|
ANION EXCHANGE STUDIES OF
FLUORIDE COMPLEXES
By
GEORGE B. KAUFFMAN
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
January, 1956
TABLE OF CONTENTS
Page
LIST OF TABLES . . . .. . . . . iii
LIST OF FIGURES. . . . . . . . iv
Chapter
I* INTRODUCTION . . . . . . 1
A. Complex Fluorides
B. Statement of the Problem
0. Previous Work
D. Ion Exchange
II. EXPERIMENTAL: CHARACTERIZATION OF THE RESIN 26
A. Resin Used
B. Preliminary Treatment
C. Preparation of Fluoride Form Resin
D. Proof of Pure Fluoride Resin
E. Thermal Stability and "Standard State"
F. Resin Bed "Density"
III. EXPERIMENTAL: ANION EXCHANGE STUDIES. . 47
A. Chemical Studies
Apparatus, Materials, and Techniques
Non-Adsorbed Transition Metal Ions
Adsorbed Transition Metal Ions
B. Radiochemical Studies
Apparatus, Materials, and Techniques
Non-Adsorbed Transition Metal Ions
Adsorbed Transition Metal Ions
IV* DISCUSSION . . .. . 11. Ill
V* SUMMARY. . . . . . . . 122
APPENDIX .. . . . . . . . . . 124
BIBLIOGRAPHY. .* . . . 128
ACKNOWLEDGIrENTS. . . . . . . . 134
BIOGRAPHICAL ITEMS . . . . . . . 136
LIST OF TABLES
Table
1. Definitions of, and Relations between,
Quantities and Terms Used in Ion
Exchange . . . . .* *
2. Proof of Pure Fluoride Resin . . o
3, Resin Bed "Density" of Standard State
R+F" . . . . . . . .
4. Ion Exchange Capacity of Standard
State R "- . . . . . . . .
5. Results of Column Experiments with
Saturated Metal Fluoride-Potassium
Fluoride Solutions . . . . .
6. Elution of Co(II) with 0.001 M KF
(Typical Case of a Non-Adsorbed
Ion) . . . . . . . . .*
7. Column Experiments with Co(II), Zn(II),
and Mn(II) . . . . . . . .
8. Column Experiments with Fe(III) in
Fluoride Solutions . . . .
9. K Values at Different Ionic Strengths. *
10. Concentrations of Species Present in
Fluoride Solutions * . .* . .
11. Calculations of Equilibrium Constants
for the 0.350 M KF Series.. ,. .
12. Values of q/r2 for First-Row Transition
Metal Ions * * *. * . .*
Page
. 16
. 40
* . 46
* .
* S
. . 89
* 95
* .*. 96
* S
S S S
lo6
113
iii
LIST OF FIGURES
Figure Page
1. Absorbancies of CrF 1 M KF and
N.BS. Standard KG2CO4 SoTutions . . . 60
2. Elution of Co(II), Zn(II), and Mn(II)
with 0.001 M KP (Typical Cases of
Non-Adsorbe6 Ions) . 0 . , 0 . . 78
3. Anion Exchange Adsorption of Fe(III)
and Ga(III). . . ., . . . 83
4. Distribution Coefficients for Fe(III)
as a Function of [Hi36] for Fluoride
Solutions* . * . . . a . 101
CHAPTER I
INTRODUCTION
A. Complex Fluorides
The term "complex compound" is subject to a wide
range of interpretation.a Neglecting the so-called "double
salts", complex compounds may be conveniently divided into
three main classes: (a) those consisting of infinite three-
dimensional complexes, which exist as such only in the solid
state, (b) those containing a finite group of atoms in the
solid state, but not in solution, and-finally (c) those con-
taining a finite group of atoms which remain bound together
as a unit in aqueous solution and possess characteristic prop-
erties.15 In spite of recent advances in the field of non-
aqueous media, most chemical reactions are still carried out
in water solution, and, therefore, this dissertation will be
restricted to finite complex ions existing as units in aque-
ous solution.
aThe following general discussion of complex fluorides
is adapted in part from:
Wells, A. F., "Structural Inorganic Chemistry", Ox-
ford University Press, 1945, and
Sidgwick, N. V., "The Chemical Elements and Their Com-
pounds", Oxford University Press, 1950.
1
2
The element fluorine occupies a somewhat anomalous
position with respect to the other halogens. This differ-
ence is sufficient to call for special treatment of the
fluorides.
Halides may be divided roughly into two main types,
ionic and covalent, with relative volatility being a differ-
entiating criterion. Although this criterion has been chal-
lenged by Pauling,82 it still possesses the advantages of
convenience and practicality. Ionic halides are character-
ized by high melting and boiling points, while covalent hal-
idea possess much lower melting and boiling points. The high
electronegativity of fluorine and the small size of its ion
favor the ionic state, and fluorides are often found to be
much more anlt-like than the other halides of the metal in
question.
Fluorides often differ greatly from the other halides
in water solubility. In cases where melting and boiling
points are quite high, as is the case with many fluorides,
low solubilities are to be expected. Consequently, in the
presence of only moderate concentrations of fluoride, the
amount of metal ion that can be retained in solution, even as
a complex, often falls below the limits of ordinary analyt-
ical detection methods. Therefore, a study of this type is
greatly facilitated by the use of radioactive tracers.
The insolubility of some complex fluorides is so
great that their precipitation has been employed in the
3
qualitative and quantitative determination of certain ele-
ments. Thus, the insolubility of cryolite in water has
suggested the use of aluminum fluoride as a reagent for
sodium salts.110 Along similar lines, Tananaev and Deich-
manl00 point out that the solid phase 2PePF3*NaF is so
slightly soluble in sodium fluoride solutions that the
latter can be used for the quantitative separation of iron
from metals whose fluorides are soluble, while Talipov and
Antipov99 have proposed the use of sodium fluoride solution
for the quantitative separation and determination of chromium
as 3 NaF*CrF3.
In spite of the prominence accorded the W'erner type
metal ammine complexes, there are still more known types of
complex halides than of any other type of complex salt; in
fact, almost as many as all the others combined. Complexes
have been described with every number of halogen atoms from
one to eight on the central atom.
It is difficult to generalize about the stability of
complex fluorides; in many cases, they exhibit greater
stability than the corresponding complex chlorides, as ex-
pected from a comparison of metal-fluorine bond energies with
those of metal-chlorine. However, sometimes they show a
lesser stability.
A survey of the literature shows that complex fluo-
rides have been described for at least forty-two of the
elements, These complexes are of more than academic in-
terest, for many of them occur as minerals, the most impor-
tant and familiar of which is cryolite, Na3AlF6, used in the
Hall process for the electrolytic production of aluminum.
In 1923, complex fluorides again proved their usefulness
when Revesy and Jantzen31 first succeeded in isolating haf-
nium from zirconium by fractional crystallization of K2HfF6
and K2ZrF6.
Even a cursory perusal of Chemical Abstracts reveals
that "complex" fluorides, especially those of aluminum,
silicon, titanium, boron, and beryllium are "heavy chemicals"
produced industrially in tonnage quantities. "Complex"
fluorides are used as fluxes, polymerization catalysts,
sources of metals, and glass additives, as well as in elec-
troplating baths, to mention only a few of the applications.
Literally hundreds of papers and patents can be found describ-
ing the properties, crystallography, and uses of these com-
pounds. However, practically all of these studies concern
the solid state or molten salt systems.
Sidgwick97 cautions against acceptance of the compo-
sition of solid phases as evidence of complex constitution
and suggests that crystal structure be regarded as the cri-
terion for complexity. As stated previously, this study is
concerned with complexes in solution, and if this further
restriction is superimposed upon Sidgwick's criterion, the
5
number of elements forming proved anionic complex fluorides
is reduced from the previously mentioned forty-two to nine,
viz., beryllium, aluminum, scandium, manganese, iron, zir-
conium, niobium, hafnium, and tantalum. This large discrep-
ancy between the present knowledge of solid complex fluo-
rides and of complex fluoride anions in aqueous solution
should not imply that few complex fluorides exist in solu-
tion, but rather that a great amount of work remains to be
done in this field.
B. Statement of the Problem
Although many compounds containing apparently com-
plex fluoride anions have been described in the literature
during the last half century, most of the information is
based on studies of solid phases, very little work having
been done in aqueous solution. The present study represents
an attempt to establish the existence or non-existence of
complex fluoride anions in solution together with quantita-
tive data on their properties. The recently developed ion
exchange method was used throughout, and this investigation
was limited to the first-row transition metal ions, specif-
ically, chromium(III), manganese(II), iron(II) and (III),
cobalt(II), nickel(II), copper(II), and zinc(II), with par-
ticular emphasis on iron(III). A brief study of gallium(III)
was also undertaken.
C. Previous Work
Since this dissertation constitutes a comparative
study of the complex fluorides of first-row transition metal
ions, a brief review of previous work is best treated by
considering each metal ion individually. In general, refer-
ence has been made only to studies of aqueous systems or of
solid phases crystallized from such solutions. Also, refer-
ence has not been made to studies where the metal in question
is part of a complex fluoride cation, except in cases where
such studies concern species in solution and thus may shed
some light on the problem at hand.
Since this study deals exclusively with specific
oxidation states of particular metals, other oxidation states
of these same metals also fall outside the scope of this dis-
cussion, It may be mentioned, however, that "complex" fluo-
rides of chromium(II),(IV),(V), and (VI), and of manganese(III)
and (IV) have been described.
Throughout this section, the author has taken the
liberty of translating the double salt formulas into Werner
coordinate formulas, merely in order to classify "complexes"
formally into types. It should be borne in mind, however,
that in most of the studies of solid phases, little, if any,
distinction has been made between double salts and true com-
plex salts, again emphasizing the need for the present study.
7
A comprehensive list of "complex" fluorides, with
references, is given by Gmelin,23 but is out of date. A
more detailed and slightly more recent discussion is given
by Kausch,39 while the most recent compilation is provided
by Sidgwick.97
Chromium(III).--Two main types of complex compounds
have been described, M 31CrF6] and M21[CrPF.H20], where MI
is a positive metal ion or an organic base. Among the former
type may be listed some of the "simple" chromium(III) fluo-
rides.107 Compounds of the type M31(CrF6] have also been
prepared with the alkali metals,10'94 ammonium ion,19#29,80'
84,103 thallium(I),10 and guanidine.106 The latter type,
M2 CrF5 H20], includes compounds of the alkali metals,3,'10
80,103 ammonium ion,103 and divalent first-row transition
metals.32,84
Although studies by Scheffer and Hammaker93 and
Wilson and TaubeI1 are mainly concerned with complex cat-
ions, they are of particular interest here, since they deal
with species in aqueous solution.
Scheffer and Hammaker applied Vosburgh and Cooper's
modification102 of Job's method of continuous variations38
to a spectrophotometric study of the system Cr(NO ) -NaF-
HN03 -H20 and concluded that the only complex formed in such
8
acid solutions was CrF++.a It may be mentioned here that by
similar experiments with divalent copper, cobalt, and nickel,
Scheffer and Hammaker found no evidence of fluoride com-
plexes, and suggest that only trivalent ions are capable of
forming fluoride complexes.
While Scheffer and Hammaker established the formula
of the predominant chromium(III) fluoride species, it re-
mained for Wilson and Taube to determine stability constants
for all the species present in acid solutions, viz., CrP++,
CrF2+, and CrF3. The average number of fluoride ions per
chromium(III) ion was shown to increase as the logarithm
[HF]/H 30+] increased. Thus, at sufficiently high pH, this
number might exceed three, i.e., complex anions might be
formed. In short, the fact that some investigators have
found no evidence for complex chromium(III) fluoride anions
in acid solutions does not necessarily exclude the possibil-
ity of the existence of such anions under different condi-
tions of concentration and pH.
Manganese(II).--"Complex" fluorides of divalent
manganese are neither numerous nor very stable, and all are
of the type MI[MnP 3. Such compounds of the alkali metals
precipitate when solutions containing the constituent ions
are mixed.9'79 A German patent of 192665 utilized the
aln general, coordinated water molecules have been
omitted from ionic formulas. It is assumed that enough
water molecules are present to fulfill the coordination num-
ber of the central ion.
9
thermal instability of such compounds in the preparation of
pure manganese(II) fluoride by heating NH4MnF3. Although
the structures of such compounds are unknown, they probably
have the perovsklte structure, because of the small size of
the manganese(II) ion, and thus they should resemble the
"complex" fluorides of magnesium, zinc, and nickel. Such
compounds are not considered complexes from the point of
view of this study.
Iron(II).--Very little work has been done on the
fluoride complexes of divalent iron, the investigation being
limited to the work of Wagnerl03 over fifty years ago. The
salts, prepared by mixing solutions of the components, are
of two types, MICFeF3] and M2I[FeF4]. Traces of iron(II) can
be detected in the presence of large concentrations of
iron(III) after conversion of the latter to colorless hexa-
fluoroferrate(III) ion.20 The fact that the iron(II) remains
unaffected by the potassium fluoride used seems to imply the
absence of stable iron(II) fluoride complexes. This view
is confirmed by Enss16 who has shown that the color of glass
containing iron(III) is decreased by adding sodium fluoride
to the melt, while the color intensity of glass containing
iron(II) is unaffected.
Iron(III).--The fluoride complexes of iron(III) have
been so thoroughly explored, and the amount of work, both on
solid phases and solutions, is so voluminous that reference
10
can be made only to the more important highlights. The
intricacy of the complex chemistry of iron(III) is empha-
sized in a review article64 in which it is pointed out that
in a water solution of iron(III) chloride, the complete
series of ions ranging from pure Fe(H2O)6+++ to FeC16 is
present. That iron(III) fluoride solutions parallel this
behavior is known, but the extent of complex ion formation
is not too clear, Among the anomalous properties of iron(III)
fluoride solutions are the following: (1) only one-third of
the fluoride is capable of taking part in metatheses;88 (2)
some of the iron is present in the iron(II) state, according
to Deussen,12 whose claim is contested by Weinland, Lang,
and Pikentscher;l06 and (3) the solutions lack the typical
iron(III) color.87 These properties have been "explained"
on the basis of complex formation, with different formulas
proposed by different investigators.
