solutions. Suction on a
fritted-glass funnel was applied to separate the mother liquor from the
Distillation of Anmonium Carbonate and Precipitation of
Calcium Carbonate.-Pure annonium carbonate, Mallinckrodt Analytical
Reagent, was distilled in a glass apparatus and condensed in a flask
with ice-cold water. The fresh a.monium carbonate solution was added
to the calcium nitrate solution, and the precipitated calcium carbonate
was filtered and washed with plenty of distilled water. The product
was preserved in a platinum dich in which it was first heated to 105$C.
for a couple of hours and then at 00C. for a short time.
Ignition for Formation of Tricalcium Phosphate.-Calciu- meta-
phosphate and calcium carbonate were mixed in stoichiometric proportions
and heated in a platinum shell at 1150C, for five hours. There was
always some loss of phosphorus durinC the ignition, thus the Ca/P ratio
in the product was usually somewhat higher than the theoretical value,
The ratio can be corrected by an addition of a right amount of
calcium metaphosphate to the product and by the repetition of furnace
The necessary quantity of calcium metaphosphate can be calculated
from the following equation:
W(Ca) + r(Ca) X 1.906
W(P) + r(P) X
W(Ca) weight of calcium in the sample by analysis
W(P) weight of phosphorus in the sample by analysis
r(Ca) ratio of Ca/Ca(P03)2 0,20238
r(P) ratio of P2/Ca(PO3)2 0.31290
X amount of calcium metaphosphate necessary
After rearrangement and insertion of the numerical values the
formula has this expression:
w(Ca) 1.9406 W(P) ,.
For example, in one case 0.060 gr. of calcium metaphosphate had
to be added to five grams of tricalcium mretaphosphate to obtain the
correct Ca/P ratio which is 1.9406.
Precipitation of Calcium Fluoride.-Crystalline calcium fluoride
was produced by adding a slight stoichiometric excess of hydrogen fluoride
in aqueous solution to a suspension of precipitated calcium carbonate.
'Ieighings and other operations were carried out in platinum dishes. The
product was washed with abundant water, dried at 1100C. for several hours,
and then heated to about 7000C. until the weight remained constant.
At higher temperatures, calcium fluoride volatilizes in notice-
able quantities. Thus the loss in grams per sq. cm. of surface for 24
hours of calcium fluoride heated in air has been reported (2) to be 6.3
ag. at 12500C. and 23.5 mg. at 1400C.
Ignition for Formation of Fluoroapatite.-The heating unit in use
was a Burrell High Temperature Electric Furnace, Model L-G-M, maximum
Stoichiometric mixtures of tricalcium phosphate and calcium
fluoride were heated in a platinum shell for about an hour at 13500C.
The charge was covered by another platinum shell which in its inverted
position rested with the edge on the even floor of the furnace, whereas
its bottom touched the upper edge of the first shell. Thus, there was
formed a doubly isolated space with hindered diffusion between the space
above the charge and the furnace atmosphere. The first shell with the
charge was adjusted a little above the floor of the furnace, supported
by pieces of firebrick, and underneath it was a platinum cover with
calcium fluoride in order to provide a vapor phase of the salt at high
temperatures. The heating proceeded gradually in the course of about
ten hours. Nitrogen, dried by the passage through a long tube of silica
gel was led into the furnace during the ignition, primarily for the
purpose of keeping the furnace atmosphere as dry as possible.
The results of various samples indicated that some fluorine-
containing compound had been volatilized during the treatment, and that
usually the phosphorus content was also lowered as compared with the
Three mixtures were then prepared containing an excess of calcium
fluoride 105, 110, and 115 per cent respectively, as compared with
the stoichiometric requirement. The following table presents the
compositions of the three preparations.
COMPOSITION OF PURE FLUOROAPATITE SAMPLES
Percentage by Weight of the Component Elements
Calcium Phosphorus Fluorine
O0.10 1 0.10 10.08
105% CaF2 39.94 18.32 3.56
110% CaF2 io.o0 18.20 3.74
115% CaF2 40.25 18.18 3.96
Theoretical 39.735 18.428 3.767
By adding ten per cent excess calcium fluoride, the Ca/P ratio
has to be newly adjusted by a proper amount of calcium metaphosphate.
This quantity can be calculated from a formula expressing the Ca/P ratio
The initial formula is quite long; after the substitution of
numerical values and contraction it is:
0.0090866 W(Ca-P) X
W(Ca-P) = weight of tricalcium phosphate in grams
X additional amount of calcium metaphosphate
needed in grams
A Simplified Procedure for Fluoroapatite Synthesis
The procedure described previously for the synthesis of
chemically pure fluoroapatite yields good results, but the preparation
of starting materials is laborious. Need was felt for a simplified
procedure which would facilitate the preparation of fluoroapatite in
sizeable quantities with good quality and acceptable purity.
A new procedure was developed which utilized the commercially
available quality chemicals and dispensed with the recrystallization of
Materials,-The following materials were used monocalcium phos-
phate monohydrate, Mallinckrodt's Analytical Reagent, was held in dry
acetone over a period of three months. First, the solvent was changed
every other day; afterwards once a week. Acetone was removed carefully
from the salt by drying at 500C.
Mallinckrodt's calcium carbonate (300 g.) and Baker's calcium
fluoride (200 g.), both Analytical Reagents, were treated for three
months rith approximately twenty liters of water, in each case, at room
temperature. After the treatment, the salts were dried at 140OC. and
calcium fluoride was further ignited at 6000C. for two hours.
Results.-Using these three salts as starting materials, five lots
of fluoroapatite were prepared following the standard heating procedures.
The appearance and chemical properties of the new products were identical
with those of the previous preparations after the first ignition. The
lots were designated by letters a-e. The analysis results are presented
in the following table,
OF FLUOROAPATITE SAMPLES PR PARED
THE SIMPLIFIED PROCEDURE
Percentage by Weight of the Component Elements
P Calcium Phosphorus Fluorine
o0.10 1 0.10 10.03
a 2.137 39.42 18.36 3.42
b 2.145 39.47 18.41 3.42
c 2.184 39.11 17.50 3.60
d 2.179 39.66 18.20 3.67
e 2.203 39.78 18.06 3.65
retical 2.1563 39.735 18.428 3.77
Identification of Fluoroapatite.-*-icroscopic observations
rendered positive proof concerning the identity of the products, and gave
an indication regarding the uniformity of the materials.
The available instrument was a Bausch & Loob Petrographic Micro-
scope, Model LC, with accessories.
Fine grains of the material were irnersed in various oils of
known refractive indices, and the movement of Becke lines was observed.
The small crystal size prohibited the determination of the interference
figures, but the characteristic birefringence could easily be followed.
Weak birefringence persisted uniformly throughout the material.
All preparations exnibited identical crystallographic ana optical
properties. The mwas of the material possessed a refractive index in the
range of 1.630 to 1.632, which agrees with the values,,/u= 1.6325 and
,-r 1.630, reported by R. Nacken ($4) for artificial fluoroapatite.
Minor amounts (a fraction of one per cent) of different crystals
were observed, the optical properties of which suggested the presence of
some unroacted calcium fluoride and tricalcium phosphate.
The material can be cleaned by treating it first with sodium car-
bonate solution (2 g. in 100 ml.) at aboat 700C. for several hours, with
subsequent standing in a two per cent citric acid solution at room tem-
perature. The total loss of the material is small, and the product has a
uniform appearance under the raicroscope as clcn fluoroapatite.
ANALYSEB OF FLUOROIPATITE
The analysis of fluoroapaLite constitutes a problem of deter-
mining quantitatively phosphorus, calcium, and fluorine in the presence
of one another, and practically in the absence of other elements except
in the smallest quantities.
A very extensive literature has been accumulated on the subject
of the analysis of phosphate rock and phosphatic fertilizers. The
described procedures are quite involved because of the necessity of
taking account of the numerous interfering elements.
In the case of pure fluoroapatite, one is not concerned with
such a multitude of co*mronents, and it was hoped that simpler and less
time-consinuln procedures could be found. For tnat purpose, the favor-
able features of various methods were combined and tested on the phos-
phatic materials. Only volumetric and graviretric determrtnations were
considered. Due attention was paid to the widely used official or
recornended procedures of the Bureau of Standards and of the Association
of Official Agricultural Cacmists.
Preparation of Analysis Standards
Calcium,-Calcium carbonate, C.P. Reagent, can be used with
advantage as a standard substance after washing with ample distilled
water and drying at 1100C. to constant weight.
Phosphorus.-Potassium dihydrogen phosphate is being widely used
as a standard substance for preparing solutions of known phosphate con-
centration. Purification beyond the reagent grade can be effected by
recrystallization (37). As a test for the purity of the material, the
loss on ignition should not be less than 13.13 per cent nor more than
13.33 per cent.
Fluorino.-Sodium fluoride, Reagent Grade, dried for one hour at
1050C., offers a satisfactory standard for fluorine determinations.
Practically all methods for calcium determination, whether by
volumetric or gravimetric procedures, start with the precipitation of
calcium as calcium oxalate monohydrate. It has been shcwn experimentally
that calcium ion can be quantitatively precipitated in the presence of
phosphate and fluoride ions (up to about 5% fluorine), and some other
elements as they usually occur in the phosphate rock.
Seven methods of calcium precipitation were recorded, and five
of them (25, 35a, 36b, 46, 72) tested. The procedures differed from
each other in the volume of the sample in the concentrations of the rea-
gents, the indicators used, and the temperature of ammonium oxalate
during the addition.
Volumetric Determination.-Precipitation (l4a) followed by titra-
tion according to the description by Kolthoff and Sandell (l4b), is a
simple and recommendable procedure. The accuracy is about one per cent
or somewhat better.
The procedure by Lingano (46) is the latest and claims the
highest accuracy, about 0.1 per cent. The method is relatively fast, but
needs more experimental verification.
In determining the proper endpoint of the titration with perman-
ganate, the recommendations of Fowler and Bright (22) should be observed.
Gravimetric Determination.-Under suitable precipitation condi-
tions, calcium can be weighed in the form of calcium oxalate monohydrate
dried at room temperature, according to Sandell and Kolthoff (41b).
Accuracy is about 0.2 per cent.
washburn and Shear (72) dried the precipitate at 105C. and
obtained good results as confirmed by the experiments in this laboratory.
Accuracy is approximately the same as in the previous case.
The most accurate weighing form for calcium ion is considered to
be calcium carbonate. The accuracy amounts to 0.1 per cent or less in
the absence of interfering ions (7h). The precipitate has to be heated
at about 5000C. for one hour.
Calcium sulfate is a good weighing form for calcium (l4a) if it
can be directly precipitated from the solution free from other earth
alkali elements. The ignition temperature does not have to be closely
observed (the range is from 1050 to 8900.).
This method was used most frequently in this laboratory with
good results which ;ere checked by the carbonate method. Accuracy was
about 0,1 per cent.
In the presence of calcium, two courses are open for the preci-
pitation of phosphorus in tho form of orthophosphate.
1) The precipitation of maWnesium ammonium phosphate may be
made in the presence of much ammonium citrate, which prevents the pre-
cipitation of calcium and other phosphates.
2) Phosphorus may first be separated from the interfering
elements by precipitation as ammonium phosphomolybdate in acid medium.
This precipitate may then be dissolved in ammonia and magnesium amumonium
The first method as described by Hoffman and Lundell (36a)
promises to be a more rapid procedure and gives (h7) equally accurate
The second method has been in most wide use in many practical
procedures for separating phosphorus from a multitude of interfering
elements. Besides, it is the basis for the ordinary volumetric deter-
mination of phosphorus (1).
Gravimetric Determination.-The most accurate results, within 0.1
per cent of the total phosphate, are obtained by igniting the ammonium
magnesium phosphate at high temperatures to pyrophosphate (35b).
