Sie sind auf Seite 1von 10

Basin Research (2006) 18, 221–230, doi: 10.1111/j.1365-2117.2006.00290.

Sub-surface precipitation of salts in supercritical


seawater
Martin Hovland n, Tatyana Kuznetsovaw, Håkon Rueslåttenz, Bjørn Kvammew, Hans Konrad
Johnsenz, Gunnar E. Fladmark‰ and Andreas Hebachw
n
Statoil, Stavanger, Norway
wDepartment of Physics and Technology, University of Bergen, Bergen, Norway
zNumericalRocks,Trondheim, Norway
zStatoil, R&D Department, Rotvoll,Trondheim, Norway
‰Department of Mathematics, University of Bergen, Bergen, Norway

ABSTRACT
Extremely low solubility of typical seawater salts within certain supercritical sections of their
pressure^temperature composition space is a proven experimental fact. Its consequences are often
referred to as either ‘shock crystallization’ or ‘out- salting’. Our alternative model for the formation of
salt deposits hypothesizes that high temperatures and pressures characteristic for the high heat- £ow
zones of tectonically active basins may bring submarine brines into the out- salting regions and result
in the accumulation of geological- scale salt depositions.
To con¢rm the laboratory observations, molecular- scale simulations (molecular dynamics) have
been employed to study structural changes in a model seawater system where temperature increased
from ambient to near-critical and supercritical. Fluid properties and phase transition regions
extracted from the simulations were then used as input parameters for a reservoir simulator program
to model out- salting in a simple hydrothermal geological system. Both numerical simulations and
laboratory experiments con¢rm that supercritical out- salting is a viable process of geological
signi¢cance for the formation and accumulation of evaporites.We suggest two regions where
hydrothermally associated salts may be depositing today: Atlantis II Deep, in the Red Sea, and Lake
Asale, Ethiopia.

INTRODUCTION ‘circulating brines’. This problem is addressed in the pre-


sent work.
The widely accepted hypothesis for the formation of geo -
Salt formation by hydrothermal processes is also known
logically signi¢cant evaporites on Earth (mainly halite, an-
to have contributed salt volumes of geological signi¢cance
hydrite and gypsum) is the solar-induced evaporation of
(Lowell & Germanovich, 1997, in review by Warren, 1999),
seawater. There are, however, numerous paradoxes and
but the basic hydrothermal processes involved have not
unresolved problems associated with this model, as dis- been seriously addressed in this context, i.e. (1) salt preci-
cussed by Warren (1999), Wilson (2003, 2004) and Talbot
pitation when seawater attains supercritical conditions
(2004) that clearly illustrate a lack of fundamental data,
and (2) salt precipitation by submerged or buried boiling
especially from the deepest portions of the salt basins, to
of seawater. This work discusses these processes in terms
verify this evaporite hypothesis as the general model for
of laboratory experiments and theoretical molecular mod-
salt formation. The following citation from the most re-
elling for water and brines under sub- and supercritical
cent and authoritative Geological Encyclopedia (Selley,
conditions. A basin modelling study has also been carried
2005) also underlines this conundrum: ‘As the name sug-
out to study the behaviour of seawater circulating in the
gests, it was once thought that evaporites formed exclu- proximity of a shallow heated sill body in the Earth’s crust.
sively from the drying out of enclosed marine basins.This
These results and facts are discussed in relation to natural
required improbably large volumes of seawater to provide
environments where analogue processes may occur. It is
the resultant evaporites. It is now realized that many eva-
concluded that hydrothermally associated salts may be
porites actually form in sabkhas (Arabic for ‘‘salt marsh’’)
more abundant than hitherto realized, and consequently,
from the replacement of pre- existing rocks, principally
more signi¢cant from the geological viewpoint.
carbonates, by circulating brines. Evaporites should thus
more correctly be termed ‘‘replacementites.’’ ’ Selley
(2005) avoids, however, to suggest a possible origin for the Supercritical water
The critical point (CP) for distilled water is 374.15 1C and
Correspondence: Martin Hovland, Statoil, N-4035, Stavanger, 221.2 bar. Beyond this point, the physical and chemical
Norway. E-mail: mhovland@statoil.com properties of water change fundamentally; e.g. the dielec-

r 2006 The Authors. Journal compilation r 2006 Blackwell Publishing Ltd 221
M. Hovland et al.

