OOlh-7037.X01101-168350200.

The kinetics of silica-water reactions

J. D. RIMSTIDTand H. L. BARNES
Ore Dcpos~t Research Section, The Pennsylvania State University, University Park, PA 16802, U.S.A.

Ahstrnct A differential rate equation for silica-water reactions from G3OO’C has been derived based on
stoichiomctry and activities of the reactants in the reaction
SiO&) + ZH,O(I) = H,SiO,(aq)

(~u,,,s,“,~~r),.~,,W.. = (A:‘M)~~,,~s,o,)(k+us,02U~~” - k~%l&O~)

where (A. M) = (the relative interfacial area between the solid and aqueous phases/the relative mass of
water in the system). and k, and k_ are the rate constants for, respectively, dissolution and precipi-
tation. The rate constant for precipitation of all silica phases is
logk_ = -0.707 - 2598/7‘(7:K)
and f;,,&, for this reactIon is 49.X kJ mol- ‘. Corresponding equilibrium constants for this reaction with
quarrl, cristobalitc. or amorphous silica were expressed as log K = a + bT + c/T. Using K = k+/k_,
I, t\as cypressed as log li + = a + bT + c/T and a corresponding activation energy calculated:
b c E,,,(kJ mol- I)
Quartz ,a,74 -2.02X x lo-3 -4158 61.4- 76.6
^r-Cristobalite -0.739 0 -3586 68.7
/I-Cristobalite -0.936 0 - 3392 65.0
Amorphous silica -0.369 -7.X90 x lO-4 -3438 60.9-64.9

Upon cooling a silica-saturated solution below the equilibrium temperature, the decreasing solubility
of silica causes increasing super saturation, which tends to raise the precipitation rate, but the rate
constants rapidly decrease. which tends to lower the precipitation rate. These competing effects cause a
m;lxlmum rate of precipitation 25-50°C below the saturation temperature. At temperatures below that
of the maximum rate, silica is often quenched into solution by very slow reaction rates. Consequently,
the quartz geothermometer will give the most accurate results if samples are taken from the hottest,
highest flow rate. thermal springs which occur above highly fractured areas.

NOMENCLATURE ;‘, activity coefficient of i

h Planck’s constant (6.63 x 10~“ J set)
intcrfacial area (m’) k Boltsman’s constant (1.3X x 10mz3 J mol- ’ K ‘)
frcqucncy factor in Arrhenius equation k+ dissolution rate constant (set- ‘)
relative interfacIal area (= A:A” where A0 = I m’) k; apparent dissolution rate constant (set I)
specific surface area (m’ gm I) k- precipitation rate constant (set- ‘)
activation cnergq (kJ mol. I) k’ apparent precipitation rate constant (sec. ‘)
free energy of formation (kJ mol- ‘) I liquid
free energy of reaction (kJ mol-‘) I,$ concentration of i (molal)
free energy of activation (kJ mol-‘) ,li amount of i (moles)
equilibrium constant qtz quartz
equilibrium constant for the formation of the acti- r radius (m)
bated complex r, net rate of change of ai with time (set- I)
equllibrlum constant for the formatlon of the acti- r; net rate of change of mi with time (m set- ‘)
bated complex hith a degree of translational free- r+ E rate of change of thickness of a layer of material
dom rcmovcd on a surface (m set- ‘)
mass of water in the system (kg) S solid
rclati\c mass of water in the system (=M/M” t time (set)
bhere M” = I kg) t, time constant (set)
pressure (bars)
activity product
gas constant (X.318 J mol-’ Km’) INTRODUCTION
degree of saturation ( =Q./K)
temperature (K) THIS STUDY was undertaken to develop a basis for
bolumc (mzi reaction kinetics that is consistent with the thermo-
molar volume of 1(m’ mol- I) dynamic approach commonly used in geochemistry.
specific volume of water (cm3 gm _ I) Thermodynamics is a powerful tool for elucidating
activity of i
:k,UCOM
geological phenomena where equilibrium is normally
amorphous silica attained; however. many geological processes are con-
cristohallte trolled by reaction rates so that they can only be
I683
1684 J. D. RIMSTIDT and H. L. BARNES

understood in terms of kinetics. Such a situation is used in these experiments. Such techniques give
illustrated by reactions at low temperatures irl the higher values of the surface area than would be calcu-
silica-water system. Therefore. the rates of dissolution lated from geometrical measurements of individual
and precipitation of quartz and amorphous silica at grains, presumably because they are sensitive to sur-
temperatures of &3OO”C were chosen for this study. A face roughness and porosity. The relationship
rate equation based on transition state theory (EYR- between the surface areas found by geometrical calcu-
[NC;, 1935) was derived to express the measured rates lations and nitrogen BET has been quantified for
in terms of the activities of the reacting substances. quartz by LEAMNSONet al. (1969) but further work is
The determined rate constants, thus, have standard necessary to find this relationship for other solids.
states that are consistent with the equilibrium con- Thus, the difference between these two ways of finding
stants for the reactions. Also, a number of factors that the relative surface areas for a system must be kept in
influence the reaction rates in this system have been mind when extrapolating these rate constants to
identified and the extent of their influence measured natural systems. Another problem in heterogeneous
or estimated. The notation used in this discussion is reactions is that the defect structure of the surface of
listed in the Nomenclature above. the solid influences its solubility so that any disturbed
The silica-water system was chosen for this study surface layer must be dissolved away before rate con-
because it is simple and well documented, yet geologi- stants for the bulk solid can be determined. Through-
cally significant. The extensive solubility data avail- out this work the term ‘bulk solid’ refers to the
able in the literature were used to find reliable tem- properties of the solid where no surface effects
perature functions for the equilibrium constants influence its behavior.
(Table I ). Also, the effects of complexing, hydrolysis
and polymerization have been reported by various
investigators to be insignificant in the dilute. neutral
to slightly acidic solutions commonly found in geolo- DERIVATION OF A RATE MODEL