Alternatively, the ammonium magnesium phosphate hexahydrate may
be weighed directly after drying the precipitate at room temperature (19),
or at about 370C., according to NJashburn and Shear (72) who obtained good
accuracy (about 0.2 per cent). This method, however, as employed in
this work was not as reliable as the pyrophosphate procedure, primarily
because of the difficulties in obtaining a reproducible tare weight for
the Pyrex fritted glass crucibles. In the course of ordinary phosphorus
analysis, the fritted glass crucibles lost first up to 1.0 mg. in weight
per determination and later on about 0.1 to 0.5 mg. per determination.
These crucibles were also somewhat hygroscopic and gained in weight up
to one milligram standing in room temperature.
Volumetric Determination.-In the volumetric method the ammonium
pphophomolybdate is dissolved in a known amount of standard sodium hy-
droxide solution, and the excess of the latter is back-titrated with
standard acid, phenolphthalein being used as indicator. The accuracy of
the volumetric procedure is hardly any better than one per cent. Several
attempts have been made to improve the procedure, for example, Kassner
(39) claims very good accuracy, about 0.1 per cent, for his method
employing mixed indicators.
For the determination of fluorine in phosphatic materials, the
most tested and the most widely used method is that outlined by Willard
and V'inter (75), which has undergone numerous modifications by a host of
other investigators. A good summary about the modifications of the
Millard and Uinter fluorine determination method has been given by
keynolds and Hill (58). The method used in this laboratory corresponded
closely to the procedure as given by Willard and Winter and modified by
Reynolds and Hill.
THE MOLECULAR STRUCTURE OF 'L'Jr.JOA2ATITE
The unit cell and space group of the crystal lattice of fluoro-
apatite have been determined by the following investigators:
Leonhardt (h5), St. Naray-Szabo (63), Mehmel (50), Beevers and
Apatite is a typical example of the crystal class C6h; etching
experiments and Laue photographs confirm this symmetry. The unit cell
contains 42 atoms, ordinarily written in the form CaF2.3 Ca3(PO)2 .
The unit cell of the apatite structure has two equal edges inclined at
1200 to one another. The length of these edces is 9.37 Ao in the case
of fluoroapatite. The third edge is at right angles to these and has a
length of 6.88 A,
The cell possesses vertical symmetry axes of three kinds:
1) passing through the points (two-thirds, one-third) and (one-third,
two-thirds) of the cell are three-fold rotation axes; 2) passing through
the corners of the unit cell are hexagonal screw axes; and 3) passing
through the halfway points of the cell edges and through the center are
two-fold screw axes.
There are also eight centers of symmetry and two mirror planes
parallel to the xy plane.
The atomicI surroundings are as follows:
Fluorine three Ca at one level
Phosphorus tetrahedron of four 0
Calcium I (on three-fold axis) nine 0
Calcium II (on reflection planes) an irregular
polyhedron of one F and five 0
Oxygen tetrahedron of one P and three Ca
While investigating the compositions of some western phosphate
rocks Hendricks, et al. (28), carried out a number of X-ray diffraction
measurements on the group of compounds structurally similar to fluoro-
apatite. The compositions and the crystallographic properties of these
minerals and of some artificial products support the suggestion that the
fluoride ion in fluoroapatite can isomorphously be replaced by carbonate,
sulfate, silicate, or hydroxyl groups; and by oxygen, chlorine, bromine,
or iodine ions.
Furthermore, the crystallographers have provided evidence that
the calcium ion in the lattice can be replaced completely or partly by
the ions of the folloirinr elements: Be, Mg, Sr, Ba, Pb, Zn, Ni, Cd, Cu,
Fe, 1n, Al, Mo, or Cr; and the phosphorus ion by V, As, Sb, S, C, Si, or
Fluoroapatite is the stablest of all these substituted structures
both thermally and chemically. The specific feature of the fluoroapatite
structure is the arrangement of calcium and phosphate ions in such a way
that channels exist within the crystal into which the fluorine ions fit
1The elements are present in the form of ions: F- 0-2 ,
with close packing. This fact evidently accounts for the remarkable
stability of fluoroapatite, and makes it the substance to which the
numerous other structurally similar compounds can be compared in their
chemical reactivity and physical properties.
SOLUBILITY OF FL'UJ.rAP2ITE EI AQUTJUS SOLUTIONS
Fluoroapatite can be regarded as a basic substance due to its
high Ca to P ratio, and therefore it may be expected to react more
readily with acidic materials.
In aqueous solutions, the mineral acids decompose fluoroapatite
slowly at room temperature, faster at higher temperatures.
TVA is currently carrying out experiments on the solubility of
fluoroapatite in various concentrations of phosphoric acid at 2$5C. and
50C., including its rate of solution at a pH rarge between one and
The solubility of fluoroapatite in solutions of organic acids is
rather limited. No reliable data are available yet about such systems.
Alkaline solutions under ordinary conditions do not affect
fluoroapatite appreci bly.
In respect to the reactions of alkaline solutions with naturally
occurring materials, the latest developments are the studies of Japanese
investigators (38) concerning the effect of alkali treatment of phosphate
rock by extracting fluorine with sodium hydroxide solution under pressure.
Fluoroapatite is very slightly soluble in water. The solubility
product, calculated from heat capacity and heat of solution is 3.73x10-117
at 250C. This value was calculated by the TVA investigators (15).
The Effect of Sodium Carbonate Solutions on Fluoroapatite
In the course of the preparatory work in this laboratory, there
was some interest in finding a solution of suitable composition which
would dissolve crystalline calcium fluoride in preference to fluoro-
apatite. The data on the relative solubilities of these two materials
in organic acids are incomplete, ani1 since there was no intention on our
part to study the solubility of fluoroapatite in acidic solutions, con-
sideration was given to sodium carbonate solutions. A series of experi-
ments were carried out in order to determine the effect of sodium car-
bonate solutions on finely powdered fluoroapatite unde' varied conditions.
The results are presented in the following section.
Experimental Results. -The extent of fluoroapatite solubility was
followed by -, osphate analysis. The materials wre treated with sodium
carbonate solution and filtered, The phosphate was precipitated in the
filtrate with a~noni~m molybdate. After washing with dilute potassium
nitrate solution, the precipitate was titrated with standard sodium
Tro grades of fineness of apatite powder a) 250-300 mesh, and
b) finer than 300 mesh, were used for the following experiments.
Five grams of each material were let stand for three weeks at
room temperature in 100 ml. of water with two grars of sodium carbonate,
with occasional stirring The solid matter was separated from the
solution by filtration. The phosphate was precipitated in the filtrate,
and the titration gave the following values: a) 1.1 mg. and b) 1.5 mg.
The residue on the filter paper was similarly treated for five
more weeks. The respective titration results were a) 2.7 mg. and
b) 2.8 mg. phosphorus.
In the next experiment, the two solutions with 1) 1.0 g. and
2) 0.5 g. of fluoroapatite (250-300 mesh) were boiled with two grams of
sodium carbonate in 100 ml. of solution for two hours and filtered at
room temperature. The results were: 1) 0.9 mg. and 2) 0.8 mg. phos-
phorus determined in the filtrates.
Repeating the procedure with the residues after filtration and
with new solutions, the following values were found: 1) 0.8 mg. and
2) 0.6 mr. phosphorus.
A higher carbonate concentration increased the phosphate concen-
tration in the solution, and filtering of the solution while hot gave
somewhat higher values.
The amount of phosphorus determined in the solution after treat-
ment was in the order of one to two per cent of the total phosphorus
present in the fluoroapatite sample (184 mg. in one gram of fluoroapatite)
Conclusions.-The results lead to the conclusion that sodium car-
bonate solutions affect the solubility of fluoroapatite. The total
effect does not exceed two per cent with one treatment at the given con-
centration of sodium carbonate.
Several factors influenced the rate and extent of the increased
solubility of fluoroapatite, but generally there seemed to be an equili-
brium value of about two to three milligrams of phosphorus in 100 ml of a
solution containing two grams of sodium carbonate. This value was
approached faster on boiling of the solution, and slower upon standing at
Solubility of Calcium Fluoride and Tricalcium Phosphate. -The
solubility of calcium fluoride and tricalcium phosphate was also deter-
mined in similar sodium carbonate solutions.
After boiling 0.100 g. of calcium fluoride with one gram of
sodium carbonate for two hours in 200 ml. of water, 75 per cent of the
total fluorine could be titrated with thorium nitrate.
Similarly, tricalcium phosphate is partly decomposed by sodium
carbonate. A sample of 0.930 g. was boiled for two hours with two
grams of sodium carbonate in 100 ml. of solution. Phosphorus precipi-
tated. in the filtrate amounted to 13 mg., that is, 1l.3 per cent of the
REACTIONS OF FLJ,3OAPATIi'Z KITH A CASEOUS PHASE
The possibilities of fluoroapatite reactions with a gaseous
phase and the formation of gaseous products in such reactions are dis-
cussed in this chapter. A number of standard free energy changes were
calculated and are tabulated for reactions involving materials for which
the thermodynamic data were available. The major sources for the thermo-
dynamic values used throughout this chapter are listed in the biblio-
graphy (40, 43, 53, 56, 61).
First it was necessary to calculate the free energies of forma-
tion of fluoroapatite and hydroxyapatite.
Formation of Fluoroapatite and Hydroxyapatite
The values for the standard free energy of formation, and the
standard heat of formation of fluoroapatite and hydroxyapatite have not
yet been published by K. L. Elmore, et al., who have determined the heat
capacities and the solubilities (15, 16a,b) of these two apatites.
The data, however, can be calculated from the available solubility
constants according to the equations:
A Fo(aq.) -aF(c) -RT In Ksop (at 298.160K.)
6 F(c) -= Hapatite T(Sopatite- 1SO elements)
By utilizing the heat capacity data, the free energy functions
of these two apatites can be formed and used up to 15000K. For the sake
of simplicity, the numerical values for only four temperatures have
FREE ENERGIES OF FO iION OF FLUOROAPATITE AND
!IYD ZYAPATITE AT FOUR T"Ti.ATIRES
Z Ho Fo Per Gram Formula Height
298.160K. 298.160K. 50OK. 10000K. 15000K.
Fluoroapatite 3266.6 3083.4 2960.8 2646.9 2330.!5
Hydroxyapatite 3212.h 3020.2 2891.4 2561.6 2232.1
The conventional standard states were used for the elements
except that the standard state for phosphorus was the metastable solid
white modification at 298.160K., and gaseous P4 at higher temperatures.
The standard free energy changes for the reaction
3 Ca3(i'O)2 + CaF2 .F CalO2(PO4)6
at the four temperatures are:-
a Fo(kcal.): 9.6 (298.160K.), 9.2 (5000K.)
-10.5 (1000K.), 4.o (15000K.)
The standard free energy values for the above reaction exhibit a
maximum which is probably due to the phase changes of calcium fluoride
and tricalciun phosphate near 1000K.
The corresponding reaction for the formation of hydroxyapatite
3 Ca3(PO4)2 + Ca(OH)2 -.. Ca10(OH)2(PO)6
gives the following standard free energy changes:-
A oF(kcal.)t 9.6 (298.16K.), 9.6 (500K.)
-11.7 (1000K.), -19.3 (5oo00K.)
For the higher temperatures it is assumed that H of Ca(OH)2
does not change and that the change in heat capacity of reactants and
products is zero.
Reactions with Hydrogen-Containing Gases
The first group of reactants to be considered consisted of
hydrogen-containing gases, and the reaction was assumed to produce gas-
eous hydrogen fluoride. The numerical values of the standard free
energy changes of the following reactions designated by small letters,
are presented for the four temperatures in Table 4.
