The formation of three different crystalline calcium phosphates (DCPD, OCP, HAP) and an amorphous
calcium phosphate was studied as a function o f p H and supersaturation. Under the experimental conditions
the formation of H A P is always found to be preceded by one or more precursors in a sequence that is
in agreement with the Ostwald rule of stages. Homogeneous formation of H A P at low concentrations is
never observed. A semiquantitative analysis of growth curves and of relaxation times at p H 6.7 and
26°C reveals the growth of OCP to be described best by a mononuclear growth model with fourth-order
dependence on the relative supersaturation. The heterogeneous formation of H A P on OCP surfaces is
noted to shift to lower a values (reaction extent) with increasing initial supersaturation. It can be described
by a polynuclear growth model. © 1987AcademicPress,Inc.
x,, ::).-...:::~
in the precipitation sequence around p H 6.7
for the composition chosen in constructing 4
Fig. 1. This interesting observation m a y also
be of significance for the micellar calcium
phosphate in milk as the p H of milk is about
p H 6.7.
The objectives of our study are twofold.
First, we wish to study the formation of dif- ,4 6 8
6.7 7.4 > pH
ferent calcium phosphates as a function o f p H
and initial supersaturation to test the validity FIG. 1. Solubility isotherms of HAP, OCP, and DCPD
at 25°C (solid lines) and 37°C (dotted lines) as log PT vs
of the Ostwald rule of stages. pH. PT is the total phosphate concentration in solution
The supersaturation (H) with respect to a (mole/liter). The chosen Ca/P ratio was 1.16. The ionic
given calcium phosphate phase is defined as strength of the solution I = 0.15 M.
Journal of Colloid and InterfaceScience, Vol. 118, No. 2, August 1987
566 VAN KEMENADE AND DE BRUY-N
interfacial properties remain essentially con- by the first method a more monodisperse pre-
stant. Precipitation studies were performed at cipitate will form. In this way the mixing pro-
pH 6, 6.7, and 7.4 at a temperature of 26°C. cedure may greatly affect the precipitation
Raising the temperature has the effect of shift- process. Different nuclei can be formed under
ing the intersection of the solubility isotherms the same "initial" conditions making it pos-
of OCP and DCPD to a lower pH (Fig. 1). At sible for different crystal phases to develop by
pH 6.7, 37°C the order of increasing stability growth. A careful study of mixing procedures
is DCPD < OCP < HAP. A series of experi- is therefore necessary before comparing ex-
ments was therefore also performed at pH 6.7 perimental results from different sources.
and T = 37°C to study this effect. In our experiments the supersaturation was
always built up by the first method. A solution
EXPERIMENTAL of KH2PO4 ( 1-15 raM) and Ca(NO3)2 (1-15
mM) was mixed with a 0.15 M KNO3 solu-
The relaxation experiments were performed tion. During this mixing the pH is about 4.5
in a thermostated double-walled reaction ves- and therefore the solution is still undersatu-
sel developed by Vermeulen et aL (18). The rated with respect to all modifications of cal-
experiments involve building up a desired su- cium phosphate. The desired supersaturation
persaturation and then allowing the system to and pH were reached by a fast injection of a
relax at constant pH. 0.5 M carbonate-free KOH solution under ni-
The particular procedure to be followed in trogen pressure. The selected pH was always
building up the supersaturation will be of im- obtained within 1-2 rain. The pH was mon-
portance in the relaxation behavior as the for- itored and adjusted with an automatic titrator
mation and growth of nuclei during this stage consisting of an Apple microcomputer, an
cannot be ruled out. Two methods are com- Ankersmit pH meter, and a Schott automatic
monly used for mixing the components in or- buret. During the relaxation the pH was always
der to achieve the desired "initial" supersat- maintained constant to within 0.003 pH units.
uration. A fast injection of a highly concen- Amagruss type C- 1017 combination elec-
trated solution of one component into a trodes were used to measure the pH of the
solution of other components has the advan- solution. They were calibrated with Electrofact
tage of reaching the desired supersaturation buffer solutions before and after each experi-
very fast. A disadvantage, however, is that ment. Only experiments with an electrode drift
small concentrated droplets may exist in the in pH less than 0.01 were accepted. Uptake of
solution for a very short time. At their interface CO2 was prevented by bubbling nitrogen gas
with bulk solution the supersaturation can be through the solution. The ionic strength of the
so high that a considerable number of nuclei solution was kept constant during the precip-
may form. It is possible that the entire nucle- itation by the presence of 0.15 M KNO3. The
ation process is restricted to this period. Vig- magnitude of the initial calcium-to-phosphate
orous stirring is therefore required to reduce ratio (Ca/P) which is an important parameter
the lifetime of the droplets to a minimum. The in the precipitation of calcium phosphates was
other method involves a slow addition of a chosen between 1.00 and 1.67 but in most ex-
less concentrated solution in order to avoid periments was kept at 1.16 _+0.01. All reagents
the high local supersaturations and to accom- were of analytical grade and for all solutions
plish homogeneous nucleation at the intended twice-distilled, demineralized water was used.
