Eur. J. Mineral.

This paper has been presented at the ECROFI XIII

1996, 8, 987-996 Symposium in Sitges, Spain (June 1995)

Formation of primary fluid inclusions under influence

of the hydrodynamic environment

MANUEL PRIETO*, ANDRES PANIAGUA and CELIA MARCOS

Departamento de Geologia, Universidad de Oviedo, C/ Arias de Velasco, s/n,

33005-Oviedo, Spain

Abstract: When crystals grow under convection conditions, the hydrodynamic environment can promote the
formation of inclusions in the rear side of the crystals, downstream in relation to the fluid flow. An experimental
study of this mechanism of fluid trapping is presented, and the conditions under which hydrodynamics influences
growth behaviour are discussed. Crystal growth experiments have been carried out with aqueous solutions in a
convection system under controlled conditions of temperature, solution velocity V* and supersaturation Goo. Growth
rates of {011} faces of ADP crystals have been measured for different orientations, in relation to the solution flow
and for different values of o<×, and V*. It is concluded that hydrodynamics can produce an anisotropy in the
contribution of material to crystallographically equivalent faces. At low solution velocities and moderate supersat-
uration, the growth rate depends on the flow "impact" angle. In contrast, at high solution velocities (Vs > 3 cm.s 1
for Goo = 2.45 %) the hydrodynamic configuration has no influence on the growth kinetics.
Hydrodynamics can also play an important role in the formation of fluid inclusions during crystal growth.
However, this effect only occurs at low solution velocities, when the growth kinetics is controlled by the hydrody-
namic configuration. Crystals grown at low solution velocities (Vs < 3 cm.s"1 for Goo = 2.45 %) show large fluid
inclusions in the rear zone. At high solution velocities the mass-transfer is high enough to avoid the hydrodynamic
control of growth, and no inclusions are generated by this mechanism.

Key-words: fluid inclusions, crystal growth, hydrodynamics, supersaturation.

Introduction dividuals. The trapping of fluid may also occur

The various ways of fluid trapping during the
crystal growth process have been widely dis-
cussed in the literature (Deicha, 1955; Chernov
& Temkin, 1977; Roedder, 1984). The presence
of bubbles or foreign matter, different "capping"
processes, subsequent dissolution and growth,
and loss of stability by the crystal-growth form
resulting in dendritic growth are the most com-
mon causes promoting the formation of primary
intracrystalline inclusions. In general, all these
mechanisms lead to random and sparse distribu-
tions of inclusions within the crystalline in-
under certain hydrodynamic conditions which
bring about an uneven distribution of inclusions.
The formation of inclusions in the rear side of
the crystal, downstream in relation to the fluid
flow, has been noticed in crystals grown from
solution under poor stirring conditions (Carlson,
1958). Janssen-van Rosmalen et al. (1978)
observed the occurrence of liquid inclusions in
the rear side of KDP (KH2PO4) crystals growing
from solution at atmospheric pressure. These
authors placed crystalline seeds on a tree rotating
in a KDP mother solution, and carried out growth
experiments at different rotational speeds and at