The conversion of fluoroapatite into hydroxyapatite upon reaction
with water vapor
(a) CalOF2(PO4)6 + 2 H20 -- Cal(OH)2(PO4)6 + 2 HF
evidently does not proceed in the given temperature range, a conclusion
which is supported by experimental evidence. The addition of acidic
oxides iould facilitate the decomposition of the original structure, for
example, by using silica:
(b) CaloF2(P04)6 + H20 + SiOC2 CaSi03 + 3 Ca3(PO4)2 + 2 HF
The actual decomposition of fluoroapatite with silica in the
presence of water vapor takes place above 1500K. The other oxides, the
addition of which may exert similar beneficial effect are: Fe203, A1203,
B203, TiO2, etc.
The possible side reaction in the presence of silica (h4)s
4 HF + Si02 ---- 2 H20 + SiF4
has a positive standard free energy change at higher temperatures.
A Fo(kcal.): 17.2 (298.160K.), 13.4 ( 500K.)
2.7 ( 1000oK. ), + 8.8 (1500oo.).
In the following examples not all the reaction products could be
predicted with certainty. They were assigned arbitrarily for the pur-
pose of providing an indication about the relative tendencies for vari-
ous possible chemical reactions.
According to the arbitrary scheme
Cal0F2(PO)6 + H2Y CaY + 3 Ca3(P04)2 + 2 HF
all the standard free energy values are positive foz the hydrogen-
containing gases, including HCl(c), lBr(d), H2S(e), H3N(f), etc. I T
two Cl, two Br, one S, two-thirds N.
The above scheme refers to the improbable condition that trical-
cium phosphate remain intact. If it is assumed that it is rather calcium
fluoride which remains among the products, the strongly acidic gases
should decompose fluoroapatite at low temperatures.
The reaction with hydrogen chloride, for example, according to the
(g) CaloF2(PO0)6 + 6 HC1 CaF2 + 3 CaC12 + 6 CaRIPO
has a negative standard free energy change at room temperature.
With hydrogen fluoride, the reaction should be far more energetic
at the same temperature.
(h) CalF2(PO)6 + 6 HF r. 4 CaF2 + 6 CaHPO4
An excess of gas would probably further the reaction with the
formation of monocalcium phosphate and more calcium halide. A large
excess of the gas, however, should not readily yield the following
STANDARD FREE ENERGY CHANGES OF REACTIONS OF
FLUOROAPATITE WITH A (ASTQTJS PHASE
A Fo in Kilocalories Per Gram-Formnla Weight
298.160K. 5000. 10000K. 15000K.
(i) Ca10 2(PO )6 + 36 HG1 -U 6 PO1.3 + 9 CaCl *
18 H20 + CaF2
(j) CalF2(P04)6 + 36 HP .. 6 POF3 + 10 CaF2 + 18 H20
Experimental Tests with Gases
A number of tests were carried out in order to find experimental
verification of the above-mentioned conclusions regarding the reactions
with hydrogen-containing gases.
Several samples of finely -round (300 mesh) synthetic fluoro-
apatite were subjected to the atmosphere of hydrogen and ammonia in the
temperature range of 500 to 10000C. for the duration of ten to twelve
hours in an electrically heated tube furnace. Similar experiments were
performed in the nitrogen atmosphere, which was saturated with water
vapor at room temperature, with and without the presence of silica dust
added to the sample.
Traces of fluorine-containing components could not be detected
in any case in the exit gases which were passed through an alkaline
The sensitivity of the test was about 8xlO'5g. fluorine, i.e.,
1 to 4000 of the total fluorine available in the sample.
Reduction of Fluoroapatite by a Gas Phase
The next group of reactions refers to the reduction of fluoro-
apatite in contact with an active gas phase. The numerical values of
the free energy changes are presented in Table 4.
Molecular hydrogen gives strongly positive standard free energy
changes in the given temperature range for the reaction
(k) CalOF2(PO4)6 + 15 H2 CaF2 + 9 CaO + 5/4 P4 + 15 H20
Molecular chlorine is also ineffective in the decomposition
of pure fluoroapatite if the reaction is written as follows:
(1) Cal0F2(P04)6 + 18 C12 .-9,6 POC13 + 9 CaC12 + 9 02 + CaF2
In contact with atomic chlorine (m), or hydrogen (n), the re-
duction yielding the same products as given above should occur readily.
An efficient low-temperature decomposition of fluoroapatite
with a mixture of chlorine and carbon monoxide will probably take place
according to the following schemes
(o) CaloF2(PO4)6 + 18 CO + 18 C12 -. POC13 + 9 CaC12 +
18 002 + CaF2
For the reaction of fluoroapatite with hydrogen-containing
gases, it is improbable that hydrogen fluoride will be formed as a
reaction product below 100000C.
The decomposition of fluoroapatite structure may take place with
strongly acidic gases at low temperatures, whereas fluorine remains in
the solid phase, most probably as calcium fluoride.
The reduction of fluoroapatite without additives is not easy
to effect under 10000C., and takes rather drastic conditions as the use
lit has been reported (11) that natural phosphatic nodules can
be defluorinated in the presence of elementary chlorine at about l000 to
6000C. The volatile products are silicon tetrafluoride and phosphoryl
chloride. In the presence of carbon and silica defluorination can be
carried out at 5OO0C.
of atomic hydrogen, or chlorine.
In special cases a mixture of reactive gases, e.g., chlorine
and carbon monoxide, might be efficient for a low temperature reduction.
RACTIOUDS 1 T'Mi SOLID STATE
Types of Solid State Reactions
Many types of solid state reactions of matter are mnown (10).
The simplest and most frequently occurring type is the additive
reaction A + B ---qtA, where A rand B may be elements or compounds. A
large group of reactions in this category are the combinations of an
acidic reactant with a reactant of basic nature to yield a salt. For
example, CaO + SiO2 --~ CaSiO3.
The second mode of reaction between two solids is by exchange
of constituents according to one of the following schemes:
(a) A + BC AC + B (b) AB + CD -~- AD + BC
(c) ABX + CB -- CBX + AB
In these reactions which involve diffusion through two product
layers, it is possible to have a considerable variety of phase arrange-
ments determined by the miscibilities of product and reactant phases,
and by the reaction mechanism.
Examples of reactions according to scheme (a) are found in the
exchange of the metal of a salt for a more electropositive one.
Reactions conforming to scheme (b) are exemplified by the
following: CaO + NiC2 CaC12 + NiO; CaO + PbC12 ->. CaCl + PbOj
or in combination with other reactions.
Reactions according to scheme (c) comprise a large group of
reactions in which an oxygen-containing acid group is exchanged between
one basic oxide and another: MeO + Me'XOn -- Met0 + MeXOn.
The reaction temperatures in the last group, as in the others,
show a correlation with the basic nature of the oxide involved; moreover,
in contrast to the other types described, these reactions have in common
the fact that the reaction temperatures for a given basic oxide are
approximately constant and independent of the nature of the other reac-
tant (carbonate, sulfate, phosphate, silicate).
The above principles were considered with a Tew to possible
application on fluoroapatite by way of inducing permanent changes in the
apatite lattice. In effect the application of these principles means
that calcium should combine with acidic components other than fluoride
and phosphate, for example, silicates and sulfates. In the presence of
additional cations, e.g., sodium and magnesium, mixed salts and new
phases could be formed.
Except for cases of miscibility of reactants and product phases,
equilibrium is usually not possible in reactions in the solid state, and
they proceed exothermally until at least one of the reactants is
completely consumed. In practice, of course, complete reaction is
difficult to achieve.
In a number of cases, the comparison of free energies of the
reactants and of the products may give an indication about the tendency
for a solid state reaction. Thermodynamic data were compiled for various
groups of compounds over a range of temperatures for the purpose of free
energy calculations. Awareness, however, was maintained of the limita-
tions of such an approach because of incomplete data, and for other
reasons. For example, even if the free energy relationships should be
favorable, the thermodynamic requirements are by no means a safe cri-
terion for determining which phases actually will be formed in a
reacting system. Since the particle motion in solids is strongly
resisted and requires large energies of activation, unstable phases
might form and coexist with other phases for practically unlimited
times if the temperature is sufficiently low. Even when the reaction
products are the thermodynamically required phases, they may appear in
unstable states characterized by an excess energy content. Nevertheless,
the intention was only to establish a broad scale for various groups of
compounds in their relative readiness to react with fluoroapatite, inso-
far as there wore proper data available. The results are summarized
in one of the following sections.
Apart from some preliminary tests, laboratory experimentation
was not carried out with regard to the solid state reactions for lack of
suitable analysis equipment and facilities for X-ray diffraction
Studies of the solid state reactions could profitably be com-
bined with the determinations of diffusion and migration of the ions in
the fluoroapatite lattice as measured by the electric conductivity of
Exchange Reactions between a Pair of Salts
A large number of solid state reactions are known in which an
exchange of cations (or anions) takes place between a pair of salts at
moderate temperatures with a measurable rate (10).
It is conceivable that the addition of some suitable salts to
the apatite system might induce changes in the initial structure of the
material whether by solid or liquid state reactions. It is possible that
new phases would result which show increased solubility or enhanced sus-
ceptibility to the gas phase reactions. Fron this point of view, various
possibilities were considered and many standard free energy changes were
compared. The problem is essentially to find a salt, the anion of which
would readily combine with calcium, and the cation of which would have a
strong affinity toward fluoride or phosphate.
Amonf the most suitable compounds for the above-mentioned
purposes are chlorides, sulfates, and sulfides; furthermore, silicates,
borates, etc. First the tendency for fluoride exchange was considered
in the simplified manner for calcium fluoride.
Chlorides.-The general reaction is:
CaF2 + XC12 -> CaC12 + X
Of the nonvolatile chlorides, only beryllium and magnesium chlo-
ride exhibited negative standard free energy changes for the given
JeCl2 --AF(kcal.): 6.7 (298.160K.), 16.6 (500K.),
IMgC12 -- AF(kcal.) -12.3 (298.160K.), -13.5 (500K.),
-16.7 (10000K.) -16.9 (15000K.).
Other cations considered were: Li, Na, K, Sr, Ba, Fe, Po, Cu,
Co, Ni, Sn, and Al.
Herein must be included the possibility that the two gaseous
chlorides, silicon tetrachloride and boron trichloride, might effect an
exchange of fluoride for chloride in the solid phase, at least according
to the standard free energy equations. The respective values are per
one mole of calcium fluoride:
SiCG --AF(kcal.): -10.1 (298.160K.), -16.0 (5000K.).
BC13 -aFo(kcal.): -15.1 (298.160K.).
Sulfates.-The exchange of fluoride for sulfate between a pair of
salts is generally unfavorable, except for magnesium sulfate. The
cations considered were: Na, K, Mg, Zn, Fb, Fe, Mn, Cu, and Al. The
free energy chance for the reaction
CaFG + XiSO4-1 CaS04 + XF2
is positive in the order of 8 to 15 kcal. (for magnesium sulfate, it is
negative by 2.2 kcal.) at room temperature per one mole of calcium
fluoride, and 0 to 9 kcal. at 1000K. The values decrease regularly
with increasing temperature. All, 'o and Cu offered the lowest positive
free energy changes.
Sulfides.-For most sulfides the thermodynamic data for higher
temperatures are deficient. At room temperature only Na, K, and Mg-
sulfides exhibit negative standard free energy changes in the order of
h to 6 kcal. for the reaction
CaF2 + XS -- XF2 + CaS
The cations considered were: Ma, K, Be, Mg, Ca, Ba, B, Al, t1, Fo, Co,
Ni, Cu, Zn, Sn, and Pb.
1An aqueous solution of aluminum sulfate undergoes double
decomposition with calcium fluoride (24).
Orthophosphates.-The meager data available on ort'.ohos siteste s
prohibited the extension of sirnilar comparicons to the exchange of phos-
phate ion for other anions.2
Conclusion.-The possibilities for solid state reactions remain
inconclusive until proper experimental testing, or until more thermo-
dynanic data will become available.