"initial" pH and supersaturation. The time Aliquots (100-500 ml) were withdrawn from
before achieving the desired supersaturation solution and separated from the solution phase
however can be very long and heterogeneous by filtration through 0.65- to 0.10-t~m filters.
nucleation cannot be ruled out. This will result The samples were not washed because of pos-
in a polydisperse collection of particles whereas sible preferential dissolution of part of the solid
Journal of Colloid and Interface Science, Vol. 118, No. 2, August 1987
PRECIPITATION FROM CALCIUM PHOSPHATE 567
(7). The precipitates were dried over silica gel solution composition will alter in a compli-
using an oil pump and were stored over silica cated manner. This will affect the saturation
gel. The solid phase was characterized by X- concentration c(oo) of all phases. Furthermore,
ray diffraction (Guinier-de Wolffcamera, En- as the different calcium phosphates have a dif-
raf Nonius, employing C u K a radiation) and ferent alkalinity (DCPD < OCP < HAP), the
the patterns were compared with those tabu- relationship between the amount of base
lated in the JCPDS chart system, by infrared added and c(t) will be different for each phase.
spectroscopy (KBr pellet technique, Perkin- Thus, OH(oo) will differ for each phase.
Elmer spectrophotometer), and by scanning As an operational measure for the progress
electron microscopy techniques (Cambridge of the precipitation we shall use the quantity
stereoscan) for morphology studies. For a OH(t) or its normalized equivalent OH(t)/PT
semiquantitative analysis of calcium and where OH(t) is the cumulative amount of base
phosphate content of the samples the micro- added (in moles/liter) at time t. Assuming that
scope was equipped with a Link X-ray micro- only one phase is growing we can then cal-
analysis system. culate the amount of base, OH(oo) or OH/PT,
necessary to reach the saturation limit for that
RESULTS phase for each experiment at a fixed temper-
ature, pH, and initial concentration c(0).
For each precipitation series performed at In Fig. 2a a typical example of a relaxation
a given pH and temperature the effect of initial curve for a single growing phase is shown. As
supersaturation was studied. As precipitation a result of an increasing surface area and a
of the solid phase is accompanied by a drop decreasing supersaturation during the relaxa-
in pH, the extent of the reaction a(t), where tion a sigmoidal shape of the curve is expected
0 ~ a(t) ~< 1, is measured by the cumulative with only one inflection point. At this point,
amount of base added to maintain a constant where the growth rate reaches its maximum
pH. For a pure phase growing from a stoi- value the relaxation time tR, a highly repro-
chiometric solution the extent of reaction is ducible parameter in precipitation kinetics, is
defined by defined. In addition a so-called lagtime tL is
c(O) - c(t) OH(t) used for analyzing precipitation kinetics. This
a(t) = c(0) - c(oo) --=O H ( w ) ' [5] time parameter is defined as the period before
the first base uptake, but is less reproducible
where c(t) and c ( ~ ) are the concentrations of and its magnitude will also depend on the ex-
a chosen component in solution at time t and perimental detection limit (sensitivity for reg-
at infinite time (saturation-concentration). istering a change in pH). Another relaxation
c(0) is the initial concentration in solution. parameter to be used in this study is the total
In this complex-precipitating system, how- amount of base or (OH/Pa-) added.
ever, it is not possible to define a single rela- For a complex-precipitating system where
tionship between the amount of base added the precipitation involves growth of more than
and a(t). As the solution and the to be precip- one phase, the relaxation curve will be more
itated solid phase have different stoichiome- complicated in form and its analysis will not
tries, c ( ~ ) will be a function of c(0) and the be straightforward. Figures 2b and 2c illustrate
initial supersaturation. For each experiment such complex, constant pH curves. In Fig. 2b,
with different initial concentrations, c ( ~ ) or a two-stage precipitation is shown, where the
O H ( D ) must be calculated separately. An- formation of phase I is followed by the sub-
other problem arises if different phases with sequent formation of phase II. This results in
different chemical compositions are growing two sigmoidal curves where we could define a
during one experiment. Depending on the relaxation time tR for each precipitating phase.