gic environments. These effects are summarized in The following discussion is based on absolute rate
RIMSTIIIT (I979). Furthermore. there is no solid solu- theory (EYRING, 1935) and is limited by the assump-
tion in the silica phases and silica dissolves con- tions implicit in this theory. The partition functions
gruently to form H,SiO,(aq). Finally. silica and water used in the Eyring formulation have been rewritten
are two very common and important constituents of into free energy terms (WERT and ZENER, 1949). Al-
the earth’s crust so this work has a wide range of though this formalism is adequate here, it obscures
applicability to geological problems. the subtleties of absolute rate theory and thus, it
A number of difficulties, which must be considered would be beneficial to consult, for example, CHRIS-
in any kinetics study, are illustrated in this work. TIAN(1975) for a more fundamental approach to kin-
First, the thermodynamic properties of the system etics.
must be well known before its reaction kinetics can be The basic assumption used in this derivation is that
investigated. Because the rate of a reaction depends the reactants must pass through a free energy maxi-
upon the chemical potential driving it, any uncer- mum before they are converted to products (Fig. 1).
tainty in the equilibrium constant leads to an error in The state defined by this energy maximum has been
the calculated driving force for the reaction; this error labeled for convenience as the activated complex
adds to the uncertainty of the derived rate constants. (EYRING, 1935) although it may not be a discrete
For heterogeneous systems, it is necessary to quantify species. The rate of reaction is assumed to be directly
the extent of the system, ie.. the ratio of the relative proportional to the concentration of the activated
surface area to the relative mass of water in the sys- complex.
tem (A/M). The (A/M) term must be separated from Thus. a differential rate equation can be derived to
the measured apparent rate constants in order to pro- describe the reaction
duce fundamental rate constants that are comparable
from one experiment to another. Argon and nitrogen SiO,(s) + 2H,O(l) = H,SiO,(aq) (1)
BET measurements were used as standard techniques
for determining the relative surface areas of the solids which consists of the two opposing reactions shown

Table I. Temperature functions of equilibrium constants for silica-water reac-

tions. See RIMSTIDT(1979) for data sources

Temperature function
Phase (7: K)

Quartz log K = 1.881 - 2.028 x lo- 3 7 - 1560;T

r-Cristobalite log K = -0.0321 - 988.2.7
P-Cristobalite log K = -0.2560 - 793.617
Amorphous silica log K = 0.3380 - 7.889 x 1o-4 T - x40.1/7-
 The kinetics of silica- water reactions 1685

in (2) and (4). where (kT/h) represents a translational degree of free-

dom of the system along the reaction coordinate, and
SiO, + 2H,O = (Si02.2Hz0)* (2) K: and K 5 are KT and KT with this degree of free-
(Si02.2H20)* --t H4Si04 (3) dom removed. The rates defined by (12) and (13) are
for a system with no limits for the mass of water in
and which the H4Si04 is dissolved or the interfacial area
H,Si04 = (SiO,‘2H,O)* (4) between the reacting solid and the aqueous solution.
Therefore, it is convenient to define a standard system
(SiO,.2H,O)* + SiOz + 2Hz0 (5) by considering two basic requirements of these kin-
Reactions (3) and (5) proceed at equal rates when the etics:
system is at equilibrium. Note that this form for the
(1) The rate of reaction between two phases is di-
activated complex has been assumed in order to sim-
rectly proportional to the interfacial area (A) between
plify further calculations. Thus, although this model is
the phases.
quite adequate for the simple silica-water system. it
(2) For a fixed rate of addition of solute, the rate at
may be necessary to assume a different form of acti-
which the concentration increases in a system is in-
vated complex to explain rate data for systems with
versely proportional to the mass of water (M) in the
very different salinities or pH’s. For a given system
system.
AG*- = AGP.,s~J.~H>o,*- AG&o, - 2AG&,o (6) The two variables A and M are thus necessary to
define the extent of the system of interest. A system
AGf = AGP,(sioJ-ZH,o)*- AGP,H4sio4 (7)
with one square meter of interfacial area and 1 kg of
as shown by Fig. 1 and water will be selected as a standard system. Then for
any system under consideration, the relative inter-
AC*, = -RTlnK*, (8) facial area can be defined as
AG? = -RTlnK*_ (9) A = A/A0 where A0 = 1 mz (14)
The constants K$t and KT are thermodynamic equi-
librium constants so that and relative mass of water as