Under actual experimental conditions, the formation of mixed
salts, like CaC1F and NaCaPO4 and the existence of solid solutions
must be considered, which further complicates the thermodynanic evalua-
tion of possible solid state reactions.
2t has been proved experimentally (26) that Sr and Ba replace
Ca in phosphates (also in sulfates and silicates) in the solid state
reactions between 5000 and 600C.
REACTIONS OF FLUOROAPATITE IN MELTS
A number of materials could be used as additives for high tem-
perature reactions for the purpose of inducing changes in the fluoro-
apatite structure. The pure synthetic fluoroapatite has not yet been
used for such experiments, but the fusion of phosphate rock with addition
agents has been the object of many investigators.
Among the materials used there have been: furnace slag, alkali
silicates, alkaline earth silicates, sodium carbonate, silica, alumina,
magnesia, alkalies, magnesium silicates, magnesium sulfate, langbeinito,
polyhalite, etc., and the mixtures of these. The individual references
are listed in the following sources (7, 18b). The reaction temperatures
are usually above 10000C., and no defluorination takes place.
Various experiments were carried out in this laboratory with
synthetic fluoroapatite in salt melts.
Melts of Neutral Salts
The neutral salt melts can possibly induce partial replacements
in the fluoroapatite lattice.
For example: one gram of fluoroapatite with five grams of sodium
chloride kept in a platinum crucible for an hour at 90000C yielded eight
per cent of its total fluorine after leaching the product with water. A
comparable amount of chlorine was found in the water-insoluble residue.
Melts of Basic Salts
The basic salt melts decompose fluoroapatite at high temperatures.
Fluoroapatite-sodium carbonate mixtures, with and without the
presence of powdered silica, were heated to the maximum temperature of
925C. in the course of six hours. The fusion products were leached
with water, and the solutions were analyzed for phosphate. The extrac-
tions contained about one third to one fourth of the total phosphate in
the sample. The value for released phosphate in the melts with silica
was higher than in the melts without silica.
Fluoroapatite-sodium tetraborate mixtures at 10000C. effected
the decomposition of the starting material, and about ten per cent of
the available fluorine was volatilized (possibly as boron trifluoride).
Melts of Acid Salts
The acidic melts should offer the best promise for the decom-
position of fluoroapatite at medium temperatures.
First of all, metaphosphates may be considered because of their
relatively strong acidity and convenient melting temperatures. The
natural relationship of metaphosphates to the other phosphate systems
would allow easier interpretation of the results, insofar as such systems
have been observed in other studies. The high thermal stability and
ease of preparation of metaphosphates speak in favor of their use.
In consideration of the above-mentioned advantages the latter
part of this work will be devoted to the study of metaphosphate inter-
actions with fluoroapatite, and other fluorine-containing materials.
INTERACTION OF SODIUM METAPHOSPHATE -'ITH FLUORIDES
In the course of the studies about the reactions of fluorides in
melts, it ras observed that an interaction takes place between polyphos-
phate and fluoride ions with the formation of a phosphorus-fluorine
containing gas (or Gases), in addition to the evolution of hydrogen
The initial system observed was calcium fluoride in sodium meta-
phosphate melt. Soon, however, it became evident that other fluorides,
e.g., sodium and potassium behaved similarly.
A series of preliminary experiments was then carried out with
calcium fluoride and sodium fluoride mixed with sodium metaphosphate and
heated in dry nitrogen atmosphere for the purpose of determining the
major factors affecting the rate and extent of the formation of volatile
The reaction started considerably below the melting point of any
single component present, and was easily noticeable because of the
immediate formation of white dense fumes as soon as the gas from the
furnace came in contact with water or atmospheric moisture.
The survey of literature revealed that the above-mentioned
fluoride-metaphosphate systems have not been systematically studied,
except that the addition of fluorides to metaphosphates, usually in
stoichiometric amounts, is the standard procedure for the production of
the respective monofluorophoophates (29) which are known to decompose at
higher temperatures with the reduction in the fluorine content of the
salts (30). According to the information received from the Ozark-
Mahoning Company concerning the fluorophosphate salts (73), neither the
decomposition rates of these salts nor the reaction products have been
Tarbutton, et al. (65), let calcium fluoride react with phos-
phorus pentoxide at temperatures ranging from 5000C. to 100000., and
identified phosphoryl trifluoride (PO3) as the predominant volatile
product of the reaction, along with some small amounts of phosphorus
Gerard Montel (51) studied the same system, CaF2-P205, over a
wider range of compositions, and confirmed the results of the previous
These authors were primarily interested in the end products of
the reaction and did not carry out studies of melts related to the rate
of the formation of volatile fluorine compounds.
Objectives of Study of the Metaphosphate Melts
It was here attempted to determine systematically the major
factors which affect the rate and extent of the evolution of volatile
fluorine compounds from the metaphosphate melts containing fluorides.
The further intention was to determine whether the same relation-
ships are valid for the decomposition of fluoroapatite in contact with
molten metaphosphate, and for the subsequent volatilization of fluorine-
The major variables were composition and temperature. The other
factors considered were: duration of heating, presence of water vapor
in the furnace atmosphere, size of charge, surface of the melt, rate of
gas flow above the melt, pressure, typo and mode of preparation of the
used salts, particle size, and way of packing.
For the fluoroapatite tests the effect of added silica, sulfates,
and sodium tetraborate to the metaphosphate melts was observed.
Use was made of three furnaces and experimental arrangements for
obtaining the necessary data.
1. The vacuum furnace was constructed for the purpose of
finding out the following information:
a) whether it was possible to remove all of the bound
water from the materials below the reaction temperature;
b) the absolute pressure above the melt and the rate of
gas evolution from the nelt;
c) the identity of the gases involved.
2. The tube furnace was used to measure the rate of gas evolu-
tion at constant temperature with varying composition and in controlled
atmosphere. It was also used for heating the closed nickel tube which
was attached to the vacuum system.
3. The crucible furnace served for the visual observation of
the melts, and for the determination of the molting range of the mixtures.
Also larger amounts of salts could be used for varying the depth of the
The use of each method, and the actual experimental conditions
are to be described separately in the following sections.
Selected Metaphosphate Systems
From among the numerous possibilities the following systems were
chosen for a closer study. The reasons for the particular choices are
indicated after each system.
1. NaF-NaPO Convenient temperature range. The single
components have the following melting points:
NaPO3 6200C., NaF appr. 9900C, Both salts
are nonhygroscopic, provided that the meta-
phosphate is in the form of crystalline tri-
metaphosphate. Much is known about sodium
metaphosphate melts as such in respect to
the molecular weight distribution, density,
surface tension, and viscosity (69).
2. CaF2-NaPO3 -The presence of two types of cations.
NaF-Ca(PO3)2 Effect of a bivalent cation replacing the
monovalent cation. Relatively low reaction
temperature, about 5000 to 6000C.
3. Fluoroapatite-A special case of the previous system.
Na and The effect of crystal structure on the
Some additional systems were also tested, e.g., calcium fluoride-
calcium metaphosphate, sodium metaphosphate various metal fluorides.
In preparing the materials for reaction, the mixing had to be
unifrom and the particle size defined. The standard sieves were used,
250 and 300 mesh, from the Newark Wire Cloth Company, Newark, New Jersey.
Characteristics of the Evolved Gases
General Observations.-The gas evolved from the melt strongly
etched a silica tube and after passing through water, made it acidic.
The presence of fluoride (or fluorosilicate) ions as well as phosphate
ions could be proved in such a solution.
The solid adsorbents (soda lime, Ascarite, etc.) did not prove
to be reliable for the total capture of the gases at a moderate flow
rate, but a dilute alkaline solution cooled to ice temperature, was
satisfactory for the absorption of the gas mixture for analysis.
The ratio of phosphorus to fluorine was relatively small in the
analysis solution. Most of the fluorine was evidently evolved as
hydrogen fluoride. Since the atmosphere above the melt was dry nitrogen,
the hydrogen-containing component must have been provided by the ever-
present chemically bound water in the melt.
The phosphorus-fluorine ratio in the trap was directly related
to the amount of moisture present in the furnace. If the atmosphere
above the melt was saturated with water vapor, the phosphorus content was
very small, although sufficient for a qualitative proof, about 0.1 mg.
If the charge was heated gradually over a period of several
hours, the P to F ratio became higher but not more than about one to
seven, expressed in mole ratio. The gradual heating in the stream of
dry nitrogen evidently removed some of the surface moisture of the con-
stituent salts before the reaction started.
Separation of the Gaseous Phosphorus-Fluorine Containing
Component.-Attempts were made to find a suitable adsorbent which would
separate hydrogen fluoride from the rest of the gas mixture.
In one experiment, the gas from the furnace was passed through
a 3-cm. depth of phosphoric acid (85%, Reagent Grade) at the gas flow
rate of 80 ml. per minute. About seventy per cent of the total fluorine
had been absorbed but the P to F ratio still remained very low. Con-
centrated sulfuric acid used under the same conditions let more gas
through unabsorbed and the P to F ratio was approximately the same as
Reagent Grade sodium fluoride powder was moistened and baked in
an oven at 110C. for 24 hours. The dry aggregate was powdered to 250
mesh and filled into a narrow copper tube 100 cm. long. At a slow rate
of gas flow, 25 ml. per minute, all phosphorus and fluorine-containing
components of the gas mixture were adsorbed.
Next, the same narrow copper tube was filled with sodium fluoride
pellets and an increased gas flow rate was applied, 100 ml. per minute.
The source of the gas was 1.5 g. of calcium fluoride-sodium metaphosphate
mixture heated between 4000 and 7600C. for three hours.
The analysis of the trap gave 1.0 milli-equivalents of fluorine
and 0.33 milli-equivalents of phosphorus. A duplicate confirmed the
result of a phosphorus-fluorine ratio one to three.
This proves that sodium fluoride pellets adsorb hydrogen
fluoride preferentially and at a moderate gas flow rate, a part of the
phosphorus-fluorine compound passes through the adsorbent.
There are three known phosphorus-fluorine containing gases1:
POF3, PF 3, PF5; PO2F is only assumed to exist (62).
Since it is highly improbable that the found P to F ratio
results from a correct mixture of PF5 and PO2F, only the first two
gases nay be considered.
Phosphorus Trifluoride.-Phosphorus trifluoride (PF ) is
absorbed by aqueous bases producing fluorophosphate which reacts readily
with potassium permanganate or bromine. Repeated attempts with these
two reagents failed to yield evidence for a reduction-oxidation reaction,
and therefore the only possible conclusion is that the major component
of the phosphorus-fluorine containing gas is phosphoryl fluoride, even
though the possibility of tracks of other gases need not be denied.
Phosphoryl Trifluoride.-Phosphoryl trifluoride hydrolyzes (42)
by steps according to the following scheme:
HR20 20 H20
POF. -- P02F2 HPO3F -- H3P
In an alkaline medium the respective salts which are readily
soluble will be forced. Aqueous solutions of the alkali difluorophos-
phatoos react neutral but become acidic upon longer standing even at low
temperatures. Heating accelerates the hydrolysis. The first products
of deco.'iposition are monofluorophosphates and they may be obtained in
quantitative yields by boiling aqueous solutions of alkali difluorophoe-
phates with very dilute solutions of the corresponding hydroxides.
excluding other elements, except oxygen, from the composition
The alkali monofluorophosphate solutions are neutral toward
phenolphthalein but alkaline to methyl orange. Neutral or weakly
alkaline solutions are not hydrolyzod on boiling for one hour. Heating
in strongly alkaline solution results in rapid hydrolysis, and fast
decomposition also occurs within a few minutes while heating in a
strongly acidic solution.
As an example of phosphoryl trifluoride neutralization in dilute
sodium hydroxide solution the following equation is presented:
POF3 + 4 NaOH -- Na2PO3F + 2 NaF + 2 H20
One molecule of phosphoryl trifluoride reacts with four equi-
valents of alkali. Phcnolphthalein is used as an indicator. In acidic
solution the hydrolysis would go one step further.