In general, the fluoroferrates(III) are similar to
the chloroferrates(III). Although the latter are intrinsi-
cally less stable than the former, they are much easier to
prepare, since iron(III) chloride is about a thousand times
more soluble in water than iron(III) fluoride.
The numerous solid-phase complex fluorides are of
three main types, MI[FeF4],106 M21[FeF5.H20],10,77,78,105,106
and M3I[FeF6].18,25,29,42,71,76,103 Most of these have been
prepared by precipitation from solutions of the component salts.
11
Many studies of the fluoride complexes of iron(III)
have been made in aqueous solution, but usually acid solu-
tions have been employed. Babko and Kleiner,,1 in spectro-
photometric studies of such dilute solutions, determined the
dissociation constants for the species FeF+, FeF2+ FeP3,
FeF4", and FeFi and concluded that at any fluoride concen-
tration above 10"6 M several of these forms are present
simultaneously. Transference experiments showed that
iron(III) was present as a cation at fluoride concentrations
below 2 x 103 M, but as an anion above this concentration.
Another spectrophotometric studyl01 has shown that the
iron(III) complex with fluoride is more stable than that
with thiocyanate, a fact well known from the failure of po-
tassium thiocyanate reagent to detect iron(III) in the pres-
ence of fluoride.20 Stability constants for the first three
complexes of iron(III) with fluoride ion have been determined
by kinetic studies in acid solution.34
The completing of iron(III) with fluoride has found
many applications in analytical chemistry. Many other com-
plex systems have been studied potentiometrically by measur-
ing their effect on the iron(III) fluoride equilibria.67^*
11,13,30,81 Many spectrophotometric studies of other sys-
tems, utilizing the change produced in the absorbancy of the
iron(III) fluoride complex, have also been carried out.40,
41,111
12
Thus, the existence of fluoride complexes of iron(III)
in aqueous solution seems well established.
Cobalt(II).--The few "complex" fluorides of cobalt(II)
that are known fall into two general classes, M21[CoF4] and
MI0oPF3'H20], and are similar to the corresponding nickel(II)
compounds. Both types can be prepared by mixing solutions
of the constituents. Such solutions must be saturated, since
the double salts formed are rather soluble. Sodium, potas-
sium, and ammonium salts of both series have been described.29'
63*86,103 The instability of these compounds seem to indi-
cate that they are double salts, rather than true complex
fluorides. Spectrophotometric studies93 have shown no evi-
dence for cobalt(II) fluoride complexes in solution.
Nickel(II).--Only a few "complex" nickel(II) fluo-
rides have been described, and all fall into three general
categories, MI[NiF*120], w21 ii], and M21[NiP-.H20]. Po-
tassium and ammonium salts of all these types have been pre-
pared.29,86,103
Investigation of the systems KF-NiF2-H20 and
H P-NiP2-H20 showed no definite double salts occurring as
solid phases in the former case, while in the latter, the
compound (NH )2NiPF-2H20 was observed. KNiF3, like the cor-
responding compounds of zinc(II) and magnesium(II) has the
perovskite structure, and, hence, should be considered not a
true complex but a double salt. The structures of the other
13
nickel(II) fluoride compounds would be expected to be simi-
lar. Spectrophotometric studies93 have shown no evidence
for nickel(II) fluoride complexes in solution.
Copper(II).--Sidgwick97 claims that this ion coor-
dinates readily with halide ions as shown by the color
change occurring in solutions of the halides at high concen-
trations. However, little work has been done on copper(II) -
fluoride complexes since the early studies of Helmolt29 and
Hass,28 which described compounds of the general types
M21[CuF4] and M ICuF3] where M is ammonium, potassium, or
rubidium ion. Since these salts, which are prepared by mix-
ing solutions of their constituents, are all pale blue in
color, and since oopper(II) fluoride is less soluble in
alkali metal fluoride solutions than in water,37 it would
seem that such compounds should be regarded as double salts,
rather than true complexes.
However, a fairly recent spectrophotometric study of
the relative stabilities of copper(II) complexes in solu-
tion75 has shown the fluoride complex to be more stable than
the chloride complex, but less stable than the bromide com-
plex. This seems at variance with the spectrophotometric
study of Scheffer and Hammaker,93 which showed no evidence
of copper(II) fluoride complexes in solution. Another re-
cent study81 employing electromotive force measurements of
aqueous solutions gives the value
14
K.5 = $5.0 1 x 103
for the reaction
Cu++ + P" CuF+
at 25C. Thus, the amount of even a positive fluoride com-
plex of copper(II) in aqueous solution is very small.
Zinc(II).--All the so-called "complex" compounds of
zinc are of two types, MIZnF3] and M2 1ZnF4]. By reaction
of aqueous solutions of zinc(II) fluoride and alkali metal
fluorides, many sparingly soluble compounds have been pre-
cipitated.29"63#861I03 All these compounds might more cor-
rectly be called "double" fluorides rather than "complex"
fluorides. KZnF3 has been shown to have the perovskite
structure4 consisting entirely of K+, Zn +, and F" ions,
and, hence, is not considered a complex from the point of
view of this dissertation. The structures of the other com-
pounds are not known but would be expected to be similar.
A recent study8l employing electromotive force meas-
urements of aqueous solutions gives the value
o0.$ = 5.4 1 x 10-3
for the reaction
Zn++ + -, ZnF+
at 2500. Thus, the amount of even a positive fluoride com-
plex of zinc(II) in solution is very small.
Thus, a review of the literature on the "complex"
fluorides of the metal ions to be investigated in this study
15
leads to the conclusion that anionic complexes of manga-
nese(II), iron(II), cobalt(II), nickel(II), and zinc(II) are
not expected to be stable in solution, while the possibility
of such complexes of chromium(III) and copper(II) exists.
The existence of anionic fluoride complexes of iron(III) is
a virtual certainty.
D. Ion Exchange
This study was carried out more from an inorganic
than a physical chemistry viewpoint. Consequently, the sub-
ject of paramount importance was the nature of the fluoride
complexes themselves, rather than the ion exchange techniques
used in their investigation. The study thus represents the
application of the ion exchange method as a tool in the
elucidation of the properties of fluoride complexes. With
this viewpoint in mind, the general discussion of ion ex-
change will be limited, its primary object being to render
the experimental work intelligible.
The phenomenon of ion exchange has been extensively
investigated in recent years, and several books dealing with
the subject have been published.61974,90 The annual review
articles by Kunin52"60 provide a comprehensive coverage of
the current literature.
Definitions of the more important quantitative con-
cepts used in ion exchange are found in Table 1.
TABLE 1
DEFINITIONS OF, AND RELATIONS BETWEEN, QUANTITIES
AND TERMS USED IN ION EXCHANGE
Column Experiments
(1) Elution constant:
E = dA
V
(2) Volume distribution coefficient:
D = amt. of metal ion/ml. resin bed
amt. of metal ion/ml, soln.
(both amts. in the same units)
(3) E 1 ; E = 1 forDv >i
D + D E Dv
Equilibration experiments
(.) Weight distribution coefficient:
D = amt. of metal ion/g. resin
amt. of metal ion/ml. soln.
(both amts. in the same units)
(5) Uptake (Ideally 50%):
U = amt. of metal ion in resin phase at equilibrium
amt. of metal ion in soln. initially
(both amts. in the same units)
(6) Loading (Ideally <1%):
L -mr x 100%
(7) For 50% uptake, ml. soln. =D (from (4) and (5))
g. resin
(8) Dv =pD
d = distance (cm.) the band maximum moves after passage of
17
TABLE I--Continued
V cm.3 of eluent through a column of cross-sectional
area A cm.2
i = fractional interstitial space in column; ims0.4
Column volume, C = d A
interstitial or void volume = i x column volume
P= bed "density", i.e., g. rosin per ml. bed
n = charge on adsorbedion
mr = concentration of metal ion in resin phase
c = capacity of resin (units same as mr)
The recent development of "strong base" anion ex-
change resins has introduced a valuable method for studying
negatively charged ions in aqueous solution. A particle of
such resin can be visualized as an elastic three-dimensional
hydrocarbon network to which are attached a large number of
ionizable groups. The high molecular weight organic portion
is insoluble because of its large size. It possesses a posi-
tive charge and therefore attracts anions, which it thus ef-
fectively removes from solution. Conversely, it repels cat-
ions, and these are not adsorbed.
The generally established use of the term "adsorp-
tion" to describe the retention of ions by an ion exchange
resin is an unfortunate choice. While undoubtedly some pure-
ly physical adsorption does occur, the phenomenon is primari-
ly a chemical exchange.
In general, complex anions are more strongly adsorbed
by an anion exchange resin than are the corresponding simple
anions. This important fact, more often implied than stated
directly, forms the basis for the use of anion exchange
18
resins in the study of complex anions. There are four
general techniques for investigating complex anions by this
method.
1. Column effluent analysis.--A very small sample
containing the metal ion in a large excess of a supporting
solution of the completing ion is allowed to seep onto the
top of a resin column which has already been equilibrated
with the same supporting solution. The column is then
eluted at a slow flow rate with more of the supporting solu-
tion. If the metal is not present as a complex anion, no
adsorption occurs, and it appears in the effluent as soon as
it has travelled through the spaces between the resin beads
in the column. This fractional interstitial space, I has
been found to be remarkably constant for a given resin, be-
ing roughly four-tenths of the total column volume for Dowex
1 in the chloride form.49 If any of the metal is present as
an anionic complex, it is adsorbed or "held up", but is
eventually eluted by the large excess of simple completing
ions, which compete for the functional sites on the resin.
In the case of colored complexes present at fairly high con-
centrations, the metal ion can actually be observed to travel
down the column as a "band". The color of the ions of many
of the first-row transition metals was one reason for their
selection for this study, another being their known pro-
nounced complex-forming tendencies. From a knowledge of the
19
volumes at which the metal ions are eluted, as detected by
spot tests or radiochemical means, elution constants and
distribution coefficients can be calculated. The smaller
the value of E, the elution constant, the larger the volume
of solution required for elution, i.e., the greater the ad-
sorption.
The column effluent analysis method, which was the
one predominantly used in this work, is particularly suit-
able for the study of complex ions that are not strongly ad-
sorbed, i.e., ions with low distribution coefficient values.
2. Column scanning.--This method varies from the
preceding in that direct scanning of the column with a suit-
able detector is used to locate the metal ion. Thus, elution
constants and distribution coefficients can be calculated
without waiting for the metal ion to be eluted from the col-
umn, making the method useful for the investigation of ions
that are strongly adsorbed, i.e., have high D values. The
method of detection may be a suitable counting device in the
case of radioisotopes or direct visual observation in the
case of colored ions. Kraus and Moore described the radio-
chemical modification of this method in 1951.47 Since that
time, it has been superseded by the simpler and more con-
venient equilibration method.
3. Equilibration or "shaking experiment" method.--A
known volume of a solution containing the metal ion in a
20
large excess of the completing ion, is shaken with a known
veoiht of resin until equilibrium is reached. From two
analyses, one giving the initial and the other the final
concentration of the metal in the solution phase, distri-
bution coefficients can be calculated. Elution constants
and distribution coefficients are related as shown by
equation (3) of Table 1, the relation being a simple inverse
one for large values of Dy.
Since the results of equilibration experiments are
most accurate for 50% uptake, a series of "successive
approximation" experiments are usually run with the goal of
eventually arriving at this uptake. If at the same time,
the loading of the resin is maintained below 1%, the value
of D should be constant and independent of both loading and
uptake.
The equilibration method is particularly suited to
the investigation of ions that are strongly adsorbed. It
was not extensively employed in this study, for reasons to
be mentioned later.
4. Special method for very strongly adsorbed ions.--
This is a combination of the equilibration and column ef-
fluent analysis methods. The radioactive metal tracer is
first adsorbed uniformly on the resin, and then the concen-
tration of solutions in equilibrium with it are measured.
Pa'.ilibrium may be achieved either by passing solutions
21
through a column containing tracer-loaded resin or by agi-
tating resin and solution together. Placing the tracer on
the resin first, permits attainment of equilibrium in each
individual experiment in much shorter times for high D val-
ues than when adsorption is carried out with the tracer
originally in the solution. This method was used in this
study in the investigation of the gallium(III) fluoride
system.
Much useful information, both practical and theo-
retical, can be derived from a study of the adsorbability
of complex anions on anion exchange resins. The most sig-
nificant practical applications lie in the field of separa-
tions. One of the advantages of ion exchange chromatography
is the fact that ions of very similar properties can some-
times be separated, if their elution constants are suffi-
ciently different. Thus, by means of cation exchange resins,
similar cations can sometimes be separated. However, ele-
ments which, as cations, ordinarily possess very similar
elution constants, are often found to possess widely differ-
ing elution constants when adsorbed as complex anions on an
anion exchange column because of their different complex-
forming tendencies. The accomplishment of the classically
difficult separation of hafnium from zirconium by Kraus and
Moore46 serves as an excellent example of this application.
In addition to the choice of a completing agent,
another variable that can be employed to magnify the
22
differences in adsorption between two or more ions is the
concentration of the completing ion. The amount of com-
plexing and, therefore, the extent of adsorption of dif-
ferent ions are often found to be markedly dependent upon
this concentration. Thus, by choosing a concentration of
completing ion at which these differences in adsorption are
most marked, the best possible separation can be achieved.