amounts of the different phases forming, the As the formation of the two phases will not be
Journal of Colloid and Interface Science, Vol. 118, No. 2, August 1987
568 VAN KEMENADE AND DE BRUYN
t (OH/PT)max (OH/PT)max
o
o 0
(OH/PT)~O× Y
A/ T
independent, a simple analysis of relaxation arate formation of each pure phase. At low
time, tRII, will be difficult. Figure 2c represents supersaturations (Fig. 3d) a two-stage precip-
a precipitation curve where over a certain pe- itation curve of the type presented in Fig. 2c
riod both phases I and II are precipitating si- is measured.
multaneously. This will result in inflection Detailed IR and X-ray studies show OCP
points in the overall growth curve which can- to be the main precipitating phase at medium
not be interpreted as relaxation times for supersaturations (curves b and c, Fig. 3). SEM
growth of the phases. The lagtime could also photographs of samples taken in this region
be influenced by the simultaneous growth of show the characteristic globules with the plate-
two phases in the early stages of the precipi- like fine structure well known for OCP (Fig.
tation. 5c). In experiments at low supersaturation (see
Fig. 3, curve d) additional large amounts of
Relaxation Times and Phase Analysis brushite are present as large platelets. Figure
at p H 6.7 and 26°C 5 (electron micrograph a) illustrates the pres-
Typical examples of relaxation curves are ence of DCPD whereas electron micrograph
given in Fig. 3. In all cases a measurable lag- b shows the presence of OCP globules in the
time is seen to precede the uptake of base. At same sample. Both OCP and DCPD are al-
high to medium initial supersaturation (curves ready present before the first inflection point
a-c) a one-stage precipitation takes place is reached. X-ray microanalysis also indicates
where the relaxation time is defined by the that the particles have Ca/P ratios of 1.30
inflection point of the curve. The lagtimes and _ 0.05 and 1.00 _ 0.05 in agreement with
relaxation times are seen to increase with de- those of the solid phases OCP and DCPD (Ta-
creasing initial phosphate concentration. Fig- ble I). At high supersaturations the OCP glob-
ure 4 illustrates the sharp increase in tR with ules are overgrown with a more alkaline ma-
initial phosphate concentration. From this plot terial (Ca/P > 1.5) as is shown in Fig. 5d. At
the kinetic precipitation boundary can be de- still higher phosphate concentrations HAP
termined. The extrapolated value of 3.9 m M could be detected by X-ray diffraction and IR
for the critical phosphate concentration of this analysis.
boundary is in agreement with earlier work of Additional information about the nature
Feenstra (10). For all experiments the satu- of the precipitating phases can be obtained
ration boundaries with respect to HAP, OCP, from calculated saturation limits, O H ( ~ ) / P T .
and DCPD were calculated assuming the sep- Above the saturation limit of a given phase,
Journal of Colloid and Interface Science, Vol. 118, No. 2, August 1987
PRECIPITATION FROM CALCIUM PHOSPHATE 569
0.2 t R(min)
, I
200 400 600
I I t
> time (rain) 1200
FIG. 3. Relaxation curves, OH/PT vs time (min), at dif-
ferent initial supersaturations at a constant pH of 6.7, an
ionic strength of 0.15 M, a Ca/P ratio of 1.16, and a tem-
perature of 26°C. Initial concentrations of phosphate in
solution: (a) PT = 6.29 mM, (b) PT = 5.39 mM, (c) PT
800
= 4.79 mM, and (d) Px = 4.20 mM. The dotted lines
indicate the saturation boundaries (~ = 1) with respect to
OCP and DCPD assuming that only the pure modifications
are formed. The HAP saturation is reached at a OH/PT
value of 1.40. Arrows point to inflection points in the
curves. 400
200
uptake of base can be ascribed only to the pre-
cipitation of a phase with a lower solubility.
At high phosphate concentrations (Fig. 3a) the 3 4 5 6 7
relaxation curve is seen to rise above the sat- > PT(raM)
uration boundary for OCP (aocp = 1). This FIG. 4. Relaxation times at pH 6.7, 26°C as a function
observation suggests that, at least in this stage of total initial phosphate concentration PT (mM), I = 0.15
of the reaction, HAP must be forming. At me- M, Ca/P = 1.16.
Journal of Colloid and Interface Science, Vol. 118, No. 2, August 1987
570 VAN KEMENADE AND DE BRUYN
S-shaped curve is obtained and in some ex- ficient apatites. Increasing the magnification
periments an extra inflection point is notice- by a factor 10 showed the same structure with-
able at a low OH/PT value. At lower super- out revealing more detail which could possibly
saturations (Fig. 6b), a three-stage relaxation indicate a fractal structure. A sufficient reso-
behavior is noted. Uptake of base occurs im- lution to study the morphology of individual
mediately, but then slows down to a near- crystals could not be obtained. The HAP crys-
standstill before slowly rising again and to rise tallites (20) are often very small (ranging from
more steeply afterward. The shape of the re- 5 to 50 rim). At lower supersaturations (Fig.