KT = la(s,o*.2~~0)*l/lla~io,3la,>o12 (10) M = M/M0 where MO = 1 kg (15)

K’ = r~~Si0*~2”~0~*1/C~H~~i0~1 (1 1) Henceforth in this derivation, A and M will refer to

the dimensionless quantities defined in (14) and (15)
The rate that reactants pass through the activated
and (AIM) will represent the extent of the system.
complex to form products is given bq
Returning to (12) and (13). all of the constants can be
(an H+‘5i04/ar)P,T,( +) collected into one term-the rate constant

= IkVl W :~s,o,~~~o1!C~~~io,.2~~o~~l (14 k+ = [kT/hl [K3!C7si0~.2~~0~*1 (16)

and and

(dnH4Si041dt)P,T,( -1
k- = LkVhl CK %‘Crcsio,.2~,0~-1 (17)
= - CWhl CK’~,,~io,l/C~~~i0,.2~~0,~1 (13)
So for a system with relative interfacial area, A

(anH4SioJ~t)P,T,~+)
= Ak+~io,d,o (18)

(%d~~b,~,~ - 1= - Ak-QH4Si04 (19)

and the observed net rate is the sum of (18) and (19)

(&IH4S10JWP.T = A(k+as,o&,o- k_aH4Si04) (20)

This equation can be recast in terms of aHqSio4because

mH8i04 = (nH,SiO,)/@‘f) (21)

quartz
then

REACTION COORDINATE ‘lH4s,o, = (m,,,,,,)(M) (22)

Fig. I. A schematic illustration of the free energy maxi-
so expanding the left-hand side of (20) gives
mum through which reactants must pass to become
products. Here the activated complex is assumed to be the (2n H4S104/af)P,T = M(8mH,Si0,/d~)p,T,M
same for both quartz and amorphous silica: this will be
demonstrated later. + “‘H4Sio4(~M/~f)P.T,m (23)
1686 J. D. RIMSTIDT and H. L. BARWS

Ass~lm~ng that the mass of water in the system is The term ( j - S) is a measure of departure from eqtn-
constant allows the reaction rate to be expressed rela- fibfium. disequilibrium. Note that when S = f the
tive to the rate in a system containing 1kg of water system is at equilibrium and the net rate of reaction is
[see eqn (15)]. zero.
It is also useful to integrate the differential rate
equation in order to obtain a function which will give
the degree of saturation (S). and therefore au4sio4. as a
function of time. To do this (30) can be rearranged
and integrated between states at times. f = 0 and I. at
constant P. 7: !M. and ;

Evaluation between limits gives

t = -(l/k’_) In ([k’+ - k’-(~~~~~~~)],

Assuming that yn,io, is constant converts (24) to
IX-‘- - k~(aH,sict,)ol: (39)
(~u~~sio~~~t)~~~,~,; combining (30) (3 t ) and (39) gives
= (AjM)(y,,sio,)(k+asio,a~p r = - (ljk’- ) In C(r,~sio,)l(r”,sio,)ol (40)
- k-an4sio4) (27) and from (36) we see that
which is the differential rate equation for the silica-
1 = -(l!kL)ln[(l - S)/(l - S,j (41)
water reaction in terms of activities.
Equation (27) can be simplified for evaluation of This form of the integrated rate equation is most use-
the rate constants by the following manipulations. ful in interpreting experimental data. Note that when
Factors taken to be constant in (27) can be combined starting from zero concentration of H,SiO, in solu-
into apparent rate constants, k’+ and k’, where tion, S, = 0 and (41) reduces to

k’+ = (A,iM)?;,,sio4~io~f~~~ok+ (28) r = -(l/k’_)in(l - S) (41)

k’_ = (~~.~~j~*~i~~k- (29)

This converts the differential rate equation into the EXPERIMENTAL DESIGN
simple form
In order to test the validity of the above derivation, it
(%,sioal% was necessary to obtain data on concentration versus time
.T.M.,-, = k’+ - kLaH4sio4 (30)
from systems where P, 7: A. 134.and yHISioI were known.
Because of our choice of the form of the activated Some of these data were obtained from other investigations
complex k‘+ and k’- are constants for reactions and the remaining were’fitled in by experiments performed
in the present study. High temperature experiments were
between silica and relatively pure water. However, it
run in either Barnes type rocking autoclaves (BAKNES.
should be remembered that k; and k’_ may neverthe- 1971) or in a system for circulating hydrothermal fluids
less be complex functions of pH and salinity. For con- (BARNES and Dowus, rn preparation). These systems were
venience, the net rate of reaction will be designated as loaded with a known mass of sand of nearly homogeneous
rH&Oqr where grain size (0.1-1.0 mm) furnished by Corning Glass Works.
State College, Pennsylvania. The specific area of this sand
rH&iOJ = (daH,SiO,lc?t)P,T,M,~ (31) was 0.92 m2 gm-’ as determined by argon BET (SOLOMON.
1978). The systems were then evacuated and filled with a
Also, realizing that known mass of distilled water. Thus. the initial (A/h,f) ratio
was established in each case. The autoclaves were then
K = k+/k_ (32) heated to and maintained at the selected run temperature
(i5”C) while samples were withdrawn at convenient inter-
allows (30) to be rearranged to
vals. These samples were weighed and then preserved for
later analysis by adding I part of 5 m NaOH solution for
rnasioq = k’+(l - Q/K) (33) every 10 parts of sample to yield a final concentration 01
Where Q is the activity quotient of the system, NaOH of about 0.5 m. The high pH of these solutions
caused ionization of the H,SiO,, and thus, a correspond-
Q= (“H,Si0,)/(asio,)(a”,o)2 (34) ingly higher solubility of silica. Samples were stored in
polyethylene containers until they could be analyzed, at
Then defining the degree of saturation (S) as which time the rapid addition of 10 ml of 0.5 m H,S04 to
I ml of the sample caused the ionized species to revert to
S = Q/K (35) H,SiO,.
gives The concentration of silica in the acidified samples was
determined by the molybdate blue method (GOV~TT. 1961)
rH4Si04 = k’,.(l - S) (36) using standards prepared by diluting a 1000 ppm Si stock
 The kinetics of silica-water reactions 1687