e-mail: mprieto@ asturias.geol.uniovi.es

0935-1221/96/0008-0987 $ 2.50
© 1996 E. Schweizerbart'sche Verlagsbuchhandlung. D-70176 Stuttgart
988 M. Prieto, A. Paniagua, C. Marcos
a constant supersaturation. As a result, they only
observed the formation of inclusions when the
crystals grew at low rotational speeds. Gits-Léon
et al. (1978) studied by means of X-ray topogra-
phy the effect of stirring on the quality of potash-
alum crystals grown from solution, observing the
formation of inclusions in the rear part of the
crystals but also dislocations in the front part.
The effect was demonstrated to be especially ap-
parent at low rotational speeds.
Therefore, the hydrodynamic environment
seems to give rise to the formation of fluid inclu-
sions under certain growth conditions. At present,
however, there is no systematic study on the cou-
pling between hydrodynamics and supersatura-
tion that produces configurational effects and the
trapping of fluid. Here, we present an experimen-
tal study on the influence of both factors, on the
assumption that this mechanism of inclusion
development, although requiring very specific
crystallization conditions, could play a role in
some natural environments.
With this aim, ADP crystals were grown from
solutions in a controlled convection system. Al-
though these compounds have no natural coun-
terpart, Kihle & Johansen (1994) enumerate a set
of reasons for the use of KDP-type crystals to
investigate fluid inclusions in the laboratory. The
main reason for chosing these compounds in the
present study lies in the facility to grow large
crystals at low temperature and atmospheric pres-
sure, conditions that are necessary if one does not
want unattainable experimental designs to control
the hydrodynamic environment.
Experimental procedures
and starting materials
Crystal growth experiments were carried out in
a controlled convection system. In such a system,
a saturated solution is pumped from a thermo-
stated reservoir to a crystallization cell (Fig. 1)
Fig, l. Scheme of the crystallization cell.
which is thermostated at a lower temperature. As
a result, the solution becomes supersaturated, and
a crystalline seed, located in this cell, will grow.
Finally, the solution returns to the saturation
reservoir. The growth cell is made of glass, and
the crystal seed is held on a platinum wire. The
cell is mounted on the plate of a photomicro-
scope, in such a way that the growth rate can be
measured by taking photographs at regular inter-
vals of time.
The bulk relative supersaturation in the growth
cell is given by:
_ Co (Γi) - Co (72) (j)
Co (Γ 2 )
where c0(T\) and c0(Ti) are the saturation con­
centrations at the temperature T\ of the mother
reservoir and at the temperature Ti of the growth
cell, respectively. Obviously, in order to maintain
a constant supersaturation level, crystals of solute
must always be dissolving in the saturating ves­
sel. The pumping system works at different circu­
lation velocities which can be measured by
means of an "on-line" flow meter. The apparatus
therefore allows us to control the bulk supersat­
uration Goo, the solution velocity Vs and the crys­
tallization temperature (± 0.052C).
ADP (NH4H2PO4) crystals were grown at Ti =
252C and at their natural pH. The supersaturation
was modified by using different saturation
temperatures (T\ > 25 9 C). Moreover, for each
specific supersaturation a set of experiments was
carried out at different solution velocities. Idio-
morphic single crystals (4-5 mm long) were used
as crystal seeds. KDP-typecrystals belong to the
tetragonal space group I42d. The ideal mor­
phology consists of prismatic {010} and py­
ramidal {011} forms. Sometimes, moreover, the
crystals show high-index {Okl} forms that are as­
sociated to the so-called "tapering" effect (Dam
et al., 1986). The crystals are slightly elongated
along [001] as a consequence of the higher
growth rate of the pyramid faces. At low super-
 Formation of inclusions under hydrodynamic influence 989