laxation curve, characterized by three inflec- 6, curve b) X-ray diffraction and IR studies
tion points, could be accounted for by either could not detect any crystalline phase during
the successive or the simultaneous growth of the first stage of the reaction. This suggests
two phases as sketched in Figs. 2b and 2c. that the growth of ACP be responsible for the
Feenstra ( 10, 11) also performed relaxation uptake of base. In the second stage, the region
studies at this pH and used the same experi- between the first two inflection points in Fig.
mental conditions except for a Ca/P ratio of 6b, OCP could be detected by X-ray diffraction
1.67. For comparative purposes we performed while in the late stage only HAP was noted.
a few experiments at Ca/P ratios of 1.33 and Beyond the last inflection point only HAP
1.67. The shape of the relaxation curves at growth ocurs. Similar results were obtained by
these Ca/P ratios does not deviate markedly Feenstra (10).
from that shown in Fig. 6 except that the near- From Fig. 6 we note that the precipitation
plateau reached in the first stage at the lower of HAP is apparently terminated long before
supersaturations (see Fig. 6b) falls at slightly the calculated saturation boundary for HAP
higher levels with increasing Ca/P ratio. This is reached and even before the saturation limit
latter observation indicates that a larger for OCP. This behavior cannot be explained
amount of the first phase to form is present. by the earlier precipitation of some OCP. A
Solid phase analyses show that at high su- possible explanation for this behavior must be
persaturations OCP was present in the early found in the existence of a kinetic boundary
stages of the reaction. Its presence was mainly for growth or to a very slow growth of HAP
detected by X-ray diffraction which showed a in this range of supersaturation. DCPD was
double line at 9.37/9.05 A so characteristic of never detected over the whole supersaturation
OCP (100 and 110 planes, respectively). In a range at pH 7.4.
later stage only HAP could be detected by X-
ray diffraction or electron microscopy. These Relaxation at p H 6.0 and 26°C
observations suggest that caution be exercised
in using relaxation times as a quantitative The precipitations can all be characterized
measure for the homogeneous growth of HAP. by a single-stage S-shaped relaxation curve.
SEM photographs taken in the late stages of However, at high supersaturations (PT >~ 15
precipitation (Fig. 7) show the precipitate to raM) the relaxation curves rise above the
have a spongy texture with a Ca/P ratio be- DCPD saturation boundary. Although the
tween 1.57 and 1.65. Ratios on the order of only detectable phase at the end of the precip-
1.50-1.65 are often reported for calcium-de- itation was DCPD, an amount of OCP could
Journal of Colloid and Interface Science, Vol. 118, No. 2, August 1987
FI6. 5 - - C o n t i n u e d .
572
Journal of Colloid and Interface Science, Vul. 118, No. 2, August 1987
PRECIPITATION FROM CALCIUM PHOSPHATE 573
/
O-0C p :1
0.6 tation. T h e p r e c i p i t a t i o n o r d e r at these con-
T ditions is thus O C P ~ D C P D . D C P D takes
00.4 over growth without detectable changes in rate
CtDCPD= 1
o f base uptake. A t low s u p e r s a t u r a t i o n s (Pa-
0.2 ~< 12.5 m M ) o n l y D C P D c o u l d be detected.
T h e curves rise t o w a r d the s a t u r a t i o n b o u n d -
ary for p u r e D C P D growth.
0.6 Ct Ocp=l
R e l a x a t i o n curves at p H 6.7 a n d a t e m p e r -
0.2 :1
ature o f 3 7 ° C are p r e s e n t e d in Fig. 8. T h e
curves show the s a m e general features as n o t e d
1 o 200 at p H 6.7 a n d 2 6 ° C (Fig. 3). A t the higher
> t i m e (rain)
s u p e r s a t u r a t i o n s (Figs. 8a a n d 8b) the curves
FIG. 6. Typical relaxation curves (OH/PT vs time (min)) again are S-shaped with o n l y one inflection
as a function of initial supersaturation at a constant pH point. T h e m a i n difference is seen at low su-
of 7.4. I = 0.15 M, T = 26°C, and the Ca/P ratio = 1.16. p e r s a t u r a t i o n s (curve c). A g a i n a two-stage
Initial phosphate concentrations: (a) PT = 3.28 raM, (b)
PT = 2.84 raM. Arrows point to inflection points in the precipitation is indicated, b u t the stages a p p e a r
relaxation curves. The saturation limit of HAP is reached to be better separated. This o b s e r v a t i o n sug-
at a OH/Px value of 1.15. gests that two phases form a n d grow separately
FIG. 7. SEM photograph of an apatitic phase developed at pH 7.4 and high supersaturation (see Fig. 6,
curve a). The measured Ca/P ratio (X-ray microanalysis, Link system) of the solid varies from 1.57 to 1.65.