solution obtained from Spex Industries, Inc., Metuchen, available in RIMS~~DT (1979). The extraction of rate
New Jersey. Atomic absorption spectroscopy was not used constants is illustrated with data from Runs 2E and
because the presence of sodium in solution causes anomal-
2F (Fig. 2). Referring to (4l), we see that it is a linear
ously high silica values to be determined (DEVINE and
SUHR, 1977). The total error from the volumetric measure- equation of the form y = mx where m = - I ik’.
ments and differences in the absorbances of the cuvettes for Therefore, if In [(I - S)/(l - S,)] is plotted versus
the molybdate technique was less than l.S%, yet occasional elapsed time, the slope of the line will be - I JL. Note
high values were found in the samples (though never in the that for this calculation S = c/c,, where cVq = 58 ppm
standards). These high values were attributed to small par-
ticles of silica that were occasionally entrained in the
for quartz at 105”. Figure 3 shows a straight line fit
sample during extraction from the experimental system. for Run 2E was obtained as predicted by the rate
This nroblem is discussed bv BARNES (1971). The large model but Run 2F gave two distinct line segments
amount of NaOH subsequently added to the sample with two different slopes. The initial line segment with
caused these particles to dissolve rapidly. yielding an
the steep slope is due to the dissolution of a disturbed
anomalously high silica concentration.
layer from the quartz surface, as will be discussed
later. After about 20 hr, the reaction in Run 2F is
RESULTS
controlled by dissolution of bulk quartz and the slope
Calculation of the rate corlstmt fbr the precipitation of the line segment is nearly the same as Run 2E. This
?WXYiOfl smaller slope will be used in subsequent calculations.
The concentration of H&O, as a function of time A linear least squares fit of the data shows that for
data was used to calculate the rate constant for the Run 2E, kL = (7.65 & 0.92) x 10m3 hr-’ and for Run
precipitation reaction (k-) using equations 35, 41 and 2F, kL = (3.46 + 0.42) x IO-” hr-’ (errors reported
29, in order. Both the literature data and the data as two standard deviations). Referring back to (29) we
produced by the experiments discussed in the last see that
section are from experiments where P, 7; and (A/M)
were held constant and the systems contained distil- k- = (k’-)j(AiM)(?;,,sio,) (43)
led-deionized water. For a summary of run con- In this dilute solution, ywqsioj = 1.0, is assumed. The
ditions, see Table 2 and a complete list of data is value for (A/M) is 261 for Run 2E and 265 for Run

Table 2. Run materials used in rate experiments

Particle
size As, Measurement
Number Material (pm) (m*/gm) method Notes

I Quartz sand 125~1000 9.2 x IO-* Ar BET (a, b, c)

2 Disturbed surface on 125-1000 9.2 x IO-* Ar BET (a, b, c)
quartz sand
3 Fused silica powder <I40 27 NaOH titration (a, b, d)
4 Silica gel (Ventron) <70 3.4 X IO2 NaOH titration (a. d, e)
5 Silica gel (Ventron) <58 9.3 x 10’ NaOH titration (a, d, e)
6 Silica gel (Fisher) <loo 1.1 x lo3 NaOH titration (a, d, e)
7 Fused silica powder <I 11.4 N,(?) BET (b. f)
8 Disturbed surface on <20 1.0 Estimated (b. g, h)
quartz powder
9 Porous leached glass - 1.75 x lo* (‘?) (i)
10 Disturbed surface on 315400 7.8 X 1o-3 N, BET (b, j)
quartz sand
II Disturbed surface on 20&250 1.45 x IO-’ NZ BET (b, j)
quartz sand
I2 Disturbed surface on 2O(f250 1.72 x IO-* N, BET (b. j)
quartz powder
I3 Disturbed surface on IO&l60 3.90 x 10-l N, BET (b. j)
quartz powder
14 Disturbed surface on 63-100 6.12 x IO-’ Nz BET (b. j)
quartz powder

(a) Material used in this study.

