THEORY OF COPRECI PI TATI ON.

THE FORMATI ON AND

PROPERTI ES OF CRYSTALLI NE PRECI PI TATES
-__
BY I. M. KOLTHOFF
1. Introduction
It has been known for a long time that crystalline precipitates separating
from a solution are not, as a rule, quite pure but usually contain imperfections
of mother liquor with foreign constituents. A vast number of empirical facts
pertaining to the presence of various impurities in crystalline precipitates may
be found in the analytical literature. Unfortunately, these data are of not
much use in the interpretation of the so-called coprecipitation or carrying
down, mainly for the reason of poor description of experimental conditions.
As will be shown later in this paper, the latter are of primary influence as
regards the kind and amount of coprecipitation. bloreover, it has been a
general custom in analytical chemistry to use the words, coprecipitation,
carrying down, occlusion, inclusion and adsorption as collective names,
meaning nothing else but the establishment of the fact that impurities are
carrisd down with or in a precipitate. I n a systematic treatment of the
problem, it should be emphasized, however, that three different phenomena
mai nl y account for the presence of i mpuri t i es i n the precipitate and in a study
of the problem, it is necessary first to find out what kind of coprecipitation
we are dealing with. I n this paper, three cases are distinguished:
I n this case the impurities are in-
corporated in the crystal lattice and they do not change tile regular structure
of the latter. The amount of mixed crystal formation depends as in case C
upon adsorption phenomena during the growth of the precipitate.
I n this case the impurities are not incorporated in the
crystal lattice, but t hey are atiao,rDed duri ng the grozt h of the crystals and give
rise to the formation of imperjcctlon.? in the crystal. (Hohlraume or Locker-
stellen (Smekal) ; or centra of activity-(H. S. Taylor).) Here adsorption
phenomena during the growth of the crystals are mainly responsible for the
amount of occlusion. I t is especially with this kind of coprecipitation that
we are concerned in this paper.
e. Surf ace udsorpt i on by the precipitnte after i t has been formed or separated.
This kind of coprecipitation is only of practical importance when the precipitate
has a large surface,i.e. when it behaves like a flocculated colloid. If the precipi-
tate has a definitely micro-crystalline character (as observed under the micro-
scope) the amount of coprecipitation caused by surface adsorption is, as a rule,
of no practica1 significance. Confusion is caused in the anaIytica1 literature
by the fact that certain phenomena have been attributed to coprecipitation (an
expression often used to indicate that a precipitate is not quite pure), but
which have nothing to do with it. This may be illustrated by two examples.
a. T h e f ormot i on of mi z ed crystals.
b. Occlusion.
THEORY OF COPRECI PI TATI OS 86 I
I n most text books of analytical chemistry, it is mentioned that magnesium is
coprecipitated with calcium oxalate. Although there is actually a slight co-
precipitation of this element, the presence of magnesium in calcium oxalate,
if the latter is precipitated from solutions containing much magnesium salt,
is mainly due to the slight solubility of magnesium oxalate. If relatively
much magnesium is present and an excess of oxalate is added, the solution
becomes supersaturated with respect to magnesium oxalate. First of all,
calcium oxalate precipitates and then on standing magnesium oxalate crys-
tallizes out slowly. Therefore, we are not dealing here with a case of co-
precipitation, but of l~ost-precipitation, the crystals of calcium oxalate being
not at all or only slightly contaminated by magnesium. The magnesium
oxalate crystallizes out as a separate phase. A detailed investigation of co-
precipitation with calcium oxalate is being made in this laboratory by E. B.
Sandell ; the results will be described in his doctors thesis and communicated
later. Another example of post-precipitation is the so-called coprecipitation
of zinc with copper sulfide. I n an extensive investigation it, has been shown
that here again we are dealing with a case of post-precipitation. I n relatively
weakly acid medium a solution saturated with hydrogen sulfide is super-
saturated with respect to zinc sulfide. First of all, the copper sulfide precipi-
tates according to the laws of fractional precipitation; on standing zinc
sulfide separates out slowly. Even in the more recent literature, this post-
precipitation of zinc sulfide is described as a coprecipitation although it has
nothing in common with it. Of course, it is quite possible that the primary
precipitate has a promoting effect upon the separation of the secondary
precipitate as in the copper sulfide-zinc sulfide case, the precipitation of zinc
sulfide is enhanced at the surface of copper sulfide. It should be clearly
understood however, that the secondary precipitate is not carried down by
the primary precipitate and that we are not dealing here with a real case of
coprecipitation.
d. A case which rarely occurs is that coprecipitation actually has to be at-
tributed to the formation of a definite chemical compound. The so-called
coprecipitation of alkali oxalate with lanthanum oxalate is caused by the
formation of a double oxalate as has been shown by I. 11. Kolthoff and R.
Elmquist.* From experiments of Z. Karaoglanov and B. Sargotschev,3 it
appears that coprecipitation of lead bromide and lead chloride with lead
oxalate is due to the formation of a double salt (PbX)*Ox. However, the
formation of such definite chemical compounds is very seldom encountered in
analytical work, although it should not be overlooked as a possible interpre-
tation of the presence of much impurity in the precipitate.
I n this paper an attempt is made to formulate a general theory of co-
precipitation, especially of the kind specified in Sub. b. The hope is expressed
that such a theory will not only contribute to the understanding of the forma-
tion of impurities in precipitates formed under analytical conditions and
I . M. Kolthoff and E. Penrson: J . Phys. Chern., 36, j49 (1932).
Kolthoff and Elmquist: J . Chem. Soc., 53, 1232 (1931).
Karaoglanov and Sargotschev: Z. anorg. allgem. Chem., 199, 7 (1931).
862 I. M. KOLTHOFF
thereby the improvement of the exact precipitation procedures, but will also
have a wider bearing on the problems of mineralogical formations and the
properties of slightly soluble micro-crystalline precipitates.
It should already be mentioned here that in recent years various authors
have tried to treat the problem of coprecipitation from a more general view
point. However, no sharp distinction has been made between the various
kinds of coprecipitation as is done in this paper.
I n the following a short discussion will be given of the formation and
properties of precipitates which is of importance with regard to the general
problem. More detailed information will be given in consecutive papers
which will contain the results of experimental studies being made at the
present time in our laboratory.
2. The Thermodynamic Potential of Crystals
An ionic lattice, according to the precipitation rule of Paneth-Fajans*
has a strong adsorbing action for ions which form a slightly soluble or slightly
dissociated compound with ions of the lattice of opposite charge.
It seems advisable to distinguish between primary and secondary adsorp-
tion of ions by an ionic lattice. If silver chloride, for example, is shaken with
an alkali chloride solution, there is a primary adsorption of chloride ions.
Owing to the fact that the system must be electrically neutral an equivalent
amount of any kind of foreign cations has to be adso-hd as well (secondary
adsorption). A preferential adsorption of those cations will take place, whose
compounds with the primary adsorbed ions are slightly soluble or slightly
dissociated. This rule should be correlated with the following: The higher
the valence of the ions with a charge opposite to that of the primary adsorbed
ions, the stronger they will be adsorbed. This may be inferred from measure-
ments made by L. Imre4a on the adsorption of actinium, thorium B and
radium by negatively charged silver iodide and could be expected already
from the similarity between flocculation of colloids and adsorption of ions by
a particle. If the lattice surface does not contain an excess of its own ions
there may be a primary adsorption of foreign ions. These will be held by
much less stronger forces than the lattice ions and as a rule will easily be
replaced by the latter.