saturations, the growth rate of the prismatic faces

is difficult to assess unless the experiments go on
for a long time. For this reason in the present
work the growth-rate study was devoted to the
pyramid faces.
Although the experiments were carried out
over a wide range of solution velocities, the flow
in the growth cell was always laminar. This fact
was checked by means of flow visualization
experiments. Small amounts of glycerine were
added to the solution to trace the bulk flow in the
growth cell as well as the "eddy" zone formed
behind the crystal, around the face placed in the
"shade position" in relation to the flow.
After the growth process, the crystals were re-
covered from the cell and characterized by opti-
cal microscopy and scanning electron micro-
scopy (SEM). The volume fraction Vv of
inclusions (volume of inclusions per crystal
volume unit) was determined from surface meas-
urements by a stereological method (Weibel,
1989). Regularly spaced sections of the crystals
were observed by SEM and for each section /,
0.25
the surface fraction (AAM of inclusions (relative
area of inclusions per unit area of the section)
was determined by image processing. Finally, the
volume fraction of inclusions was estimated from
0.20-1 0
0.15
the surface fractions corresponding to a given
crystal, according to the expression: E 0.10 V s (cm.s- 1 )
N
E
Vv = ÄA = {\IN)Σ{AA)i, (2) <P 0.05^
i=i
where N is the number of crystal sections con­ 0.00
sidered. This equation states that the average o 1
areal fraction AA, determined on sections through Fig. 2. Influence of the solution velocity on the growth
a volume, represents an estimate of the volume rate of faces {011} of ADP crystals (Squares, circles,
fraction Vv of the phase under investigation. The and triangles represent faces in position 1, 2, and 3,
statistical error in the estimation of Vv depends respectively), a) α<×> = 2.45 %. b) Goo = 1.71 %. c) Goo =
on the number of interceptions of inclusions ob­ 1.2 %.
served (DeHoff & Rhines, 1968), and will be
shown graphically for each specific experiment.
is in the "shade position" in relation to the_bulk
stream. As the pyramid faces (011) and (Oil) of
ADP crystals are nearly perpendicular (89.609),
Hydrodynamics and growth kinetics the three orientations can be studied simul-
taneously by arranging the crystal as shown in
Growth rate measurements Fig. 1. For each specific value of Goo and Vis, the
growth rates RGI, RG2 and RG3 were measured,
Linear growth rates of pyramid {011} faces of corresponding to orientations 1, 2 and 3. The
ADP were measured for different values of Goo results are shown in Fig. 2. In these plots, each
and Vs. Measurements were carried out for three point represents the mean value of 5 experimen-
different orientations in relation to the flow. tal runs. Dispersions from the mean values are
Orientations 1 and 2 are, respectively, perpendic­ shown in the figure by means of error bars .
ular and parallel to the flow, while orientation 3 Fig. 2a represents the dependence of the
990 M. Prieto, A. Paniagua, C. Marcos
growth rate on the solution velocity at a constant
bulk supersaturation (Goo = 2.45 %). At low so­
lution velocities the growth rate depends on the
face orientation, according to the ranking RGI >
RG2 > RG3. However, as Vs increases, the values
of RG for the three orientations tend to be equal.
Moreover, at high solution velocities, the curves
RG-VS show an asymptotic trend: the growth rate
increases with the solution velocity up to a cer­
tain level, after which further increases in Vs have
no important effect on RG.
The results are analogous at a lower supersat­
uration (Goo =1.71 %), but in this case the curves
corresponding to RGI, RG2 and RG3 are closer
together (Fig. 2b). Finally, at a very low super-
saturation (Goo = 1.20 %), it is impossible to dis­
tinguish the three curves (Fig. 2c), i.e., the width
of the range RGI-RG3 decreases as the supersatu­
ration decreases. Moreover, by comparing Fig. 2a,
2b and 2c, it is clear that the asymptotic value of
RG decreases as the bulk supersaturation
decreases.
Discussion: growth rate and flow configura­
tion
We can conclude from Fig. 2 that, when a
crystal grows under convection conditions, the
hydrodynamic configuration can produce an ani-
sotropy in the contribution of material on crys-
tallographically equivalent faces. At low solution
velocities and moderate supersaturations, the
flow "impact" angle has an important bearing on
the linear growth rate. In contrast, at high solu­
tion velocities the hydrodynamic configuration
seems to have no influence on the growth kinet­
ics.
These results are not surprising. Crystal growth
from solution is a complex process that involves
several interrelated subprocesses. First of all,
crystallization involves the transport of solute
from the solution bulk to the crystal interface.
However, unlike other mass-transfer processes
where only the volume mass-transfer determines
the rate of the process, in crystal growth there are
surface-diffusion phenomena that can be rate-
determining. In a general way, the linear growth
rate RG of a crystalline face depends on the sur­
face roughness or surface-entropy factor α
(Bourne & Davey, 1976; Bennema & van der
Eerden, 1977), the stirring or solution velocity Vs,
and the bulk supersaturation Goo. For specific
values of α and Goo, it has been proved that with
the increase of the solution velocity, the growth
rate increases up to approaching a certain con­
stant value at high Vs values (Mullin & Garside,
1967; Garside, 1971; Rodriguez-Clemente, 1976;
Garside, 1991). This means that at solution
velocities higher than a certain level, the growth
is no longer controlled by the volume transport
but rather by the surface processes. Moreover, the
value of the solution velocity required in order to
reach the asymptotic level of constant growth
rate increases as the supersaturation increases.
Both effects are clearly shown in Fig. 2.
Obviously, hydrodynamics will only be relevant
for growth phenomena at low solution velocities,
when convection is the rate-controlling process.
For this reason RGU RG2 and RG3 tend to con­
verge as the solution velocity approaches the
asymptotic level, and the process gradually
becomes surface-integration controlled. For a
specific solution velocity, the higher the super-
saturation, the higher the effect of the hydrody­
namic environment on the growth kinetics is (the
higher the difference between RGI, RG2 and RG3)-
At very low supersaturation (Fig. 2c), the growth
is controlled by the surface processes, even at
low solution velocities, and the face orientation
has no influence on the growth rate.
The dependence of the growth rate on the
flow-impact angle at low solution velocities can
be explained by boundary-layer effects. The in­
corporation of growth units to the solid involves
a decrease of the solute concentration near the
interface, and the subsequent development of a
"concentration boundary layer" of thickness bc
around the crystal surface. Outside the boundary
layer the fluid is homogeneous in concentration,
but within the layer there is a concentration
decrease from the bulk value Coo to an interface
value a. Therefore, one must distinguish between
bulk supersaturation Goo and interface supersatu­
ration G/.
The thickness of the concentration film de­
pends on the degree of agitation of the system,
and can be formalized for simple flow configu­
rations by defining another boundary layer, viz.
the "hydrodynamic or velocity boundary layer".
This is a thin region near the solid-liquid inter­
face where the flow becomes reduced owing to
the influence of the surface, i.e., where the solu­
tion velocity undergoes a continuous change
from its value Vs in the solution bulk to stagna­
tion. The thickness 0// of this layer depends on
both the solution viscosity and the specific flow
configuration around the crystal surface. So, as
 Formation of inclusions under hydrodynamic influence 991