(b) Particle size measured by sieving.
(c) A,, measured by G. Solomon (see SOLOMON, 1978).
(d) NaOH titration technique developed by SEARS (I 956).
(e) Reported particle size is the average size of the particles incorporated into the gels
(measured by the manufacturer).
(f) Material used by STOBER (1967).
(g)A,, estimated by the technique reported by LEAMSONet al. (1969).
(h) Material used by BECKWITH and REEVE (1969).
(i) Material used by ELMER and NORDBERG (1958).
(j) Material used by USIXIWSKI (personal communication. 1974).
1688 J. D. RIMSTIDT and H. L. BAR~FS

2E

01 1 1 I I

0 40 80 120 160
ELAPSED TIME, hours

Fig. 2. Concentration of H,SiO, vs elapsed time for Runs 2E and 2F containing quartz sand and
distilled water at 105’C. In both reactions cH4s,oq is approaching 58 ppm which IS the equilibrium
solubility under these conditions.

2F. Thus, in Run 2E, k_ = (8.15 + 3.25) x 10e9 also been indicated by heat of immersion experiments of
and in Run 2F, k_ = (3.62 k 0.46) x 10m9 WAHLEK (1961) and by the adsorption experiments of
set-’
WADE et al. (1961). This evidence leads to the conclusion
set-I. The values of k- found from both this ex- that quartz surfaces contain defects which are a result, per-
perimental data and for published data are listed in haps, of mechanical abrasion and it was the rapid dissolu-
Table 3. tion of these higher solubility zones that was initially
observed in these rate experiments.
771~disturbed surface ofquartz grtcins Because the solubility of this disturbed layer appears to
An initial steep slope (see Fig. 3, Run 2F, for example) be nearly the same as bulk amorphous silica, its rate of
was observed in all runs where dissolution was the domin- dissolution has been used to calculate rate constants. In
ant reaction. This effect implies the presence of a higher this case the solubility of amorphous silica (Table 1) was
solubility (relative to quartz) surface layer which dissolves used to calculate the degree of saturation (S). These rate
away rapidly until only bulk quartz remains to control the constants were found to be nearly the same as those of
dissolution rate. quartz and amorphous silica (Fig. 4).
The presence of a higher solubility surface on quartz
grains is best shown by the low temperature solubility
The Arrhrnius equation and the activatim energy
measurements of BECKWITH and REEVE (1969). In these ex-
periments, it was found that the concentration H4Si04 in The Arrhenius equation states
water containing a suspension of ground quartz was
IO&120 ppm at 25°C. These authors showed that most of k_ = (A’)~~‘“.,‘,‘RT’
(44)
the coating responsible for this anomalous, high solubility
could be removed by treatment with HF so that sub- so
sequent solubility measurements gave values much nearer
those expected for bulk quartz. The high solubilities found Ink- = (E,,,/R)(l/T) + In (A’) (45)
in these experiments are suspiciously close to the solubility
of amorphous silica at 25°C which is 116 ppm based on where A’ is a ‘frequency factor’ related to the geo-
Table I. The disordered nature of many quartz surfaces has metry of the activated complex. Its exact meaning is

ELAPSED TIME, hours

Fig. 3. In[(l - S):( I - S,)] vs elapsed time for Runs 2E and 2F.
 The kinetics of silica-water reactions 1689

Table 3. Run conditions and calculated values of k- for rate experiments

Run Dominant Number

Run Temperature material reaction of (A/M) km Error
number (‘C) (a) (b) samples (c) set- ‘) (d) Notes

A 18 D 4 1.35 x IO2 I.51 x lo-9 2.5 x 1om9 w)