However, in addition to its use in practical separa-
tions, such a knowledge of the variation of adsorption of
metal ions with variation in completing ion concentration
is of fundamental theoretical importance. From the view-
point of this behavior, i.e., change in adsorption with
change in concentration of completing ion, metal ions can be
divided into three broad groups.
The first includes those which show no adsorption at
any concentration of completing ion. This is interpreted as
presumptive evidence that no anionic complexes are formed in
the concentration range investigated.
The second group includes those ions which show in-
creasing adsorption with increasing concentration of complex-
ing ion. This is interpreted as increasing complex anion
formation; the complex is not yet fully formed. The ions in
this group form weak complexes with the completing ion in
question and show strong adsorption only at high concentra-
tions of completing ion.
23
In the third group are found ions that show decreas-
ing adsorption with increasing concentration of completing
ion. Here the anionic complex is interpreted as being fully
formed, and the excess completing ions have an adverse ef-
fect on the adsorption of the complex; the simple completing
ions are competing with the complex ions for the functional
sites of the resin. The ions in this group form strong com-
plexes with the completing ion in question.
Although some ions fall into only one of these
groups, it is possible for a given ion to exhibit two or
even all three of these different types of behavior, depend-
ing upon the particular concentration range of complexing
ion. Thus, an ion may exhibit no adsorption at a very low
or very high concentration of completing ion but in the in-
termediate range may show both increasing and decreasing ad-
sorption with a maximum in the adsorption curve at the transi-
tion point. Since D (distribution coefficient) and E elutionn
constant) are inverse functions, this maximum would corre-
spond to a maximum on a graph of D versus concentration of
completing ion or a minimum on the corresponding graph of E.
In addition to elucidating the extent of complex ion
formation, the slope of the plot of the logarithm of D vs.
the logarithm of the concentration of completing ion can be
used to determine the charge on the complex ion in the range
where the complex is already fully formed.
24
The adsorption equilibrium of the complex ion B"n
with fluoride form resin can be represented by the equation:
(1) nPr" + B"n sBrp-n + nF-
where subscript r refers to the resin phase, and the absence
of this subscript refers to the solution phase. By approxi-
mating activities by concentrations, the equilibrium con-
stant for this reaction can be written:
(2) K = [Bpn][F]n
[FrP ]nB-n]
Moreover, assuming that the complex is fully formed,
[Br-n]
(3) D = k
(B-n]
where k is a proportionality constant. For low loading,
W[Fr] is constant and equal to the capacity c. This, in
combination with equations (2) and (3) yields:
(4) K D p[F]n
k on
Solving for D:
S) K4
($) D = .. ..
where
K' = Kkcn
(6) '. log D = log K'-n log (IF].
This is an equation of the general form,
y g + hx,
25
where h is the slope and g is the intercept.
(7) d log D -n
d log [F"]
or in general:
(8) d log Dn/a
d log[A"a]
where A"a represents the completing ion of charge -a, and -n
is the charge on the complex ion.
This equation was used (Chapter III) to determine the
charge on the complex ions present in the iron(III) fluo-
ride and gallium(III) fluoride systems.
Finally, a knowledge of the variation of the logarithm
of Dv with pH for the Fe(III)-KF-HF system was used in this
study to evaluate the third ionization constant of H3FeF6.
CHAPTER II
EXPERIMENTAL: CHARACTERIZATION OF THE RESIN
A. Resin Used
The resin used in these studies was the "strong
base" quaternary ammoniumr anion exchange resin Dowex 1.
The polymerization of styrene yields a linear or
two-dimensional polymer, while the copolymerization of sty-
rene with divinylbenzene yields a cross-linked or three-
dimensional polymer. Dowex 1 is prepared from such a sty-
rene-divinylbenzene copolymer, which confers water-insolu-
bllity upon the resin, The functional group is the benzyl
trimethylammonium ion. Thus, Dowex 1 may be represented as:
1-CH2N+(CHH3)3X
C6H5 C6H5 06H5S
...-CH-CH2-H-CH2 CH-CH2 -CH-CH2-CH-CH2-...
O -CH2N (CH3)3X"
... -CH-CH.-CH-,H. H-CH- --OH-CHl-CH-CHCH-...
27
where X" represents a monovalent exchangeable anion, while
the rest of the diagram represents the high molecular weight
polymer which contains ionic groupings as integral parts of
the polymer structure.
There are two variables in the preparation of sty-
rene-divinylbenzene polymers which influence the ion exchange
properties: first, the cross-linkage or the amount of
divinylbenzene, and second, the mesh size of the particles.
In the case of styrene-divinylbenzene resins such as
Dowex 1, cross-linkage is arbitrarily defined in terms of
the initial percentage of divinylbenzene, since there are no
known analytical techniques for determining the cross-linkage
in a fully polymerized resin. The divinylbenzene content of
a Dowex resin is indicated by an "X" number following the
number of the resin. Although two different batches of
Dowex 1 were used in this study, both were of 10% divinylben-
zene cross-linkage, i.e., Dowex 1-X10. Cross-linkage affects
the moisture content, capacity, porosity, and equilibration
rate of a resin.
Mesh sizes for Dowex resins refer to U. S. standard
screens and are based on the dry copolymer before the func-
tional groups have been introduced. The particle size of a
resin affects the equilibration rate, flow rate, and pressure
drop across an ion exchange column.
The two different batches of resin, differing only
slightly in particle size, which were employed in this study,
28
will henceforth be referred to as Resin A and Resin B.
Their characteristics are summarized below.
Resin A (used in preliminary studies): Dowex 1-XO1,
+160 mesh (on a wet basis) provided by Dr. Kurt A. Kraus of
the Oak Ridge National Laboratory. This resin was original-
ly obtained from the Dow Chemical Company, Midland, Michigan.
Resin B (used in later studies): Dowex l-X10, 200-
400 mesh (dry basis), medium porosity, total capacity ap-
proximately 3.0 0.3 milliequivalents per dry gram ("dry"
not defined), moisture content approximately 33-39% by
weight, lot number 3684-11, order number 177008, date 4-13-55.
This resin was obtained directly from the Dow Chemical Com-
pany, TTidland, Michigan.
Both Resins A and B were obtained in the usual com-
mercial chloride form.
B. Preliminary Treatment
Since the presence of very fine particles makes it
difficult to separate the resin from the solution in equili-
bration experiments, the crude resin was shaken with dis-
tilled water, the larger and heavier particles allowed to
settle for a short time, and the finer and lighter particles
decanted. This process was repeated until a satisfactory
separation was obtained. The resin was then transferred to
a large sintered-glass funnel fitted to a filter flask, where
29
it was mixed thoroughly with methanol'a, the latter removed
by aspirator suction, more methanol added and the process re-
peated until a clear filtrate showed that all methanol-solu-
ble organic impurities had been removed. Thorough removal
of methanol by washing with distilled water completed the
preliminary treatment.
C. Preparation of Fluoride Form Resin
Since the resin is marketed in the chloride form and
since in a study of complex fluoride ions extraneous ions
should be absent or kept at a minimum, the preparation of
the fluoride form of Dowex 1 was the first object in this in-
vestigation. Clearly, two methods of preparation were pos-
sible: (1) conversion of R+C1" to R+OH" with excess sodium
hydroxide solution followed by neutralization of R+OH" with
hydrofluoric acid, or (2) conversion of R+C01" directly to
R+FP with excess potassium fluoride solution.b The latter
aTechnical data concerning chemicals and solutions
have been relegated to the Appendix. Whenever such informa-
tion is to be found there, an asterisk has been placed after
the name of the substance at its first mention in the body
of the text.
bThroughout this dissertation, for the sake of sim-
plicity, the symbol "R+1" will be used to represent the large,
polymeric, organic, cationic portion of the resin. Thus,
R+C1-"should be read as "the chloride form of Dowex 1".
The "+" and "-" signs are used to emphasize the fact that
the attraction between the cation and anion is electrostatic,
and that a covalent bond is not involved.
30
procedure was decided upon since it involved one step in-
stead of two and since it did not require the use of the
more toxic hydrofluoric acid with its concomitant need for
polyethylene equipment.
Dowex 1 possesses distinct but different affinities
for different anions, and Wheaton and Baumanl08 have meas-
ured the selectivity of this resin for seventeen monovalent
anions. This selectivity pattern has been attributed to
differences in the sizes of the hydrated anions involved.
For ions of the same charge, more work is required for large
anions to enter the resin network, resulting in a resin
selectivity or preference for small anions. In the halide
series, the hydrated ion size decreases from fluoride
through iodide, and in keeping with the large size of the
hydrated fluoride ion, this ion appears at the very bottom
of the selectivity series.
Consideration of the foregoing, together with
Le Chatelier's Principle, leads to the conclusion that in
order to force the reaction
R+C1- + F" q=Y= R+F" + C01
to go to the right, a large excess of fluoride ion is re-
quired, together with the removal of the chloride ion formed.
Both these conditions can be met in two ways, leading to
slightly different modifications (1) the equilibrium or
batch method, and (2) the continuous flow or column method.
31
Since both these methods were used at various times in the
course of the work, it seems appropriate to describe them
in some detail at this point.
Equilibrium or batch method.--A portion of R1Cl- was
placed in an Erlenmeyer flask, together with more than
enough potassium fluoride" solution to completely cover the
resin. The mixture was covered with a watch glass and
stirred with a magnetic stirrer for an hour or two. It was
then poured onto a sintered-glass funnel fitted to a filter
flask, and the solution separated from the resin by aspira-
tor suction. The Erlenmeyer flask was rinsed with distilled
water, and these rinsings were used to wash the resin in the
funnel. The air-dried resin was then returned to the Erlen-
meyer flask and covered with fresh potassium fluoride solu-
tion* The entire procedure was repeated until a negative
chloride test was obtained on the washingsd A drop of the
filtrate and washings was acidified with one drop of concen-
trated nitric acid* and tested for chloride ion with one
drop of approximately 0.1 N silver nitrate* in a black spot
plate. Experimental investigation of the sensitivity of
this test showed the limit of detection to be about 10"' M
chloride. When a negative chloride test was obtained, the
resin was assumed to be completely converted to R+F", and it
was then washed free of interstitial potassium fluoride solu-
tion with distilled water, either by the batch or column
32
method. In either case, the filtrate was spot tested for
fluoride20 with zirconium alizarinate solution*, and washing
continued until a negative test was obtained. The limit of
detection of the fluoride test was found experimentally to
be in the range 10"3-10-4 M.
The batch method becomes increasingly inefficient as
the number of individual equilibrations increases. In prac-
tice, a half dozen or so equilibrations were performed, the
partially converted resin placed in a column and the rest of
the chloride ion removed with potassium fluoride solution.
Finally, the interstitial potassium fluoride solution was
removed with distilled water.
Continuous flow or column method.--A tuft of glass
wool was placed in a burette just above the stopcock and the
burette filled with distilled water. The stopcock was then
opened, and as the distilled water drained from the bottom,
a slurry of R+Cl- and distilled water was poured in at the
top and the resin particles allowed to settle, forming a bed
or column. Sodium fluoride* solution was then allowed to
flow continuously through the column until a negative chlo-
ride test was obtained on the effluent, at which point the
sodium fluoride solution was replaced by distilled water,
which was allowed to flow through the column until a negative
fluoride test was obtained on the effluent.
33
Impurities often present in the solutions and in the
distilled water collected on the top of the column, and this
portion of the resin was discarded.
The column method was found to be more efficient
than the batch method, in terms of materials and labor re-
quired, and, accordingly, this method was used in all but the
initial stages of the work. The position of the influent
solutions was arranged so that a continuous, automatic flow
through the column was obtained. Roughly 3 1. of 1 M sodium
fluoride at a flow rate of about 1 cm./min.a were required
to completely convert about 50 ml. (wet volume) of R+Cl" to
R+F" by this method. A saturated solution of sodium fluo-
ride (about 1 M) was routinely used for conversion of R+C1"
to R+F".
D. Proof of Pure Pluoride Resin
Except for a single equilibrium constant value
between the chloride and fluoride form of Dowex 1,108 vir-
tually nothing was known about the fluoride form of the
resin. The problem was to show first that a pure fluoride
resin could be prepared, and then to investigate its proper-
ties, with special emphasis on stability.
Hydrofluoric acid, unlike the other hydrohalogen
acids, is a weak acid, and, thus, the fluoride ion possesses
aFlow rates are usually reported in units of
cm.3 cm."2 min.-l This reduces to cm./min.
34
basic properties. However, the reaction of the fluoride ion
with water, shown by the equation
P" + H20 OH" + HF,
goes only to a small extent.
Dowex 1 has about the same selectivity for both hy-
droxide and fluoride ions,108 and in concentrated sodium
fluoride or potassium fluoride solutions such as those used
for preparing the fluoride form resin, the concentration of
fluoride ion is very much greater than that of hydroxide ion.
For these reasons, the possibility of contamination of R+F"
with R+OH" seemed fairly remote. Nevertheless, this possi-
bility was investigated.
First a rough qualitative test was run. A l-g. sam-
ple of R+WF (Resin A) was shaken for a half hour in a test
tube with about 10 ml. of 1 M potassium nitrate,* the mix-
ture centrifuged, and a drop of phenolphthalein solution*
added. Since the nitrate ion is very much higher than the
hydroxide ion in the selectivity series,108 any hydroxide
ion present in the resin would have been replaced by nitrate
ion, and the hydroxide ion released in the solution would
have given a basic reaction with the indicator. The com-
plete absence of any pink tinge showed the absence of hy-
droxide or other strongly basic ions, e.g., carbonate, both
in the solution and in the resin.
35
The quantitative method involves the determination
of both the chloride and fluoride capacities of the same
sample of resin; if the fluoride form resin is free of for-
eign ions, these capacities should be equal. Several meth-
ods are available for these quantitative determinations.