Fig. 3. Hydrodynamic boundary layers. Configurations for flow (a) perpendicular and (b) parallel to the crystal
interface.

shown in Fig. 3a, when a laminar fluid flow
impinges perpendicularly to a crystal interface
the value 8// is uniform, and is given by
(Rosenberger, 1979):
λ l/2
8//= 2.4
a (3)
where v is the kinematic viscosity and a is a con­
stant.
The situation is similar when the flow im­
pinges parallel to the crystal face, but in this case
the thickness of the hydrodynamic boundary
layer is not uniform. It increases with the
distance x, downstream, from the leading edge
( F i g . 3 b ) , a c c o r d i n g to the e x p r e s s i o n
(Rosenberger, 1979):
x l/2
8/f = 3 vx (4)
Vs
For the simple configurations described above,
the relative magnitudes of 8// and 8 c may be
roughly estimated from (Mullin, 1993)
ç / \l/3
OH
(5)
D
which is the ratio of the kinematic viscosity to
the diffusivity D of the solute in the solution. For
aqueous solutions v/D » 1 , which means that 8//
exceeds 8 c considerably. From equations (3) to
(5), expressions for the value of the boundary-
layer thickness can be deduced. So, for a face
with orientation 2, parallel to the fluid flow, Sc
decreases with xl/2 from the leading edge:
fDWr / Λ1/2
8c = 3 (6)
Vs
Obviously, when the growth process is control­
led by the volume mass-transfer, a thick concen­
tration boundary layer involves a slow growth
process, and vice versa, i.e.:
RG oc 1/8C. (7)
 992 M. Prieto, A. Paniagua, C. Marcos
Fig. 4. Flow configuration in the proximity of the rear
side of a crystal (Vs = 1.5 cm.s 1 ).
It is clear from equations (3) to (6) that both
3H and ô c depend on the specific flow configura-
tion around the crystal surface. These values can
be predicted theoretically if the nutrient flow is
fully understood, as has been shown for faces
with orientations 1 and 2. For example, for KDP
crystals in saturated solution at 36 2 C, D =
7.8 × 10-10 m 2 s - l ; v = L 1 1 3 × 1 0 -6 m 2 s - l 5 v /£)
1427.17, and ô///ôc - 11.26. Therefore, for Vs =
0.25 cm.s 1 , at 1 mm from the leading edge of a
face parallel to the flow, ô// ~ 2002 µm and 5C ~
178 µm. These values decrease as the solution
velocity increases, being 67/ ~ 409 µm and ô c ~
36 µm at Vs = 6 cm.s 1 , which means an impor-
tant reduction in the mass-transfer resistance.
The value of ô c for a face in the "shade"
position is not so easy to calculate. Behind the
crystal, at low solution velocities, a large, stable
and closed eddy zone is present. As the solution
velocity increases, the eddy zone becomes open,
4. mm
Fig. 6. Fluid inclusions in the rear side of a crystal of ADP (GOO = 1.7 %; Vs = 1.5 cm.s"1).
Fig. 5. Fluid inclusions in a crystal of ADP. (Ooo = 2.45 %;
Vs= 1.5 cm.s" ). The contour of the crystalline seed can
be observed in the center of the crystal.
and the eddies become free. As an illustration,
Fig. 4 shows this region as observed in a flow
visualization experiment. Obviously, a formal
expression of 8// for such a configuration cannot
be calculated. Anyway, the thickness of the
concentration boundary layer seems to depend
= on the face orientation, according to the ranking
ôc/ < &c2 < ôcj, and consequently, the ranking
RGI > RG2 > RG3 is to be expected. This fact
could give rise to an obvious morphologic effect:
homologous faces can show unequal develop-
ments as a result of their orientation in relation
to the fluid flow.
Generation of fluid inclusions
Experimental findings
The u n e q u a l d e v e l o p m e n t of crystal-
lographically equivalent faces is not the only
 Formation of inclusions under hydrodynamic influence 993