B I8 4 D 6 1.70 x IO3 3.22 x 10m9 1.64 x 1O-9 (e. f)
c 18 D 5 4.66 x IO” 2.80 X 10-‘O 2.03 x IO-‘” (e. f)
D 24 6 D 5 5.44 x IO3 6.35 x IO-I0 3.16 x IO-‘” (e. f)
Sl 25 D I2 5.02 9.12 x IO-’ 3.98 x lo-* (!a
S2 25 D 11 30.1 2.32 x IO-* 1.06 x IO-” (g)
s3 25 D I0 90. I 1.10 x lo-* 0.33 x lo-” (P)
B&RI 75 8 D 6 1.00 x IO2 9.87 x lo-: 10.0 x lo-: (h)
B & R2 25 P 4 1.00 x IO2 I.01 X lo-’ 0.87 x 1om4 (h)
WBIA 65 P 17 92 3.81 x IO-” 3.24 x IO-” (e)
E&N 95 9 D 3 1.12 X lo3 1.95 X 1o-8 1.37 x 10-R (i)
2D 105 2 D 3 2.65 x IO’ 4.45 x 1om8 3.12 x IO-’ (e)
2E 105 P IO 2.61 x 10’ 8.15 x 1o-9 3.25 x IO-” (2)
2F 105 D 14 2.65 x IO’ 3.62 x IO-’ 1.46 x IO-’ (e)
2F 105 2 D 3 2.65 x IO2 4.06 x 10m9 2.38 x lo-” (e)
2c 145 D 5 1.42 x 10’ 1.36 x IO-’ 0.77 x lo-’ (4
2M 145 4 P 9 1.47 x IO2 3.03 x IOmR 1.03 x lomR le)
20 145 2 P II 1.47 x IO’ 2.92 x lO-8 1.06 x IO-’ (e)
2K 170 D 9 2.24 x IO’ 7.68 x IO-* 117 x lo-”
_.__ (2)
Ul73A I 73 IO D 4 1.73 X 10-l 1.21 x 10mh 1.84 x IOmh (i)
ti173B 173 II D 4 3.22 x IO-’ 9.10 x lo-’ 0.29 x IO-’ II)
U173C 173 12 D 4 3.82 x IO--’ I.21 x 10-b 0.36 x l0-h Ci)
U173D 173 I3 D 4 8.67 x IO-’ 7.01 x lo-’ 2.57 x lo-’ (1)
Ul73E 173 14 D 3 1.36 6.39 x IO-’ 1.66 x lo-- (I)
2N 187 2 D 8 1.47 X IO2 1.27 x lo-: 0.58 x lo-’ (e)
U202A 202 10 D 4 1.73 x IO’ 3.21 x IO-” 0.91 x lo-” C.i)
U202B 202 11 D 4 3.22 x 10~’ 2.51 x 1O-6 I.56 x IO-’ (.I)
U202C 202 I’‘. D 4 3.82 x IO-’ 2.11 x lom6 1.31 X lomh 0)
U202D 202 I? D 4 8.67 x IO-’ 1.38 x lom6 0.45 X lomh (.i)
U202E '02 I4 D 4 1.36 9.96 x IO-’ 3.34 X lo-’ (.I)
R3D 213 P 6 8.58 1.05 x lo-b 0.54 x IO-(- (e)
U223A 223 IO D 5 1.73 X 10-l 7.95 x lo-6 1.24 X l0-h (1)
LJ223B 223 D 4 3.22 x IO-’ 6.69 x IO-’ 2.71 x 10m6 (J)
LJ223c 723 12 D 5 3.82 x IO-’ 4.28 x lO-6 1.63 x lO-h (J)
U223D 223 I3 D 4 8.67 x IO-’ 2.24 x lo-6 1.12 X l0-h
U223E 223 I4 D 3 1.36 2.24 x lO-6 1.56 x lo-’ lli
R3C 265 D 4 7.27 I.85 x 1O-6 1.23 x l0-h (e)
R3B 305 D 3 6.70 3.69 x lO-6 1.68 x lO-6 (e)

(a)See Table 2 for a summary of the characteristics of the run materials and RIMSTIDT (1979) for a listing of the
concentration versus time data for each experiment.
(b) P = precipitation and D = dissolution.
(c)A0 = 1 rn’ and M” = I kg.
(d) Reported errors are +20 for kL and + 10% for (A/M).
(e) Results of this study.
(f) Runs B, C and D gave values of k- that are much too large because of the excessive rate of dissolution of the very
small particles in the gels. These data were excluded from the calculation of the temperature function for k-.
(g) Experiments performed by STOBER (1967). Here high pH, sodium carbonate solutions were used so the rates of
dissolution were much higher than expected for pure water. These results were excluded from the calculation of the
temperature function for k-.
(h)Experiments performed by BECKWITH and REEVE (1969). These data were not used to calculate the temperature
function for k_ because of the large uncertainty in the estimated value of (A’M).
(i) Experiments performed by ELMER and NORDBERG (1958).
(j) Experiments performed by USDOWSKI (personal communication, 1974).

one of the subtleties lost by assuming that the parti- these variables see CHRISTIAN (1975). The rate con-
tion functions of the Eyring formulation are equal to stants from Table 3 have been plotted on an Arrhe-
free energy functions. It apparently has little or no nius plot (Fig. 4). From (45), it can be seen that the
temperature dependence and thus, has been con- result should be a linear function if the reaction mech-
sidered a constant in this discussion. The important anism is unchanged over this temperature range.
variable here is the activation energy (E,,,) which From the slope of this line (-&JR), the activation
should not be confused with the free energy of acti- energy for the precipitation reaction has been calcu-
vation (AC*). For a derivation of the relationships of lated (Table 4).
1690 J. D. RIMSTIDT and H. L. BARKS

IQuartz
-6.0 _ AAmorphous
*at2 Sfc
I -7.0
Y

$ -6.0

3.5 2.5 1.5

1000/T, K-’
Fig. 4. Arrhenius plot of experimentally determined precipitation rate constants for silica-water reac-
tions. Qtz sfc indicates values calculated from the dissolution rate of the disturbed layer on quartz
grains. The line IS a least squares fit of the points shown where each was weighted by the number of
concentration-time determinations for each experiment.