All of the methods considered for the determination
of chloride involved the precipitation of silver chloride.
The gravimetric method95 was found unsatisfactory since
nitric acid must be used in order to prevent the precipita-
tion of silver oxide by excess silver nitrate solution. In
the presence of fluoride ion, this acid produces hydrofluoric
acid, which attacks the Gooch or sintered-glass crucibles
used for filtration and weighing.
Of the volumetric methods, the adsorption indicator
method43,95 as well as the Volhard method95 proved unsatis-
factory because fluoride ion was found to interfere here
also. The Mohr method95 was finally chosen, since it alone
successfully circumvented this difficulty.
For the separation of fluoride ion from solution,
the precipitation of lead chlorofluoride95 was chosen.
After precipitation and washing, fluoride can be determined
either by direct weighing as lead chlorofluoride95 or by
dissolving the lead chlorofluoride in hot nitric acid, deter-
mining the resulting chloride by a suitable volumetric meth-
od, and then calculating the amount of fluoride ion.95
36
None of the above-mentioned volumetric methods for chloride
proved feasible here; in this case, the Mohr method could
not be used because of the possible precipitation of lead
chromate. Therefore, the method involving the simple,
direct weighing of lead chlorofluoride was employed. This
precipitate is ideal in that it is about fourteen times
heavier than the fluoride which it contains.
Before the actual determinations on effluent derived
from the resin, reagent solutions had to be standardized and
runs made on weighed samples of known composition.
Chloride determination.--The silver nitrate solution
was standardized by titrating eight accurately weighed sam-
ples of sodium chloride,* ranging from 0.5-0.8 g., that had
been dried to constant weight at 120-1300C. In order to re-
produce conditions to be encountered in the titration of the
effluent, duplicate samples were dissolved in two 100-ml.
portions of each of the following solutions: (1) distilled
water, (2) 1 M potassium fluoride, (3) 5 M potassium fluo-
ride, and (4) 10 M potassium fluoride. Four drops of satu-
rated potassium chromate* solution was used as the indicator,
and the end point was determined by comparing the color with
that of two mixtures in white casseroles, one containing
about 150 ml. of distilled water, four drops of saturated
potassium chromate solution and a few grams of calcium carbo-
nate,* and the other containing the same plus one drop of
37
the silver nitrate solution used in the titration. The cal-
cium carbonate was used to reproduce the turbidity of the
silver chloride formed in the titration, but, unlike the
latter, is not sensitive to light. The amounts used in the
controls corresponded to the amounts to be used in the actual
titrations of the effluent. The end point is more easily ob-
served if the titration is performed under a yellow light.
The normalities of the standard silver nitrate solu-
tion, calculated on the basis of these titrations, were
found to decrease slightly with increasing fluoride ion con-
centration. This was probably due to the effect of the lat-
ter on the ionic strength of the solution, which in turn
decreases activity coefficients, thus increasing the amount
of titrant needed for precipitation. The following "apparent"
normality values, each the average of two determinations,
were plotted against the corresponding fluoride ion concen-
trations: 0.2110 at 0.00 M, i.e., distilled water; 0.2103
at 1 M; 0.2100 at 2.5 i; 0.2091 at 5 M; and 0.2087 at 10 4.
From this plot, which levelled off at high fluoride ion con-
centrations, normality values were chosen for the particular
fluoride ion concentration present in the effluent, contain-
ing chloride and fluoride ions.
Fluoride determination.--Two accurately weighed sam-
ples of 0.8-0.9 g. each of sodium fluoride, dried to con-
stant weight at 120-1300C., together with two 50-g. portions
38
of potassium nitrate (to approximate the conditions to be
encountered in the effluent, containing fluoride and nitrate
ions), were dissolved in distilled water to a total volume
of 100 ml, each. The determination was carried out accord-
ing to the directions in Scott's "Standard Methods".95 The
solution was tested for complete precipitation by adding a
small amount of hydrochloric acid* and observing that no
further precipitation of lead chlorofluoride occurred. Zir-
conium alizarinate reagent could not be used for this purpose
since the amount of fluoride ion in a saturated lead chloro-
fluoride solution was sufficient to give a positive test. Ad-
dition of a large excess of hydrochloric acid will cause pre-
cipitation of lead chloride, which is easily recognized by
its characteristic needle-like structure. The precipitate
was washed several times with a saturated lead chlorofluoride
solution, but still the results were about 1% low due to
solubility losses.
Analysis of the effluent.--Fifty milliliters (wet
volume) of 0+C1" (Resin A) was treated columnwise with about
10 column volumes of 1 M hydrochloric acid to make certain
that the resin was completely in the chloride form. This
was followed by continual washing with about 6 column vol-
umes of distilled water, until all the interstitial hydro-
chloric acid had been removed, as shown by a negative chlo-
ride test on the effluent. The resin was air-dried on a
39
sintered-glass funnel by aspirator suction, dried at 110 C.
for several hours, and powdered lightly in a mortar. It
was then dried for thirty-six hours at 600C. in a vacuum
desiccator over Anhydrone,* as suggested by Kraus and
Yoore.48
Three accurately weighed samples, ranging from 2-5
g. each, of this dried resin, were treated batchwise with
5 M potassium fluoride in a half dozen equilibrations. The
partially converted resin samples were then completely con-
verted columnwise to the fluoride form with 2.5 M potassium
fluoride until a negative chloride test was obtained on the
effluent. The reaction is
R+C-l + F" -+ R+F" + C1.
The combined effluent, containing chloride and fluo-
ride ions, was evaporated to dryness, brought up to a total
volume of 100 ml. with distilled water, and analyzed for
chloride ion by the Mohr method with the previously men-
tioned silver nitrate solution, which had been standardized
by the same method. The resin, now completely in the fluo-
ride form, was washed continuously with distilled water un-
til a negative fluoride test was obtained on the effluent,
showing the absence of interstitial fluoride ion.
The resin was then eluted with 1.5 M potassium ni-
trate until a negative fluoride test was obtained on the ef-
fluent. The reaction is
40
R+F" + NO- 3 R+NO 3 + F_.
The combined effluent, containing fluoride and ni-
trate ions, was evaporated down to 200 ml. and analyzed
gravimetrically for fluoride by precipitation as lead chloro-
fluoride.
The results of this capacity study are shown in
Table 2.
TABLE 2
PROOF OF PURE FLUORIDE RESIN
Samples
1 2 3
Weight R+Cl-, g. 3.o651 2.3604 4.3150
Vol. AgN03 soln., ml. 50.81 38.78 70.57
Normality AgNO soln., mM/ml. 0.2088 0.2089 0.2087
Total amt. Cl", mM 10.49 8.100 14.72
Capacity Cl-, mM/g. 3.421 3.431 3.411
Deviation, capacity Cl" -3.98 -3.70 -4.27
from Kraus and Moore's
value (3.563 mM/g.), %
Weight PbClP, g. 2.7122 2.0912 3.8062
Total amt. Fp, mMa 10.36 7.998 14.55
Capacity F", mM/g. 3.382 3.388 3.371
Total amt. F" considering 10.60 8.240 14.80
PbCFlP solubility, mM
Capacity F" considering 3.461 3.490 3.431
PbClF solubility, mM/g.
Ratio mM P"/mM Cl" 0.9880 0.9874 0.9885
TABLE 2--Continued
Samples
1 2 3
Ratio mM F'/mM Cl1, 1.011 1.017 1.005
considering PbCIF
solubility
aPbClF contains 3.822 mM FV/g.
The deviation of "Capacity Cl"" from Kraus and
Moore's value48 is probably due to incomplete drying of the
resin. It should be remembered, however, that the primary
purpose of this capacity study was not to duplicate this val-
ue for R+Cl", but to establish the purity of the R+F" pre-
pared.
The "Ratio mM F/mM Cl"" is only about 15 less than
1.000. The low value is undoubtedly caused by loss of lead
chlorofluoride due to solubility. When the load chlorofluo-
ride gravimetric method was used previously on sodium fluo-
ride samples of known weight, the results were also about 1%
low. If allowance is made for lead chlorofluoride equilib-
rium solubility (0.0325 g./l00 g. of water at 18C.)95 the
"Ratio mM FP/mM Cl1" becomes about 1% greater than 1.000.
The true value is probably somewhere between these values,
showing that the resin obtained is a pure fluoride resin, or
at most deviates from this by 1%.
E. Thermal Stability and "Standard State"
In order to obtain quantitative data, such as distri-
bution coefficients from equilibration experiments, the
weight of the resin used in each experiment must be known.
Since resin particles contain distinct but different amounts
of water under different conditions, resins are usually
dried to constant weight under a specific set of conditions,
e.g., drying in vacuo over Anhydrone at 600C. used by Kraus
and Moore48 for R+Cl".
When a small sample of air-dried R+F'(Resin A) was
placed in an oven at 1100C. (a condition often used for dry-
ing R+Cl") for only a short time, a pronounced darkening and
loss of weight was noted. While minor color changes some-
times occur during conversion from one ionic form of a resin
to another, profound color changes, such as the one observed
above, rarely occur while drying one particular ionic form.
The darkening of R+F" is probably indicative of decomposition.
The only way to confirm this is by a series of capacity
studies or nitrogen analyses carried out at periodic inter-
vals, a time-consuming and complicated procedure, which was
not undertaken in this study.
Since the study of the thermal stability of R+F" was
not of primary importance in this work, continual loss in
weight and darkening on heating were chosen as convenient
criteria of decomposition.
Using these criteria, progressively less and less
drastic drying conditions were investigated, among them, in
order of decreasing decomposing action: vacuum desiccation
with Anhydrone; desiccation with Anhydrone without vacuum;
oven drying at 110C.; oven drying at 600C.; oven drying at
500C. and desiccation with calcium chloride.* A closed
bottle of cream-colored air-dried resin was used as a color
control. The only one of these treatments that led to the
attainment of constant weight (within 0.5% in 24 hours) with
only a small amount of decomposition was oven drying at 600C.
These studies, carried out over a period of several
months, led to the conclusion that not only is RPF" unstable
toward heat, but after a certain percentage of water is lost,
either by heating or desiccation, the resin begins to decom-
pose. Since even the most favorable of the above methods
resulted in some degree of decomposition, it was decided to
select a reproducible set of conditions which would result
in attainment of constant weight without decomposition, even
though this operational "standard state" would not yield
"dry" resin. Several hygrostats were investigated. In the
one finally adopted, R+F" was placed in a desiccator contain-
ing a slurry of calcium nitrate* in contact with its saturat-
ed solution at 25 2C. These conditions yielded a rela-
tive humidity of 51%,27 which was sufficiently high to
prevent decomposition of the resin, but low enough to allow
free flow of the particles, thus facilitating handling and
weighing.
A series of weighing on an initial 2-g. R+F" sample
showed that constant weight, and therefore, equilibrium was
attained to within 013% after a week in this hygrostat. A
simple moisture content experiment showed that in order to
convert from a given weight of hygrostated R+F" to the
weight which this resin would have if dried to constant
weight at 600C., one has only to subtract 30.1 0.5% from
the original weight.
F. Resin Bed "Density"
Since equilibration experiments usually yield data
in terms of resin weight, while column experiments yield
them in terms of resin volume, it is convenient to have a
conversion factor between the two quantities. The weight
per unit volume of resin bed, a "density" factor, was deter-
mined as follows.
Two columns were made from slurries prepared from
accurately weighed samples of standard state R+F" (Resin B)
(calcium nitrate hygrostated at 25 20C.). Ten-milliliter
burettes were used for this purpose, and in addition to the
glass wool plug, enough glass beads were added to the empty
burettes to bring the resin completely within the graduated
portions. After the particles had settled, the volume of
45
resin was read; the results are shown in Table 3.
TABLE 3
RESIN BED "DENSITY" OF STANDARD STATE R+F"
Columns
1 2
Weight R+F" sample, g. 4.430 4.250
Volume resin bed, ml. 8.980 8.735
P, resin bed "density",
g. standard state RFi'/ml. bed 0.4933 0.4867
l/p, bed volume per unit weight resin,
ml. bed/g, standard state RT"- 2.027 2.055
P, average,
g. standard state R+F-/ml. bed 0.4900a
1/P, average,
ml. bed/g, standard state R+F- 2.041a
aAverage deviation = 3.3 parts per thousand85
G. Ion Exchange Capacity of Standard State Resin
The resin (R+P", Resin B) in both of the columns men-
tioned in the previous section was converted into the nitrate
form by elution with about 200 ml. of 0.1 M potassium nitrate.
Complete conversion was insured by obtaining a negative fluo-
ride test with zirconium alizarinate reagent on the effluent.
This effluent, containing fluoride and nitrate ions,
was analyzed gravimetrically by the lead chlorofluoride meth-
od,95 The results are shown in Table 4.
TABLE 4
ION EXCHANGE CAPACITY OF STANDARD STATE R+F"
Columns
1 2
Weight R'+F", g. 4.430 4.250
Weight PbC1F, g. 3.160 3.029
Total amount F", mMa 12.07 11.57
Capacity F*, mM/g. 2.725 2.724
Capacity F", average, mM/g. 2.725
Average deviation85 (parts/thousand) <1
aPbCIF contains 3.822 mM F"/g.
The two columns were then washed with distilled
water until free of nitrate ion, employing the familiar
"brown ring" test.20 In the course of this washing, the
first column was disturbed by air bubbles, and at one time
the entire column shifted position. However, the final vol-
ume of the resin in Column 2 was measured, and the shrinkage
on conversion from R+F" to R47IO3 (both water-equilibrated)
was found to be 20.10%.
The relationships between the weight, bed volume,
and capacity of a given amount of R+F" have been experimen-
tally measured. From any one of these three quantities, the
other two can be readily calculated on the basis of the data
given in this chapter.