K. Fajans and W. Frankenburge? give a very clear picture of what occurs
when a crystal lattice is in contact with a solution containing an excess of
one of its own ions.6 The crystal lattice adsorbs one of the kind of ions
constituting it. The adsorption forces are here identical in nature with those
holding the adsorbent together . . ., If we assume, for simplicity, a
difficultly soluble salt, both of whose ions are equally strongly adsorbed on the
F. Paneth: Physik. Z., 15, 924 (1914); K. Horovits and F. Paneth: Z. physik. Chem.,
89, 513 (1915); Wien. Ber., 123, I P, 1819 (1914); K. Fajans and P. Beer: Ber., 46,. 3486
(1913); Fajans and K. Richter: 48, 700 (1915); 0. Hahn, 0. Ersbacher and N. Feichtinger:
Ber., 59, 2014 (1926); 0. Hahn: Naturwissenschaften, 14, 1196 (1926).)
(a L. I mre: Z. physik. Chem., 153 A, 127 (1931.))
SK. Fajans and W. Frankenburger: Z. physik. Chem., 105, 255 (1923).
6 Ref. 5, page 270.
THEORY OF COPRECI PI TATI ON 863
same crystal lattice, i.e. in contact with the saturated solution containing
both ions in equal concentration, there arises no potential difference between
salt and solution. If we add an excess of one of the ions, say the cations, the
adsorption equilibrium of this ion on the anions of the lattice will be disturbed
and consequently the cations will be deposited on the anions of the lattice
covering them in part and giving the lattice a positive charge. The equilib-
rium with the anions in the solution is thereby also disturbed as can easily be
seen from the fact that the anions of the lattice which are covered by positive
ions are removed from kinetic contact with the solution and that the surface
now carrying an excess of positive charge exerts an increased attraction on
anions of the solution. Fresh anions must, therefore, be deposited on the
crystal lattice from the solution until a new state of equilibrium is set up with
a smaller concentration of anions. This means that the solubility of the salt
is lowered by an excess of cations. I t is clear that a closer investigation of
the whole adsorption isotherm of both ions must give a quantitative connec-
tion with the law of solubility product. Two points can clearly be seen: Since
the adsorption of cations in the case considered is greater, the greater the
excess added to the solution so must the amount of anion precipitated in-
crease likewise and the solubility decrease corresponding entirely quantita-
tively with the law of mass action. I n excellent papers J . A. V. Butler has
already shown that there is no reason to expect that a crystal lattice will
adsorb cations and anions equally well. I t may obviously happen that the
tendency of one of the ions to go in solution may be greater than that of the
other ion owing either to a smaller attraction by the lattice or a greater
attraction of the solvent for this ion. To this statement may be added the
fact that the conditions under which the thermodynamic potential will be
equal to zero will be different in different solvents, since the activities of
various ions change in a different way in going from one solvent to another.
H. R. Kruyt and P. C. van der Willigens in harmony with Butlers state-
ment found that crystals of silver iodide in contact with the saturated solu-
tion in water assume a negative charge, thus showing a preferential adsorp-
tion of the iodide ions. Butler succeeded in showing that the expression for
the solubility product holds only by considering the kinetic equilibrium be-
tween solution and solid,
From the above it is evident that any crystal in equilibrhm with its solu-
tion always will adsorb ions of its own kind, if they are present in excess in
the solution; only under one condition, either in a saturated solution in water
(if the adsorption potential of the anions equals that of the cations) or in
presence of a slight excess of anions (adsorption potential of anions smaller
than that of cations) or in presence of a slight excess of cations (adsorption
potential of anions larger than that of cations) will the thermodynamic
potential be equal to zero.
J . A. V. Butler: J . Phys. Chern., 28, 438 (1924); Trans. Faraday SOC., 19, 659, 723,
734 (1924).
8 H. R. Kruyt and P. C. van der Uilligen: Z. physik. Chem., 139, 53 (1928).
864 1. M. KOLTHOFF
I n considering the adsorption by a crystalline precipitate during its growth
it will be seen that this thermodynamic potential primarily determines the
amount of coprecipitation. Quantitatively there is a simple relation between
the thermodynamic potential and the corresponding ion concentration in the
solution as has been shown by F. Haber.g The same expression is found as the
one that holds for the potential difference between an electrode and a solu-
tion containing ions which the electrode can send into solution (Kernsts
equation) ;
E = - -1n alon + Const.
RT
nF
alon is the activity of the ion.
I t is singular that no analytical application has been made of this relation
between thermodynamic potential of a crystal lattice and a solution contain-
ing ions which are the building stones of the lattice. It may be expected, for
example, that a barium sulfate electrode will behave as a specific electrode
for barium-and sulfate ions. With the present development of the technique
of measuring the electromotive force of cells with extremely high resistances,
a wide field of potentiometric measurements may be anticipated.
The relation between thermodynamic potential and the amount of ions
adsorbed has been determined in the case of silver iodide by E. Lange and
R. BergerO who found that the change of the thermodynamic potential is
proportional to In C,,, in the solution and up to a certain limit to the con-
centration of adsorbed ions on the surface.
3. Solubility and Particle Size
For many years it has been known that the solubility of crystals of very
small size is larger than that of crystals of large dimensions owing to the
greater surface energy of the former. If the crystal size becomes smaller
than I to z p the solubility quite generally increases with increasing surface
development. The most extensive and interesting experiments in this field
have been carried out by 11. L. Dundon and E. hIack,l* who also have given
corrected values for the classical data of G. A. Hul etP on the solubility and
crystal size of gypsum. Although the order of magnitude of the surface
tension of various crystals as calculated by Dundon is probably correct no
claim is made for great exactness of the figures. The particle size was approxi-
mated by microscopic measurements, involving a relatively large uncertainty
at small dimensions, and it was assumed that the crystals were quite com-
pact, i.e. the presence of capillaries (inner surface) was not considered. More-
@ F. Haber: Ann. Physik, 47, 26,947 (1908); F. Haher and Z. Klemensiewicz: 2. physik.
Chem., 67, 385 (1909).
E. Lange and R. Berger: 2. Elektrochemie, 36, 1 7 1 (1930).
For literature review compare H. Freundlich: Kapillarchemie, 2nd Ed., 207-21 I
(1922); R. A. Gortner: Outlines of Biochemistry, I j o (1929); T. B. Smith: Analytical
Processes, 239-278 (1929).
M. L. Dundon and E. Mack: J . Am. Chem. Soc., 45, 2479 (1923); hl. L. Dundon:
45, 2658 (1923).
I 3G. A. Hulett: Z. physik. Chem., 37, 385 (1901); 47, 357 (1909).
THEORT OF COPRECI PI TATI OS 865
over the following paradoxical phenomenon has not been adequately ex-
plained: If coarse crystals of b:iriuni sulfate wr e added to a saturated solu-
tion in equilibrium with crystals of very small size the solubility decreased
gr:idually anti finally became equal to the inacroscopic solubility (normal
solubility), although by microscopic examination a great number of particles
of original small size appe:ired to he present. I n order to explain this peculiar
f:ict Ilundon considers one of the two altrrnntives; either that originally
rii:iny particles had been present which were too small for microscopic mea-
sureinent (which mould mean th:tt none of the data in the following table are
correct) or that soon after a sxturated solution in equilibrium with the small
p:irticlcs has been formed the solubility of the smaller particles is decreased
by some adsorption effect such as the acquisition of il charge. According to
Ilundon, the latter assuniption seems to be in harmony with the facts. How-
ever it cannot be denied that this esp1nn:ition is far from being satisfactory.