Mechanism of’ reaction we need only add them to get

The large activation energies for silica-water reac- log k, = (a, + a*) + b, T + (c, + c,)/T (49)
tions (Table 4) indicate that the rate-limiting step for
these reactions is the breaking of strong Si-0 bonds. The temperature functions for k, are listed in Table 4.
For comparison, the activation energies for dissolu- The h term has been retained in these functions,
tion reactions in aqueous systems where diffusion even though it is not statistically significant, in order
away from the surface is the rate limiting step are in to maintain consistency with the equations that
the range of 17 to 25 kJ mol- ’ (4-6 kcal/mol). express the equilibrium constants for these reactions.
The control of the rate of the silica-water reaction
by bond breaking has an important implication. It APPLICATIONS
means the rate of precipitation of H,SiO, is con-
A few simple applications of our silica-water kin-
trolled by the strength of the Si-0 bonds. The acti-
etics model are illustrated below. Table 5 lists some of
vated complex must be the same for quartz, cristoba-
the variables to be considered in extrapolating experi-
lite and amorphous silica reactions so their reaction
mental rate measurements to natural conditions. This
rates are simply related by the differences in their free
table is intended only as a summary because these
energies (see Fig. 1). Therefore, k_ must be the same
relationships have been discussed at length in RIM-
for quartz, cristobalite and amorphous silica.
STIDT (1979). Laboratory measurements can only be
Calculation of‘ the rate constunt jar dissolution used to predict the behavior of natural geologic sys-
It is an easy task to find k, for the reactions of tems when the variables controlling the phenomenon
silica phases with water provided the equilibrium con- of interest, in this case rHdsio4. have been identified
stant and the rate constant for the precipitation reac- and quantified. Thus, such applications should be
tion are known. From (32) approached with caution. Notice especially that
Table 5 also indicates the limits beyond which the
k+ = WV-) (46)
extrapolation of this work is unreasonable.
Because the equilibrium constants and the precipita-
tion rate constant are given by respective equations of Rate of‘eyuilibration
the form A fundamental expression of how quickly a system
IogK = a, + blT + c,/T (47)
approaches equilibrium is given by the time constant
log k_ = a2 + Q/T (48) (t,) of the system. It is important to recognize that,

Table 4. Temperature functions of the rate constants for silica-water reactions

Activation energy
Temperature function (7; K) (kJ/mol) (kcal/mol)

log k_ (all phases) = -0.707 - 2598/T 49.8 (11.9)

log k, (qtz) = 1.174 -2.028 x 10-j T - 4158/T 67.4-76.6 (16.1 -18.3)
log k, (n-crist) = -0.739 - 3586/T 68.7 (16.4)
log k, (fl-crist) = -0.963 - 3392,/T 65.0 (15.5)
log k, (am sil) = -0.369 - 7.890 x 10m4 T - 3438/T 60.9.-64.9 (14.5-l 5.5)
 The kinetics of silica-water reactions 1691

Table 5. Factors that influence the rate of silica-water reactions

Variable Effect on reaction rate Valid extrapolation range

Temperature Exponential dependence ~300°C

(Arrhenius equation)
Pressure Very little effect C-500 bars
Extent of system Rate directly proportional (A/M) < IO4
to (A) and inversely
proportional to (M)
Activity of H&O, Rate directly proportional At very iarge values of (Q/K)
to (I - Q/K) higher free energy phases may
nucleate
Mechanism Rate controlled by breaking At low (A!,M) nucleation controls
of strong Si-0 bonds the rate of precipitation; at
temperatures above 300 ‘C
diffusion may control the rates
Silica phase Determines K and therefore S This model valid for quartz.
present cristobalite and amorphous silica
PH (?) No indication of any Slightly acid to neutral systems
effect on reaction mechanism
near neutrality
Salts Reduces aH20and thus silica (?) No evidence of any effect
solubility on reaction mechanism
Particle size Very small particles have For particles smaller than 0.1 pm,
higher solubilities than K should be corrected for surface
macroscopic grains free energy

because S approaches unity asymptotically according constant of a system as a function of the temperature
to eqn (41). there is no particular time when the sys- and the extent of the system (A/M) (Fig. 5). Because
tem achieves an ideal equilibrium state. Therefore, it this function depends upon only the precipitation rate
is legitimate to state only how quickly S approaches constant (k_), which is the same for quartz. cristoba-
one. This can be found by raising (41f to a power of e Iite and amorphous silica, this diagram is valid for the
and rearranging the result to obtain equiIibration rate of all of these phases with water.
The realization that most systems have effectively
(I - S)/(l - S,) = em’-’ (50) achieved equilibrium after 5t, allows some immediate
Or deductions about the nature of metastability of the
silica-water reaction. Figure 5 shows, for example,
S = (s,)(emL”) - (em”‘) + 1 (51) that a system with an *(A/M) ratio of 100, which is
The properties of the time constant are most easily approximately equivalent to a solution in a fracture
illustrated by considering a system where initially 20pm wide, will effectively reach equilibrium in less
there is no silica in solution, i.e. So = 0. In this case than one day at 300°C. On the other hand, at 25°C it
(5 1f becomes would take this same system nearly IO yr to achieve
equiIibrium. This means that very high concentrations
S = 1 - emk’g (52) of silica can be maintained in solution at low tem-
peratures for relatively long periods of time provided
The time constant for the reacting system is defined as the (A/M) ratio of the system is not increased by nuc-
I, = l/k’_ leation and growth of new solids.
(53)
so that when t = t,: Rate of reaction