Fig. 7. Development of downward macrosteps on a face

(010) parallel to the flow.
50-

^ 40« %
morphologic effect of the hydrodynamic environ­ -| o„=2.45%
ment. ADP crystals grown at low solution veloci­ "5
ties show large fluid inclusions in the rear zone, ^,30-
(O
beyond a critical point on the surface in relation
to the fluid flow. The front part of the crystals is
§ 20-
'5>
X
free of this kind of inclusion. The phenomenon
is shown in Fig. 5 and 6, which also display the
arrangement of the crystals during the growth
•§ io-

process.
A more detailed observation of Fig. 5 enables
us to distinguish two different types of inclu­
sions: the solitary inclusion labeled "a", and the
bunch labelled "b". The inclusion "a" is shallow,
unfolds parallel to a prismatic face, and has a
large surface development. Generally, this kind
of inclusion is associated with downward macro-
steps, which develop downstream on the {010}
faces (Fig. 7).
Type "b" inclusions are situated in the inner
part of the rear half of the crystals. The arrange­
ment of these inclusions seems to retain a certain
parallelism to the {011} faces. They can form
parallel, eventually interconnected, sets (Fig. 6),
or display more irregular shapes (Fig. 5). The
shape of these inclusions, as well as their size,
depends on the growth conditions, i.e., on both
solution velocity and super saturation.
The influence of both parameters on the rela­
tive amount of inclusions is shown in Fig. 8,
where the ordinate refers to the volume fraction
of inclusions Vγ in the rear half of the crystal. As
shown in Fig. 8a, at a constant solution velocity
(Vs = 1.2 cm.s-1) , the volume fraction of in­
clusions increases as supersaturation increases up
to a limiting value, nearly 50 vol. %. For a con­
stant supersaturation (α<×, = 2.45 %), the amount
oJ— ^ F — i I —-*F—^
X
i *•* • • 1
a
• 1
0 1 2 3 4 5
Vs(cm.s-1)
Fig. 8. Volume fractions of inclusions of ADP crystals.
a) Influence of the supersaturation (Vs = 1 . 2 cm.s"1).
b) Influence of the solution velocity (Goo = 2.45 %).
of inclusions decreases as the solution velocity
increases (Fig. 8b), until at high solution veloci-
ties (Vs > 3 cm.s"1), the volume of inclusions be-
comes negligible.
It is worth noting that no correlation is made
in Fig. 8 between RG and the volume fraction of
inclusions. This is because growth rate measure-
ments were carried out in separate experiments
and for a different arrangement of the crystals
(see Fig. 1 and 5). However, some additional
conclusions can be drawn by comparing Fig. 2a
and 8b with the same supersaturation (Goo =
2.45 %). Fig. 2a indicates that for Vs > 3 cm.s-1
the growth rates approach the asymptotic level
(RGI = RG2 = RG3), and the volume fraction of
inclusions becomes negligible. In contrast, at
low solution velocities RGU RG2, and RG3 differ
and Vv takes high values. Moreover, from Fig. 2c
and 8a, it is clearly seen that at Vs = 1.2 cm.s 1
994 M. Prieto, A. Paniagua, C. Marcos
and Goo = 1.2 % the volume fraction of inclusions
is negligible and RGI = RG2 = RG3. That is the
reverse for higher supersaturations (Fig. 2a
and 2b) where RGI > Ra > RG3 and Vγ takes
values approaching 45 vol. %.
One can therefore conclude that the coupling
between low solution velocity and moderate
supersaturation promotes the formation of fluid
inclusions. The effect is so determining that,
under suitable conditions, the hydrodynamic con­
figuration may even lead to the formation of open
holes.
Discussion: hydrodynamic mechanisms of in­
clusion formation
Hydrodynamics seems to play an important
role on the formation of fluid inclusions during
crystal growth. However, this effect is only
important under certain conditions, when the
growth process is controlled by volume mass-
transfer. This is clear from Fig. 2a and 8b, which
show the influence of the solution velocity on
both the amount of inclusions and the growth
rate: as Vs increases, the growth kinetics becomes
controlled by the surface processes, and the
influence of the hydrodynamic environment
becomes negligible, whereas at low solution
velocities the growth kinetics is mainly deter­
mined by the volume transport, and two different
mechanisms of inclusion formation seem to
operate.
Inclusions of type "a" develop on a face par­
allel to the flow of solution. For such an arrange­
ment, both 0// and Oc increase downstream (Eqs. 4
and 6), and the interface supersaturation