tric constant declines from the normal-water value of 80 to low and above their CP.The researchers observed the pro -
only 2 at 250 bar and 450 1C (Tester et al., 1993). Similarly, cess through a sapphire window in the pressure chamber,
the ionic dissociation constant declines from the normal- and when the brines were passing into the supercritical re-
water value of10  14 down to10  23. In the near-critical re- gion at a temperature of 405 1C and 300 bar pressure (the
gion, the density of water changes rapidly with respect to CP of seawater), they could see a ‘cloud’ formed by the on-
both temperature and pressure and at CP it attains a value set of ‘shock crystallization’of NaCl and Na2SO4.The sud-
of 0.3 g/cm3 (Tester etal.,1993; Bellissent-Funel, 2001).This den phase transition occurred as the solubility of the salts
strongly in£uences the inter-molecular behaviour of water, declined to near-zero over a temperature range of only a
and Raman spectra of deuterated water in the supercritical few degrees C. The resulting solids were found to consist
region show only remnants of hydrogen bonding left of amorphous microscopic particles of sizes between 10
(Franck, 1976; Kohl et al., 1991). Consequently, the super- and100 mm.The reason for the observed shock crystalliza-
critical water behaves essentially as a non-polar low-den- tion is discussed below.
sity £uid, with its solvation properties resembling those of
low-polarity organic £uids. Under certain pressure/tem- Supercritical brines in nature
perature conditions, supercritical water is also unable to
dissolve common sea salts (Tester et al., 1993; Bellissent- Natural occurrences of supercritical water and associated
Funel, 2001). These dramatic deviations from ‘normal’ processes are well hidden from direct observation by hu-
water behaviour make supercritical water highly corrosive mans, deep below the Earth surface and inside the deep-
to a number of compounds, a property of importance sea hot vents (Von Damm et al., 2002; Kawada et al., 2004).
when studying deep hydrothermal systems. In the ocean, the critical pressure for seawater corresponds
to a depth of 42800 m. Given that the world average ocean
depth is more than 3000 m, there is a great potential for the
Supercritical brines
formation of supercritical water in hot and fractured ocea-
When salt is added to pure water, one more degree of free- nic crust. Deep sea drilling of oceanic crust has also shown
dom is added to the system (in accordance with the Gibbs regional porosities of up to 25% (Becker, 1999; Davis et al.,
phase rule), with the CP changing along a line connecting 2004), and consequently, a large potential for the circula-
the CP of the two pure components (salt and water). De- tion of £uids. At these depths, the exposure of seawater to
pending on the local salinity, supercritical conditions in strong heat sources; e.g. intrusive basalts of typically 800^
seawater and brines will be reached at pressures and tem- 1200 1C, will potentially cause the water to become super-
peratures that di¡er signi¢cantly from those of pure water. critical (Kawada et al., 2004; Hovland et al., 2005; M. Hov-
The onset of critical behaviour in normal seawater with land, H. Ruesltten, H.K. Johnsen, B. Kvamme & T.
salinity 3.5 wt% occurs at around 300 bar and 405 1C (Bis- Kutznetsova, submitted). A fundamental knowledge of
cho¡ and Rosenbauer, 1989). If boiling takes place, a higher the behaviour of water and salt at various temperatures
£uid column (steam and water) is necessary in order to and pressures encountered in the Earth’s crust must be
reach a su⁄cient hydrostatic pressure to attain the CP the starting point for understanding the natural processes
(Fig.1). Some experimental studies of supercritical brines associated with supercritical water.
have been reported in the literature; e.g.Tester et al. (1993),
who examined the phase behavior of synthetic brines be-
SOLUBILITY OF SALT ^ THEORETICAL
BACKGROUND
The dipolar nature of water
The ability of water to dissolve salts is due to its highly di-
polar nature, allowing the water molecule to form shell-
like structures around the dissociated salt ions. These
shells screen the ions’ charges and signi¢cantly attenuate
their Coulomb interactions with each other (the e¡ect is
not unlike the Debye shielding in plasma, although the
screening entities are electroneutral in case of hydration).
The dipolar moment of water and its orientational selec-
Fig. 1. Location of possible supercritical and boiling tivity is directly re£ected in the strength of the hydrogen-
environments according to depth (pressure): Far left: fresh water bonded network.
and steam column above a deep-lying heat source (4374 1C) in a
lacustrine situation (e.g. on Iceland). Depth corresponding to the The SSIP and CIP regimes
critical point exceeds 3 km due to the low density. In the ocean,
the potential critical point (PCP) lies at around 2800 m (pressure At low to moderate temperatures, the ions are e⁄ciently
of about 300 bar), provided that a su⁄cient heat source screened from each other by water molecules. This is
(4405 1C) is located under the sea£oor. known as the solvent- separated ion pair (SSIP) regime.

222 r 2006 The Authors. Journal compilation r 2006 Blackwell Publishing Ltd, Basin Research, 18, 221^230
Sub-surface precipitation of salts in supercritical seawater

390 bars, there will be two intersection lines. These lines


are called lower and upper solidus lines. At 250 bars, they
are located at 450 and 720 1C, respectively.Temperatures in
between the two solidus lines will correspond to the region
where the only coexisting phases are vapour and solid salt.
At lower temperatures, the NaCl^water system is a mixture
of liquid and vapour phases.

Solubility of NaCl around the CP


The solubility of NaCl in the liquid close to the CP could
be as high as 40 wt%, and consequently, the liquid phase
can contain a large amount of salt. When the temperature
of the salt- saturated £uid rises above the lower solidus line
(450 1C at 250 bars), the £uid can only exist as either vapour
or solid salt. Given the extremely low solubility of salt in
vapour, almost all the salt previously dissolved in the water
will fall out. This is the so - called ‘out- salting’ or ‘shock
crystallization’. Temperatures above the upper solidus line
Fig. 2. P^T projection of the monovariant solid^liquid^vapour
(or even lower in case of low salt concentrations) will be
saturation curve (solidus) for the NaCl^water system based on
Hodes et al. (2004a). (Note that pressure increases downwards, as
able to support a liquid phase again, resulting in sharply
in earth and marine systems.) Arrows indicate the two points of increased salt solubility.We should note here, however, that
intersection with the section at 250 bar (lower and upper solidus although dissolved in water, this salt remains undisso -
boundaries). See also Fig. 6 and text for further details and ciated, a condition that drastically a¡ects its properties.
discussion.