The adsorption process 1:rkes place rsther quickly and it is hard to accept
the view that the particlcs do not acquire a charge until a saturated solution
has been formed. J I oreover experimentnlly the influence of the charge of a
particle on its solubility has never been shown although IV. C. AIcC. Lewist4
and L. F. Knapplj in more or less mathenintical papers show that the charge
opposes the surface tension and hence tends to decrease the solubility. The
charge increases with decreasing size and the solubility reaches a maximum at
certain dimensions, and decreases with increasing surface development.
Without entering into a detailed discussion it should be realized that the
solubility decreases with increasing thermodynamic potential of the particles,
as has been mentioned in the preceding paragraph; experimentally however it
has never been shown that the electrokinetic potential exerts a simi1:tr in-
fluence; and on thermodynamic grounds this is not to be expected.
There is much more reason to assume that on account of the charge or
the presence of an ion atmosphere around the particle the speed of exchange
of ions between surface and solution (the kinetic equilibrium between solution
and deposition velocity) is materially inhibited. This may give an explana-
tion of the peculiar fact that the crystals of small dimensions disappear so
slowly if macroscopic crystals are added to their saturated solution. At the
surface of the large crystals, the solution and deposition equilibrium is estab-
lished much more quickly than at the surface of the charged particles of small
size. If the large and small crystals are both present as solid bodies the state
of equilibrium between the former and the solution is readily established;
the small crystals very slowly send more ions into the solution, which then
are deposited on the surface of the large crystals; in other words the latter
grow Since the solution
rate of the latter is so sniall it will require a long time before all small particles
have disappeared. This explanation based on the difference between solution
and deposition rate at the surface of small and large crystals readily accounts
slozclu at the cost of the small sized particles.
I4 JV. C. McC. Lewis: Kolloid-Z., 25, 91 (1909).
L. F. Knapp: Trans. Faraday SOC., 14, 457 (1921/ 22).
866 I. M. KOLTHOFF
for the fact that a normal solubility is found if the small particles are present
with the larger size crystals.
This interpretation is in harmony with recent experiments of P. S. RollerG
on the relative rate of solution of gypsum. Down to a diameter of z 5p he
found the rate of solution to be proportional to the specific surface, and the
dissolution factor was a constant. Below z5p the rate of solution increased
more rapidly than the surface exposed until at a size of 2 . 8 ~ the dissolution
factor reached a maximum. At smaller dimensions it decreased again,
probably on account of acquisition of a charge.
Assuming, then, that Dundons data at least give the order of surface
tension of various crystals it is possible to explain with their aid some facts
which so far have been more or less obscure in analytical chemistry. I n the
following table a summary of Dundons results is given.
TABLE I
Surface Tension of Some Substances according to M. L. Dundon
Substance M Mol. Vol. r (p) % Increase T,emp. u Hardness
PbI2 461 6.16 74.8 0.4 2 30 130 Very Soft
CaS04.aH20 172 2.32 74.2 0.2-0.5 4.4-12 30 370 1.6-2
Ag2CrO4 332
5 . 5 2 60.1 0.3 IO 26 575 Appr. 2
PbF2 245 8.24 29.7 0.3 9
25 900 2
SrS04 184 3.96
46.4 0.25 26 30 1400 3.0-3.5
BaS04 (Hulett) 233 4.5
5 2 0.1 80 25 1250 2. 5- 3. 5
CaF2 78 3.18 24.6 0.3 18 30 2500 4
M denotes molecular, weight, E density of crystal; Mol. Vol. molecular volume:
r is radius of particle expressed in microns (microscopically measured), u
is surface tension of particle, Yc Increase Sol. is per centum increase of solu-
bility with regard to massive crystals.
If we consider slightly soluble salts the saturated solutions of which are
completely dissociated into the ions and whose activity coefficients can be
put equal to I the relation between increase of solubility on the one hand
and the size of crystals and their surface tension on the other, can be repre-
sented bv the equation
Solubility C
(Dundon) 0.2 90 30 3000
R is the gas constant, T the absolute temperature, M the molecular weight,
Sr the solubility of particles with a radius r, S the same of normal crystals,
u the surface tension and 5 the density.
By means of this equation and Dundons data the ratio Sr/S for barium-
sulfate, silver chromate and lead iodide at a size of 0.04~ (r= 0 . 0 2 ~ ) was
calculated.
6P. S. Roller: J. Phys. Chem., 35, 1133 (1931).
THEORY OF COPRECIPITATION
so. 02
BaS04- = 930
S
Ag,CrOl = 4.0
PbI2 = 1.38
Whereas at this small size the solubility of barium sulfate is about 1000
times larger than that of the large crystals, the solubility of silver chromate
under the same condition has increased only 4 times and that of lead iodide
only 1.4 times.
These differences explain the fact that substances of about the same
solubility and precipitated under analogous conditions behave in an entirely
different way. Silver chloride and barium sulfate have a solubility product
of the same order of magnitude; in spite of this, silver chloride is always
precipitated as a flocculated colloid, barium sulfate as microcrystals under
analytical conditions. It is easy now to account for this difference. Suppose
a barium solution is added to a sulfate solution and silver ions are added to
a chloride solution under such conditions that the macroscopic supersatu-
ration of barium sulfate and silver chloride are the same. The silver halides
form soft crystals and their solubility is more or less independent of the
crystal size: in other words, the solubility of the particles first formed in the
solution (nuclei) is about the same as that of large size crystals. For barium
sulfate (strontium sulfate, lead sulfate, etc.) the case is quite different. The
solubility of the primary particles is much larger than that of the large
crystals: therefore the solution is much less supersaturated with respect to
the small particles of barium sulfate than of silver chloride. The velocity
of formations of nuclei and growth of the latter to large crystals increases
with increasing supersaturation. Therefore the formation of nuclei is much
more spontaneous in the case of silver chloride than of barium sulfate; by
the rapid formation of so many nuclei the solution is soon exhausted and
no ions are left in solution to contribute to a growth of the small particles;
the silver chloride precipitates as a flocculated colloid. I n the case of barium
sulfate much less nuclei are formed, which grow at the cost of the ions left
in the solution; the slightly soluble substance finally settles as a micro-
crystalline precipitate. From the above it is evident that Bottgersl* relation
between sensitivity of a precipitation reaction and solubility:
E =L +S
in which E represents the sensitivity, L, the solubility and S the visibility
of the particles has no general validity. I t will only holdig if the micro
and macro solubilities of the substance are approximately the same as
in the case of silver halides. But if this condition is not fulfilled, which is
very often the case (barium sulfate, strontium sulfate, calcium oxalate)
there is no simple relation between solubility and sensitivity of the reaction.
I Comp. A. Rei8 and L. Zirnrnerrnann: 2. physik. Chern., 102, 299 (1902).
18Comp. W. Bottger: Chern. Ztg., 33, 1003 (1909); M. Gorski: 2. anorg. allgem. Chem.,
81, 315 (1913); W. Bottger: Chem. Ztg., 36, 1097 (1912); 2. angew. Chern., 25, 1992 (1912).
IQ I. M. Kolthoff: Bottger Festschrift, 2. anal. Chem., 86, 34 (1931).
868 I . ni. KOLTHOFF
The velocity of recrystallization of finely divided precipitates is also
determined by the relation between solubility and surface tension. I t is
well known that finely divided precipitates of barium sulfate, calcium oxalate,
etc., which are formed in not too dilute solution on standing crystallize to
large crystals; with silver chloride, such a recrystallization does not take
place, because the difference in solubility between the small and large size
crystal is too small. Only if the supersaturation is made very slight is i t
possible to obtain large crystals of silver chloride (recrystallization from
ammonia).