DISCUSSION
(
.~OCP=I
J Precipitating Sequence
0.,9
The precipitating sequences observed under
0.6
the different experimental conditions are
0.4 Ct DCPD= 1
summarized in Table II.
0.2 A comparison of the information in Table
II with the solubility isotherms in Fig. 1 shows,
C/'OCP = 1 without exception, that the order within which
0.8 the solid phases precipitate is also that of in-
&"- 0.6
creasing phase stability under the specified ex-
I perimental conditions. The reversal in the
o
o,4 precipitation of the first phase at pH 6 and
~O'DCPD= I
T 0.2 26°C compared to that at pH 6.7 and 37°C,
especially, provides a striking confirmation of
this general observation. At pH 7.4 where one
O'OC P= 1
0,8 would expect DCPD to be the first precipitat-
0.6 ing phase, the much less stable ACP takes over
as precursor. The Ostwald rule of stages (19)
0.4
is thus well obeyed by precipitation from su-
0.2 O~Dc pD= 1 persaturated calcium phosphate solutions. A
satisfactory explanation of this rule, which
1(30 2 0I 0 300 clearly must be based on kinetic arguments,
-~> t i m e (rain)
is les s readily available, when one considers
FIG. 8. Typical relaxation curves at pH 6.7 and 37°C that the thermodynamically most stable phase
as a functionof supersaturation.I = 0.15 M, Ca/P = 1.16. will have the highest degree of supersaturation
Initial phosphateconcentrations:(a) PT = 4.5 mM, (b) PT
= 3.9 raM, and (c)PT = 3.0 raM. Arrowspoint to inflection in a solution supersaturated with respect to all
points in the relaxation curves. The saturation limit of possible solid modifications.
HAP is reached at a OH/PTvalue of 1.31. Stranski and Totomanov (21), Gutzow and
Toschev (22), Feenstra (11), and Van Straten
(12) provided kinetic arguments largely based
and in succession (compare with Fig. 2b). The on classical nucleation theory (23, 24) and in-
first stage just reaches the DCPD saturation duction times derived from an analysis of
boundary and we may conclude tentatively nonsteady-state nucleation to explain observed
that only DCPD grows during this stage. This precipitation sequences. Clearly growth rates
conclusion is in agreement with SEM photo- (24-26) should also be taken into account, es-
graphs taken in the early stages of the precip- pecially when at relatively high supersatura-
itation, where only DCPD crystals could be tions the decision as to which phase would
detected (Fig. 9a) in contrast to the observa- eventually develop cannot be taken in the nu-
tions at 26°C (Figs. 5a and 5b). Electron mi- cleation stage.
crographs taken during the second stage of As already mentioned it is very likely that
precipitation show both OCP and DCPD to nucleation occurs during the mixing stage.
be present (Fig. 9b). The OCP crystals appear Experimental indications for this assumption
to form only on the surface of DCPD crystals. were obtained from relaxation time studies
At intermediate supersaturation (Fig. 8, curve which showed a dependence of tR on mixing
b) only OCP could be detected whereas at high speed and concentration of titrant. If nucle-
supersaturations (curve a) HAP is the domi- ation is indeed restricted to the mixing period,
nating phase. a flash-like formation of very small critical
Journal of Colloid and Interface Science, Vol. 118, No. 2, August 1987
FIG. 9. SEM photographs taken at p H 6.7, 37°C at low supersaturation (PT = 3.0 raM). (a) Early stages
of precipitation where only D C P D crystals are detected. (b) Late stages where both D C P D and OCP are
present. The OCP crystals seem to have developed only on DCPD surfaces.
575 Journal of Colloid and Interface Science. Vol. 118,No. 2, August1987
576 VAN KEMENADE AND DE BRUYN
n
a(tR) = n +-----m" [8]
3
In Table IV we illustrate that a(tR) has a single
characteristic value for a diffusion-controlled
growth mechanism but, as expected, will have
%2
different values depending on the magnitude
of m for the other growth mechanisms. -71
In order to identify the growth mecha-
nism(s) responsible for the observed relaxation
behavior we use Eq. [7] to fit the relaxation
lo
curves to straight lines. This may be clone by -1
plotting log(a -n. &) versus log(1 - ~) for n = ~,
3, and 4 and assuming m to be constant. The -2 i i i i
latter assumption is justified in the case of the 0.2 0.4 0.6 0.8
layer growth mechanisms (mono- and poly- > -log ( l - a )
nuclear) if the concentration of the precipi- FIG. 10. A typical plot of -log(c~-". A) vs - l o g o - a)
tating species does not change by more than for a relaxation curve at pH 6.7, PT = 5.39 mM, and a
one decade in the relaxation experiment. In temperature of 26°C for different values of n: (A) n = ],
(ll) n = ~,(O) n = 4 .
our experiments the concentration varied
by less than a decade, mostly by a factor of
3to4.