decreases with JΓ1^2 from the leading edge. This
implies that growth steps generated at the leading
edge decrease their advancement rate as they pro­
gress downstream on the surface. As a con­
s e q u e n c e , the steps will tend to pile up
downstream, beyond a critical point on the sur­
face. This could explain the presence of large
downward macrosteps on the faces parallel to the
flow (Fig. 7). At the same time, these macrosteps
can give rise to the development of type "a"
inclusions on these faces. A downward macro-
step produces a sort of "shade" effect for the
stream lines, which promotes the establishing of
a small "dead fluid" region in the vicinity of the
step rise. During growth, the nutrition of the step
rise is better in the upper part than in the lower
part, and a layer overhanging the growing face
will appear. This overhang will give rise to a
flat inclusion parallel to the growing face, i.e., to
a type "a" inclusion.
Although both kinds of inclusions are formed
under the same growth conditions, type "b"
inclusions seem to have a different genetic mech­
anism. According to Janssen-van Rosmalen et al.
(1978) this kind of inclusion is due to solute-
concentration depletion in the proximity of
the pyramidal terminations of the growing crys­
tal. Behind the crystal, at low solution velocities,
a large and stable eddy is present. Since the eddy
is closed, exhaustion of its solution occurs, which
will give rise to the formation of fluid inclusions
along the rear side of the crystal. The situation
will be similar at intermediate solution velocities,
when the eddy zone becomes open, but in this
case the concentration gradient in the vicinity of
the rear side will be smaller, and the effect will
be less pronounced. Finally, at high solution
velocities the mass transfer will be high enough
to avoid the hydrodynamic control of growth,
and no inclusions will be generated by this mech­
anism.
It is worth noting that the control of growth by
the volume mass-transfer does not mean that the
surface phenomena are not operative. Crystal
growth depends on the surface roughness, and
always involves the surface diffusion of growth
units until they can enter a growth kink on the
crystal surface. However, we can only speak of
kinetic control by the surface processes when the
transport of solute molecules from the bulk fluid
to the crystal interface is high enough to supply
all the incorporated growth units. The concentra­
tion at the crystal surface is then assumed equal
to that of the bulk fluid, and hydrodynamics
becomes irrelevant to the growth process.
Conclusions and geological implications
In this paper we have outlined two different
mechanisms of primary fluid inclusion formation
and discussed the factors which enable the hy­
drodynamic environment to influence the growth
kinetics. The flow configuration seems to
influence the formation of inclusions and other
morphologic effects during crystal growth from
fluid phases. The conditions under which these
effects appear have been established from
growth-rate measurements on faces with different
flow "impact" angles. The effect is so critical
that, under suitable conditions, the hydrodynamic
 Formation of inclusions under hydrodynamic influence
environment may even lead to the formation of
open holes in the rear side of the crystals.
Our results are obviously limited to laboratory
experiments with KDP-type crystals, but their
high reproducibility allows us to suppose that
these phenomena can be effective over a wide
range of substances. If a similar asymmetrical
distribution of inclusions is observed on crystals
whose growth conditions are unknown, for
example on natural crystals, a flux direction
could be suggested.
Although the experiments described in this
paper were carried out under specific steady-state
conditions, the conclusions can be extended to
natural systems with configurational similarities,
regardless of the specific mode of supply of min-
eralizing material. In a general way, it is easy to
imagine flow configurations in natural environ-
ments that involve unidirectional mass-transfer:
advection and directional diffusion through
porous media, convection in magmatic fluids,
flow in hydrothermal veins or in low-temperature
hydrothermal systems in which the flow direction
is controlled by the orientation and permeability
of the strata, etc.. Examples of morphological
phenomena induced by the flow direction are