CHAPTER III
EXPERIMENTAL: ANION EXCHANGE STUDIES
A systematic study of the anion exchange behavior
of many of the transition metals in hydrochloric acid solu-
tion has been carried out by Kraus and Moore.49,73 In
carrying out such a study of the fluorides, a number of dif-
ficulties not encountered with the chlorides arise, as a re-
sult of the anomalous behavior of the fluoride ion compared
with the other halide ions.
First, the relative insolubility of transition metal
fluorides may cause them to precipitate from solution or on
ion exchange columns. Secondly, since hydrofluoric acid is
a weak acid, the basicity of the fluoride ion may cause simi-
lar precipitation of insoluble hydroxides or basic fluorides.
Thirdly, in the case of chloride ion, hydrochloric
acid was used, but the use of acid in the case of fluoride
ion involves several difficulties. Hydrofluoric acid being
weakly dissociated, hydrogen fluoride and acid fluoride ion
are present, together with the possibility of complexes other
than pure fluoride complexes. In addition, the presence of
the above undesirable species makes it necessary to measure
or calculate the pH in order to obtain the fluoride ion
47
concentration, while in the case of a "neutral" fluoride
solution, this concentration is given directly by the nomi-
nal concentration. Finally, hydrofluoric acid attacks glass,
and polyethylene columns become necessary.
The above difficulties, together with the fact that
most previous solution studies of fluoride complexes had
been carried out in strongly acid media, caused the selec-
tion of potassium fluoride solutions for these studies.
As pointed out in the Introduction, the mere fact
that a complex ion exists as a unit in the crystalline solid
does not imply that it will be stable in solution. Moreover,
complex ions which may exist in a concentrated solution may
decompose when the solution is diluted. Because of this
dependence of the stability of a complex ion on its environ-
ment, the concentration of the solution is a parameter that
will be carefully specified in this dissertation.
A. Chemical Studies
In general, an extensive series of column experi-
ments was carried out in order to determine which of the
transition metals investigated [chromium(II), manganese(II),
iron(II), iron(III), cobalt(II), nickel(II), copper(II) and
zinc(II)] form complex fluoride anions and which do not.
Then more intensive equilibration experiments were carried
out on those metal ions which showed evidence of complex
formation, i.e., those rhich were adsorbed in these column
experiments.
Apparatus, Materials, and Techniques
Solutions.--Reagont grade chemicals were used through-
out the work unless otherwise specified. In agreement with
the results of Savchuck and Armstrong,92 it was found that
alkali metal fluoride solutions reacted with glass containers
if allowed to remain in contact with them for extended peri-
ods of time. Therefore, polyethylene equipment or waxed glass
jugs were routinely used whenever necessary.
Potassium fluoride solutions.--The potassium fluo-
ride solutions were prepared, in very large quantities to in-
sure uniformity, by quantitative dilution of an accurately
prepared 10 M solution. Four of these solutions (0.001 M,
0.01 M, 0.1 M, and 1 1) were analyzed for fluoride spectro-
photometrically by the zirconium-alizarin sulfonate method.67
The solutions were diluted to the range 1.5-2.5 p.p.m., and
absorbancies were measured at 5$2 mu using a Beckman Model
DU spectrophotometer with matched 1-cm. Corex cells.a The
experimentally measured concentrations agreed with the nomi-
nal concentrations to within 10, which is the expected ac-
curacy of the method.
Two series of solutions were used in the column ex-
periments: (1) saturated metal fluoride potassium fluoride,
aThe author is indebted to Mr. John R. Wilson of the
University of Florida for these measurements.
50
and (2) 10"3 M metal ion-potassium fluoride.
Saturated metal fluoride potassium fluoride solu-
tions.--In order to avoid the presence of extraneous anions
and thus make certain that any complexes formed would be
fluoride complexes, metal fluorides were used in the prepara-
tion of these solutions. Of the fluorides used, chromium(III)
fluoride* and zinc(II) fluoride* were commercial samples.
The fluorides of iron(II) and iron(III) were prepared by
metathesis, i.e., reaction of solutions of the sulfate* and
nitrate,' respectively, with potassium fluoride solution.
The remaining fluorides, those of manganese(II), cobalt(II),
nickel(II), and copper(II), were prepared by the action of
excess hydrofluoric acid* on the metal carbonates*14062*109
and separated from solution by addition of 95% ethanol* as
suggested by Balbiano.2
These fluorides were thoroughly rinsed with both
ethanol and ether, and then air-dried. Samples of each fluo-
ride were shaken for a week with each of the following solu-
tions: (1) distilled water, (2) 0.01 M potassium fluoride,
(3) 0.1 M potassium fluoride, (4) 1 M potassium fluoride, (5)
5 M potassium fluoride, and (6) 10 M potassium fluoride, and
the undissolved metal fluoride was removed by filtration. In
order to prevent air-oxidation, the iron(II) fluoride was
shaken for only a few hours with potassium fluoride solutions
which had first been boiled to remove dissolved air; these
$1
solutions were not allowed to stand but were used immediate-
ly.
10"3 M metal ion-fluoride solutions.--In the detailed
investigation of copper(II) discussed below, it was discov-
ered that the copper(II) fluoride was not pure but was con-
taminated with the hydrofluoric acid used in its preparation.
Therefore, in order to obviate this difficulty for all the
metal fluorides, and at the same time to control the transi-
tion metal ion concentration, a second series of solutions
was prepared containing 10"3 M transition metal ion in the
same potassium fluoride solutions used in the first series.
A sufficient excess of fluoride ion was always pres-
ent to provide for possible completing of the metal ion.
Thus, even in the most unfavorable case, that of 0.01 M po-
tassium fluoride, the fluoride ion concentration was ten
times the transition metal ion concentration. For the same
reason, the amount of extraneous anion (nitrate or sulfate)
was negligible in comparison with the amount of fluoride
ion,
The solutions were prepared by pipetting 0.1 ml. of
a 0.1 M solution of each of the transition metal nitrates*
into 10 ml. of each potassium fluoride solution. In addi-
tion, two similar series of solutions were prepared, one us-
ing the same concentrations with ammonium fluoride* solutions,
and the other using 0.01 M, 0.1 M and saturated sodium
fluoride solutions.
In three cases, salts other than nitrates were used:
copper(II) sulfate* in the case of copper(II); Mohr's salt*
in the case of iron(II); and Mohr's salt oxidized with 30,1
hydrogen peroxide in the case of iron(III). The usual pre-
cautions were taken against air-oxidation in the case of
iron(II).
Solubility studies of 10-3 M metal ion fluoride
solutions.--The second series of solutions was agitated for
a week and observed for precipitation. In every potassium
fluoride solution, precipitation occurred at or below the
fluoride ion concentration calculated from solubility-pro-
duct constants (K p). The latter were obtained from solu-
bility values given by Seidell,96 assuming activity coeffi-
cients of unity. All the data listed for these metals in
potassium fluoride and ammonium fluoride solutions showed a
decrease in solubility with increasing fluoride ion concen-
tration.
At the same time, solubility-product constant calcu-
lations indicated that in most cases, at concentrations above
1 M, the amount of metal ion in solution would be too small
to be detected by spot tests. This was later confirmed ex-
perimentally.
Thus, on the basis of these approximate experiments,
not much evidence for the formation of soluble complexes is
indicated.
The solubility of the transition metal fluorides in
ammonium fluoride solutions was always greater than that in
potassium or sodium fluoride solutions, in which cases the
fluorides showed similar solubilities. This is probably due
to the fact that the pH's of ammonium fluoride solutions are
lower than those of potassium or sodium fluoride solutions.
The possibility that the ammonium salt of the complex might
be intrinsically more soluble than the potassium or sodium
salts seems unlikely since in general, potassium and ammo-
nium salts exhibit similar properties, among them solubility.
The phenomenon might be investigated by solubility studies
of metal fluorides in potassium or sodium fluoride solutions
of various pH's.
Column techniques.--Preliminary experiments to es-
tablish the presence or absence of complex anions were car-
ried out by the column effluent analysis method (Chapter I,
D). Short columns, about 10 cm. high, were constructed of
2-mm. inside diameter Pyrex tubing, the cross-sectional area
ranging from 0.035-0.040 cm.2, calibrated by measuring the
lengths of known weights of purified mercury* contained in
the columns. R+F- (Resin A) was used, together with retain-
ing plugs of glass wool, and the resin bed was covered with
a thin layer of spherical glass beads* to prevent floating
when the dense 5 M and 10 M potassium fluoride solutions were
used. The columns were fastened with "Scotch" tape to meter
sticks, which served the dual function of a support and a
graduated scale,. The eluent was added from burettes fastened
to the columns with Tygon sleeves. These burettes ranged in
capacity from 1-50 ml., depending upon the amount of adsorp-
tion.
Small samples (about one drop or 0.05 ml.) of the
various metal fluoride potassium fluoride and 10"3 M metal
ion potassium fluoride solutions were permitted to seep
onto the columns, which had been equilibrated with the same
molarity potassium fluoride solutions as used in the samples.
Long polyethylene capillary pipettes controlled with hypo-
dermic syringes were found useful for this purpose. Flow
rates were controlled by the burette stopcocks and were in
the range 0.5-1.5 cm./min. The experiments were carried out
at room temperature, 25-30OC.
Effluents were collected dropwise in white porcelain
spot plates, and the metal ion was detected by appropriate
spot tests. Detection of the maximum concentration of the
metal ion at about one interstitial volume ( i x column vol-
ume or about 0.4 column volume) was interpreted as lack of
adsorption. This was regarded as presumptive evidence of no
negative fluoride complex formation. Spot tests were run on
both blank solutions (potassium fluoride without metal ion)
and on potassium fluoride solutions with the metal ion to
55
establish the appearance of negative and positive tests,
respectively.
In general, with increasing potassium fluoride con-
centration, the spot tests appeared to become less sensitive.
This was due to decreasing metal ion concentration rather
than interference by fluoride ion. At fluoride concentra-
tions above 1 M, many of the transition metal ions could not
be detected at all; these results were in agreement with
previous solubility-product constant calculations.
Results of the column studies of saturated metal
fluoride potassium fluoride solutions are summarized in
Table 5. Similar column studies were run with those 10*3 r
transition metal ion potassium fluoride solutions which re-
mained free of precipitate, i.e., those below 1 M fluoride.
The results here coincided with those obtained from the first
series of solutions and, therefore, will not be tabulated.
Howovcr, some discrepancies were observed with copper(II),
which will be discussed later in this chapter.
On the basis of these column experiments, the metal
ions may be divided into two broad groups: (1) those which
show no adsorption, and (2) those which are adsorbed.
Non-Adsorbed Transition Metal Ions
Manganese(II), iron(II), cobalt(II), nickel(II), and
zino(II) showed no adsorption and were presumed not to form
complex fluoride anions at the concentrations investigated.
TABLE 5
RESULTS OF COLUMN
EXPERIrMENTS WITH SATURATED METAL FLUORIDE -
POTASSIUM FLUORIDE SOLUTIONS
Cr(III) Mn(II) Fe(II) Fe(III) Co(II) Ni(II) Cu(II) Zn(II)
0.01 M KF + + -- +
0.1 M KF + + + "
1MKF + + -- +
SKP + x # x x x + x
10 K # x # x x x #
0 *t 00
SpotTest o o 2 0
Reagents** Pd 4-3
"-' '0 04S 00
____o______ i __ a___3_' __ P P
Adsorption.
No adsorption.
Not performed.
Not detectable
by spot test.
Adsorbed Transition Metal Ions
Of the three cases in which metal ions showed adsorp-
tion, chromium(III), copper(II), and iron(III), not one
could definitely be attributed to complex anion formation.
Mere adsorption is not sufficient evidence for complex anion
formation. In many cases, especially the ones in question
where the fluorides and hydroxides are insoluble, precipita-
tion may be the cause of this apparent "adsorption". The
adsorption must be reversible before it can be accepted as
presumptive evidence of complex anion formation. Thus, a
metal ion must be both adsorbed and eluted by the use of the
same solution. Since these metals do not fulfill this cri-
terion, their behavior cannot be ascribed unequivocally to
the presence of complex anions.
If a metal ion is adsorbed on a column and is not
eluted within a reasonable volume (about 20 column volumes),
equilibration experiments are performed, since the column
method becomes tedious as well as less accurate for high val-
ues of D. Such equilibration experiments were carried out
but did not meet with any great degree of success.
Chromium(III) .--Chromium(III) failed to be eluted
with potassium fluoride solutions, but required 1.5 M potas-
sium nitrate for elution, during which it was observed to
travel down the column as a faint yellowish-green band. Such
behavior seems indicative of precipitation rather than true
adsorption.
The color alone of the saturated chromium(III) fluo-
ride potassium fluoride solutions was sufficient to indi-
cate the occurrence of some type of complex formation (not
necessarily anionic). Water, 0.01 M potassium fluoride, and
0.1 M potassium fluoride solutions were a deep grass green,
while the sediment was bright yellow. Solutions of higher
fluoride ion concentration (1 M, 5 M, and 10 M) were a very
light yellow color, while the sediment was light green.
Since equilibration experiments involve determination
of the amount of metal in solution both before and after
equilibration with resin, a quantitative analytical method
for chromium had to be devised. Potassium chromate solution
has been suggested as a spectral transmittancy standard by
Mellon,69 and therefore spectrophotometric determination as
chromate after oxidation with sodium peroxide* was the pro-
cedure selected.