As the temperature increases, the structure in the solution SALT CROSSOVER BEHAVIOUR AND
will gradually decay. The ions’ Coulomb forces are much BRINE PROPERTIES: MOLECULAR
stronger than the forces acting between the water dipoles; SIMULATIONS
thus, the electrostatic attraction between the ions will per-
Model system and parameters
sist at temperatures where the hydrogen-bond network
has already deteriorated to such a degree that it is no long- State- of-the art models describing atomistic-level inter-
er able to shield the salt ions from each other.The resulting actions between water molecules have been compiled in
e¡ect is that the salts are no longer soluble in water (which recent reviews by Nieto -Draghi et al. (2003) and Guillot
now is in a vapour state). When the temperatures are ele- (2002). The SPC/E model is chosen for describing the
vated even higher, a regime will emerge where the ions are water^water interactions in our simulations, and the mod-
soluble again as electroneutral ion pairs surrounded by el by Smith and Dang (1994) is chosen for the ion descrip-
water molecules.This regime is called the contact ion pair tions. Cross interactions were calculated using the
(CIP) regime (undissociated salt, the upper solidus in Lorenz^Berthelot mixing rules (Allen & Tildeslay, 1987).
Fig. 2). Between these two regimes (SSIP and CIP) of salt Our model system consists of 512 water molecules and six
solubility, there is a two -phase equilibrium region where molecules of sodium chloride, corresponding approxi-
solid salt coexists with £uid containing very small mately to 3.7 wt% salt. Simulations were performed in a
amounts of salt: the solid- £uid ‘out- salting’ region. closed system at constant pressure and temperature. The
thermostat parameters in the Nose^Hoover formulation
were ¢xed at 100 femtosecond (fs). The pressure control
Lower and upper solidus line
parameter was set to 8000 fs, and for the rotational degrees
Consider the P^T projection of the monovariant solid^ of freedom we used an implicit quaternion scheme (Finc-
liquid^vapour (S^L^V) saturation curve (‘‘solidus’’) for ham, 1992). The long-range electrostatic forces were
NaCl^water in Fig. 2 (Hodes et al., 2004b). The S^L^V handled using the Ewald summation approach (Allen &
curve describes a surface in the pressure^temperature^ Tildesley, 1987).
salt concentration (P^T^x) space where three phases, solid We have found our simulated results to be in excellent
salt, salt- saturated liquid and salt- saturated vapour can agreement with the previous numerical studies of similar
coexist. This surface separates several regions where only model systems (Lyubartsev & Laaksonen, 1996; Driesner
two phases coexist. If we cut the S^L^V curve horizontally et al., 1998; Chialvo & Simonson, 2003) and the experimen-
(at a constant pressure), we will obtain a temperature^ con- tal investigations of the solubility regions for sodium
centration (T^x) section. Given the shape of the S^L^V chloride in water as functions of temperature and pressure
curve in Fig. 2, one can see that at any pressure below (Hodes et al., 2004a being the most recent).

r 2006 The Authors. Journal compilation r 2006 Blackwell Publishing Ltd, Basin Research, 18, 221^230 223
M. Hovland et al.

Pair-correlation functions and hydrogen


bonding
The existence of the hydrogen-bonded structure dis-
cussed in the previous section can readily be inferred from
the so - called pair- correlation functions. A pair correla-
tion function describes the probability of ¢nding a pair of
particles at a certain distance from each other, relative to
the probability expected in a random distribution of parti-
cles at the same density. Pair- correlation functions can be
directly deduced from experiments probing £uid struc-
tures (see for example neutron scattering results: Soper,
1986; Soper & Phillips, 1986) or computed by means of
molecular simulations. The average number of water
neighbours around a speci¢c ion (the aforementioned co -
ordination shell) can be estimated by integrating the pair-
correlation function. As the oxygen atom dominates the
water molecule, a fair approximation of the number of
water molecules, na, around a speci¢c ion, a, can be found
by

naoxygenwater ðr1 ; r2 Þ
Zr2
ð1Þ
¼ 4prwater gaoxygenwater ðrÞr2 dr
Fig. 3. Pair- correlation functions for oxygen^ oxygen (a) and
r1
oxygen^hydrogen (b) in brine at di¡erent temperatures.

where r1 and r2 de¢ne the boundaries around ion a. Inte-


gration from r1 5 0 to r2 corresponding to the ¢rst mini-
mum of the pair- correlation function will yield the
average number of water molecules in the ¢rst coordina-
tion shell around the ion.
The numerically simulated pair correlations for oxygens
and hydrogens in water as functions of temperature at
300 bars are shown in Fig. 3a and b. The centre of mass in
the water molecule is very close to the geometric centre of
oxygen. In the absence of any orientational preference
among the water molecules, one would expect to ¢nd the
¢rst peaks of the oxygen^hydrogen pair- correlations at a
separation close to the size of the oxygen atom. The fact
that the ¢rst peak in the oxygen^hydrogen pair correlation Fig. 4. Sodium^ chloride ions (Na1^Cl  ) pair- correlation
function lies closer than the corresponding peak in the functions for sub- and supercritical temperatures at 300 bar.The
oxygen^ oxygen pair correlation function is the evidence inset shows higher temperatures.
of the hydrogen-bonded structure (preferential orienta-
tions of oxygen and hydrogen molecules). As the tempera-
ture increases, two phenomena emerge: ¢rstly, the
decreasing height of the oxygen^hydrogen peak is directly
related to the strength of hydrogen bonding. Secondly, we for the transition to a di¡erent solvation structure is found
should note the lack of any longer-ranging structures at in the oxygen^Na1 and oxygen^Cl  pair- correlation
high temperatures, meaning that the hydrogen-bond net- functions (Fig. 5a and b). One can ¢nd the average number
work deteriorates. In this situation, water will essentially of oxygen molecules surrounding each ion by integrating
behave as a non-polar £uid and loses the ability to dissolve the pair correlation function up to its ¢rst minimum and
salts as ions (indicated above). then multiplying by the average solution number density.
The behaviour of the Na1^Cl  ion pair in terms of The oxygen coordination numbers around the small Na1
their pair- correlation functions at temperatures outside ion decreased from 5.5 to 3.8, whereas the corresponding
the two -phase (out- salting) region is described in Fig. 4. number for Cl  increased from 7.3 to 12.3 for the same
Particularly important are the dramatic changes in the range of temperatures. From 980 K and upwards, the indi-
heights of the ¢rst (CIP) and second (SSIP) peak of the vidual coordination numbers for both Na1 and Cl  de-
Na1^Cl  correlation function. Supplementary evidence crease, as does the density of water.