I t is not very easy to find suitable examples for the demonstration of
validity of the mass law expression. A saturated solution of silver chloride
or lead iodide gives almost instantaneously a precipitate with excess of one
of the two ions; the experiment does not succeed with lead sulfate, strontium
sulfate, calcium oxalate, etc. V. Chlopinzo reports that he was able to prepare
stable lead sulfate solutions which were more than 1000 times supersaturated.
The explanation is simple again, the solution is supersaturated only if inocu-
lated with large crystals, but it is not necessarily supersaturated with respect
to the primary particles formed at the beginning of crystallization.
Considering then crystallization from a solution it may quite generally
be said that supeysaturation is a relatize conception: the supersaturation can-
not be expressed in an absolute figure unless crystals of the normal solubility
are present as solid body.
4. Filterability of Precipitates
I t is the usual practice in analytical chemistry to heat precipitates and
mother liquor after precipitation for some time in order to obtain a more
readily filterable precipitate by recryst,allization.21
I n the preceding chapter however we have seen that the process of recrys-
tallization of very small particles is extremely slow. DundonlZ for example
boiled a suspension of barium sulfate (size 0.2 to 0 . 3 ~ ) under a reflux con-
densor for a week with no visible change of particle size even when seeded
with large crystals. From experiments made by Professor Bigelow, H. M.
Trimblezz expected that the rate at which larger crystals of a nearly insoluble
substance grow at the expense of smaller ones in contact with the saturated
solution must be very low, probably too low to account for coalescence of
the precipitate as it occurs in analytical chemistry. Experimentally this
was shown to be true. Finely divided barium sulfate digested for a few hours
in the presence of some hydrochloric acid at 100 was readily filterable if
not stirred during digestion, Microscopic examination showed that the
average size of the small particles (between I and 4p) remained unchanged
during digestion. If the mixture was stirred during the digestion, no clear
filtrate was obtained even after three to four days.
2o V. Chlopin: Naturwissenschaften, 17, 959 (1929).
21 Cornp. Wi. Ostwald: Grundlagen der analytischen Chemie, 14, 22 (1894).
rzH. M. Trirnble: J . Phys. Chem., 31, 601 (192;).
THEORY OF COPRECI PI TATI OS 869
From these and analogous experiments Trimble inferred that the co-
alescence of an unfilterable precipitate to give a filterable one, cannot be
explained in terms of growth of the larger particles at the expense of the
smaller ones. During the digestion without stirring the particles come into
close contact with one another and form aggregates which adhere firmly.
As digestion proceeds, barium sulfate either from supersaturated solutions
or from dissolution of very small particles, deposits upon these aggregates
and cements them together. This effect is enhanced when the solution is
allowed to cool. On the other hand, when digestion is accompanied by
stirring, all crystals grow alike at the expense of any material which may
be separating from solution; there is no opportunity for cemented aggregates
to form. With E. B. Sandell, the writer has made various experiments with
B C.
A .
f l ovab/ a Do u b i a L a y e r Fi oc L u J a t c d F l a c c o l a t a d
( c o l l o i d ) ( X / n I A t t l C . ) ( Pr o b a b f . c a s e )
FIG. I
suspensions of barium sulfate, calcidm oxalate monohydrate, silver chloride,
etc. I t was observed that after digestion with or without stirring for one to
two days, a distinct growth of crystals took place. Even in a silver-chloride
suspension it was possible to observe small crystals under the microscope
after heating for a day. Still it is true that in any case, after prolonged
heating, a vast amount of the small particles is present with the original
size. I n the preceding chapter it has been shown that recrystallization of
charged particles is a slow pro~ess.2~ On the other hand it is also true that
suspensions which originally did not yield filterable precipitates gave clear
filtrates after digestion. Under the microscope one gets the impression that
the small particles of the precipitate are held together in a film after digestion.
The problem of filterability then is reduced to the question, why the particles
of a flocculated sol glue together. I n textbooks on colloid chemistry this
problem is not discussed. A priori it is not to be expected that small particles
will grow t,ogether. If a flocculated particle possessed a pure crystal surface
one could accept the view of K. Fajans and von Beckerat,h2 that the small
particles will grow together with loss of energy, the negative ions in the
surface of one lattice being attracted by those of opposite sign of the other
lattice. However, it is hard to assume that the picture is as simple as this.
Suppose in Fig. I A represents a particle of silver chloride, which is kept
in colloidal solution by an excess of chloride ions. After destruction of the
diffuse electrical double layer the picture is that represented by B or C. I t is
23 Many examples may be found in P. P. von Weimams monograph: Die Allgemeinheit
des Kolloideustandes. Kolloides und kristalloides Losen von Niederschliigen (1925).
24 K. Fajans and von Beckerath: Z. physik. Chem., 97, 478 (1921).
870 I. M. KOLTHOFF
not to be expected that every kind of flocculating ion (X) can be incorporated
in the crystal lattice (C), in other words forms mixed crystals and therefore
C represents the most usual case after the flocculation. The surface of the
other flocculated particles will have a similar appearance and there is no
direct reason why the flocculated particles will lay their surfaces together
(cement together). I t is much more reasonable to assume that at the moment
of flocculation the particles will repel each other; an impression which is
actually obtained in observations under the ultramicroscope. This problem
will be studied more extensively in Professor Gortners laboratory. The
flocculating ions X (Fig. I ) which are adsorbed at the surface of the floccu-
lated particle are still able to exert a polarizing effect upon the solvent, in
this case water, which means that they are more or less hydrated. It is a
well known fact that a flocculated colloid contains large amounts of water.
If the hydration is great enough a water layer will be formed around the
particle, which, on account of its high curvature will have a tendency to
decrease its surface. This may be accomplished by aggregation of various
particles which in this way share their water jacket, thus giving the impres-
sion of a film formation. I n this manner it is also possible to explain why
the rate of solution and of deposition of ions from the solution on the surface
is so strongly inhibited, for the surface of the flocculated particles is more or
less isolated.
Before drawing further conclusions, an experimental study is necessary
to show whether the view offered here gives a true account of the facts. I n
a quantitative treatment, the orientation (polarization) of the water molecules
around the particles should be considered.
5. The Formation of Precipitates. Amorphous and Crystalline Precipitates
An exhaustive treatment at this place of the speed of formation of nuclei
and of growth of small crystals would require more space than is allowable.
There is a specisl lack of exact data in the literature on the speed of formation
of nuclei from supersaturated solutions; it may be expected that this speed
will be proportional to the degree of supersaturation if the latter is expressed
with regard to the solubility of the nuclei (v. i.). More is known of the
speed of growth of crystals, especially from studies made by physi ~i sts.~~
The speed of growth of crystals decreases on going from the corners to
the edges and from thence to the planes. The speed of growth is not deter-
mined by the rate of diffusion but by the speed with which the ions are
adsorbed at the crystal surface. Constituents which change the adsorption
equilibrium may have a tremendous influence on the speed of growth of
various pla6es and the crystal habit finally obtained. Special reference is
25 Comp. M. Volmer and I. Estermann: Z. Physik, 7, 13 (1921); 9, 193 (1922); Volmer
and A. Weber: Z. physik. Chem., 119, 277 (1925); Volmer: Z. Elektrochemie, 35, 555
(1929); Volmer and M. Marder: Z. physik. Chem., 154 A, 97 (1931); K. S angenberg: Z.