Absorbancies were measured at 375 mj& with a Beckman
Model DU spectrophotometer using 1-cm. Corex cells and tung-
sten lamp. Measurements on standard potassium chromate solu-
tions showed that the system obeyed Beer's law perfectly
within the range investigated (1 x 10"5-4 x 10-4 M), and the
oxidized chromium solutions were diluted to bring them within
this range. The chromium concentration in the saturated
chromium(III) fluoride potassium fluoride solutions
59
ranged from 1 x 10"3-2 x 10"1 M, decreasing with increasing
fluoride ion concentration.
The results of these equilibration experiments were
not reproducible; moreover, the D values obtained showed no
correlation with the maximum E values obtained from column
experiments. Low D values of about 20 indicated only very
weak adsorption in equilibration experiments, while low E
values indicated strong adsorption on columns. The two
situations are evidently not comparable; chromium(III) hy-
droxide is probably precipitating on the columns.
Although no significant D values were obtained by
equilibration experiments, the presence of two distinct
species, indicative of complex formation, was demonstrated
by spectrophotometric studies of the unoxidized saturated
chromium(III) fluoride potassium fluoride solutions. A
plot of A. (absorbancy) vs. X (wavelength) for the green
solutions (0.01 M and 0.1 M potassium fluoride) showed dis-
tinct absorbancy peaks at'430 m, and 625 mpi. A similar
plot (Figure 1) for the yellow solutions (1 M, 5 M, and 10 M
potassium fluoride) showed distinct peaks at 270 mk and
375 mt. For the value in the ultraviolet region (270 mA),
a hydrogen lamp and silica cells were used. That the latter
two peaks were not due to the presence of chromate ion,
which gives absorbancy maxima at practically the same wave-
lengths (275 mu and 375 mu ), was shown by two observations.
60
250 300 350 400
WAVELENGTH, X (mp)
ABSORBANCIES OF
CrF3
-IMKF
AND
N. B. S.
STANDARD
KGr04
SOLUTIONS
FIGURE
1.000
0.900
0.800
0.700
0.600
0.500
0.400
0.300
0.200
0.100
61
First, although the positions of the As peaks are practical-
ly the same for both species, the 270 mtA peak is higher
than the 375 m1A peak for the saturated chromium(III) fluo-
ride potassium fluoride solution, whereas for the potas-
sium chromate solution, the reverse is true. Secondly, a
series of precipitation tests employing six different cat-
ions [barium(II), copper(II), lead(II), mercury(I), mer-
cury(II), and silver(I)] which form characteristic colored
precipitates, failed to show the presence of chromate.
Although ion exchange methods failed to definitely
establish their presence, chromium(IIl) fluoride complexes
are indicated by visual observation and spectrophotometric
measurements.
Copper(II).--The column behavior of copper(II)
showed a number of discrepancies. With all the saturated
copper(II) fluoride potassium fluoride solutions, some
"adsorption" was observed. With the 1 M and 5 M potassium
fluoride solutions, strong "adsorption" occurred, 2.5 M
hydrochloric acid being required for elution. On the other
hand, the 0.01 M and 0.1 M potassium fluoride solutions
showed only very slight "adsorption", being eluted with
about 2 and 4 interstitial volumes of potassium fluoride
solution, respectively. However, when these experiments
were repeated with 10"3 M copper(II) potassium fluoride
solutions, the 0.01 M potassium fluoride solution showed no
62
adsorption, while the 0.1 M potassium fluoride solution re-
quired 2.5 M hydrochloric acid for elution. The results are
not sufficiently reproducible to be significant.
Even more discrepancies were encountered in the
equilibration experiments. The analyses were carried out by
an electrodeposition method85 at a current of about one-half
ampere, using a Sargent-Slomin electrolytic analyzer with
platinum gauze cathodes and rotating platinum anodes. That
the method was both precise and accurate was shown by electro-
analysis of duplicate samples of an accurately prepared
standard copper(II) sulfate solution. However, analysis of
a saturated copper(II) fluoride solution revealed a copper
concentration forty times as great as that given by Carter.8
Carter also reported a pH value of 5.9 for such a solution,
while the p11's of the above solution and of some of the
copper(II) fluoride potassium fluoride solutions were
found to be about 3 with Hydrion paper, indicating the pres-
ence of excess hydrofluoric acid. When small portions of
these solutions were "neutralized" to pH 6 with 1 v potas-
sium hydroxide, most of the copper(II) fluoride precipi-
tated. It was this discovery of the presence of hydrofluo-
ric acid in the transition metal fluorides that led to the
substitution of 10-3 M metal ion potassium fluoride solu-
tions for the previously used saturated metal fluoride po-
tassium fluoride solutions.
63
Since the electroanalytical method required large
samples to insure a reasonable degree of accuracy, and since
each analysis required at least several hours, other methods
were investigated, with no success. A spectrophotometric
method70*98,112 involving measurement of the absorbancy of
the copper(II) ammonia complex at 580 mtL was experimental-
ly explored, but was discarded as not sufficiently sensitive.
The sodium diethyldithiocarbamate* method98 was explored
using the AC Model Fisher Electrophotometer with 2.5-cm.
cylindrical cells and a 425 B blue filter, but the results
obtained with this method were not reproducible because of
interference by fluoride ion.
Equilibration experiments were therefore carried out
by the reliable but cumbersome electroanalytical method. No
adsorption was found consistently in the case of the 10"3 M
copper(II) 0.01 M potassium fluoride solution, in agree-
ment with the results obtained by column experiments. Defi-
nite adsorption was observed with the 10"3 M copper(II) -
0.1 M potassium fluoride solution, but D values varied from
222-605, no agreement being obtained with similar samples.
The electroanalytical method was vindicated when it was dis-
covered that slow precipitation of copper(II) fluoride had
been occurring in both storage bottles and polyethylene shak-
ing containers. Thus, the constant possibility of precipi-
tation is a complicating factor in ion exchange studies of
64
transition metal fluoride complexes, and makes "adsorption"
difficult of interpretation.
Iron(III).--All three saturated iron(III) fluoride -
potassium fluoride solutions used in column experiments
showed strong "adsorption", but none were eluted with potas-
sium fluoride solutions. The 1 M potassium fluoride solu-
tion was eluted with 1.5 M potassium nitrate, while the
0.01 M and 0.1 M potassium fluoride solutions required 0.5 M
hydrochloric acid for elution, after 1.5 M potassium nitrate
had merely precipitated the iron as a reddish-brown deposit
at the top of the resin bed.
Precipitation difficulties were also encountered in
equilibration experiments. Several series of such experi-
ments were performed with erratic and inconsistent results
until it was realized that these results were being caused
by precipitation. A solution with as low an iron(III) con-
centration as 10"4 M precipitated in the presence of as small
a fluoride concentration as 0.05 M. The insidious point
about these precipitations is their slowness. Solutions
which had remained clear for a month or more and had thus
been considered "stable" and suitable for equilibration ex-
periments, suddenly began to precipitate. This then cast
doubt upon all equilibration experiments run with such solu-
tions.
65
Eventually a series of six stable solutions was pre-
paredt
1.00 x 10"4 M iron(III) 5.00 x 10" MA potassium fluoride
1.00 x 10-4 M iron(III) 1.00 x 10-3 M potassium fluoride
1.00 x 10"4 M iron(III) 5.00 x 10-3 M potassium fluoride
1.00 x 10,"4 m iron(III) 1.00 x 102 M potassium fluoride
4.00 x 104 M iron(III) 5.00 x 10-2 M potassium fluoride
3.00 x 10"-5 iron(III) 1.00 x 10-1 V potassium fluoride
The concentration of iron(III) in the last two solu-
tions was necessarily less than 1.00 x 10i4 M because of in-
solubility factors, but these two solutions were definitely
not saturated. The iron(III) was added from an accurately
prepared standard stock solution of 0.01 M iron(II) ammonium
sulfate (Mohr's salt) which had been oxidized with 30% hy-
drogen peroxide, any excess peroxide having been expelled by
boiling. In addition to the "extraneous" sulfate and ammo-
nium ions from the Mohr's salt, hydronium and more sulfate
ions were present from the 2.8 ml. of concentrated sulfuric
acid* added to a liter of this solution to prevent hydroly-
sis.
In each equilibration experiment, the procedure was
to weigh "standard state" R+F" (Resin B) samples into indi-
vidual polyethylene bottles, pipette in the required amount
of iron(III) potassium fluoride solution and shake the re-
sulting mixture on a rotating wheel. The loading was kept
below 1%.
After the required equilibration time of about
twenty-four hours had elapsed, the solutions were filtered
and analyzed for iron by a colorimetric method modified from
that of Mehlig and Shepherd,68 employing ammonia,* hydroxyl-
amine,* and thioglycolic acid.* A Fisher Model AC Electro-
photometer was used with a 5$2 B filter and matched 5-cam
rectangular glass cells. Each resin-equilibrated sample was
analyzed in duplicate. On the basis of a number of experi-
ments on known solutions the analytical method had given re-
sults reproducible within a few percent, and no fluoride
interference had been observed.
D (distribution coefficient) values varying from
about 40 to as high as 1400 were obtained, but, contrary to
theoretical expectations, for any given solution the D val-
ues were not constant but were apparently strongly dependent
upon the uptake, decreasing with decreasing uptake. This
effect was much too great to be caused by analytical errors
arising from uptakes differing from the optimum condition of
50%. Another series of similar equilibration experiments
yielded similar results and confirmed this apparent depend-
ence of D values on uptake.
It was found by plotting the logarithm of D vs. the
logarithm of the iron(III) concentration, that the values
of D were essentially independent of fluoride ion
67
concentration, at least up to about 0,01 M, and were not
strongly dependent above that concentration. Also, the
relatively large concentration of sulfuric acid (5 x 10"4 M)
present in these solutions to prevent hydrolysis, was con-
siderably in excess of the iron(III) concentration. In the
light of these facts, a guess as to the reason for the un-
usual results can be hazarded.
Due to the much higher position of sulfate with
respect to fluoride in the ion selectivity series for Dowex
1,108 it is quite probable that the resin was largely in the
sulfate form. The invariance of D with fluoride ion concen-
tration at fluoride concentrations less than or equal to
0.01 M is probably evidence that the complex being studied
was predominantly an iron(III) sulfate, rather than an
iron(III) fluoride complex. Thus, the D's that were meas-
ured were values for the distribution of an unknown mixture
of iron(III) ion, and complexes of iron(III) hydrofluoric
acid (due to low pH), iron(III) -rsulfate, and iron(III) -
fluoride between a solution phase and a resin phase which
was predominantly R2+SO$ with perhaps a small amount of
R+F". Not only is the system too complex to be of much theo-
retical significance, but it falls outside the scope of this
study.
Thoms and Gantz101 have reported that for iron(III)
the fluoride complex is more stable than the sulfate one.
68
Measurements of the pH's of the iron(III) potassium fluo-
ride solutions, made with a Coleman pH Electrometer, showed
that in the lower molarity fluoride solutions much of the
fluoride was present as hydrofluoric acid. This may account
in part for the apparent lack of fluoride completing and the
invariance of D with fluoride ion concentration.
In order to eliminate the presence of sulfate ion,
solutions of 3.00 x 10"-5 iron(III), both with and without
fluoride ion, were prepared from a stock solution of 0*01 M
iron(III) perchlorate, together with varying amounts of
perchloric acid* to prevent hydrolysis. Even in solutions
where the concentration of added perchloric acid was greater
than that of the sulfuric acid in the previous solutions,
hydrolysis occurred as evidenced by the formation of a yel-
low golden sol which exhibited the Tyndall effect. The in-
escapable conclusion is that sulfuric acid keeps iron(III)
in solution not by a simple pH effect, but by a completing
action.
Solutions of 3.00 x 10-5 M iron(III) in hydrofluoric
acid with a wide range of acid concentration (0.00286-28.6 m)
were prepared and found to be quite stable. In the previous
solutions, the high concentration of hydronium ion from the
perchloric acid repressed the fluoride ion concentration,
whereas in hydrofluoric acid solutions, a greater fluoride
ion concentration is possible. Thus, by the use of
69
hydrofluoric acid, it is possible to keep iron(III) in solu-
tion over a wide range of fluoride ion concentrations. How-
ever, as stated previously, it was desired to study the com-
plexes of transition metals in "neutral" fluoride solutions,
without addition of acid, so this expedient seemed unaccept-
able.
It has been shown that most of the difficulties en-
countered in this work were caused by precipitation and the
presence of "extraneous" ions. By the use of extremely low
metal ion concentrations (tracer amounts) it was hoped to
circumvent these difficulties and to prepare and study sys-
tems consisting simply of the metal ion in potassium fluoride
solutions, yet not be troubled by colloid formation, hydroly-
sis, fluoride precipitation, etc. Since the available radio-
isotopes of the first-row transition metals are all either
quite "hot" or short-lived, health monitoring devices and a
convenient, close source of supply became practically manda-
tory. For these reasons, the following radiochemical studies
were undertaken at the Oak Ridge National Laboratory.
B. Radiochemical Studies
The essentials and advantages of the application of
radioisotopes to chemical problems have been adequately dis-
cussed elsewhere.21,89,104 However, it does not seem inap-
propriate to mention briefly the advantages of this method
as they apply to the problem at hand.
70
1. Speed and easb of analysis.--The large number of
quantitative and semi-quantitative determinations necessary
in studies of this type are greatly facilitated by simple
and rapid analytical methods. By the use of radioisotopes
and counting equipment, a direct, complete quantitative
analysis, without the necessity for any preliminary separa-
tions, can be made in the short time required for counting,
usually one minute.
2. Low loading of the resin.--For a given solution,
adsorbability and consequently D values, may decrease strong-
ly with increasing loading. However, this effect is small
if the loading is kept below 1%, as can readily be done with
radioisotope tracers. Thus, adsorbability can safely be as-
sumed to be independent of loading, leaving the concentra-
tion of the solution as the only variable.