224 r 2006 The Authors. Journal compilation r 2006 Blackwell Publishing Ltd, Basin Research, 18, 221^230
Sub-surface precipitation of salts in supercritical seawater

SIMULATION OF OUT-SALTING IN
SEDIMENT BASINS
The ‘ATHENA’ reservoir simulator
The reservoir simulation program ATHENA (see detailed
description in Garrido et al., 2004) was applied to model
seawater £ow and out- salting e¡ects in the supercritical
(SC) region of a simpli¢ed hypothetical sediment basin.
Primary input parameters to the ATHENA simulator are
temperature, water pressure and molar masses for each
£uid component, including dissolved salts. ATHENA uses
¢nite-volume space discretization and a backward Euler
scheme for time to determine the £uid pressure and tem-
perature. The mass-balance equations are solved using an
implicit scheme.
The objectives of the basin modelling were to:

(1) provide a ¢rst approximation of the produced salt vo-


lumes (anhydrite and halite) in a very simple hydro -
thermal system;
(2) establish constraints on the geometry, mass balance
and circulation processes in the system; and
(3) suggest a three-dimensional deposition model for the
Fig. 5. Oxygen^ sodium (a) and oxygen^ chloride (b) pair- salts produced by boiling and passage of brines and
correlation functions for 300 bar at various temperatures. seawater through the CP.

The basin model used


The geological basin model used for this simulation is to
model out- salting after a hot sill has been injected into se-
diments at high pressure, i.e., more than 300 bars, total sta-
tic pressure. For this purpose, we consider a static, hot sill
at a constant temperature of 1400 K and of unspeci¢ed
thickness (more than about 10 m). The following simple
one-dimensional geological model has the following geo -
logical constraints:

 Water depth 3000 m (corresponding to a pressure of


Fig. 6. Density of water and brine as a function of temperature 30 MPa).
along the 300 bar isobar. Legend: 1 5 IAPWS-95 (Wagner &  A hot sill at 1400 K, buried 3220 m below sea level.
Pruss, 2002); 2 5 EoS from Phillips et al. (1981); 3 5 solidus
 Three layers of sediments resting upon the sill: Two
boundaries; 4 5 MD results for pure water; and 5 5 MD results
for brine.The two-phase region (or ‘out- salting region’) is
sand layers with a shale layer ‘sandwiched’ in between.
between the two vertical lines (3). Thicknesses: upper 20 m, sand layer, 100 m shale and
lower sand layer, 100 m thick.
 The sand was assigned a porosity of f 5 0.35 and per-
meability of k 5 230 mD.The shale was assigned a por-
Density of brine as a function of temperature
osity and a permeability of f 5 0.1025 and k 5 2.9 mD.
The density of brine as a function of temperature at  The boundary conditions included a constant pres-
300 bars, determined by means of constant T- constant P sure of 32 MPa at the sill intrusion level, and 30 MPa
molecular dynamics, is shown in Fig. 6. Also shown are at the sea£oor.
the corresponding data from Phillips et al. (1981) and the  The simulation added or removed brine from the top
out- salting region (solidus lines from Hodes et al., 2004b). or bottom to keep the pressure gradient constant.This
The modelled densities agree very well with the experi- resulted in an upward £uid £ux.
mental values. This information, combined with the cor-  Temperatures were maintained ¢xed at roughly1400 K
rect behaviour of the pair- correlation functions, allowed at the sill, and 2801K on the sea£oor.
us to use the results of the simulations as input parameters  The simulation model was subdivided into cells of 5 m
for the ‘ATHENA’ reservoir simulation program. height.

r 2006 The Authors. Journal compilation r 2006 Blackwell Publishing Ltd, Basin Research, 18, 221^230 225
M. Hovland et al.

 As the model was treated as a single-phase model,


there was no need for relative permeability and capil-
lary pressure data.
 The basin model allows only vertical £uid £ow.