Krist., 59, 403 (1924); H. Brandes: Z. physik. Chem., 126, 196 (1926); W. %ossel: Natur-
wisaenschaften, 18, 901 (1930); I. Stransky: Z. physik. Chem., 136, 259 (1928); Il B, 342
(1930); D. Balarew: Kolloidchem. Beihefte, 30, 249 (1930); 32, 205 (1931); for eneral dls-
cussion comp. A. E. von Arkel and J . H. de Boer: Chemische Binding als Eleftrostatiach
Verschynsel, 248-286 (1930).
THEORY OF COPRECI PI TATI OS 87 1
made to the work of R. Marc and collaboratorsz6 and more particularly to
that of W. G. France and his studentsz7 on the influence of dyestuffs on the
habit of various growing crystals. The problem of crystal growth is of ex-
treme importance for that of coprecipitation because the amount of copre-
cipitated substance is greatly dependent upon the speed of formation of the
crystals. This will be shortly discussed in the following chapter.
The form, shape, and size of a precipitate depends upon the experimental
conditions. From all that has been said above, it is evident that a simple
relation between supersaturation on the one hand, and the speed of growth
and form of crystals on the other, as has been advocated by P. P. von Wei-
marnz8 cannot exist.
Although i t cannot be denied that interesting statements and experiments
of qualitative nature may be found in von Weimarns numerous publications,
his general equations are too simple and cannot be accepted. The velocity
of formation of nuclei (condensation) during the first stage of the precipi-
tation is formulated thus:
in which W is the initial rate of precipitation; Q the total concentration of
the substance that is to precipitate, S the solubility of coarse crystals of the
substance; Q - S =P the amount of supersaturation and U the percentage
supersaturation at the moment precipitation begins. Von Weimarn recog-
nized that the velocity W of the first stage of precipitation could not be
measured in actual practice and therefore, he introduced a specific coefficient
called the precipitate form coefficient or dispersity coefficient X, which
is given by the expression:
S =P S K,I, Kad K bd K,, Z
in which Z is the viscosity, and Knb Kcd etc. represent the physical and
chemical association of the substances AB, CD etc., which enter into the
reaction AB (in solution) + CD (in solution) = AC (precipitate + BD
(in solution). H. B. WeiserZ9 remarks: The significance of physical asso-
ciation is known, but it is not clear what von Weimarn means by chemical
association.
The growth of the nuclei depends not only on the degree of supersatura-
tion, at a given moment, but also according to von Weimarn upon the diffu-
sion coefficient :
V = D/ d 0 (C-S)
zeR. Marc and Wenk: Z. physik. Chem., 61, 385 (1908); 68, 104 (1910); 73, 685 (1910);
75, 710 (1911); 79, 71 (1912).
4 7 T. S. Eckert and W. G. France: J . Phys. Chem., 34, 72 (1930); F. G. Foote, F. C.
Blake and W. G. France: 34, 2236 (1930); W. G. France: Col?oid Symp. Ann., 59 (1930);
especially C. H. Saylor: J . Phys. Chem., 32, 1441 (1928).
2u P. P. von Weimarn: Zur Lehre von den Zustknden der Materiel Bd I Text, Bd I1
(1913); Die Allgemeinheit des kolloiden Zustandes (192 j).
2 8 H. B. Weiser: The Colloidal Salts (1928); for a discussion of von Weimams theory
see also T. B. Smith: Analytical Processes, 263 (1930).
872 I . M. KOLTHOFF
in which D is the diffusion coefficient, d the thickness of the adherent film,
0 the surface, C the concentration of the solution and S the solubility of the
disperse phase.
Without entering too much into details, it is evident that von Weimarns
equations have no general quantitative bearing:
(a). I n the equations of the speed of formation of the nuclei S represents
the solubility of coarse crystals. I n a preceding chapter we have seen that
S is a function of the size of the crystals, and even the order of its magnitude
may be quite different for nuclei and for larger crystals; the percentage
supersaturation has no definite value.
The equation for the speed of growth of the crystals cannot be
accepted, since this rate does not depend upon the speed of diffusion but
upon the speed of adsorption. All factors which influence the speed of
adsorption of the ions during the growth may materially change the speed
of growth and the habit of the crystals obtained.
I t is to be expected that the speed of formation of nuclei will be
dependent not only upon the relative supersaturation but also upon the
actual concentration of the reacting ions. The formation of nuclei in a solu-
tion ten times supersaturated with regard to calcium sulfate is much more
rapid than in that of barium sulfate of the same supersaturation. That the
actual concentration of the reacting ions is of primary significance may also
be inferred from the studies of Sven OdBnzg8 on the formation of precipitates.
I t should be mentioned that in most studies on the speed of precipitation
the supersaturation was calculated on the basis of the solubility of the sub-
stance in pure water. Even if the difference between micro and macro
solubility could be neglected the repression of the solubility by the common
ion effect should be considered. On account of this omission most of these
studies have no quantitative significance.
Of much more importance are the views of F. HabeP on the formation
of precipitates. He considers primarily the following two factors: aggregatzon
eeloczty, (called by Haber Hdufungsgeschwindigkeit) and orientatzon neloczty,
(Ordnungsgeschwindigkeit). If the solubility limit is exceeded, the molecules
or aggregates of molecules, will have a tendency to lay themselves together
and to accumulate to give larger aggregates. This aggregation velocity is
a function of the supersaturation; the larger the latter, the less regular the
separated aggregates will be. Besides the supersaturation the absolute
concentration of the reacting ions will also be of significance. The aggregates
formed, in which the molecules are mixed in a more or less arbitrary
manner, are not stable. By loss of energy they tend to reach a state of
equilibrium, in which the mass is ordered in a regular way in a crystal lattice.
The speed with which this process takes place is called the orientation ve-
locity. I t is evident that the form in which a precipitate separates depends
upon the competition between the aggregation and orientation velocity.
(b).
(c).
z9* Sven Odh: Ark. Kemi Mineral. Geol., 7, No. 26 (1920); 9, No. 23 (1925); No. 32
(1926).
3 O F. Haber: Ber., 55, 1717 (1922).
THEORY OF COPRECI PI TATI ON 873
If the supersaturation is extremely large the aggregation velocity will domi-
nate and the separated particles do not show an X-ray spectrum, in other
words they are amorphous. On standing, (aging) the amorphous precipitate
will slowly transform into a crystalline modification. The phenomena de-
scribed can be observed in the precipitation of various extremely slightly
soluble hydrous metal oxides and metal sulfide^.^'
Strongly
polar substances, such as silver chloride, for example, will have a high orien-
tation velocity, comparable with the crystallization velocity of this kind of
molecule from the vapor state. The more completely the central ion of a
molecule is surrounded by the ions of opposite charge the weaker the electric
field outside the molecule will be, because its electric effect is mostly com-
pensated by the surrounding ions. Such molecules will show a small orienta-
tion velocity. From this point of view it is clear why precipitated cadmium
hydroxide, for example, is never obtained in an amorphous form; hydrous
ferric oxide, on the other hand, if precipitated from cold solutions is amor-
phous, but is transformed on aging, especially on heating into a crystalline
product. Hydrous oxides of the quadrivalent cations, like those of thorium,
cerium, zirconium, in which the central ion is more or less completely sur-
rounded by the anions, always precipitate in an amorphous form and show
on aging very little or no tendency toward crystallization.
I n the slow preparation of colloidal suspensions according to the conden-
sation method, crystalline sols are obtained; if the precipitation takes place
quickly as under analytical conditions, amorphous precipitates may be formed.
I n the older literature, a sol was usually considered as an intermediate state
in the process of formation and growth of Precipitates, this appears not to
be true; the sols are crystalline whereas it is quite possible to obtain many
slightly soluble substances in an amorphous form.