Difluorophosphoric Acid.-Luring the experiments with the tube
furnace at lower temperatures, it Tas noticed that small droplets of a
colorless liquid had condensed on the valls of the polyethylene tubing
near the end of the nickel tube. These droplets were rinsed with small
volumes of cool nitronI solution and thus removed from the tube. After
standing for a few hours at ice-temperature, a slight colored precipi-
tate was formed. The precipitates of several runs were combined, dried,
weighed, and analyzed for fluorine. The weight ratio of fluorine to the
whole sample (14.8 ng.) was about 1 to 10. The theoretical ratio for
the compound nitron-difluorophosphoric acid (one molecule) is 10 to 109.
It may be assumed that the condensed material consisted primarily
of difluorophosphoric acid2 which is the first decomposition product of
phosphoryl trifluoride with water,
INitron is a shorter name for diphenyl-endo-anilo-hydro-triazole,
which is an analytical reagent.
2The boiling point of difluorophosphoric acid (5) is appr. 1100G.
Construction of the Vacuum Furnaces
The Furnace for Crucibles.-A vacuum furnace was constructed
which could accommodate a platinum crucible. Since the design was
unique for a furnace of this size, the details of construction will be
Nichrone V resistance wire (B&S 24) was wound around a specially
made nonporous porcelain core in a helix one-eighth inch in diameter and
nine inches long (when closed). The total resistance was 20 ohms and
the current was, for example, four amperes at 7000C. The dimensions of
the core were: the diameter inside one inch, outside one and three-
eighths inches, height two and three-quarters inches. The core was
surrounded by two concentric radiation shields made of nickel sheet,
one-one hundred twentieth inch thick. The diameter of the inner shield
was two and one-quarter inches, of the outer shield two and eleven-
sixteenths inches, the height three and three-quarters inches. The two
shields were separated by porcelain spheres through which short pieces of
nichrome wire held them together. There were also horizontal shields,
two above and two underneath the core. The core and the shields were
suspended by nichromo wire from three steel support rods eight inches
high and one-eighth inch in diameter. Two of these wires served also for
the conduction of current to the furnace. The support rods were screwed
into the base which was a brass plate five-sixteenths inch thick. The
connections through the base for the heating current as well as for the
thermocouple wires, were made with the help of Stupakoff Kovar-Glass
Terminals, Type FC, soldered through the plate.
The whole furnace was enclosed by a glass dome three and seven-
eighths inches in diameter and ten inches high which rested in a groove
in the brass plate. The dimensions of the grove were outside diameter
four and one-eighth inches, one-quarter inch wide, one-eighth inch deep.
The vacuum seal between the plate and the glass was achieved by using
the vacuum wax Apiezon W, which softens at about 800C.
The platinum crucible rested in the middle of the furnace on a
hollow porcelain rod through which the thermocouple junction was led up
to the crucible without touching it. Through the center of the plate
went a copper tube, three-eighths inch in diameter and fifty inches long
which connected the furnace with the rest of the vacuum system. The
furnace itself rested on a thermally insulated tripod which was firmly
attached to the desk.
The furnace rendered good service at low pressures and reached
high temperatures readily. The bottom plate did not become warmer than
5oC., even when the furnace was at 8000C. for several hours. The tem-
perature control was affected noticeably by the pressure changes, and
was not better than SoC. Baking of the furnace over a longer period of
time at moderate temperatures did not remove the adsorbed gases completely
and with an increase in temperature, slight degassing continued. Intro-
duction of hydrogen into the system before opening it to the atmosphere
improved subsequent evacuations.
According to the available data regarding phosphoryl trifluoride
(65), it was not supposed to react with the furnace materials under the
given conditions. Some etching of the porcelain core, however, became
noticeable after the experiments, and silicon tetrafluoride was found to
be present in the liquid air trap. Since it was possible that the de-
corposition of the phosphorus-fluorine containing gas took place because
of high temperature, and the presence of siliceous materials, another
type of furnace was made use of for a number of experiments.
The Nickel-Tube lurnace.-An all-nickel tube (International Nickel
Co., Mark:53965) was placed into a Hoskins electric tube furnace (Type
FH 303 A, 15 V, 37 A) and connected to the copper tube leading to the
vacuum system with Apiezcn W wax. From the closed end of the nickel
tube, a smaller tube of the same material, silver soldered to the larger
one, led up to the middle of the furnace and housed the thermocouple.
The dimensions of the tutes were as follows: larger tube had the length
thirty-three and one-half inches, and the inside diameter fifteen-
sixteenths inch; tho smaller tube had the length nine and one-half inches,
and the inside diameter three-quarters inch.
A Sorgel electric. transformer (Type IFS, sec. volts 17; 32.3 A)
belonged to the tube furr.ace as an accessory.
The thermocouples were chromel-alumel junctions standardized
against the melting points of pure inorganic salts, and compared to the
other couples. The potertiometer used with the thermocouples was from
Wheelco Instruments Co., Model 310.
The now apparatuE had a smaller volume, 365 cc., as compared to
1600 cc. of the former. It also offered better dogassinp characteris-
tics and closer temporatire control. The disadvantage was that the
size of the charge was limited by the capacity of the platinum boat.
The platinum boat was placed into a larger zirconia boat so as to avoid
the contact of platinum iith the nickel tube.
This furnace operated in a satisfactory manner and an example
of the measurements is presented in the next section.
Method of Measurement with the Vacuum Furnaces
When the furnace was below 300C., a vacuum of the order of
103 anm. was reached in a couple of hours. At higher temperatures it
took considerably longer, about a day.
The vacuum system on the rack consisted of a McLeod gauge, a
mercury manometer, three traps, and provisions for removing material
from the system. The one-stage mercury diffusion pump was backed by a
Welch Duo-Peal vacuum pump. As a safety precaution, the water flowing
through the cooling jacket of the diffusion pump operated a relay
regulating the heating current of the diffusion pump, thus securing safe
over-night operation of the pump without supervision.
The rate of the gas evolution was determined in the following
manner:- 'Thile the system was open to the pumps, the pressure was
measured with the McLeod gauge. Then the pumps were cut off by closing
a stopcock and letting the system stand for five minutes (or some other
known period of time), whereupon the pressure was again measured. The
liquid air trap was opened simultaneously with the closing of the furnace.
After 5 to 15 minutes, the pressure was determined again. The difference
between the two last measurements made it possible to calculate the
amount of condensible gases evolved from the melt in the given time
period at a known temperature. After such a series of measurements, the
system, including the furnace, was pumped empty through the other liquid
The volumes were: the McLeod gauge 282 cc., the rest of the
glass system minus the traps 312 cc., a single trap 18 cc., and the
furnace 365 cc.
Experimental Results of the Vacuum Measurements
The results showed that water was held very tenaciously by the
melts, and that even prolonged heating (12 hours and longer) under
reduced pressure (10-3 mm.) in the temperature range from 3000C. to
5000C., did not remove the hydrogen-containing constituent completely.
Then the temperature was raised above 500., the reaction
gradually started and gases condensible in the liquid air trap were
present in the vacuum system.
The rate of gas evolution did not present any sharp maxima in
its temperature dependence. It usually showed somewhat higher values
immediately after the rise in temperature but then levelled off to a
rate which decreased slowly at a given temperature.
For example, a charge of 5.0 grams of the standard sodium
fluoride-sodium metaphosphate mixture with the platinum boat in the
nickel tube was preheated under vacuum for twelve hours. V;ith a gradual
rise in temperature above 500C., the following rates were observed:
RATE OF GAS EVOLUTION FROM .ETAPH.JGPHATE ITLLTS
IN THE VACUUM FURNACE
Temperature Time Period Above Rate Expressed in Remarks
OC. 5000C. in Minutes Moles of Gas per
i5 Minute x105
510 60 1.7 Steady Temp.
530 90 3.7 Steady Temp.
S40 120 3.8 Steady Temp.
580 150 3.9
594 160 5.6 Increased Temp.
628 270 5.0 Steady Temp.
630 300 5.0 Steady Temp.
713 380 11.2 Increased Temp.
716 480 3.9 Steady Temp.
790 560 5.6 Increased Temp.
800 630 2.8 Steady Temp.
These rates refer to the mixture of gases the pressure of which
was measured in the system. The analysis showed that silicon tetra-
fluoride was the major component in the gas phase, with about ten per
cent of a phosphorus-containing component.
Traces of hydrogen-containing compound (probably in the form of
water) still present in the melt evidently interfere with the assumed
formation of phosphoryl trifluoride, and produce hydrogen fluoride
instead which in turn attacks the siliceous parts of the apparatus with
the formation of silicon totrafluoride and water. This water may attach
itself to phosphoryl trifluoride, causing the decomposition of the
latter to difluorophosphoric acid and hydrogen fluoride.
Thus, even the smallest traces of water vapor may decompose most
of the phosphoryl trifluoride, the total quantity of which is rather
small to begin with in th3 order of about 7x10-3 moles of gas for the
given charge of five gram3.
In the above example, the trap contained 1.18x10"3 moles of gas.
The analytical determinatLon gave the following values: fluorine 4.6x10-3
equivalents, phosphorus 0.13x103 equivalents, and silica.
No etching was visible in the other parts of the glass apparatus
except in or near the liquid air trap where most of the decomposition of
the gaseous components seemed to have taken place.
The other experiments in the vacuum furnaces gave similar
Experimental Procedure at Atmospheric Pressure
Furnace.-The same tube furnace as described previously served for
the series of experiments presented in the following text. First a
Sillimanite tube, Coors Porcelain, was used, but since there was some
contamination of the outcoming gas with silicon tetrafluoride, it was
replaced with a nickel tube. The dimensions were length twenty-five
and one-half inches, inside diameter fifteen-sixteenths inch; to this
another smaller nickel tube was attached: length fourteen inches, inside
diameter one-half inch. A copper nipple was silver-soldered to the free
end of this tube. It was provided with threads for the adjustment of
polyethylene tubing, twenty inches long, which led to the bottom of a
one-liter polyethylene flask. Another tube through the cap led to the
second trap. The other end of the larger nickel tube was closed with a
hard rubber stopper through which passed a short piece of nickel tubing
of a small diameter, On the furnace side of the stopper, asbestos tape
was wrapped around the incoming end of the small nickel tube in order to
protect the rubber from the heat radiation.
The temperature was measured with a protected chromel-alumel
thermocouple in a porcelain sheath. After a steady temperature state
was reached, the thermocouple was withdrawn for the duration of the
The furnace atmosphere was formed by Linde's dry nitrogen, which
was passed through a four foot column of silica gal before entering the
Treatment of the Gaseous Reaction Products.-The outcoaing reaction
products diluted with nitrogen entered the polyethylene bottle which was
immersed in ice water, and contained 50 to 75 ml. 0.1 normal sodium hy-
droxide solution. The absorption of the acidic components of the gas
took place very readily, and only small amounts of it reached the next
flask. The collection of the gas continued at least half an hour beyond
the time of the removal of the boat from the furnace, since it was
observed that some of the acidic components of the reaction products
condensed or were adsorbed on the walls near the cooler outgoing end of
the nickel tube. Heating this end with gas flame removed practically all
the adsorbed materials, and an additional 5 to 15 per cent of the total
was usually recovered.
The sodium hydroxide solution in the traps was back-titrated
with a dilute nitric acid solution, and the number of equivalents neu-
tralized by the furnace gas was thus determined. All the solutions
with condensed products were collected and analyzed for fluorine and
phosphorus. Frequent blank runs were carried out between the experiments
for checking purposes.
Standard Mixtures and Conditions
On the basis of the preliminary tests, a set of standard con-
ditions ias chosen and defined with which the effect of varied experi-
mental factors could be compared.