Input parameters to the modelling and


simulation
The ‘ATHENA’ reservoir simulator needs densities, speci-
¢c enthalpies and bulk viscosities of the £uid phases as in-
put parameters to calculate the mass and heat balances.
The values are tabulated, and the simulator performs line- Fig. 7. Simulated injection of supercritical brine into a three-
ar interpolations to ¢nd the intermediary values. Pure layer reservoir: temperature pro¢le for di¡erent combinations of
water was used in the ¢rst simulation run, and the results scaling factors for porosities and permeabilities relative to
were analysed for consistency, and subsequently compared original sandstone values. Legend: 1 5 porosity of 10  3 and
with those for brine £ow to study the in£uence of out- salt- permeability of 10  6; 2 5 porosity of 1 and permeability of 10  6;
ing. The input properties for pure water were obtained 3 5 pure water.
from equations presented by the International Association
for the Properties of Water and Steam (aka IAPWS-95 for-
mulation; Wagner and Preuss, 2002), and were tabulated at
temperature steps of 50 K.
The brine enthalpy and density data were extracted
from the numerical simulations that also yielded the pair
correlation data. It should be noted, however, that
although the numerical simulations provide brine densi-
ties that can be used directly, only the residual enthalpy gi-
ven by the molecular dynamics can be related to the
experimental values in a straightforward manner. To ¢nd
the brine- speci¢c enthalpies required by the ‘ATHENA’
simulator, we have calculated the enthalpy di¡erences be- Fig. 8. Simulated injection of supercritical brine into a three-
tween simulated pure water and brine, and adjusted the layer reservoir: pressure pro¢le for di¡erent combinations of
IAPWS values accordingly. Pure water viscosity was used scaling factors for porosities and permeabilities relative to
for both water and brine. original sandstone values. Legend: 1 5 porosity of 10  3 and
permeability of 1; 2 5 porosity of 10  3 and permeability of 10  2;
A quasi- single-phase system was simulated instead of a
3 5 porosity of 10  3 and permeability of 10  5; 4 5 porosity of 1
complicated two -phase system. The following scenario
and permeability of 10  6; 5 5 porosity of 10  3 and permeability
was therefore proposed and backward-validated: out- salt- of 10  6; 6 5 porosity of 10  3 and permeability of 10  8; and
ing may start to occur only at temperatures lower than the 7 5 pure water.
upper solidus line for pressures equal to 30^32 MPa (980^
1020 K, see Fig. 6). Salt is soluble above this line but re-
mains largely non-dissociated. Thereafter, the brine will
be salt-free (pure water) until the gradual cooling of the
upward £ow crosses the lower solidus line and water re-
gains the ability to dissolve salts (conventional solubility).
The salt content of brine in this region will be restored
due to mixing with the existing brines in the sediments
and/or dissolving existing salt deposits. The precipitation
of salt at the upper solidus line will block the pores and re-
sult in decreased porosity and permeability of the given
cell.To incorporate the e¡ect of out- salting, the cell poros-
ity and permeability were scaled up by constant factors as
Fig. 9. Simulated injection of supercritical brine into a three-
described in the captions of Figs 7^9. Allowing the poros- layer reservoir: £ow velocity pro¢le for di¡erent combinations of
ity to vary during the simulation required a slightly longer scaling factors for porosities and permeabilities relative to
time to reach a steady state. original sandstone values. Legend: 1 5 porosity of 10  3 and
permeability of 1; 2 5 porosity of 10  3 and permeability of 10  2;
Simulation results 3 5 porosity of 10  3 and permeability of 10  5; 4 5 porosity of 1
and permeability of 10  6; 5 5 porosity of 10  3 and permeability
A wide variety of combinations of porosity and permeabil- of 10  6; 6 5 porosity of 10  3 and permeability of 10  8; and
ity of sandstone and shale layers are tested. None of these 7 5 pure water.

226 r 2006 The Authors. Journal compilation r 2006 Blackwell Publishing Ltd, Basin Research, 18, 221^230
Sub-surface precipitation of salts in supercritical seawater

a¡ected the temperature pro¢le; it remained essentially transported upwards as brines that vent on the sea-
linear and identical to that of pure water as shown in Fig. £oor.
7. No drift of the out- salting front has been observed
either, validating our earlier assumptions. On the other Another observation by Von Damm et al. (2002) at the East
hand, we found that unlike temperature, the pressure and Paci¢c Rise showed extraordinary segregation in £uids
£ow velocity are very sensitive to variations in porosity and from a multi-funneled vent complex, the Brandon Vent:
permeability, as shown by the reservoir simulation results ‘‘Of the ¢ve ori¢ces on Brandon, sampled, three were vent-
displayed in Figs 8 and 9. The e¡ects of the out- salting are ing £uids with less chlorine than average seawater, and two
best illustrated in Fig. 8: the blocking of pores caused by were venting £uids with greater content of chlorine than
salt deposition leads to the establishment of a stable low- average seawater. The liquid-phase ori¢ces contain 1.6^1.9
permeability layer within the sandstone. This layer will times the chloride content of the vapors.’’ The di¡erent
give rise to a substantial pressure rise over a relatively £uids were venting within metres of each other, with the
small region (20 m) and its existence may have drastic prac- densest high- chlorinity £uids emanating from the lower-
tical consequences; e.g. unexpected pressure build-up most chimneys.
during exploration drilling. The £ow velocity pro¢le of These observations can also be explained similar to
Fig. 9 is also quite instructive. It demonstrates that those above. An interesting issue is the short distance be-
although the out- salting clearly has an impact on the tween vents with very di¡erent salinities. This may be ex-
velocity gradient, the e¡ect will manifest itself further plained by the precipitation of salt (or other minerals e.g.
downstream from the out- salting front. silica and sulphides) along the migration routes of the
high- salinity brines that block communication to the
neighbouring vents.

Environments for hydrothermal salt formation


DISCUSSION OF CONSEQUENCES
The processes we study here have very much in common
Manifestations of supercritical brines in
with mineral chimney formation and layered ore-genera-
nature
tion by hydrothermal processes. The necessary prerequi-
The fact that boiling and supercritical conditions a¡ect site conditions for such structures and deposits are:
the salinity distribution in hydrothermal systems has been
 long-lived sub- surface-focused heat sources
observed at hydrothermal vents (Ramboz et al., 1988; Low-
 hydrothermal circulation of water and brines.
ell & Germanovich, 1997). One good example shows a
change in chlorinity over time. During the period from During extensive studies of buried hydrothermal systems,
1991^1994,Von Damm (1995) observed a change in chlori- such as those conducted in several Ocean Drilling Pro -
nity from £uids emanating at the same deep- sea vent: gram (ODP), ‘sedimented ridges’ campaigns, it has been
from having been very low in chlorinity in 1991, directly clearly demonstrated that seawater circulates deep into
after a volcanic eruption, to attain about 1.5 times seawater the sediments, down towards the buried heat sources that
chlorinity 3 years later.These observations were explained may lie within the oceanic crust (Becker, 1999; Galley &
by the venting of vapour-phase £uids subsequent to the Koski, 1999; Davis et al., 2004). This circulation occurs
volcanic eruption, followed by brine-dominated £uids despite the intervening sediments consisting mainly of
that at ¢rst were stored within the oceanic crust, and that low-permeable shales.
vented when the hydrothermal system had cooled down 3 In their work on hydrothermal venting, Goodfellow and
years later (Von Damm, 1995). Similar observations are re- Zierenberg (1999) discussed £uid recharge and discharge
ported by Butter¢eld et al. (1997) from the Juan de Fuca through low-permeable sediments.They conclude that re-
ridge and the East Paci¢c Rise. charging of seawater (in the Escabana trough) takes place
These observations can easily be interpreted from the either locally associated with a heat source, or regionally
theoretical discussions above: near the margins of the rift. Fluid discharge is in£uenced
by the position of the heat source, but lateral £ow may oc-
 The venting of low- salinity water indicates that sea-
cur along fractures in the basaltic basement or along more
water coming into contact with the hot rock is brought
permeable sand and silt layers in the sediment package.
into the supercritical ‘two -phase region’ between the
They also suggest that £uid discharge through low-
upper and lower solidus line (SSIP and CIP). Salt pre-
permeable sediments takes place in the form of hydrother-
cipitates, and the supercritical vapour £ows upwards.
mal £uid diapirs that migrate upwards along zones of
At some point along the migration route, the vapour
structural weakness.
temperature drops below the CP, and conventional
boiling occurs. A further cooling will lead to conden-
Preservation of salts
sation, and this water is venting on the sea£oor.
 When the hydrothermal system cools down (after 3 For the preservation of precipitated salts, there is another
years) the CP is displaced downwards to where the salt important prerequisite: the produced salts have to be pro -
was precipitated initially. Salt will be re-dissolved and tected from re-dissolution by seawater.We therefore theo -