I t is of practical importance to consider the solubility of amorphous
precipitates somewhat more fully. I t does not seem justifiable to speak of
an amorphous rnod?fication if by some wild growth an amorphous precipi-
tate is obtained; at least if a. modification is considered as a chemically
homogeneous individual. The amorphous form consists of aggregates which
:ire grown together in a more or less arbitrary manner; it is not stable but is
undergoing continuous transformation into a more stable crystalline state.
The soliibility of the amorphous form therefore, will not only be quite different
from that of the crystalline phase, but what is more important, the solubility
will no longer be constant but depend upon the state in which the amorphous
form happens to be at a certain moment. This inference is quite important
because in analytical problems, such as the precipitation of slightly soluble
hydrous oxides and metal sulfides, the mass action law as a rule is applied.
The latter, however, holds only for stable modifications when there is equilib-
The orientation velocity will vary for various substances.
For esampies comp. F. Haber: (ref. 27) ; J . Bbhm and H. Siclassen: 2. anorg. allgem.
Chem., 132, I (1923); G. F. Huttig and collaborators: 187, I ; 190, 353, 364; 191, 161; 192,
1 x 7 , 2.25; 193, 81, 93, I OO i rggo). Kolloidchem. Beihefte, 31, 347 (1930); and especially
R. Fricke: 2. anorg. allgem. Chem., 166, 244 ( 1927) ; Kolloid-2. 49, 229 (1929).
874 I. M. KOLTHOFF
rium between solution and solid phase. The amorphous form is not in
internal equilibrium and, therefore, neither with the solution. If an amor-
phous hydrous oxide is precipitated under various conditions, the same
solubility product can hardly be expected. The water present in the amor-
phous precipitate will also have a great influence upon the solubility. Sup-
pose for example, that a hydrous aluminum oxide is precipitated by mixing
aquo aluminum ions Al(HB06)+++ with hydroxyl ions. With the large
aggregation velocity, the aluminum ions will carry part of their water into
the amorphous precipitate, and the hydroxyl ions probably will keep part
of the water molecules oriented by polarization. The water is a powerful
dielectric, and by its presence in the precipitate, will diminish the force by
which the aluminum and hydroxyl ions attract each other in the solid.
Therefore] owing to the presence of water between the ions, the solubility
of the amorphous hydrous aluminum oxide must be much greater than that
of the crystalline modification. It cannot be expected that the solubility
will remain a constant during the transformation of the amorphous into the
crystalline form; i t will gradually decrease until the entire precipitate is
present as the crystalline modification. Neglect of these considerations has
caused some unfortunate confusion in the literature.
I n the rapid formation of precipitates, another point must be considered.
If various crystalline modifications of a substance exist, there is always a
possibility that a metastable form will separate out first. On standing the
labile form will be transformed more or less rapidly into the stable modifi-
cation thus causing an entire change of the internal structure. Berthel ~t, ~~
for example, claimed that freshly precipitated silver iodide undergoes a trans-
formation on standing, a fact not confirmed by the experiments of J . W. A.
van Henge1.33 I n agreement with Berthelot, the latter found however, that
freshly precipitated barium carbonate undergoes a structural change on
aging. E. B. Sandell, in experiments carried out in this laboratory found
that calcium oxalate di or tri hydrate was separated under conditions under
which only the monohydrate is stable. I n the interpretation of the change
of the amount of coprecipitated substance with the time of standing, such
allotropic changes must be considered.
6. The Theory of Coprecipitation
I n the introduction i t was emphasized that the expression carrying
down used as a collective term to indicate that a precipitate contains foreign
constituents is misleading, and that first of all i t should be decided whether
the impurities are incorporated in the crystal lattice or form imperfections
in the interior of the crystals or finally are adsorbed at the surface of the
precipitate. The second group in which the impurities are present as imper-
fections in the crystals is the most common one and will be defined as real
coprecipitation.
r 2 Berthelot Ann. Chim. Phys. 151 4, 181 ( 1875) ; 29, 242 (1883).
33 J . W. A. van Hengel. De metastabiliteit der stof. Praecipitatie-reacties, Thesis Utrecht
(1931). On theory see espec.ally C. H. Saylor (ref. 27).
THEORY OF COPRECIPITATION 875
Mixed crystal formation: Although the phenomena of isomorphism and
mixed crystal formation have been known for almost a century, it is only of
relatively recent date that more general relations have been discovered
between size of ions and mixed crystal formation. H. G. Grimm and collabo-
rators34 formulated three conditions which must be fulfilled for the formation
of mixed crystals of polar compounds:
(I ), The chemical building type must be the same.
( 2) . The lattice types must be similar.
(3). The lattice constants must be of the same order of magnitude.
Grimm already showed that the following systems form mixed crys-
tals: SrSOl + KMn04; BaSeO, + KMnO,; BaCrO, + KMn0,; BaS04 +
KBF4; KBF, + KMn0,. Grimms studies are supplemented by the beau-
tiful investigations of V. M. Goldschmidt35 on the relation between crystal
structure and lattice properties on the one hand and chemical constitution
on the other. The laws of isomorphism and mixed crystal formation are
laid down, especially in his seventh Isomorphism he defines as the
phenomenon that substances of analogous chemical formula show analogous
crystal structure. Analogy of chemical formula means analogous brutto
formula in respect to total number of ions and to the number of positive
and negative building stones (ions). Analogy of crystal structure means
that both substances possess a geometrically symmetrical elementary paral-
lelopiped, in which a same number of atoms is arranged in a geometrically
similar fashion in such a manner, that the kind of charge (positive or nega-
tive) of the individual crystal building stones correspond with one another
in both structures. Isomorphism occurs if the relative size of the crystal
building stones and the relative strength of their polarizability within certain
limits are identical, assuming of course that the brutto formulas of both
substances are the same. If not only the relative size, but also the absolute
size of the building stones are the same, the conditions of mixed crystal
formation are created, other conditions being the same. There is a certain
tolerance with regard to similar size; it seems that mixed crystal formation
is still possible if the radii of the ions does not differ more than 157,. With
these rules it is understandable why BaS04 - KMnO,; BaSOl - PbS04;
CaC03 - NaN03 can form mixed crystals, Relatively little is known of
the propertiesof mixed crystals, which are formed in analytical processes;
especially of their stability. From studies of 0. Ruff and E. A ~cher ~~ one
would infer that mixed crystals are very unstable if one of the constituents
is slightly soluble and the other readily soluble. Thus in the case of CaC03
- NaN03 Ruff and Ascher found practically no sodium nitrate in the crystals,
34 H. G. Grimm and G. Wagner: Z. physik Chem., 132, 131 ( ~ 6 ) ; H. G. Grimm: 98,
istall., 57, 574 (1922); 353 (1921); Z. Elektrochemie, 28, 75 (1922); 30, 467 (1924); Z.
Handb. Physik, 24, 581 (1927).
35 V. M. Goldschmidt: esp. Geornetrische Verteilun sgesetee der Elemente VI1 Die
Gesetze der Kristallochernie, Videnskapssel Skrifter I &fat. Naturw. Klasse Det Norske
Videnskaps Akad. i Oslo I 1926; No. 2 ; Utgitt for Fridtjof Nansens Fond.
3E 0. Ruff and E. Ascher: 2. anorg. allgem. Chern., 185, 369 (1929).