Standard Conditions.-The standard conditions were as follows:
A charge of 1.50 g. in a platinum boat was held in the hot zone of the
furnace, at 700oC. (- 5$C.), for one hour while a stream of dry nitrogen
swept the furnace at a rate of 110 to 130 ml. per minute. The platinum
boat (length 7.5 cm., depth 1.0 cm.) was lying in a zirconia boat
surrounded by a cylindrical sheath of thin nickel sheet, to which was
attached a piece of nichromo wire for the centering and fast removal of
the charge. The surface of the molten charge was about 6 sq. cm., and
the depth about 3 to 5 mm.
Standard Compositions.-There were two standard compositions one
with sodium fluoride and the other with calcium fluoride. In terms of
equivalents, the compositions were identical in respect to the fluoride
and the metaphosphate content, viz., 18.9 mole per cent of fluoride and
31.1 mole per cent of notaphosphate, counting the anions and cations
separately in forrdna one hundred per cent. The original mixture, which
was also used for the vacuum furnace experiments, had resulted from
mixing one grati of sodium fluoride with four grams of sodium metaphos-
The natorials were finely ground and intimately mixed. W
mixture was transferred into the platinum boat, the powder was not
pressed in too tightly, only the top was smoothed with a spatula. The
charge was dried for about thirty minutes with the boat in an oven at
1500C,, and was then directly placed into the hot ,one of the furnace.
Samples of Sodiun Metaphosphate.-Three differently prepared
samples of the same salt, Mallinckrodt's monosodium orthophosphate mono-
hydrate, Analytical Reagent, were used for comparison. According to the
mode of dehydration, the three preparations were designated as (a), (b),
and (c) samples.
NaPO3 (a) Heated at 600-620oC. for two hours; stored in
a screw-capped bottle for four months.
Heated gradually up to 8000C., let stand at this
temperature for twelve hours, cooled slowly to
about 600OC., held there for two hours. The
salt was removed from the furnace at 4000G. and
Heated slowly in the course of three hours to
500C. The product was used immediately.
Experimental Results of the Measurements at 7000C.
rwith the Standard Co-:rposit.ions
The purpose of the first series of experiments was to follow the
ef ects of various factors, other than temperature and composition, on
the reduction in the fluorine content of the samples, and to establish
the limits of r eroducibility of the data. Table 6 presents the results,
expressed in terrn of equivalents neutralized in the traps.
There is a marked difference in the values of sodium and calcium
fluoride compositions, -hich runs consistently through all tests, the
mixture with calcium fluoride always offering a higher yield. Since this
difference is too large to be ascribed to the influence of varying
moisture content, or the presence of impurities in the respective salts,
it must result from a stronger effect of the calcium ion to further the
The mode of sodium metaphosphate preparation may also change the
yield considerably. However, it is not so much the length of heating of
the compound as the aging of the material which seers to play a role.
Much, undoubtedly, depends upon the way of cooling of the product, since
this determines the crystallinity and the hygroscopic properties of the
resultant sodium metaphosphate.
The presence of water vapor in the atmosphere exerted benefidal
influence upon the reduction in the fluorine-content of the samples.
Doubling of the reaction time did not improve the yield significantly;
and the rate of gas flow seemed to be immaterial in the rivcn range.
The reproducibility of the single runs ranged from two to four
per cent deviation from the moan. Usually the accuracy was somewhat
better for the mixtures with the calcium component than with sodium.
All sodium netaphosphate relts showed spr cding of the melt on the walls
of the boat, IUtt ith sodiu~ fluoride-sodiui mi '-hspht t.o mixtures the
te ndncy was particularly pronounced. Several erratic values, usually
coinciding -A th spilli, j or strong spi ~ i,. of the 'alt,had to be
The yield was not proportional to the size of the charge and
generally suffered a relative decrease when the charge increased.
Fluorine analysis in the trap solution cave usually a quite
close value to the nuliber of equivalents neutralized by nitric acid,
although at tines it was slightly lower. The tro values could be con-
sidered practically equal after the correction for a blank was made.
The presence of the phosphorus-con' ai:in i component vwa evident
in all runs, although the phosphorus-content in the trap was rather low,
seldom exceeding 0.2 milli-equivalcnts, and sometimes only sufficient
for a qualitative proof.
Variation in Composition at 7000C.
The next series of experiments dealt with the variation of com-
position of the melts. The results rere expressed in percentages of
fluorine expelled from these melts. The total fluorine content of the
sample counted as one hundred per cent.
The previously used salts, sodium fluoride and calcium fluoride,
were nixed in varying proportions. In two cases calcium metaphosphate
replaced sodium metaphosphate. For comparison, use was made of a
commercial disodium monofluorophosphate salt.
EFFECT OF YXPrTfIlEUIAL CONDITIONS AT 7000C. ON THE EVOLUTION OF
FLUORINE-CONTAIfING GASIS FROM METAPHCSPHATE MELTS
Composition of Milli-equivalents Special
the Charge Titrated in Traps Conditions
5.63 1 .14
3.90 1 .15
Presence of water vapor
Double Gas Velocity
Composition of Milli-equivalents Special
the Charge Titrated in Traps Conditions
The sources for the used chemicals were:
NaF Herck, Reagent Grade, dried at 1100C.
CaF2 "Baker Analyzed", rinsed with water, ignited at
8000C. for an hour.
NaPO3 Preparation (a), previously described.
Ca(PO3)2 Prepared from monocalciun phosphate monohydrate,
Fisher Certified Reoaent. Described previously.
Na2PO3F Ozark & Mahoning Company, dried at 1100C.
The tests were carried out under standard conditions as defined
in the preceding section. At least duplicate samples wore tested of each
composition and Table 8 presents the mean values of the reduction in the
fluorine-content of the samples.
The results revealed that an increasing P03 to F mole ratio
furthered strongly the fluorine volatilization. The absolute quantity
of the removed fluorine was smaller for the composition of the mole ratio
1 to 1, than either for a moderate excess of sodium fluoride or of
sodium metaphosphate. The reason must be related to the salt formation
at this mole ratio. The tendency for any reaction of the fluoride ions
is lowered in such a case, since fluoride forms a part of the salt
The ready-made compound, obtained from Ozark & Mahoning Co.,
gave a yield of volatile fluorine-containing compounds comparable to that
of a salt mixture of equal composition. A somewhat higher value of the
yield from the former may have been due to a different moisture content,
The difference in the behaviour of molts containing calcium and
sodium, as observed before, persisted throughout the varying compositions.
COMPOS ITIOI OF THE EXMPE ITAL :L TTAPHOSPHATE-FLUORIDE MIXTURES
Per Cent in Elements Milli-equivalents
Equivalents and PO3 Per Cent in "ei-Lht of Fluorine in a
of the Salt Component Equivalents Per Cent 1.5 g. sap1le
Per Cent in Elements Milli-equivalents
Equivalents and PO3 Per Cent in Weight of Fluorine in a
of the Salt Component Equivalents Per Cent 1.5 g. sample
EVOLUTION OF FLUORINE-CONTAINING OASES FROM METAPHC6PHATE MELTS
AT VARIOUS COMPOSITION' I
Composition Milli-equivalents Per Cent by Yeight
Per Cent in Milli-equivalents of Fluorine Present of Fluorine Removed
Equivalents Titrated 0.15 in a 1.5 g. Sample 1 3 (average)
The stoichiometrically analogous mixtures made up from the salt
pairs CaF2-NaPO3 and Ca(P03)2-NaF gave rather close values in respect to
the reduction in the fluorine-content of the samples. This is an
indication that the effective composition of the melt is significant
rather than the combination of initially added salts.
Even the calcium metaphosphate-calcium fluoride system lost part
of its fluorine-content at this temperature, in spite of the high melting
points of the individual components. In this case no melting had taken
place; the powder was slightly sintered, however.
The increase in temperature generally enhanced the reaction, and
the fluorine-content of the samples was successively decreased under the
conditions of the standard procedure.
Table 9 shows the respective mean values for the experiments
with the standard mixtures of calcium and sodium fluoride with sodium
metaphosphate. The melts with calcium fluoride exhibit higher yields in
the fluorine-containing gases.
Reactions Involving Other Cations
Since the interaction between fluorides and metaphosphates was
assumed to be a general reaction, it was considered interesting to let
sodium metaphosphate react with fluorides other than sodium and calcium
Some of the available materials wore tried and samples were
prepared which corresponded to the composition of the standard samples in
respect to fluorine equivalents. The experimental conditions remained
EVOLUTION OF FLUORINE-CONTAINING GASES FR~i I4ETAPHOSPHATE
MELTS AT VARIOUS TEIPERATUPJS
Percentage by eight of Fluorine Removed
600C. 700C. 80000.
NaF NaPO .
a a62.8 5.0 78.8-3.0 85.5+3.0
2 3 72.514.0 92.212.0 97.012.0
The sources for the other fluorides were
KF General Chemical Division, anhydrous.
BaF2 "Baker's Analyzed", anhydrous.
AlF3 A.D. Mackay, Inc.
ZrFj A.D. Mackay, Inc.
The accompanying chemical analysis of the fluorides gave the
stoichiometric fluoride values for anhydrous potassium fluoride and
barium fluoride. Aluminum fluoride and zirconium fluoride contained an
undetermined amount of moisture and less fluoride than expected from the
formula. Therefore, the titration data for the last two salts present
actually the minimum calculated percentage for fluoride removed from the
test samples. Potassium fluoride is a very hygroscopic salt, therefore,
the weighing and mixing of this salt was performed in a dry box and
closed containers. However, while transferring the mixed charge into the
furnace, some atmospheric moisture may have been adsorbed by the salt.
EVOLUTION OF FLUORINE-CONTAINING GASTS FROM METAPH06PHATE MELTS
IN THE PRESENCE OF OTiHE CATIONS
Fluorine Present Percentage by Weight
Milli-equivalents in the Sample of Fluorine Removed
Fluoride Titrated (duplicates) (calc.) mg. :2.5
KF 5.94 5.62 135 93.0
BaF2 4.19 4.36 105 77.2
AlF3 3.82 3.78 126 53.5
ZrF4 l.30 4.18 118 64.0
CaF2 5$.2 (mean) 138 92.2
NaF 3.90 (mean) 136 78.8
The extent of the reduction in the fluorine-content of the
mentioned fluorides is comparable to that of sodium and calcium fluoride.
The results are tabulated in the above table.
INTIPLACTION OF ::;.AP?-6'SP -TE -ITH FLUCROAPATITE
As the preliminary tests had indicated, a destructive decomposi-
tion of fluoroapatite takes place when the material is in contact with
sodium metaphosphato above 55000.
An independent series of measurements was performed in order to
establish the extent of fluorine removal from fluoroapatite under
standard conditions. Subsequently, the variations in temperature and
composition, the presence of moist atmosphere, and supplementary materials
were imposed upon the system.
The standard composition in this case consisted of fifty per cent
(by weight) of sodium metaphosphate (preparation c) with finely ground
(300 mesh) fluoroapatite (preparation b).
All samples were tested at least twice, and the tabulated percen-
tages present the respective mean values. Fluorine analysis was carried
out in each case in the trap solution, as well as in the residue of the
sample. No spreading of the melt occurred with these compositions, and
the reaction product could easily be removed from the boat,
The results showed that a considerable proportion of fluorine
was removed from the 1.5 g. charge within one hour at 70000. Two other
ratios of fluorospatite to metaphosphate (see Table 11) were tested, but
COaIPOSITIONS OF METAPHOSPHATE MELTS
Percentage by Weight
Composition of Fluorine Removed
50% fluoroap. 50% NaPO3 . . . .. . 56.9+-2.0
251 fluoroap. 75% NaP03 . . . .. 43.4- 14.0
75, fluoroap. 25% NaPO3 . . . . . 33.4 14.0
25% fluoroap. 25% Ca3(PO4)2 -50% NaP03 . 53.51 1.o
25% fluoroap. 25% hydroxyap.-50% NaPO3 . . 51.22 4.O
in either case the yields were lower. The probable explanation is that
with the smaller amount of sodium metaphosphate, the acidity of the melt
is lower, and with the larger proportion of metaphosphate, the viscosity
of the melt may exert a retarding influence on the rate of fluorine
removal. The latter assumption was supported by the experiment in which
tricalcium phosphate replaced 25 per cent of sodium metaphosphato and
effected an improvement in the yield of fluorine-containing gases rather
than a decrease.