r 2006 The Authors. Journal compilation r 2006 Blackwell Publishing Ltd, Basin Research, 18, 221^230 227
M. Hovland et al.

suggested that salt can precipitate in the following do -


mains:
 At the surface, by evaporation of supersaturated
brines.
 On the sea£oor, by precipitation from supersaturated
brines in shallow pools on the sea£oor of deep con-
¢ned basins (generally 41km water depth).
 Within the central re£ux zone, immediately above the
buried heat- source(s), by ‘surpercritical out- salting’.
 In the £anking recharge zones and central re£ux zone
Fig. 10. Oceanographic and physical conditions at the Atlantis II of the sedimentary package (at temperatures between
Deep, Red Sea, according to Zierenberg and Holland (2004), 130 and 405 1C).
Faber et al. (1998), Hartmann et al. (1998a,b), Zierenberg (1990),
Danielsson et al. (1980). Note the assumption that hot brine vents
exist, indicated to the left.The arrows pointing down indicate
Furthermore, we hypothesize that salts precipitating un-
mineral (mainly sulphides and metal) precipitation onto the der supercritical conditions in the sub- surface will accu-
sea£oor (see referred literature). mulate in the pore spaces and fractures above the heat
source(s) or (depending on accommodation space) move
upwards as super- saturated brines or a slurry of salt and
rize that tectonically active environments, with high sedi- brine in a gaseous water phase. If these salts and £uids
mentation rates, may o¡er such protection where sedi- reach the CP upon entering the sea£oor, the venting will
ments e⁄ciently cap the salts. Thus, marine and even be similar to a ‘black smoker’ discharge, and the deposits
continental rift zones, for example the Atlantis II Deep in will be spread along the axial rift, controlled by the heat
the Red Sea, represent a suitably generative and protective source(s) (Goodfellow & Zierenberg, 1999).
environment (Fig. 10) (M. Hovland, H. Ruesltten, H.K. However, if the migrating £uids reach the CP within the
Johnsen, B. Kvamme & T. Kutznetsova, submitted). An- sedimentary sequence that acts as an overburden to the
other interesting tectonic setting with a modern accumu- hydrothermal system, boiling may occur (depending on
lation of salt is Lake Asale in Ethiopia. Unlike the Red Sea, whether it is pressure or temperature that ¢rst reaches
this is a terrestrial hydrothermal setting, and the hot and the CP) and the salts will precipitate within the sedimen-
dry climate ensures the evaporation of venting brines and tary sequence. Upon further cooling of the hydrothermal
the preservation of precipitated salts. system, the CP (i.e. the boiling point) will be displaced
downwards, and the condensation water will start to dis-
solve salt that was previously precipitated, and subse-
A NEW MODEL FOR HYDROTHERMAL quently transport it upwards until it ¢nally discharges
SALT FORMATION onto the sea£oor or on the terrestrial surface.This is in ac-
The new proposed model for hydrothermally induced salt cordance with the observations of hydrothermal venting
formation and accumulation can be summarized as fol- made by Butter¢eld et al. (1997).
lows (M. Hovland, H. Ruesltten, H.K. Johnsen, B.
Kvamme & T. Kutznetsova, submitted): it is based on ¢ve
main prerequisites:
CONCLUSIONS
1. A source of seawater or brine.
2. A buried, strong, local heat- source, which is either a Laboratory experiments performed in the early 1990s de-
shallow magma chamber (To1200 1C), or a set of inter- monstrated that supercritical water behaves like a non-po -
secting high-heat£ow crustal faults (T4405 1C), or a lar £uid, and that salts dissolved in the water ‘shock
combination of both (12004T4405 1C). crystallized’ upon passage into the supercritical P^Tregion
3. A system of intersecting fractures and faults above the (Tester et al., 1993). Molecular theory con¢rms why water
heat source(s) that allows the circulation of water. loses its ability to dissolve salts on passage into the super-
4. A sedimentary unit located above 3 that allows the con- critical region. We provide the theoretical basis for this
vection of £uids, i.e. that the pore-pressure is near-hy- fact. One of the natural consequences of this revelation is
drostatic from the surface down towards the heat- that brines and seawater that circulate into deep- seated
source along the ‘£anking recharge zones’, and slightly hydrothermal systems will become supercritical and that
over-pressured along an upwards ‘central re£ux zone’ the salts will precipitate, although this has not been
located above the heat- source(s). directly observed in any natural environment, so far.
5. An environment that protects the salts from re-dissol- We predict that ‘shock crystallization’ (or supercritical
ving. out- salting) occurs during the intrusion of magma bodies
in the oceanic crust and during sill intrusions into deep-
With these prerequisites satis¢ed, and where the sedimen- seated water- saturated sediments. Numerical basin
tary pile is su⁄ciently thick (up to several kilometres), it is modelling and simulation demonstrates the e¡ects on