relatively common in the literature, but only
scarce attention has been paid to the formation of
inclusions by this mechanism. Grigor'ev (1965)
has reviewed different cases of non-uniform crys-
tal growth by a unidirectional supply of the min-
eralizing material. For instance, trough-like crys-
tals of beryl in some pegmatite veins are
non-uniformly developed due to the direction of
the solution flow: the growth was inhibited in the
rear side of the crystals, where the solutions were
i m p o v e r i s h e d in c o n c e n t r a t i o n . Quartz
phenocrysts in rhyolites are frequently oriented
subparallel to the magmatic flow, and in such
cases show inclusions and open holes systemati-
cally situated on the rear side.
In recent years, an important theoretical frame-
work has been developed to characterize the
mass flow in natural systems (Foster & Smith,
1990; Tait & Jaupart, 1990; Nicholls, 1990).
Numerical modeling carried out within such a
framework provides a rigorous basis for estab-
lishing the likelihood for a postulated flow
system to develop. However, these studies are
u s u a l l y d e v o t e d to m e s o - or large-scale
processes. Micro-scale studies, such as the study
of the formation of boundary layers near the in-
terface of crystallizing minerals, are rare.
Roedder (1984) has reviewed the role of the
995
boundary-layer effects in the formation of both
aqueous inclusions and silicate melt inclusions.
The "ability" of a mineralizing fluid to form con-
centration boundary layers depends on both the
viscosity and the diffusion coefficient of the
growth units at the crystallization temperature. In
low-temperature aqueous systems, considerable
concentration gradients can form near the crystal
interface. However, such effects have not been
studied for hydrothermal crystallization at
temperatures where the diffusion is fast. Diffu-
sion rates in viscous fluids such as silicate melts
are much lower than in aqueous solutions, so the
boundary-layer effects might be expected to be
more important. Furthermore, as the ability to
form boundary layers increases with the viscos-
ity, boundary layers in rhyolitic melts should be
larger than those found in basaltic melts, because
the viscosities of rhyolites are orders of magni-
tude higher.
A final problem is the degree of supersatura-
tion required for the kinetic control of the growth
process by the volume mass-transfer. There is a
tacit assumption that natural crystallization oc-
curs near equilibrium, and there remains scepti-
cism in accepting that super saturation is relevant
to natural fluids. However, there is unequivocal
evidence that supersaturated fluids can exist over
a wide variety of geological environments. For
instance, in low-temperature sedimentary en-
vironments authigenic feldspars can crystallize
with almost any state of (Al, Si) order, depending
on the supersaturation of the fluid (Kastner &
Siever, 1979). In hydrothermal environments, the
formation of veins of adularia with (Al, Si) order
ranging from high-sanidine to maximum-micro-
cline testifies to the possible range of supersatu-
ration in the parent fluid phase. There have been
numerous studies demonstrating a relationship
between morphology and undercooling degree in
igneous melts (Muncill & Lasaga, 1988). The
presence of non-equilibrium internal morpholo-
gies, like sector and oscillatory zoning (Prieto et
al., 1993), provides another argument supporting
the existence of high super saturations in natural
fluids, either aqueous or magmatic. Putnis et al.
(1995) have outlined a range of causes and re-
viewed natural environments where the existence
of supersaturated fluids has been proved.
All these arguments support the possibility of
the volume mass-transfer control of crystal-
lization in some natural environments, which is
imperative for the formation of fluid inclusions
by the influence of flow configuration. However,
996 M. Prieto, A. Paniagua, C. Marcos
the geological implications of the issues we have
raised here have yet to be fully tested, either ex-
perimentally or in the field.
Acknowledgements: This work was supported
by DGCYT (Ministry of Science and Education
of Spain), Grant PB92-0998. Drafts of this man-
uscript were much improved by the careful re-
view by R.J. Bakker, and an anonymous referee.
References
Bennema, P. & van der Eerden, J.P. (1977): Crystal
growth from solution: development in computer