The use of hydroxyapatite in place of tricalcium phosphate did
not change the result. The amount of hydrogen which hydroxyapatite may
contribute is small (about 0.3 mg.), compared with the rest of bound
water. Hydroxyapatite was prepared according to the method of Rathje
A mixture of calcium fluoride and tricalcium phosphate was made
up with sodium metaphosphate, which was analogous to the standard
fluoroapatite-sodium metaphosphate in respect to its stoichiometric
composition. The extent of defluorination in the first sample was
larger than for the fluoroapatite charge; the difference, however, was
The relationships between the conditional factors and the extent
of defluorination in fluoroapatite melts were expected to be somewhat
different from the results of the fluoride group, primarily because of
the relatively low fluorine content (about 1.8 per cent by weight) of the
fluoroapatite charges. Besides, solid phases were always present in the
fluoroapatite mixtures, whereas previously a majority of the fluoride
melts had formed a homogeneous molten phase.
The most interesting result appeared in the temperature variation
of the experiments, where a maximum reduction in the fluorine-content of
the samples was displayed at about 6IOC. The reaction became noticeable
at about 550C., and proceeded best just above the normal melting point
of sodium metaphosphate (approximately 6200C.). With rising temperature,
the extent of defluorination of the melt decreased until 800C. was
reached. From there on the decrease in the fluorine-content of the
samples gained again gradually with increasing temperature. See Table 12.
Evidently there are two different processes involved regarding
the reduction in the fluorine-content of the melts. The first one is the
decomposition of fluoroapatite, and the gradual removal of the fluorine-
containing volatile products, e.g., phosphoryl trifluoride, from the
EVOLUTION OF FLUORINE-CONTAINING GAS0S FROM
Percentage by 'eight of Fluorine Removed
C. ...... ..--------------
5 Water Vapor Silica CaF2 and
Standard Present Present Ca3(P0)2
Conditions 3.0 5.0 o2.0
575 47.7 1 5.0
600 65.2 1 h.o
640 79.6 1 2.0 80.7 72.9 82.0
700 56.9 2.0 65.0 75.1 69.0
800 50.4 1 2.0 57.2 61.9 65.6
900 59.2 1t .o
1000 63.01 5.0
melt. This process is presumably favored by an increase in temperature.
The other reaction may be the interaction of hydrogen-containing com-
ponents (e.g. water) in the melt or at the surface of the melt with
fluorine-containing components, facilitating the formation and expulsion
of hydrogen fluoride. This effect rmst be proportional to the concen-
tration of hydrogen-containing components.
Upon fast insertion of the material into the hot zone of the furnace
at higher temperatures, part of the surface-adsorbed moisture may bo
vaporized and removed before it can participate in the reaction in the
molten phase. Just above the melting point of sodium metaphosphate a
relatively larger amount of water may be trapped by the incipient melt.
This assumption is supported by the observation that the presence
of water vapor in the atmosphere above the melt favors fluorine evolution
as hydrogen fluoride. The effect of moisture probably consists in the
acceleration of the process. For example, a charge treated for two hours
in a dry atmosphere gave about the same yield of fluorine-containing
gases as the charge which was held for one hour in a wet atmosphere.
The Effect of Supplementary Materials
Small supplements, 0.1 e. of silicic acid (containing 20 per cent
water) added to 1.5 standard fluoroapatite charge, induced an improve-
ment in the expulsion of fluorine-containing Cases above 70000. Below
that temperature such an effect was not evident. Some of the silica was
found in the trap after the reaction, consequently it had participated
in the reaction and had been volatilized as silicon ttrafluoride.
The other added salts (0.2 g. to each standard sample), sodium
sulfate and sodium tetraborate, did not further the defluorination of
the melt but rather suppressed it somewhat. Calcium sulfate o'ave a better
yicld, possibly because of the presence of some residual water of consti-
tution. The respective mean values uere: sodium sulfate 53 per cent,
sodium tetraborate 53 per cent, and calcium sulfate 72 per cent of the
total fluorine content of the sample removed at 7000C.
The sources for chemicals were:
Na2SO4 Mallinckrodt's Analytical rea;-rnt, anhydrous.
CaSOh.2 H20 Coleman & Bell Co., C.P., dried slowly and
ignited at 5000C.
Borax, Baker's Analyzed, C.P., dried first at
150C. and then at 300C.
MTallinckrodt's Analytical Reagent.
EXPERIENTS IN TIM C RUPCIBZ. FURNACE
Solubility of Salts in Molten Sodium Metaphosphate
The furnace in operation was an electric multiple unit crucible
furnace from Hevy Duty Electric Co. (Type 82, 115V, 725'). Platinum
crucibles were used in all experiments.
First the solubility of various salts in sodium metaphosphate
melt was tested.
Small amounts of calcium fluoride disappeared in the melt quite
rapidly at the beginning; later on, the process became slower, and after
seven hours at 7000-8500C., about 2.5 g. of the salt had dissolved in
7.5 g. of sodium metaphosphate.
Tricalcium phosphate displayed a similar behavior in the same
temperature range. In the course of six hours about twenty per cent by
weight had dissolved in the melt.
Sodium fluoride was very readily soluble so long as there was not
any excess of sodium fluoride. At the mole ratio 1 to 1, the molt grad-
ually disappeared and crystalline salts were formed.
Fluoroapatito was very slow in becoming a homogeneous part of
the melt. Possibly other solid phases of phosphates replaced the
original structure since the decomposition of the material took un-
For example, in a mixture of 1.23 g. of fluoroapatite and 6.13 g.
sodium metaphosphate held for two hours at 8500C., the molten part was
analyzed separately from the crystalline phase. The glassy part of the
melt contained more than half of the total fluorine and about thirteen
per cent of it had been volatilized. In this experiment, fluoroapatite
powder was laid at the bottom of the crucible, and the decomposition
products had to diffuse through a layer of 12 nn. of viscous melt before
reaching the surface.
In another experiment in which the layer was even deeper, less
than ten per cent of the total fluorine had been volatilized in the
same period of time.
If a similar melt was held overnight for sixteen hours at about
8000C., half of the total fluorine still remained in the melt.
Analogous experiments with small amounts of sodium or calcium
fluoride gave better yields of volatilized fluorine, the percentage of
which was still considerably lower than in the case of tube furnace
Weight Loss of Standard Mixtures
and the Component Salts
The rate of weight loss for various samples was determined in
another series of experiments.
The standard compositions of sodium and calcium fluoride with
sodium metaphosphate at 7000C. gave closely similar results. The total
weight loss for a 1.5 g. charge was about 120 mg.
The highest weight loss, about one fourth of the total, occurred
during the first quarter of an hour. In the last quarter, the rate was
down to about four per cent of the total. During the first hour only
half of the total weight loss had taken place. From then on there
continued a rather steady rate of weight loss for further twelve hours,
Additional tests confirmed the observation that after the first
hour, the weight loss for a given sample had reached a slow steady state
for a number of hours.
The weight loss of the standard charges upon drying in the oven
at 1500C. for two hours was small, 1.1 to 1.4 mg. per 1.5 g. charge.
The ignited calcium fluoride, which was used in making up the
mixtures, lost 0.45 per cent of its weight in 20 hours at 6000-7000C.
The respective loss for sodium fluoride was 0.39 per cent in 15 hours at
A charge of 1.2 .g of sodium metaphosphate held in the tube
furnace at 750C0. for two hours lost 0.1 nig. of its weight.
Some of the crystalline phases which separated during the course
of reaction, or upon cooling of the above-mentioned melts, were subjected
to study under a petrographic microscope.
An excess of metaphosphate forms a glass when chilled rapidly,
and on occasions the crystalline parts of a melt could be observed
embedded in the optically isotropic glass, thus obviating the need to
separate the crystals from the rest of the melt.
A variety of phases were observed, the type of which depended
primarily on the composition of the melt and the rate of cooling. For
example, in the system calcium fluoride-sodium metaphosphate the presence
of calcium pyrophosphate was positively proved,
SUMMARY OF TiH EXPERIMENTAL RESULTS WITH IIfTAPHCSPHATE MELTS
In all cases when sodium metaphosphate comes into contact with
fluorides or fluoroapatite, above the temperature of about 550C.,
reactions take place which liberate a considerable part of fluorine in
the form of volatile compounds. Most of this fluorine is in the form
of hydrogen fluoride. The hydrogen component is provided by the melt
which always maintains small amounts of water, even though the consti-
tuent salts wore dried and heat-treated before mixing. Evidence also
points to the presence of phosphoryl trifluoride in the exit gas.
The basic reactions for fluoride are assumed to be:
2 P03- + 2 F- + H20 2 HF + P22 HF
5 P03 + 3 F" POF3 + 2 P2-
The same reactions for fluoroapatite give the following equations:
8 PO3" + (PO-3)6(F")2 + H20 --> 7 P20"4 + 2 HF
28 PO3- + (Po-3)18(F-)6 -> 2 POF3 + 22 P204
In the case of the standard fluoroapatite-metaphosphate mixture
(50 to 50%) at least five per cent of the material must have remained in
the form of orthophosphate.
The Major Factors and Their Influence Upon the Reactions
The major factors which influence the extent of the reduction in
the fluorine content are composition and temperature. The rate of
defluorination of the melt is markedly affected by the depth of the layer
through which the decomposition products have to diffuse and by the pre-
sence of hydrogen-containing compounds as constituents of the melt.
For a given temperature the percentage of removed fluorine in-
creases as the acidity of the melt increases. Herein the relative
acidity of a metaphosphate melt (in respect to fluoride ion) is defined
by the concentration of P-O-P bonds as they occur in polyphosphates
(straight chains), and polymetaphosphates (cyclic structures). There is
a rather abrupt increase in the percentage of removed fluorine as soon as
the mole ratio of metaphosphate to fluoride exceeds one to one.
The rate of formation of fluorine-containing gases is highest
immediately after a mixture has reached the reaction temperature. After
the first hour the rate has slowed considerably, and under favorable
conditions more than ninety per cent of the original fluoride in the
sample has been volatilized.
Complete removal of the fluoride component of the melts was
reached in no case with the standard compositions, not even at high tem-
peratures (10000C.) or with prolonged heating (12 hours and longer) under
A large excess of metaphosphate, particularly if it formed a deep
layer, impaired strongly the rate of expulsion of fluorine-containing
gases. This effect must be due to the fact that the viscous melt impedes
the transfer of the reaction products. This assumption was also supported
by the observation that the total amount of fluorine removed was not
directly proportional to the size of the charge, but decreased as the
A rise in temperature generally favored the rate and extent of
defluorination. Special conditions prevailed at low concentrations of
fluorine in the presence of solid phosphate phases, as in the case of
fluoroapatite melts. A maximum was exhibited in the formation of
fluorine-containing gases just above the normal melting point of sodium
metaphosphate (about 620C).
The presence of water vapor in the furnace atmosphere was
efficacious in promoting the evolution of hydrogen fluoride, but what
seemed to be more significant was the concentration of the hydrogen con-
stituent in the melt.
There was a consistent and marked difference in the yield between
the melts containing sodium and calcium ions. Such a variation empha-
sizes the effect which different cations may have on the reaction, or
for example, in respect to the viscosity of the melt.