228 r 2006 The Authors. Journal compilation r 2006 Blackwell Publishing Ltd, Basin Research, 18, 221^230
Sub-surface precipitation of salts in supercritical seawater

temperature-, pressure- and £uid velocity ¢elds of such a Franck, E.U. (1976) Properties of water. In: High Temperature,
small- scale hydrothermal system. High Pressure Electrochemistry in Aqueous Solutions (Ed. by D. de
By referring to ¢eld observations from two large- scale, G. Jones & R.W. Staehle),109^116. National Association of Cor-
deep - seated hydrothermal rift-related systems, the Atlan- rosion Engineers, Houston, TX.
tis II Deep, of the Red Sea, and the Dallol area and Lake Galley, A.G. & Koski, R.A. (1999) Setting and characteristics
Asale, Ethiopia, it is suggested that much of the accumu- of ophiolite-hosted volcanogenic massive sul¢de deposits. In:
Volcanic-Associated Massive Sul¢de Deposits: Processes and Examples
lated salts and brines observed in these systems stems
in Modern and Ancient Settings (Ed. by C.T. Barrie & M.D. Han-
from hydrothermally associated boiling and supercritical
nington), Rev. Econ. Geol., 8, 221^246.
out- salting at depth. For this reason, it is concluded that
Garrido, I., Fladmark, G.E. & Espedal, M. (2004) An im-
hydrothermally associated salt formation and accumula- proved numerical simulator for multiphase £ow in porous
tion is under-rated as a contributor to evaporite deposi- media. Int. J. Numer. Methods Fluids, 44, 447^461.
tions, especially in rift-related tectonic settings. Goodfellow, W.D. & Zierenberg, R.A. (1999) Genesis
of massive sul¢de deposits at sediment- covered spread-
ing center. In: Volcanic-Associated Massive Sul¢de Deposits:
Processes and Examples in Modern and Ancient Settings (Ed.
ACKNOWLEDGEMENTS by C.T. Barrie & M.D. Hannington), Rev. Econ. Geol.,
8, 297^324.
We would like to thank Statoil for giving the necessary
Guillot, B. (2002) A reappraisal of what we have learnt during
support to carry out this work and for permission to pub- three decades of computer simulations on water. J. Mol. Li-
lish the results. We would also like to thank B. Charlotte quids, 101(1^3), 219^260.
Schreiber and one anonymous reviewer for their construc- Hartmann, M., Scholten, J.C., Stoffers, P. & Wehner, F.
tive and encouraging comments. (1998a) Hydrographic structure of brine- ¢lled deeps in the
Red Sea ^ new results from the Shaban, Kebrit, Atlantis II,
and Discovery Deep. Marine Geol., 144, 311^330.
Hartmann, M., Scholten, J.C. & Stoffers, P. (1998b) Hydro -
REFERENCES graphic structure of brine- ¢lled deeps in the Red Sea: correc-
tion of Atlantis II Deep temperatures. Marine Geol., 144,
Allen, M.P. & Tildesley, D.J. (1987) Computer Simulation of 331^332.
Liquids. Clarendon, Oxford. Hodes, M., Griffiths, P., Smith, K.A., Hurst,W.S., Bowers,
Becker, K. (1999) Permeability measurements in Hole 896A and
W.J. & Sako, K. (2004a) Salt solubility and deposition in high
implications for the lateral variability of upper crustal perme-
temperature and pressure aqueous solutions. AIChe Journal,
ability at Sites 504 and 896. Ocean Drilling Program, Scienti¢c Re-
50(9), 2038^2049.
sults, 148, 353^363.
Hodes, M., Marrone, P.M., Hong, G.T., Smith, K.A. & Tes-
Bellissent-Funel, M.-C. (2001) Structure of supercritical
ter, J.W. (2004b) Salt precipitation and scale control in super-
water. J. Mol. Liquids, 90, 313^322.
critical water oxidation ^ Part A: fundamental research.
Bischoff, J.L. & Rosenbauer, R.J. (1989) Salinity variations in
J. Supercrit. Fluids, 29, 265^288.
submarine hydrothermal systems by layered double-di¡usive
Hovland, M., MacDonald, I., Ruesltten, H., Johnsen,
convection. J. Geol., 97, 613^623.
H.K., Naehr, T. & Bohrmann, G. (2005) Chapopote asphalt
Butterfield, D.A., Jonasson, J.R., Massoth, G.J., Feely,
R.A., Roe, K.K., Embley, R.E., Holden, J.F., McDuff, volcano may have been generated by supercritical water. EOS,
R.E., Lilley, M.D. & Delaney, J.R. (1997) Sea£oor eruptions 86(42), 397^402.
and evolution of hydrothermal £uid chemistry. Philos. Trans. Kawada, Y., Yoshida, S. & Watabane, S. (2004) Numerical si-
Roy. Soc. London A, 355, 369^386. mulation of mid- ocean ridge hydrothermal circulation in-
Chialvo, A.A. & Simonson, J.M. (2003) Aqueous Na1Cl- pair cluding the phase separation of seawater. Earth Planets Space,
association from liquidlike to steamlike densities along near 56(2), 192^215.
critical isotherms. J. Chem. Phys., 118, 7921^7929. Kohl,W., Lindner, H.A. & Franck, E.U. (1991) Raman spectra
Danielsson, L.G., Dryssen, D. & Graneli, A. (1980) Chemi- of water to 400 1C and 3000 bar. Ber. Bunsenges. Phys. Chem., 95,
cal investigations of Atlantis II and discovery brines in the Red 1586.
Sea. Geochim. Cosmochim. Acta, 44, 2051^2065. Lowell, R.P. & Germanovich, L.N. (1997) Evolution of a
Davis, E.E., Becker, K. & He, J. (2004) Costa Rica Rift revisited: brine- saturated layer at the base of a ridge- crest hydrothermal
constraints on shallow and deep hydrothermal circulation in system. J. Geophys. Res., 102, 10245^10255.
young oceanic crust. Earth Planetary Sci. Lett., 222, 863^879. Lyubartsev, A.P. & Laaksonen, A. (1996) Concentration ef-
Driesner,T., Seward,T.M. & Tironi, I.G. (1998) Molecular dy- fects in aqueous NaCl Solutions. A molecular dynamics simu-
namics simulation study of ionic hydration and ion association lation. J. Phys. Chem., 100, 16410^16418.
in dilute and 1 molal aqueous sodium chloride solutions from Nieto-Draghi, C., Avalos, J.B. & Rousseau, B. (2003) Dy-
ambient to supercritical conditions. Geochim. Cosmochim. Acta, namic and structural behavior of di¡erent rigid nonpolarizable
62, 3095^3107. models of water. J. Chem. Phys., 118, 7954^7964.
Faber, E., Botz, R., Poggenburg, J., Scmidt, M., Stoffers, P. Phillips, S.L., Igbene, A., Fair, J.A., Ozbek, H. & Tavana, M.
& Hartmann, M. (1998) Methane in Red Sea brines. Org. (1981) A technical databook for geothermal energy utilization.
Geochem., 29, 363^379. Lawrence Berkley Laboratory Report 12810.
Fincham, D. (1992) Leapfrog rotational algorithms. Mol. Simu- Ramboz, C., Oudin, E. & Thisse, Y. (1988) Geyser-type dis-
lation, 8, 165^178. charge in Atlantis II Deep, Red Sea: evidence of boiling from