3. No precipitation problems.*-By the use of radio-
isotopes in tracer amounts, the solubility-product is not
exceeded, and precipitation difficulties, such as those en-
countered previously, should be eliminated.
4. Sensitivity.--In many of the previously described
solutions, especially those of fluoride concentrations above
1 M, ordinary spot tests were not sensitive enough to detect
the small amount of metal ion'in solution. By the use of
radioactive tracers, concentrations as low as 10-7 M can
readily be detected.
71
5. No extraneous ions.--By use of this method, con-
centrations of interfering foreign ions can be kept at a
negligible minimum.
One disadvantage inherent in the use of tracer
amounts is the uncertainty in the oxidation state of the
element present, especially in the case of unstable oxida-
tion states. Thus, the possibility of oxidation of tracer
amounts of iron(II) was the reason that this species was not
studied by this method.
Apparatus, Materials, and Techniques
Radioisotope tracers.--Tracers were obtained from
the Oak Ridge National Laboratory Isotopes Division and are
adequately described in detail elsewhere.33,36 Therefore,
only a minimum of information from the Oak Ridge Isotopes
Catalog is quoted below
1. "Co-60-P Processed, High Specific Activity,
Co00 A ,Y, T 5.3 years".
2. "Z -65-P-1 Processed, High Specific Activity,
Zno5 <3,y, T, 250 days".
3. "Mn-5254-P Processed, Carrier-Free, Mn52 (3,y,
Tj 5.8 days, Mn54<3,Y, T, 310 days".
4. "Fe-59-P Processed, Enriched, Fe59 (3, ,
T V 46.3 days",
5. "Ga-72-I Irradiated Unit, Ga?2 (3, Ti
14.25 hours",
The manganese tracer was about two years old, so
that practically all the Nn52 had decayed, leaving almost
72
pure Mn54. The Ga72 tracer was prepared by neutron bombard-
ment of reagent grade Ga203 in the Oak Ridge National Labora-
tory Low Intensity Training Reactor (LITR). All the above
radioisotopes were in the form of chlorides dissolved in
dilute hydrochloric acid (well below 1 L), except for the
Ga72, in which case the irradiated oxide was dissolved in
hydrochloric acid.
These concentrated isotope solutions were evaporated
just to dryness to volatilize excess hydrochloric acid. The
only case in which this precaution was absolutely necessary
was that of zinc(II). All the other elements show negligible
adsorption, and therefore no apparent completing in dilute
hydrochloric acid solutions.49,5l Stock solutions were pre-
pared by dissolving the residue in an amount of the appro-
priate potassium fluoride solution, such that the final
tracer concentration yielded a convenient and accurate count-
ing rate.
Counting equipment.--Samples were counted in 10 x 75
mm. Kimble glass culture tubes, while similar tubes of Lus-
teroid were used for solutions containing hydrofluoric acid*
Since all the radioisotopes used were gamma-emitters, the
samples were counted by placing the tubes directly in a well-
type scintillation counter. This device consists of a thal-
lium-activated sodium iodide crystal cemented to the flat,
sensitive end of an RCA 5819 photomultiplier tube. The tube
73
and crystal were contained in a light-tight aluminum can
mounted in a standard lead "castle", and the scintillation
pulses were detected and counted by a Nuclear Measurements
Corporation Model SU-1 scaler.
Column techniques.--Column techniques were essen-
tially the same as those described previously, except for
the modifications described below.
The columns were much larger, being prepared from
5-ml. graduated Mohr glass pipettes. Fluorothene tubing of
about the same diameter, plugged with Lucite shavings,
served as columns for acid solutions, in which case poly-
ethylene burettes and pipettes were used. R4F" (Resin B)
was used throughout. Samples were collected in small cul-
ture tubes of glass or Lusteroid and analyzed radiometri-
cally. Flow rates were kept well below 1 cm./min. In al-
most all the experiments, symmetrical Gaussian elution
curves were obtained, showing that the columns were operated
essentially under equilibrium conditions. Tracer samples
were placed on columns with micro pipettes. The experiments
were carried out in an air-conditioned room at 25 20C.
In most cases a few lambdasa of the stock tracer
solution were mixed on a paraffin-coated microscope slide
with about 50 ~ of the desired solution. This sample was
allowed to seep onto a column previously equilibrated with
aOne lambda equals 0.001 ml.
74
the same solution without the tracer. Then, before elution
was started, an additional 50-100 X of the solution without
tracer was allowed to seep onto the column to prevent back
diffusion of the tracer. In some of the iron(III) experi-
ments, the tracer was added to the column in a solution in
which it was strongly adsorbed, and the iron(III) was then
eluted with a solution of a different composition.
Non-Adsorbed Transition Metal Ions
It was decided to perform column experiments with
several of the ions that had shown no adsorption when chemi-
cal spot tests were used, in order to: (1) confirm the non-
adsorbability of these elements, since spot tests usually
yield only approximate results, (2) accurately determine the
value of i the fractional interstitial space, and (3) be-
come familiar with the radiochemical method and make certain
that experiments were performed under equilibrium conditions.
Because of the long equilibration time required for chro-
mium,111 and the very short half-lives of copper (Cu64) and
nickel (Ni65), radiochemical studies of these elements were
not carried out.
The fractional interstitial space, i has been stated
to be fairly constant for a given resin, being about 0.4 of
the total column volume for Dowex 1 in the chloride form.49
However, i should not be regarded as a true constant, since
75
small differences in i values between different columns
are observed. Thus, although approximately constant, the
value of i is sensitive to the method of preparation of the
columns, their age and history, and to density differences
between the resin and solutions. Moreover, the columns are
apparently loosely packed, since a consideration of resin
particles as close-packed spheres of uniform size leads to
a value for j of only 0.28.
Since similar results were obtained for the three
metal ions used in this series of experiments, complete data
are presented for only one typical case, cobalt in 0.001 M
potassium fluoride, in order to illustrate the method.
These data are presented in Table 6, while the symmetrical
elution curves for this case and two other typical cases are
shown in Figure 2. In addition, the data for the entire
series of experiments are summarized in Table 7.
Prom data of the form shown in Table 6, the volumes
in column (5) as abscissae are plotted against the corre-
sponding units in column (9) as ordinates (Figure 2). The
height of the elution curve peaks depends merely upon the
amount of tracer placed on the column; it is the horizontal
position of these peaks with respect to the horizontal axis
that is important. Since the abscissae have been magnified,
the differences between the peaks, and therefore between the
i values, have been exaggerated.
76
TABLE 6
ELUTION OF Co(II) WITH 0.001 M KF
(TYPICAL CASE OF A NON-ADSORBED ION)
Volumes, ml,
Burette
Readings
(2)
Volume
Difference
(3)
Total
Volume
(4)
Plotted
Volume
(Midpoints)
(5)
Tracer +
Rinsea 0.15
1 4.00-3.00 1.00 1.155 .0.655
2 ._.l0-4..0O 0.10 1.255 1.205
3 4.20-4.10 0.10 1.3.5 1.30I
-. 4.30o-..20 0.10 1. 455 1. o
5 4.4o-..30 0.10 1.555 1.505
6 i.5.4i..o- 0 0.10 1.655 1.6o5
7 4.6o- .5o 0.10 1.755 1.705
8 .70.-4.60 0.10 1.855 1.805
9 4t.80-0.,70 0.10 1.955 1.905
10 4.90-4.80 0.10 2.055 2,005
11 5.00-4.90 0.10 2.155 2.10.
12 5.10-5.oo00 0.10 2.255 2.205
13 5.255.10 0.15 2.405 2.330
1 5..35-5.25 0.10 2.505 2.455
15 ...45-.35 0.10 2.605 2.555
16 ..5-5.4 o.10 2.705 2.655
aTracer = 10 X Co60 stock solution mixed
KP. Rinse = 100 ), 0.001 M KF.
with 100 X
Sample
(1)
0.001
TABLE 6--Continued
Concentrations
Counts/Min. Counts/Min. ( Counts/Min.
Counts/Min.b (Background,0-4, (Decimal /0.1 ml.
Large-Small Subtracted) Equivalents) (Arbitrary Units)
(6) (7) (8) (9)
0 28 0 + 24/64 0.38 0.04
0 0 + 1/64 0.02 0.02
0o -9 o + /6 0.08 0.08
0 4 0.00 0.00 0.00
I 7 1 + 3/64 1.0- 1.05
9 37 9 + 33/64. 9.52 9. 52
II 2 11 1 + 48/6 11.75 11.75
.3 7 3 + I3/64 3.67 3.67
0 .42 0 + 38/64. 0.59 0.59
0o 10 o + 6/6 0.09 0.09
o -6 0 + 2/6. 0.03 0.03
So + 1/64 0.02 0.02
0 0 + 1/64 0.02 0.01
o 5 o + 1/6 0.02 0.02
0___ 0 + L1/64 0.02 0.02
0 4 0.00 0.00 0.00
bone large "count" = 4096 counts/min. One small
"count" = 64 counts/min. or 1/64 large "count".
NUMBER
0.30
78
OF COLUMN
0.35
VOLUMES
0.40
.6 1.7
VOLUME (ml.)
ELUTION OF Co(n),Zn(U),
AND Mn (II) WITH
0.001 M KF.
(TYPICAL CASES OF NON- ADSORBED IONS)
FIGURE 2
TABLE 7
COLUMN EXPERIMENTS WITH Co(II), Zn(II), AND Mn(II)
Experiment Eluent Column Peak I E = 1/{
Number Metal Ion M KF Volume Volume Peak Volume Elution
ml. ml. (Column Volumes) Constant
9 Co(II) 0.001 .90 1.67 0.34 2.9
12 Co(II) 0.01 ..88 1.75 0.36 2.8
15 Co(II) 0.1 4.87 1.75 0.36 2.8
18 Co(II) 1 4.78 1.99 0..2a 2.a
10 Zn(II) 0.001 4.90 1.66 0.34 2.9
13 Zn(II)- 0.01 4.88 1.74 0.36 2.8
16 Zn(II) 0.1 4.87 1.91 0.39 2.6
19,20 Zn(II) 1 4.78 "**"* "* ***"
11 Mn(II) 0.001 J.90 1.76 0.36 2.8
.1. Mn(II) 0.01 4.88 1.78 0.36 2.8
17 Mn(II) 0.1 4.87 1.81 0.37 2.7
aSllght precipitation is probably occurring.
80
Symmetrical elution curves were obtained in all
cases; however, at higher concentrations, a slight tailing
was noticed, together with a small displacement of the
peaks toward larger volumes. This effect is shown in
Table 7. Because of the smallness of the deviations, and
in view of the sensitivity of i to many variables, these
differences should not be regarded as significant, and all
these ions should be considered non-adsorbed. However, the
fact that tailing becomes more pronounced at higher fluoride
ion concentrations seems to indicate that some precipitation
is starting to occur. This was found to be the case with
zinc(II) in I M potassium fluoride. In column experiment
19, no definite peak was observed, but the activity in the
effluent was spread out at roughly a constant concentration,
slightly above background. This was thought to be caused by
precipitation, but could possibly have been caused by strong
adsorption, spread out so much that the activity per sample
was not high enough to show a well-defined peak. The elu-
tion was repeated again in column experiment 20 using ten
times the amount of tracer used previously, but the same
result was obtained.
If the effect had been due to true adsorption, it
should have been possible to elute the zinc as a symmetrical
band by using 0.1 M potassium fluoride, since this solution
had been shown to elute zinc in one interstitial volume.
81
However, in column experiment 21, five column volumes of
0.01 M potassium fluoride failed to elute the zinc in a sym-
metrical band, although it eluted most of the activity slow-
ly. Thus, the "holdup" of zinc(II) in 1 M potassium fluoride
was probably due to precipitation rather than actual adsorp-
tion.
According to equation (1) of Table lj
E = dA/V,
but in the cases considered,
d x A = C(one column volume),
and thus
E = 1/i,
i being the fraction of a column volume required for elution
of non-adsorbed ions. This same result can be obtained by
considering equation (3) of Table 1,
E = 1 ..
i + DV
and noting that as Dv approaches zero,a E approaches the
maximum value 1/i. In these studies, i was found to range
from 0.34-0.39.b Thus, on the basis of these experiments,
aActually DEv never equals zero, since some distribu-
tion between the two phases always takes place*
bSince some very slight adsorption takes place even
for completely "non-adsorbable" ions, such as those of the
alkali metals, the most exact determinations of i can be made
with an organic dye, determining the band maximum spectro-
photometrically. Because of the observed variations in I ,
this procedure hardly seems warranted.
82
values of E from 2.6-2.9 can be used as a definition for
"non-adsorbable" ions, and cobalt(II), zinc(II), and manga-
nese(II) in 0.001-0.1 ? KF solutions fall in this class. At
higher concentrations, precipitation begins to occur.
Absorbed Transition Metal Ions
Iron(III).--During the course of these radiochemical
studies, attention was shifted from an extensive study of
the first-row transition metal ions in general to an inten-
sive study of the most interesting of these, viz., iron(III).
Thus, except for the few excursions into "non-adsorbable"
transition metal ions just discussed, most of the radio-
chemical studies dealt with this metal.
Equilibration experiments with tracer concentrations
of iron(III) in potassium fluoride solutions ranging from
0.05 to 0.2 M did not give very reproducible results. How-
ever, the D values obtained were sufficient to indicate that
the iron(III) was being adsorbed as a fully formed triply
charged complex anion, which hence should have the composi-
tion FeF6. This conclusion was based on the fact that a
plot of the logarithm of D vs. the logarithm of the molarity
of fluoride ion was linear with a -3 slope, as Ehown in
Figure 3. The relationship between the charge on the ad-
sorbed ion and this slope has been derived in Chapter I, D.