The decomposition of fluoroapatite by metaphosphate was almost
complete, judging from the comparison between the standard samples of
fluoroapatite and the respective mixtures of calcium fluoride and tri-
Small supplements of silica enhanced the defluorination by the
fornmtion of silicon ttrafluoridc; the addition of calcium sulfate gave
only slightly better results than the standard mixture, whereas sodium
sulfate and tetraborate did not have any beneficial effect under the
T 'I3ICAI NTr~T-TIT! Ii
The reactions of fluoride ions in salt melts at high termeratures
can be interpreted in the light of acid-base theory extended to non-
aqueous systems. From the point of view of the Lewis theory, the fluoride
ion acts as a strong base capable of reacting with high temperature acids,
which can be coordinatively unsaturated molecules like silica, or ions
such as metaphosphate.
Ions with medium ionic potential (B, Si, P) preferably tend to
form structures with oxygen bridges, thus building up macromolecules which
react as acids at higher temperatures. These so-called polyacids are
characterized by the linking of the single groups to large complexes by
means of oxygen bridges of the type X-O-X. The transition of a polyacid
to its corresponding base consists in breaking of an oxygen bridge, and
a simultaneous binding of an oxygen atom. The acid-base reaction is thus
connected with a disintegration of the macromolecular structure, as for
example, in the case of metaphosphates and silicates.
These macromolecules are built up of a network of PO4 and SiO4
tetrahedra, in the interstices of which are interspersed cations, such as
sodium or calcium. In the melt there is a certain degree of freedom of
motion for the individual cation, or tetrahedron, and undoubtedly there
exists a dynamic equilibrium while a continuous rearrangement and inter-
change of tetrahedra takes place (69). The occurrence of single
phosphate ions in metaphosphate melts is rather improbable.
Effect of the Type of Cations
A distinct relationship exists between the acidity of the poly-
acids and the cations present. Flood and Forland (20) working with
pyrosulfates, concluded that the stability of the oxygen bridge S-O-S is
greatly influenced by the polarizing power of the cation. The size of
the cation is the decisive factor for the number of oxygen bridges broken
per cation, provided that the polarizing power of the cation exceeds a
A similar set of relationships may be expected to exist also in
the metaphosphate melts which would explain the difference between the
effect of sodium and calcium ions. The ionic radii of these ions are
relatively close (Na0,.98; Ca1l.06), but since the polarizing power of
calcium ion is larger owing to its double charge, the ease of breaking
the P-O-P bonds should be enhanced by calcium ions. Barium ion should
not be as effective as calcium ion because of its larger size and
decreased power of polarization. Such a difference was observed experi-
mentally. Because of the presence of mdsture and uncertain chemical
composition, the tests with other cations were not reliable enough to
enable an extension of such conclusions to a variety of systems.
The problems regarding the effect of an individual cation on the
volatilization of fluorino-containing compounds, as related to its size
and charge, from salt melts, would form a special study which would
necessitate the use of refined experimental methods and a complete
exclusion of moisture from the starting materials.
Extension of the Reaction Concept
Besides cations, also anions other than metaphosphates could be
considered in systems which would exhibit similar phenomena of anion-
fluoride interaction in an acidic melt with the formation of volatile
fluorine-containing compounds. Borates and silicates evidently form such
systems. Furthermore, pyrosulfates form fluorosulfonates with fluorides,
and it is known that fluorosulfonates decompose with the formation of
sulfuryl difluoride (66).
Possibly even bichromates and retavanadates could behave
similarly; they might, for example, form the products chromyl fluoride
(Cr02F2), and vanadyl trifluoride (VOF3).
The interesting subject the possibilities and conditions of
formation of volatile covalent fluorine compounds in salt melts has
not been previously treated from a general point of view, most probably
because of lack of experimental data. It certainly would offer a new
rewarding field of experimentation, and possibly contribute to the theory
of the properties of salt melts.
The study of the various factors affecting the stability and in-
ertness of fluoroapatite toward chemical reactions as compared with
related materials, as TOll as with the component salts, should reveal
interesting relationships between structure and reactivity which may
prove to be meaningful in a wider context.
A significant feature of the fluoroapatite structure lies in the
fact that it is chemically more resistant than either one of its compo-
nent salts, calcium fluoride and tricalcium phosphate, or other minerals
of similar structure. Since the physical and chemical properties of
fluoroapatite are determined by the structure as well as by the size and
properties of the component ions, further detailed studies of those
physically determined facts may be profitable.
Recent developments in the investigation of phosphatic materials
have provided reliable methods for the synthesis of reproducible fluoro-
apatite. Thus it has become possible to study the chemical reactivity
of this material in its pure form. With the simplified method of fluoro-
apatite synthesis as described in this work, sufficient quantities of it
can be prepared in reasonable time. The quality control of the product
may be carried out by chemical analysis and microscopic observations.
The experiments in this laboratory confirmed the fact that pure
fluoroapatite is a very stable substance and chemically quite inert
toward most chemicals even at elevated temperatures. The general
possibilities of fluoroapatite reactions with chemicals in various phases
were studied, and the following conclusions were reached for each phase.
The consideration of gas-solid reactions does not give much
promise for the formation of volatile fluorine-containing compounds among
the products below 10000C, The destruction of the fluoroapatite struc-
ture, however, should be possible at relatively low temperatures in the
atmosphere of highly acidic gases or mixtures of strong reducing agents.
The addition of suitable solid reactants, for example, silica and
carbon would facilitate the rate of such reactions with a gaseous phase.
According to the available thermodynamic data, there are few
indications for solid-solid reactions below 10000C, for the modification
of the fluoroapatite structure with ordinary salts, but the possibility
of special effects by tho added materials is by no means excluded.
The melts should offer the widest variety of reactions with
fluoroapatite. The neutral salt melts can possibly induce partial
replacements in the fluoroapatite lattice. The basic salts react with
fluoroapatite at higher temperatures and the acidic melts do the same in
a lower temperature range.
Suitable combinations of two or three types of reactions as
indicated above may give rise to fluoroapatite reactions proceeding at a
rapid rate in the range of moderate temperatures. Such systems, however,
will become too complicated for the deduction of new facts regarding the
characteristics of the pure fluoroapatite structure.
The interaction of sodium metaphosphate with fluorides and
fluoroapatite was studied in greater detail and a number of factors
influencing the reactions were considered. The major variables were
composition and temperature. In all cases when sodium metaphosphate
comes into contact with fluorides or fluoroapatite above the temperature
of about 550C., reactions take place which liberate a considerable part
of fluorine in the form of volatile compounds, mostly as hydrogen
A special contribution of this work lies in the experimental
proof that at least one gas containing phosphorus and fluorine, is
released from the melt. Evidence points to phosphoryl trifluoride (POF3)
as a component among the gaseous products of the reaction. The relative
amount of hydrogen fluoride as compared to phosphoryl trifluoride depends
on the concentration of combined hydrogen in the melt and water vapor in
the atmosphere above the melt.
The interaction of metaphosphate melts with fluorides is a
general reaction independent of the type of fluoride and produces the
same major products in the gaseous phase. There are indications, however,
that the type of cation influences the rate of the reaction. For example,
calcium ions seem to be generally more effective in furthering the gas
evolution from the melt than the sodium ions.
The fluoroapatite structure is destroyed by sodium metaphosphate
above 550C., and under favorable conditions the fluorine content of the
sample is reduced by eighty per cent within an hour. The yield in the
fluorine-containing gases shows a maximum at about 6400C. The effect of
various added materials (e.g. silica) upon the reaction was determined.
The results have been interpreted in the light of the acid-base
theory according to the Lewis concept, and the possibilities for further
extension of such studies have been indicated.
1. "A.O.A.C. Methods of Analysis", 6-th Ed., (1945), p. 23.
2. Badger, A.E., and Pittman, W.C., Ceram. Ind., 28, 218 (1937).
3. Bassett, H. Jr., and Taylor, H. T., J. Chem. Soc., 101. 576 (1912);
J. Am. Chem. Soc., 37, 1691 (1915).
4. Beevers, C.A., and Mclntrye, D.B., Min. Mag., 2, 254 (1945).
5. Brasseur, H., and Dallemagne, M.J., Bull. soc. chim. France,
D 135 (1949).
6. Bredig, M.A., Franck, H.H., and Fuldner, H., Z. Elektrochem. 38,
7. Bridger, G.L., and Boyland, D.R., Ind, Eng. Chem., 5, 646 (1953).
8. Chaudron, G., and Wallaeys, R., Bull. soc. chim. France, D 133
9. Clark, N.A., J. Phys. Chem., 35, 1232 (1931).
10. Cohn, G., Chem. Rev., 42, 527 (1948).
11. Dadape, V.V., and Rao, M.R.A., Current Sci. (India) 20, 324 (1951).
12. Eisenberger, S., Lehrman, A., and Turner, 7-.D., Chem. Rev., 26,
13. Elmore, G.V., and Huffman, E.O., J. Am. Chem. Soc., 1, 4950 (1952).
1. Elmore, K.L., Chief of Research Branch, TVA, Vilson Dan, Alabama.
Private communication, April 1, 1953.
15. Private communication, August 24, 1953.
16. Elmore, K.L., Egan, E.P., and Wakefield, Z.T., J. Am. Chem. Soc.,
(a) 73 5581 (1951); (b) 2, 2418 (1950); 73, 5579 (1951).
17. Elmore, K.L., Huffman, E.0., and bolf, .W., Ind. Eng. Chem.,
32, ho (1942).
18. "Encyclopedia of Chemical Technolory,," Vol. 6, The Interscience
Encyclopedia, Inc., New York, 1951, (a) p. 425; (b) pp. 393-429.
19. Fales, H.A., and Kenny, F., "Inorganic Quantitative Analysis,"
D. Appleton-Century Company, Inc., New-York, 1939, pp. 284-290.
20. Flood, H., and Forland, T., Acta Chem. Scand., !, 781 (1947).
21. Fluorspar and Fluorine Chemicals, Rept. of Invest. No. 141, Div. of
the State Geological Survey, Univ. of Illinois, 1949, p. 21.
22. Fowler, R.N., and Bright, H.A., J. Research Nat. Bur. Standards,
15, 493 (1935).
23. Fox, F.J., Hill, s1.L., Jacob, K.D., and Reynolds, D.S., Ind. Eng.
Chem., 38, 329 (1946).
24. Friedel, C., Bull. Soc. Chim. (1) 21 241 (1874).
25. Hart, G., J. Assoc. Off. Agr. Chem., 24, 302 (1940).
26. Hedvall, J.A., Chem. Rev., 15, 139 (1934).
27. Hendricks, S.B., et al., Ind. Eng. Chem., 23, 1413 (1931).
28. Hendricks, S.B., Jefferson, M.E., and Mosley, V.M., Z. Krist.,
81, 352 (1932).
29. Hill, O.F., and Audrieth, L.F., "Inorganic Syntheses," Vol. III,
McGraw-Hill Book Co., New York, 1950, pp. 106-111.
30. Hill, O.F., and Audrieth, L.F., J. Phys. & Colloid Chem. 54,
31. Hill, '.L. and Beeson, K.C., J. Assoc. Official Agr. Chem., 1.8,
32. Hill, W.L., and Hendricks, S.B., Ind. Eng. Chem,, 28, O40 (1936).
33. Hill, W.L., et al., Ind. Eng. Chem., 39, 1667 (1947).
34. Hill, s.L., and Reynolds, D.S., J. Assoc. Official Agr. Chem.
28, 105 (1945).
35. Hillebrand, ".F., and Lundell, G.E.F., "Applied Inorganic Analysis,"
New York, Wiley, 1946, (a) p. 5011 (b) p. 565.
36. Hoffmann, J.I., and Lundell, G.E.F., J. Research Natl. Bur.
Standards, (a) 19, 59 (1937); (b) 20, 607 (1938).
37. Holcomb, R., and McKibbin, R.R., J. Am. Chem. Soc., 50, 1695 (1928).