A straightline is obtained for n = 4. Marked
A typical example of such plots for exper-
deviations from a straightline can be seen for
iments at pH 6.7 and 26°C is shown in Fig.
n = ] and n = ~ at low a. This suggests that
10. At this p H and medium initial supersat-
the growth rate is controlled by surface nucle-
urations (Figs. 3b and 3c) mainly one phase
ation based on the mononuclear growth
(OCP) has been identified which justifies the
model. All three plots yield straight lines with
proposed growth analysis. The extent of the
slope rn = 4 at higher a values when the term
reaction, a = OH(t)/OH(oo), was therefore
(1 - a)m in Eq. [7] dominates. The value m
calculated with respect to OCP with a = 1 at
= 4 agrees with that earlier observed by
OCP saturation.
Heughebaert and Nancollas (32) for seeded
OCP growth at pH 6.0. The growth rate of
OCP under these experimental conditions is
TABLE IV
thus best described by the relation
Points of Inflection according to Eq. [8] for Different
& = Ko~4/3( 1 - oz)4. [9]
Growth Models and Values of m
A check on the validity of Eq. [9] is obtained
Growthmodel n m a
when we note that the experimental relaxation
Diffusion ½ 1 0.25 curve has its inflection point at a(tR) = 0.25.
Polynuclear 2_
3 1 0.40 This value is consistent with that predicted by
2 0.25 Eq. [8] given the values for n and m. It should
3 0.18
be noted with reference to Table IV that other
4 0.14
5 0.12
combinations of n and m will also yield a(tR)
Mononuclear 43 1 0.57 = 0.25.
2 0.40 This analysis is based on the assumption
3 0.31 that the growing particles are reasonably
4 0.25 monodisperse. If the particles were initially
5 0.21
polydisperse the polydispersity should increase
Journal of Colloid and Interface Science, Vol. 118,No, 2, August1987
PRECIPITATION FROM CALCIUM PHOSPHATE 579
in time when growth proceeds by a mono- (the critical nucleus size decreases) with a
nuclear mechanism (see Appendix B) and the larger total surface area at a given a. As HAP
validity of the proposed analysis may be ques- forms heterogeneously on the OCP surface,
tioned. From Fig. 5 we note that samples taken this will encourage HAP growth and we can
in the later stages of relaxation are reasonably expect HAP to grow at a lower a value. The
monodisperse. On assuming, as indicated by break in the plots therefore shifts to lower a
the analysis, that the growth is proceeded by values as is indeed noted in Fig. 11. We also
a mononuclear mechanism, we conclude note that an initial slope of m = 4 is observed
therefore that the initial particle size distri- over the whole supersaturation range. No sig-
bution was very narrow. This suggests that a nificance should be attached to the value of
sudden burst of nucleation is responsible for the slopes of the straight lines beyond the break
the creation of the growing particles and in as the plotted a has been calculated on the
turn supports our contention that nucleation assumption that OCP and not HAP is the
occurred in a flash-like manner during the ex- growing solid phase.
perimental mixing step for building up the su- In Table V the dependence of the position
persaturation. of the break in the straightlines (Olbreak) o n su-
In Fig. 10 we note a sharp break in the persaturation is listed. We note that at the
straight line plot at high a (a ~ 0.75) to yield highest supersaturation values (experiments d
another straight line with a lower slope. and e), abreak ~ O~(/R) ~" 0.25. One would there-
This break is seen to move progressively to fore expect that when H A P already starts to
lower a values (higher supersaturation) with grow in the early stages of the relaxation pro-
increasing initial supersaturation. Three dif- cess, this should be reflected in an analysis of
ferent explanations may be offered for the ob- the relaxation time tR.
served break in the plots: (a) a change in OCP In the mononuclear growth model the
growth mechanism; (b) a change in m, not in growth rate is largely determined by the fre-
growth mechanism; and (c) the occurrence and quency with which a surface nucleus forms on
growth of a new phase. At low supersaturations a surface of known area. The linear growth
growth of a solid phase can be accounted for rate (dr/dr) (24, 26), assuming the particles to
by the presence of screw dislocations; however,
the observation that with increasing l~initia 1 the
break shifts toward higher supersaturation n =4/3 ~d
eliminates this possibility (explanation (a)).