r 2006 The Authors. Journal compilation r 2006 Blackwell Publishing Ltd, Basin Research, 18, 221^230 229
M. Hovland et al.

£uid inclusions in epigenetic anhydrite. Can. Mineral., 26, from the southern East Paci¢c Rise. Earth Planet. Sci. Lett.,
765^786. 6490, 1^4.
Selley, R.C. (2005) Mineralogy and classi¢cation. In: Encyclope- Wagner, W. & Preuss, A. (2002) The IAPWS formulation 1995
dia of Geology (Ed. by R.C. Selley, L.R.M. Cocks & I.R. Plimer) for the thermodynamic properties of ordinary water substance
Elsevier, Amsterdam, 25^37. for general and scienti¢c use. J. Phys. Chem. Ref. Data, 31,
Smith, D.E. & Dang, L.X. (1994) Computer simulations of 387^535.
NaCl association in polarizable water. J. Chem. Phys., 100, 3757. Warren, J.K. (1999) Evaporites: Their Evolution and Economics.
Soper, A.K. (1986) On the determination of the pair correlation Blackwell Science, Oxford, pp. 438.
function from liquid structure factor measurements. Chem. Wilson, H.H. (2003) Extensional evolution of the Gulf of Mexi-
Phys., 107, 61^74. co Basin and the deposition of tertiary evaporites. J. Pet. Geol.,
Soper, A.K. & Phillips, M.G. (1986) A new determination of 26, 403^428.
the structure of water at 25 1C. Chem. Phys., 107, 47^60. Wilson, H.H. (2004) Extensional evolution of the Gulf of Mexi-
Talbot, C.J. (2004) Extensional evolution of the Gulf of Mexico co Basin and the deposition of tertiary evaporites ^ reply to
Basin and the deposition of Tertiary evaporates ^ discussion. discussion. J. Pet. Geol., 27, 105^110.
J. Pet. Geol., 27, 95^104. Zierenberg, R.A. (1990) Deposition of metalliferous sediment
Tester, J., Holgate, H.R., Armellini, F.J., Webley, P.A., beneath a brine pool in the Atlantis II Deep, Red Sea. In: Gor-
Killilea, W.R., Hong, G.T. & Berner, H.E. (1993) Super- da Ridge: A Sea£oor Spreading Center in the United States Exclusive
critical water oxidation technology. In: Emerging Technologies in Economic Zone (Ed. by G.R. McMurray) Springer-Verlag, New
Hazardous Waste Management III. American Chemical Society, York, 131^142.
35^76. Zierenberg, R.A. & Holland, M.E. (2004) Sedimented ridges
Von Damm, K.L. (1995) Controls on the chemistry and temporal as a laboratory for exploring the subsurface biosphere. In: The
variability of sea£oor hydrothermal £uids. In: Sea£oor Hydro- Subsea£oor Biosphere at Mid-Ocean Ridges (Ed. by W.S.D.
thermal Systems: Physical, Chemical, Biological, and Geophysical Wilcock, E.F. DeLong, D.S. Kelley, J.A. Baross & S.C. Cary),
Interactions (Ed. by S. Humphris), Geophys. Monogr., 91, Geophys. Monogr., 144, 305^323.
222^247.
Von Damm, K.L., Lilley, M.D., Shanks,W.C. III, Bocking -
ton, M., Bray, A.M., O’Grady, K.M., Olson, E., Graham,
A. , Proskurowski, G. & the Souepr Science Party (2002) Manuscript received12 August 2005; Manuscript accepted 3 April
Extraordinary phase separation and segregation in vent £uids 2006.

230 r 2006 The Authors. Journal compilation r 2006 Blackwell Publishing Ltd, Basin Research, 18, 221^230

Das könnte Ihnen auch gefallen