876 I. M. KOLTHOFF
if the precipitate was kept under the mother liquor until no further change
took place. On the other hand, if both compounds are slightly soluble, as
in the case of BaS04 - PbSO, both cations are always present in the pre-
cipitate. From the analytical point of view, it is highly desirable to obtain
more information regarding the ratio in which the two constituents occur
in mixed crystals if precipitated under various conditions, and also of the
stability of the crystals if kept under the mother liquor. Such studies are
being made in our laboratory. I t seems that mixed crystal formation is not
materially dependent upon the manner of precipitation; i. e. whether an excess
of cations or anions is present during the precipitation. This is in striking
contrast to what occurs in real coprecipitation, where the amount of
coprecipitated foreign ions is highly dependent upon the conditions of pre-
cipitation.
Otto Hahn formulates his precipitation rule in the following way;
An ion precipitates from any dilution with a precipitate crystallizing out,
if it is incorporated in the crystal lattice; i.e. if it forms mixed crystals with
ions of the crystalline precipitate. If it does not form mixed crystals, the
ion remains in the filtrate even if its compound with ions of opposite charge
in the precipitate is slightly soluble. According to 0. Hahn3* mixed crystal
formation is possible even if it does not occur under normal conditions.
This phenomenon is called isodimorphism and occurs for example accord-
ing to Hahn39 in the cases BaCla + RaB(ThB) or RaD(ThD); PbS04
(ThB) + K2S04; T12S04 + ThBS04; PbCr04(ThB) + Ag2Cr04. Although
Hahn worked under unusual conditions (the concentration of the radioactive
constituent always was extremely small) it seems to the author that Hahn
has not proved definitely the exstence of such isodimorphism. Hahn as we
will see later, does not make a distinction between mixed crystal formation
and real coprecipitation, and there are reasons to assume that his cases of
isodimorphism actually are examples of true coprecipitation. However,
more experimental work must be done before a decision can be reached.
From the above short discussion, it follows that in all cases of coprecipitation,
the possibility of mixed crystal formation should be considered.
Adsorpt i on and Real Coprecipitation: I n a study of the purity of
precipitates, a distinction must, be made between adsorption and real
coprecipitation. If a precipitate separates as a flocculated colloid, it has
a layer of flocculating ions rigidly adsorbed at its surface. These adsorbed
ions can be partly removed by washing out or replaced by washing with
suitable electrolyte solutions which do not interfere later in the analytical
process (in gravimetric analysis : ammonium salts, acids and other electro-
lytes which volatilize on gentle ignition may be used.) If a real coprecipita-
tion takes place, the impurities are present in the interior of the crystal
and cannot be removed by a washing procedure. If a precipitate separates
3 i Otto Hahn: 0. Erzhacher and N. Feichtinger: Ber., 59, 2014 (1926); 0. Hahn: Natur-
nissenschaften, 14, I 196 (1926).
31 0. Hahn: Z. angew. Chem., 43, 871 (1930).
3 0 0. Hahn: Sitsungsber. preuss. Akad. Wiss., Physik. Math. Abt., 30, 547 (1930).
THEORY OF COPRECIPITATION 877
in a micro crystalline form, such a coprecipit'ation always seems to occur
even if there is no trace of mixed crystal formation. From scant data
in the literature, and our own experience, one would infer that colloidal
precipitates al ways are contaminated by adsorbed i ons but neuer cont ai n
the f orei gn i ons in the interior of the particles. I n order to explain this,
one has to realize that as long as the particle is in a colloidal state, the
adsorbed foreign ions are present on the mobile side of the electrical double
layer around the particle; they are not fixed at' the surface. Therefore, as
long as t,he particles are in the colloidal state, there is no reason why these
ions which have retained their mobility should be incorporated during the
growth of the particles (unless mixed crystal formation occurs). Here the
electrokinetic potential of the particles is the governing factor. After the
colloidal state has been passed, the thermodynamic potential is the main
factor in the occlusion during the growth. As has been discussed in the first
chapter, a crystal surface of a slightly soluble compound adsorbs that one
of its own ions which happens to be in excess in the solution. A growing
barium sulfate crystal for example, in the presence of a large excess of barium
chloride in solution, will adsorb barium ions at its surface. Owing to the
electroneutrality of both phases, the other ions of the electrolyte (here
chloride ions) must be dragged along with the barium ions to the surface
of the growing particle, where they will be more or less fixed. If the condi-
tions are such that the crystals will grow very slowly, the chloride ions will
be replaced more or less by sulfate ions, which fit in the crystal lattice and
the coprecipitation will be relatively small. If the crystals grow much
faster, there is no time available for a complete exchange between contami-
nating ions and those belonging to the crystal and a large coprecipitation
will result. If finally, the conditions are such that owing to an extremely
large supersaturation the particles are not given a chance to grow to bigger
crystals, a flocculent colloid will separate and the coprecipitat'ion will be
extremely or negligibly small. Various studies made in our laboratory,
which will be discussed in subsequent papers, yield results in complete har-
mony with the postulates of t,he developed theory. I n t,he cases of calcium
oxalate (E. B. Sandell) and barium sulfate, the largest coprecipitation is
observed if t'he crystals are allowed to form under such conditions that they
have a relatively large size. If the supersaturation is made so large that a
flocculated colloid settles out, no coprecipitation should be observed but the
impurities should be kept in an adsorbed state at the surface. If these small
particles are allowed to recrystallize, a very slow growth takes place and a
practically pure precipitate is obtained. If, finally, the conditions of precipi-
tation are such that the crystals acquire an intermediate size, a real copre-
cipitation takes place. We therefore, arrive at the interesting conclusion that
the largest crystals are the least pure if formed under analytical conditions.
This is not in harmony with the general rules of analytical chemistry, where
the directions usually are based on the assumption that larger crystals are
purer than small crystals. I n a recent publication, 0. Hahn38 expresses
this rule in the following way:
878 I. M. KomnoFF
Fur die analytische Chemie ergibt sich auch aus diesen Versuchen das
ja schon in der Praxis verwendete Verfahren, Niederschlage in moglichst
oberflachenarmer, gut kristallisierender und daher gut filtrierbarer Form
ahszufiihren. Al l e Adsorptionsvorgange werden dadurch stark zuriickgedrangt.
J e mehr man von diesem Grundsatz abgeht, destomehr entfernt man sich
von der Arbeitsweise des Analytikers; man kommt in das Gebiet des Kolloid-
chemikers; die Vorgange an die Grenzflachen werden ausschlaggebend.
The contradiction between Hahns and our views is explained by the fact
that Hahn considers only the adsorption at the surface after the crystals have
been f ormed, but not the adsorption duri ng the growth. For the st udy of copre-
cipitation the adsorption duri ng the growth however i s the predominating factor.
It is of interest to discuss a few other consequences of our picture of copre-
cipitation, although a more detailed discussion will be given later in connec-
tion with reports of practical work. The thermodynamic potential is deter-
minative for the kind of adsorption taking place during the growth. If
barium sulfate is precipitated from a solution containing an excess of sulfate
ions, the latter will be adsorbed during the growth of the precipitate and
drag f orei gn cations like H30+; K+, Na+, Caff etc. to the surface; if on the
other hand, barium ions are in excess during the growth, foreign anions like
chloride, nitrate, etc. will be adsorbed and will coprecipitate with barium
sulfate. Dependi ng upon the conditions of precipitation, a cation or ani on
occlusion will be predominant. Experimentally, this has been shown to be
true in the cases of barium sulfate and calcium oxalate monohydrate, although
i t should be mentioned that on account of incomplete dissociation of binary
and ternary electrolytes, an apparent cation precipitation may be found
where only anion coprecipitation is expected and also the reverse. However,
in its general form, the rule derived seems to be of general applicability. Of
special interest is the coprecipitation of ferric iron with barium sulfate. If
the precipitation is carried out in a solution containing an excess of sulfate,
a large coprecipitation of ferric iron-either as ferric ions or more probably
as positive colloidal hydrous ferric oxide which is formed by hydrolysis-
takes place. If the acidity of the solution is increased, the concentration of
positive ferric oxide particles decreases and therefore, the amount copre-
cipitated also. On the other hand, if barium sulfate is precipitated from a
solution containing an excess of barium ions, there is no coprecipitation of
ferric iron at all.