The observation that m is lower after the break, c b
which is not to be expected when the super-
saturation is decreasing, argues against expla- c•1.5
nation (b) and we are left with possibility (c). 1]- __,~f
Our experimental observations suggest that
growth of HAP in the late stages of the pre-
cipitation reaction may be responsible for this
To tJ
break. When HAP starts to grow, a disconti-
nuity in the plots of Fig. 10 is likely because
a different growth rate dependence on super-
saturation is expected. This prediction is sup-
Q1 Q2 O.3 0.4 0.5
ported by experiments at higher initial super- > -log(I-a)
saturations where increasing amounts of HAP
FIG. l 1. Plots o f - l o g ( a - " " &) as a function of increasing
in the later relaxation stage have been detected. supersaturation for n = ]. (a) PT = 4.8 mM, (b) Px = 5.4
Increasing the initial supersaturation leads to mM, (c) PT = 5.6 raM, (d) PT = 5.9 mM, and (e) Px = 6.3
the formation of more and smaller particles mM.
Journal of Colloid and Interface Science. Vol. 118, No. 2, August 1987
580 VAN KEMENADE AND DE BRUYN
tR oc [Jm" r(0)] -1 6
m r(0) -1 .A-lexp(AG2D/kT). [12]
2.C
,.--, 2[ m=,o.7-*O ~
TM
~"~11.5
c,
-T
0.5
I I I I I I I I
2 4 6 8 ~10-2 .4 8 12 16 . 1 0 -2
o > -log(I-a) b > -Iog(1-~)
FIG. 13. Plots of -log[(a')-"- &] vs -log(l - a) accordingto Eq. [15] for the heterogeneousgrowth of
HAP at two initial concentrations:(a) PT = 5.4 raM, (b) PT = 6.3 mM (curvesb and e, Fig. 11, respectively).
( i ) n = 3, (O) n = 4.
formation and growth of one or more precur- netics of complex precipitating systems which
sor phases. are often difficult to identify in experimental
At pH 6.7 and 26°C OCP was observed to relaxation curves.
form at medium supersaturation largely to the
exclusion of other phases. The growth curves APPENDIX A: CALCULATION OF
(a vs t) and relaxation times (tR) obtained un- SOLUTION CONCENTRATIONS
der these conditions were analyzed semiquan-
titatively by application of classical nucleation Solution concentrations of all possible
and growth theories• The analysis reveals that complexes are calculated by considering all the
the kinetics of formation of OCP is best de- equilibria in aqueous solutions involving cal-
scribed by a flash-like nucleation step in com- cium and the various hydrolysis products of
bination with surface nucleation-controlled phosphoric acid. On taking the logarithm of
growth based on the mononuclear growth all the association constants collected in Table
model. AI, a set of linear equations is formed (34).
In the later stages of the constant p H relax- Using the activity of H ÷ defined by p H and
ation HAP forms heterogeneously on the OCP two trial concentrations, [Ca 2+] and [HPO 42-],
surface. With increasing initial supersaturation this set of linear equations is solved. The con-
the onset of HAP formation is shifted to lower centrations of all possible complexes and the
values of the extent of the reaction (a) as is ionic strength (again in an iterative manner)
also clearly illustrated by the relaxation time are computed and used in two mass balances.
analysis. This heterogeneous growth of HAP Following Holt et al. (35) better approxima-
is best described by a polynuclear growth tions for [Ca 2÷] and [HPO 2-] are found in a
model. number of cycles to minimize the mass bal-
The combination of growth curve and re- ances to within 0.1%. Activity coefficients were
laxation time analysis is shown to be an ex- calculated with the formula proposed by Dav-
cellent method to locate transitions in the ki- ies (36).
Journal of Colloid and Interface Science, Vol. 118, No. 2, August 1987
PRECIPITATION FROM CALCIUM PHOSPHATE 583
TABLE A 1
Intrinsic Association Constants (liters mole-I) and Solubility Products (mole liter-')" at 25 and 37°C
Association constants (liters mole -~)
H÷ Ca 2+ H+ Ca 2+
mechanism where the formation of a surface models the growth rate can be described by a
nucleus is assumed to be the rate-limiting step general growth equation of the form
in the addition o f a new layer to the growing
& = KaY( 1 - a) '~, [B 15]
crystal. The rate of advance of the layer orig-
inated by the surface nucleus is orders of mag- where the term a ~ represents the dependence
nitude larger than the frequency with which of growth rate on the surface area or size of
the nucleus forms. the monodisperse particles and the term
The size-dependent rate of growth is given (1 - ~)m represents the dependence on the de-
by (24-26) creasing supersaturation during the precipi-
f = 13. r 2 • d- J2D, [B6] tation. The rate law was derived for a set of
monodisperse particles. If the precipitate is
where/3r 2 represents the surface area of one
relatively polydisperse, the applicability of the
layer with thickness d, JzD is the flux of surface
rate law is limited. The development of the
nuclei, and /3 is a shape factor. Nielsen ap-
size distribution will depend on the value of
proximates the expression for J2D by
n. I f we express the growth rate in terms o f the
J2D(t) = kc(t) m [B71 linear dimension of the isotropic particles,
then
with a constant value o f exponent m, which
t:= k'r~3n-2)(c(t) - c(oo)) m'. [B16]
is justified if the concentration c(t) does not
change more than one decade during the ex- This expression shows that for n = ~, the
periment. Using Eqs. [B2], [B6], and [B7] and growth rate will not depend on the particle
a degree of reaction defined by size and the size distribution will not change
in time during polynuclear growth. For
c(t) = c(O). (1 - a(t)) [B8]
we derive mononuclear growth (n = 4) the polydispersity
C¢ ~- g m ° 0d4/3( 1 - 0£) m, will increase in time whereas during diffusion-
[B9I
controlled growth (n = ~) the particles will
where
become more monodisperse while growth
K,n = 3k/3d. r ( ~ ) . c(O) m. [BI0]
proceeds.