Equilibration experiments with 0.1 M potassium fluo-
ride solutions yielded D values that were very high
0.01 0.1
MOLARITY
OF KF
ANION EXCHANGE ADSORPTION
OF Fe (m) AND Go (m)
FIGURE
1000
I100
z
IL
LJ
0
0
I0.00
0.001
84
(900-1000). By extrapolation of the line of slope -3 men-
tioned above, one would predict a D value of about one for
1 M potassium fluoride solutions, Such a low D value should
be readily obtainable by column experiments* However, when
such experiments were run with 1 M potassium fluoride solu-
tions, precipitation of iron(III) on the column was observed,
Once such precipitation has occurred, it cannot be reversed
by the use of lower concentrations of the same eluent, as
noted in the radiochemical studies of zinc(II). Therefore,
this precipitated iron(III) could not be eluted with potas-
sium fluoride solutions, but required 1 M hydrofluoric acid
for elution.
Thus, column experiments with 0.1 M potassium fluo-
ride solutions yielded very high but inconsistent D values.
With I M potassium fluoride solutions, in which D values
should be low enough to be determined by column experiments,
which have been shown to be reproducible, precipitation oc-
curred. Therefore, it was decided to investigate potassium
fluoride solutions of intermediate concentrations, with the
distribution coefficients lowered to a range convenient for
determination by column experiments. Hydrofluoric acid was
added for this purpose, on the basis of the following ob-
servations.
It was found, surprisingly enough, that while
iron(III) was strongly adsorbed from potassium fluoride
85
solutions, no adsorption at all was found in column experi-
ments with 1 M hydrofluoric acid solutions. This ability
of hydrofluoric acid to lower distribution coefficient
values for iron(III) in potassium fluoride solutions was the
primary reason for its utilization in the following prelimi-
nary experiments. Although iron(III) was not adsorbed from
1 M hydrofluoric acid solutions, tests of the effluent with
Hydrion paper showed that the hydrofluoric acid itself was
adsorbed, presumably because of the high concentration of
acid fluoride ion, HFg".
Preliminary column experiments with mixed potassium
fluoride hydrofluoric acid solutions in the intermediate
concentration range, i.e., between 0.1 and 1 M potassium
fluoride, indicated that the wide difference observed between
the adsorption of iron(III) in potassium fluoride solutions
and that observed in hydrofluoric acid solutions was not so
much a function of fluoride ion concentration as of hydro-
nium ion concentration. This difference could be ascribed
to weak acid formation, i.e., combination of the FeF 6 ion
with one or more protons to form less highly charged species
or even a neutral molecule according to the possible reac-
tions,
FeF6 + H30+* =: HFeF6P + H20,
HFeF6= + H30 ~ H2FeF" + H20,
and
H2FeF6" + H30+ d H3FeF6 + H20.
86
Thus, the difference in adsorption could be assumed to be
due to the variation in the ratios of these species present
in solutions of different pH's. Undissociated H3FeF6 should
not be adsorbed, while the uptake of the highly charged ion
FePF6 should be considerable.
If the fluoride ion concentration of a solution is
held constant, and the pH is varied, the distribution coef-
ficients of iron(III) in such a solution should vary in a
regular manner if weakly acidic species are being formed,
iLe., if PeF6 is combining with protons. Since the fluo-
ride ion concentration of such a solution is constant, the
concentration of fluoride complex should also be constant,
were it not for the fact that the complex anion is combining
with protons, leaving less of it to be adsorbed by the resin,
and resulting in a decrease in distribution coefficient val-
ues as the pH is lowered. Column experiments were carried
out with three series of such solutions, and from the data
obtained, tentative values of the constant K for an acid-
base reaction of the type
HnFeF6"3+n + nH 20 nH 30 + FeF6
were calculated.
The three series of potassium fluoride solutions
(0.500, 0.350, and 0.250 M KF) had pH values ranging from
about two to eight. Standard hydrofluoric acid or potassium
hydroxide solutions were used to produce the different
87
hydronium ion concentrations, and all solutions vere quanti-
tatively prepared using polyethylene graduates in the case
of acid solutions and glass volumetric equipment in the case
of neutral or basic solutions. The hydrofluoric acid solu-
tions were standardlee.;. by titration with standard sodium
hydroxide solutions, and the pot9rn- ium hydroxide solutions
were then standardized by titration with these hydrofluoric
acid solutions. In all cases, the experimentally measured
concentrations checked with the nominal concentrations to
within 1).
The equilibration method was not used in these ex-
periments. Although this method is ideal for the determina-
tion of a lIrge number of high distribution coefficients, it
did not yield conEistent results even with mixed potassium
fluoride hydrofluoric acid solutions. Moreover, because
of the adsorption of hydrofluoric acid on the resin, the
hydroniur. ion concentrations of the solutions after equill-
bration would be less than before, and this effect would
have to be calculated from distribution coefficient dcta for
hydrofluoric acid in the various solutions. The other al-
ternative of equilibrating the resin with the particular po-
tassium fluoride hydrofluoric acid solution before adding
the tracer is quite inefficient. Thus, all distribution co-
efficient values for these solutions had to be determined by
column experiments, even in the case of high values, for
88
which this method would not normally have been used. The
column techniques were essentially the same as those de-
scribed previously.
Before placing the tracer on the columns, they were
equilibrated with the eluent without the tracer. Since
hydrofluoric acid is adsorbed on anion exchange columns,
these equilibrations were checked experimentally by titra-
tion of equal volumes of influent and effluent in order to
show that within experimental error the amount of standard
base required for neutralization (phenolphthalein endpoint)
was the same. These titrations were performed only with
solutions of acid concentrations above 0.005 M. Below this
concentration, such titrations were not feasible, and col-
umns were routinely washed with about 100 column volumes of
eluent and assumed to be equilibrated.
These column experiments yielded data of the general
type shown for cobalt(II) in Table 6, from which symmetrical
elution peaks of the type shown in Figure 2 were obtained.
Of course, the greater the adsorption, the broader the elu-
tion bands, Eventually, with solutions of high pH, these
bands became so diffuse that accurate assignment of peak
positions could not be made. Moreover, in this same pH
range, precipitation was beginning to occur. Dv values
which were uncertain for either of these reasons are indicated
by question marks in Table 8.
TABLE 8
COLUMN EXPERIMENTS WITH Fe(III) IN FLUORIDE SOLUTIONS
C P P P *
Column Peak Dv =
Concentration of Volume Volume Peak Volume Volume Distribution
Solution Additional Reagent ml. ml. (Column Volumes) Coefficient
(1) (2) (3) (4)
0.500 M KF Series
1 5.00 M HF 0.89 0.60 0.67 0.31
2 2.50 M HF 0.89 0.65 0.73 0.37
3 1.25 M HP 0.89 0.73 0.82 0.46
4 0.875 .HF 0.89 0.73 0.82 0.46
5 0.500 M HP 0.89 0.93 1.04 0.68
6 .0.222 1 HF 0.89 1.55 1.74 1.38
7 0.111 4 Hp 0.95 2.40 2.53 2.17
8 5.56 x 10o2 M HP 0.98 3.65 3.73 3.37
9 2.78 x 10-2 _M HF 1.00 4.80 4.80 4.74
10 1.39 x 10-2 M HP 1.00 6.30 6.30 5.94
N
'0
TABLE 8--Continued
C P P P
Column Peak v -
Concentration of Volume Volume Peak Volume Volume Distribution
Solution Additional Reagent ml mi. (Column Volumes) Coefficient
(1) (2) (3) (4)
11 5.56 x 10-2M HPF 100 7.50 7.50 7.14
12 1.39 x 10o3 M BF 1.00 8.55 8.55 8.19 ?
13 3.47 x 10"4 M HP 1.03 9.25 8.98 8.62
14 .**.* 1.01 9460 9.50 9.14
15 1.00 x 10"6 M KOH 1.00 9.65 9.65 9.29
16 5.00 x o0"4 M KOH 1.02 10.25 10.05 9.69 ?
0,350 M KF Series
1 0.500 M HP 1.13 2.125 1.88 1.52
2 0.250 M HF 1.16 3.325 2.87 2.51
3 0.100 M-iF? 1.18 7.05 5.98 5.62
4 5.00 x 10-2 M HP 1.22 10.28 .8.402 8.06
TABLE 8--Continued
C P P iP
Column Peak "- Dv = - "
Concentration of Volume Volume Peak Volume Volume Distribution
Solution Additional Reagent ml. ml. (Column Volumes) Coefficient
(1) (2) (3) (4)
S5 2.50 x 102 M HF 1.23 16.75 13.62 13.26
6 1.00'x 10"2 M HP 1.16 18.10 15.60 15.24
7 5.00 x 10"3 M Hp 1.16 23.20 20.00 19.64
8 2.50 x 10"3 M HP 1.26 28.65 22.74 22.38
9 1.00 x 10"3 M BF 1.18 27.00 22.89 22.53 ?
10 5.00 x 10"4 M HF 1.25 31.75 25.40 25.04
11 2.50 x 10-4 M HF 1.16 31.15 26.82 26.46 ?
12 1.00 x 10"4 M HF 1.28 33.60 26.22 25.86 ?
13 ____ *"1.16 27.80 23.98 23.62
14 1.00 x 10-6 M KOH 1.07 30.50 28.50 28.14
TABLE 8--Continued
C P P P
Column Peak Dv = C
Concentration of Volume Volume Peak Volume Volume Distribution
Solution Additional Reagent ml. ml. (Column Volumes) Coefficient
(1) (2) (3) (4)
15 1.00 x 10"5 M KOH 1.26 40.25 31.94 31-94
16 .1.00 x 10"4 M KOH 1.16 29.75 25.65 25.29
0.250 M KF Series
1 1.00 x 10-1 M HP 1.oo 9.86 9.86 9.50
2 1.00 x 10-3 M HF 1.17 58.60 50.1o 49.7
3 *****1.23 103.50 84.20 83.8 ?
4 1.00 x 104 M KOH 1.26 78.80 62.60 66.2
93
Solving equation (3) of Table 1 for Dv yields
1 .
Dv -N- I
From equation (1) of the same table,
E = dA
V '
and from a knowledge that
Column volume, C = d A,
it follows that
p
where P/C is the peak volume in terms of column volumes,
i.e., the number of column volumes required for elution.
Thus, from the positions of elution peaks, volume distribu-
tion coefficients were calculated. A value for I of 0.36
was used, since this was the average value obtained in radio-
chemical column studies on non-adsorbed ions. A summary of
the results for the column experiments is given in Table 8.
Calculation of the concentrations of the species
present in potassium fluoride hydrofluoric acid solutions
involves a knowledge of the values of two equilibrium con-
stants,
[HF]
for the reaction
HF + H20 0 H3 0 + F"
and
94
k -
for the reaction
HF + F" =HPF2"
Broene and de Vries5 give the values 6.71 x 10"4 and
3.96 for K and k, respectively, at 2500C. for infinite dilu-
tion. From these data, the values of these constants at the
particular ionic strengths used (0.500, 0.350, and 0.250)
were calculated, using the Debye-Huckel expression
log Y,= -0.509 z2 1
where Y. is the activity coefficient of ion j of charge z,
and 1A is the ionic strength. These values of K are shown
in Table 9; the value of k should be independent of ionic
strength at low concentrations.
From these constants and the expressions
HF + E20 1H304' + F"
ml-q-r q m2+q-r
aThis equation is actually
log Yj= -0.509 z2 1
J [1 + aBs /
where a is the mean distance of approach of other ions
("average effective diameter") and B is a constant depend-
ing on the nature of the solvent and the temperature. At
250C., the value of B for aqueous solution is approximately
0.33 x 108; for most electrolytes, a is about 3 to 4 x 10*0
cm. Hence, aB does not differ greatly from unity.22
95
TABLE 9
K VALUES AT DIFFERENT IONIC STRENGTHS
K
0.50 0.707 1.707 0.414 1.76 x 10-3
0.35 0.592 1.592 0.372 1.60 x 10-3
0.25 0.500 1.500 0.333 1.46 x 10-3
and
FH + F HF2"
ml-q-r m2+q-r r
where
mI = nominal concentration of HP ,
m2 = nominal concentration of KF ,
q = actual [H30+] ,
and
r = actual (HF2" ,
and making the usual simplifying assumptions, the concentra-
tions of the species present in the different solutions were
calculated and are shown in Table 10. Also shown here are
the hydronium ion concentrations for solutions of pH above
seven; such pH measurements were made with a battery-operated,
potentiometric Beckman Model G pH meter, employing a calomel
reference electrode and an unshielded glass electrode used
in a shielded electrode compartment. This apparatus was
TABLE 10
CONCENTRATIONS OF SPECIES PRESENT IN FLUORIDE SOLUTIONS
0.500 M KF Series
1 5.00 M HP 4.45 o0.474 0.103 0.217 7,62 x 10o2
2 2.50 M HF 2.02 0.446 0.093 0.207 3.83 x 10-2
1.25 M HP
0.852
0.386
0.126
0.326
1.20 x 10-2 u-
4 0.875 M HF 0.529 0.340 o.166 0.488 5.50 x 0-"3
5 0.500 M HP 0.250 0.250 0.250 1.o00 176 x 1l"3
6 0.222 M HF 9.00 x 10-2 0.133 0.368 2.77 4.29 x 10"4
7 0.111 M HP 4.06 x 10-2 7.03 x 10-2 0.430 6.12 1.67 x 10'-
8 5.56 x 10o'2 m HF 1.94 x 10"2 3.61 x 10-2 0.464 12.9 7.38 x 10"5
9 2.78 x 10"2 m HP 9.55 x 10-3 1.82 x lo*2 0.482 26.5 3.50 x 105
10 1.39 x 10"2 M HF 4.67 x 10-"3 9.22 x 10-3 0.491 53.3 1.68 x 10-5
0
|