According to the above views contaminating ions are not present in the
crystal lattice, but as imperfections of the crystal lattice. They may be
identified with the Lockerstellen or Hohlraume in real crystals according
to A. Smeka140 or centra of activity, according to H. S. Taylor.41
I t should be remembered that coprecipitated foreign ions are not fixed
by the lattice and therefore, keep more or less of their water of hydration.
I n this way the presence of the solvent (mother liquor) is explained and it
4o A. Smekal: Z. angew. Chem., 42, 489 (1929), where also other literature is given;
comp. also W. J ost: Z. physik. Chem., 6B, 88 (1929); 7, 234 (1930).
4L H. S. Taylor: Proc. Roy. SOC., 108, IOj (1925).
THEORY OF COPRECI PI TATI ON 879
is a problem of interest to determine whether there is a simple relation be-
tween the size of coprecipitated ions (which determines with the charge, the
polarizing effect upon the solvent) on the one hand, and the ratio of amounts
of coprecipitated ions and solvent on the other.
I t is a well known fact that real crystals are far from ideal (Smekal);40
they seem to have a porous structure. The more they are contaminated,
the more pronounced the porous structure will be and vice versa. This may
explain the peculiar fact that if a freshly formed precipitate is kept under the
mother liquor, under such conditions that no recrystallization can take place,
the amount of coprecipitated impurities decreases on aging and reaches a
minimum after a day or so standing. This phenomenon seems to be quite
general and has been noticed in the case of barium sulfate, lead sulfate,
calcium oxalate, etc. The crystals seem to exert an action tending to perfect
their own lattice which can be done only by expelling the impurities. I n the
interpretation of the peculiar aging effect, i t should not be overlooked that
in various cases a possibility of the primary formation of a metastable modi-
fication of the precipitate exists, which on standing may change into the
stable modification thus yielding a radical internal change.
Summarizing then, i t seems that real coprecipitation must be attributed
to and is governed by adsorption phenomena during the growth of the
crystals. A purely chemical interpretation based on the formation of definite
chemical compounds between ions belonging to the lattice and contaminating
ions of opposite sign, as is still done by 2. Karaoglanow42 must be rejected.
D. Balarew43 in his extensive studies on purity of precipitates, attributes
any kind of coprecipitation to inner adsorption by crystalline salts. I n so
far as true coprecipitation is concerned, Balarews views approach those
developed in this paper; however, his picture and many of his statements
are vague and he does not mention any connection between thermodynamic
potential and kind of ion coprecipitation. As a result of coprecipitation
polar adsorption compounds like BaS04, KzS04, HzO are formed according
to Balarew. According to his opinion the occluded water still has solvent
properties, but only for one special salt. If, for example, potassium sulfate
and water are occluded by barium sulfate, no sodium or lithium sulfate can
be coprecipitated, because the occluded water has solvent properties only
for potassium sulfate. This view cannot be accepted as true, and is contrary
to experimental evidence available at the present time. Although Balarews
papers contain many interesting data, his picture of coprecipitation is not of
wide bearing and general applicability.
on fractional precipitation, formation of mixed crystals
and adsorption compounds is very important and stimulating, but again no
Ruffs
4* Z. Karaoglanow: Ber., 63, 597 (1930); Karaoglanow and B. Sargotschev: Z. anal.
Chem., 187, 273 (1930); 81, 275 (1930); 2. anorg. allgem. Chem., 194, I j I (1930); 195, Ioj
(1011).
~ . I
4J Comp. summary of his work Kolloidchem. Beihefte, 30, 249 (1930).
440. Ruff and B. Hirsch: 2. anorg. allgem. Chem., 146, 388; 150, 8j (192j); 151, 81
(1926); 0. Ruff and E. Ascher: 185, 369 (1929); 0. Ruff: 185, 387 (1929).
880 I. M. KOLTHOFF
clear picture is developed for the case of true coprecipitation; especially
adsorption during the growth of the crystals is not considered. For a quan-
titative treatment of the problem, the change of adsorption with time has
to be studied more closely. Interesting and promising investigations of this
subject are being carried out in 0. Hahns laboratory, especially by L. Imre.4;
Moreover, the change in internal structure of a fresh precipitate in con-
tact with its mother liquor has to be considered with regard to decrease of
coprecipitated impurities on standing.
summary
(I).
Impurities in crystalline precipitates are due to coprecipitation or
post precipitation. Three cases of coprecipitation are to be distinguished :
Mixed crystal formation, in which the impurity is incorporated in the crystal
lattice; real coprecipitation, in which impurities form imperfections in the
crystal and surface adsorption by the precipitate after it is formed. I n all
these cases the presence of impurities is attributed to adsorption, either
during the growth of the crystals or after their separation.
I n the case of real coprecipitation, the adsorption of foreign ions
during the growth of the particles is of primary importance. Depending
upon the thermodynamic potential of the precipitate during the growth, a
cation or anion coprecipitation may be expected. If, for example, during
the precipitation of barium sulfate, an excess of barium ions is present, a
coprecipitation of anions may be expected, in the reverse case a coprecipi-
tation of foreign cations.
(3): Up to a certain limit, coprecipitation with a slightly soluble substance
increases with crystal size; a statement which is contrary to the general rules
of analytical chemistry.
The Paneth-Fajans precipitation rule should be extended in the
following way: The higher the valence of an ion the more preferential it will
be adsorbed by an ionic lattice which contains an excess of lattice ions of the
sign opposite to that of the secondary adsorbed ions.
A precipitate which does not exceed the colloidal dimensions con-
tains impurities in an adsorbed but not in a coprecipitated state (provided
there is no mixed crystal formation). As long as the particles are in colloidal
solution, the foreign ions are present on the mobile side of the double layer
and are not occluded.
(6). The analytical significance of the relation between particle size and
solubility has been discussed. The concept of supersaturation with regard
to speed of formation of precipitates has no exact significance; it depends
upon size and surface tension of the tiny particles. Also for this reason
von Weimarns expression of the relation between the speed of formation of
precipitates and supersaturation has no general validity.
( 2 ) .
(4).
( 5 ) .
46 L. I mre: 2. angew. Chem., 43, 875 (1930); 2. physik. Chem., 153A, 127, 262 (1931).
THEORY OF COPRECI PI TATI ON 881
( 7 ) . A discussion is given of filterability and properties of a flocculated
colloid.
(8). An explanation is given of the fact observed by Dundon, that a
mixture of very small and large crystals shows the normal solubility.
(9). The expression amorphous modification has no exact significance.
The properties of an amorphous substance are more of less accidental; the
mass action law cannot be applied when amorphous precipitates are dealt with.
( I O) . A freshly formed precipitate is not in equilibrium with its mother
liquor. On standing, a change of the internal structure of the crystal takes
place. Under any conditions, the amount of coprecipitated ions decreases
with time of standing before filtration.
School of Chemistry of
The Cniuersity of Mznnesota,
Mi nneapol i s, September. 1931.