When the relative rates of formation of a
surface nucleus and its spreading across the ACKNOWLEDGMENTS
surface are on the same order o f magnitude, This investigation was supported financially by the
a polynuclear model (model c) is valid. For NIZO, Ede, The Netherlands. We also thank J. Pieters for
growth by polynuclear layers (24-26) his assistance in the electron microscopicanalysis.
9. Abbona, F., Madsen, H. E. L., and Boistelle, R., J. 23. Toschev, S., in "Crystal Growth, an Introduction" (P.
Cryst. Growth 74, 581 (1986). Hartman, Ed.). North-Holland, Amsterdam, 1973.
10. Feenstra, T. P., and de Bruyn, P. L., J. Phys. Chem. 24. Nielsen, A. E., in "Kinetics of Precipitation." Perga-
83, 475 (1979). mon, New York, 1964.
11. Feenstra, T. P., Doctoral thesis, Utrecht, 1980. 25. Nielsen, A. E., J. Cryst. Growth 67, 289 (1984).
12. Van Straten, H. A., Doctoral thesis, Utrecht, 1984. 26. Chernov, A. A., in "Modern Crystallography III,
13. Madsen, H. E. L., and Thorvardarson, G., J. Cryst. Crystal Growth" (A. A. Chernov, Ed.). Springer-
Growth 66, 369 (1984). Verlag, Berlin, 1984.
14. Meyer, J. L., and Eanes, E. D., Calcif. TissueRes. 25, 27. Nielsen, A. E., and S6hnel, O., J. Cryst. Growth 11,
209 (1978). 233 (1971).
15. Ftiredi-Milhofer, H., Bre6evi6, L., and Purgafi6, B., 28. S6hnel, O., J. Cryst. Growth 57, 101 (1982).
Faraday Discuss. Chem. Soc. 61, 184 (1976). 29. Gilmer, G. H., and Bennema, P., J. Appl. Phys. 43,
16. Koutsoukos, P. G., and Nancollas, G. H., J. Cryst. 1347 (1972).
Growth 53, 10 (1981). 30. Boskey, A. L., and Posner, A. S., J. Phys. Chem. 80,
17. De Rooij, J. F., Heughebaert, J. C., and Nancollas, 40 (1976).
G. H., J. ColloM Interface Sci. 100, 350 (t984). 31. Pei-Tak Cheng, and Pritzker, K. P. H., Calcif. Tissue
Int. 35, 596 (1983).
18. Vermeulen, A. C., Geus, J. W., Stol, R. J., and de
32. Heughebaert, J. C., and Nancollas, G. H., J. Phys.
Bruyn, P. L., J. Colloid Interface Sci. 51, 449
Chem. 88, 2478 (1984).
(1975).
33. Termine, J. D., Peckauskas, R. A., and Posner, A. S.,
19. Ostwald, W., Z. Phys. Chem. 22, 289 (1897).
Arch. Biochem. Biophys. 140, 318 (1970).
20. Nelson, D. G. A., Salimi, H., and Nancollas, G. H., 34. Feenstra, T. P., J. Chem. Ed. 56, 104 (1979).
J. Colloid lnterface Sci. 110, 32 (1986). 35. Holt, C., Dalgleish, D. G., and Jenness, R., Anal
21. Stranski, I. N., and Totomanov, D., Z. Phys. Chem. Biochem. 113, 154 (1981).
A 163, 399 (1933). 36. Davies, C. W., "Ion Association." Butterworth, Lon-
22. Gutzow, I., and Toschev, S., Krist. Tech. 3, 485 (1968). don, 1962.
Journal of Colloid and Interface Science, Vol. 1 IS, No. 2, August 1987