simulation. / . Cryst. Growth, 42, 201-213.
Bourne, J.R. & Davey, R.J. (1976): The role of solvent-
solute interactions in determining crystal growth
mechanisms from solution. I. The surface entropy
factor. J. Cryst. Growth, 36, 278-286.
Carlson, A. (1958): The fluid mechanics of crystal
growth from solution, in "Growth and perfection of
crystals", R.H. Doremus, ed. J. Wiley and Sons,
New York, 368-373.
Chernov, A.A. & Temkin, D.E. (1977): Capture of in-
clusions in crystal growth, in "Crystal growth and
materials", E. Kaldis & H J . Schoel, eds. North
Holland, Amsterdam, 3-77.
Dam, B., Bennema, P., van Enckevort, WJ.P. (1986):
The mechanism of tapering on KDP-type crystals.
J. Cryst. Growth, 74, 118-128.
DeHoff, R. T. & Rhines, F.N. (1968): Quantitative mi-
croscopy. McGraw-Hill, New York, 404 p.
Deicha, A. (1955): Les lacunes des cristaux et leur in-
clusions fluides. Masson and Cie, Paris, 285 p.
Donaghey, L.F. (1980): Hydrodynamics of crystal
growth processes, in "Crystal growth", B. Pamplin,
ed. Pergamon Press, New York, 65-101.
Foster, C. & Smith, L. (1990): Fluid flow in tectonic
regimes, in "Short Course on fluids in tectonically-
active regimes of the continental crust", B.E.
Nesbitt, ed. Mineralogical Association of Canada,
Vancouver, 1-47.
Garside, J. (1971): The concept of effectiveness factors
in crystal growth. Chem. Eng. Sci., 26, 1425-1431.
— (1991): The role of transport processes in crystal-
lization, in "Advances in industrial crystallization",
J. Garside, R. J. Davey, A. G. Jones, eds. Butter-
worth-Heinemann, Oxford, 92-104.
Gits-Léon, S., Lefaucheux, F., Robert, M.C. (1978): Ef-
fect of stirring on crystalline quality of solution
grown crystals-case of potash alum. / . Cryst.
Growth, 44, 345-355.
Grigor'ev, D.P. (1965): Ontogeny of minerals. Israel
Program for Scientific Translations, Jerusalem, 249 p.
Janssen-van Rosmalen, R., van der Linden, W.H.,
Dobbinga, E., Visser, D. (1978): The influence of
the hydrodynamic environment on the growth and
the formation of liquid inclusions in large
potassium dihydrogen phosphate (KDP) crystals.
Kristall und Technik, 13, 17-28.
Kastner, M. & Siever, R. (1979): Low temperature
feldspars in sedimentary rocks. Amer. Jour. Sci.,
279, 435-479.
Kihle, J. & Johansen, H. (1994): Low-temperature
isothermal trapping of hydrocarbon fluid inclusions
in synthetic crystals of KH2PO4. Geochim. Cosmo-
chim.Acta, 58, 1193-1202.
Muncill, G.E. & Lasaga, A.C. (1988): Crystal-growth
kinetics in igneous systems: isothermal H2O-satu-
rated experiments and extension of a growth model
to complex silicate melts. Am. Mineral., 73, 982-
992.
Mullin, J.W. (1993): Crystallization. Butterworth-
Heinemann, Oxford, 527 p.
Mullin, J.W. & Garside, J. (1967): Crystallization of
aluminium potassium sulphate: a study in the
assesment of crystallizer design data: I: Single
crystal growth rates, II: Growth in a fluidized bed.
Trans. Inst. Chem. Engrs., 45, 285-295.
Nicholls, J. (1990): The mathematics of fluid flow and
a simple application to problems of magma trans-
port, in "Modern methods of igneous petrology:
understanding magmatic processes", J. Nicholls &
J. K. Russel, eds. Mineral. Soc. Am., Reviews in
Mineralogy, 24, 107-124.
Prieto, M., Putnis, A., Fernández-Diaz, L. (1993):
Crystallization of solid solutions from aqueous so-
lutions in a porous medium: zoning in (Ba, Sr)SO4.
Geol. Mag., 130, 289-299.
Putnis, A., Prieto, M., Fernández-Diaz, L. (1995): Fluid
supersaturation and crystallization in porous media.
Geol. Mag., 132, 1-13.
Rodriguez-Clemente, R. (1976): Crystal growth kinet-
ics of sodium chloride from solution, in "Industrial
crystallization", J.W. Mullin, ed. Plenum Press,
New York, 187-199.
Roedder, E. (1984): Fluid inclusions. Mineral. Soc.
Am., Reviews in Mineralogy, 12, 664 p.
Rosenberger, F. (1979): Fundamentals of crystal
growth I. Macroscopic equilibrium and transport
concepts. Springer-Verlag, Berlin, 530 p.
Tait, S. & Jaupart, C. (1990): Physical processes in the
evolution of magmas, in "Modern methods of ig-
neous petrology: u n d e r s t a n d i n g magmatic
processes", J. Nicholls & J.K. Russel, eds. Mineral.
Soc. Am., Reviews in Mineralogy, 24, 125-190.
Weibel, E.R. (1989): Measuring through the micro-
scope: development and evolution of stereological
methods. / . Microsc., 155, 393-403.
Received 22 September 1995
Modified version received 1 April 1996
Accepted 23 May 1996