S = (1 - e-‘) = 0.63 The rate of silica precipitation at a particular point

(54)
in a natural system is predicted by the differential rate
or the concentration of silica in solution is 63% of the
equilibrium c~~ncentration. The times to other frac- Table 6. Extent of equilibration at mul-
tions ofequi~ibrium can readily be found by substitut- tiples of the time constant of a system
ing multiples of t, into eqn (52) (see Table 6). How-
ever, the important point here is that when the Elapsed
time Ce-lk’ ““I [, _ e-Ik’ ““]
elapsed time equals 5t, the system has proceeded to
99.37: of equilibrium concentration. Thus, for all 1f, 0.368 0.632
practical purposes, we can consider the system to 2C 0.135 0.856
have attained equilibrium after 5t,. 3r, 0.050 0.950
The rate of equilibration of silica-water reactions 4t, 0.018 0.982
% 0.007 0.993
can be conveniently portrayed by expressing the time
1692 J. D. RIMSTIDT and H. L. RAKNES

0 100 200 300

TEMPERATURE,“C

Fig. 5. The time constant for silica+water reactions. (A/M) is the extent of the system

eqn (36). For this calculation the temperature, the ac- ratio is obtained. The factor, IOOO,is necessary to
tivity of silica in solution, and the (A/M) ratio at the convert the units of V,, (cm3/gm) to the units of
site of deposition must be known. A”/Mo (m3/kg). Since the specific volume of water
The more difficult part of this calcuIation is esti- only varies between 1.00 and 1.40 in the temperature
mating the effective (A/M) ratio. The first step in find- range of O-300°C this correction will often be small
ing this value is to find the ratio of the area of mineral relative to the possible error in the A/V ratio so that
surfaces (A) to the volume of open space (V) which the in most cases the specific volume can be reasonably
solution can occupy. A/V ratios have been determined approximated as one. However. in cases where the
for some simple, common geometries illustrated in most accurate calculation is desired the specific
Fig. 6. These values are for idealized cases. The pres- volume of liquid water along the liquid--vapor curve
ence of surface roughness on fracture walls and min- can be calculated from the function given by HAAS
eral grains or the presence of smaller particles in (1976). The (A/M) ratio must have units of m2/kg so
sediments will increase the A/V ratio. On the other that it can be compared to the standard system of
hand, some minerals may not act as growth surfaces I rn’ interfacial area (A’) and 1 kg of water (MO) for
for silica so the effective A/V ratio is less than which the rate constants from Table 3 apply. Thus,
expected. In these calculations it has been assumed the ratio (A/M)/(A’/M’) gives (A/M) which indicates
that growth surfaces for silica are furnished by sili- how much faster or slower the reaction will proceed
cates and iron oxides and that they are pervasive in in the system of interest relative to the standard sys-
the system. As the temperature of the system in- tem.
creases, the density of the solution decreases so that a Once the (A/M) of the system has been established
given mass of water contacts a larger surface area. the rate of reaction of silica can be found by rewriting
This means that the A/V ratio varies somewhat with (30) in terms of Q and K:
temperature.
By multipIying the A/V ratio by V,,/lOOO the (A/M) rH&iO~ = (~~~)(~~~~~i~~~ (~+~,~~u~~~(l - QiK) (55)

CYLINDRICAL PIPE RECTANGULAR PIPE

f!III
------I- +------,-

A/V= 2nrllnr21= 2r A/V= 2l(h+w)/whl=Z(h+w)/hw

A/M= 2Vap/tOOOr A/M=2~p~h+w)llOOOhw

FRACTURE CLOSE PACKED SPHERES

A/V= Phl/whl= 2/w A.W=l2.5?r=l1.4?r’

A/M = 2V~911000w AJM=8.55VspilOOOr

Fig. 6. (A/V) and (A/M) ratios for some simple geometries. A and V values for various packing configur-
ations of spheres were derived by GKATON and FRASER (1935). All dimensions are in meters and the units
of V,, are cm”,/gm.
 The kinetics of silica-water reactions 1693

TEMPERATURE, “C

Fig. 7. The rate of quartz precipitation as a function of temperature in a system where aHZo, asio2, ~~~~~~~
and (AIM) arc one, The solid lines are for constant values of the activity product, Q, while the dashed
lines are for constant values of the saturation ratio, S.

To provide a simple example, the subsequent calcula- quartz precipitation decreases. At temperatures below
tions will consider the rate of precipitation of quartz 100°C the rates are extremely low even though the
from a supersaturated system where (A/M) is one and degree of supersaturation may be very high. Conse-
the salinity is low enough that ymSio4 and aw,o are quently, at low temperatures, very high degrees of
effectively one. Under these circumstances (55) can be supersaturation can be quenched into solution.
reduced to the rate of change of concentration of sil- The rate that a layer of silica changes in thickness
ica with time can also be determined using the differential rate
equation. The right ordinate of Figs 7 and 8 are
(56)
scaled in terms of the rate a quartz surface would
Temperature functions for K and k, can be found in increase in thickness if it were in contact with a silica
Tables 1 and 4. Using eqn (56), the rate of quartz solution under the temperature and concentration
precipitation (r;14sio4 ) from supersaturated solutions conditions specified by the plotted functions. The rate
can be found as a function of temperature (Fig. 7). that the surface layer thickens (rsrc) is related to r;14sio4
This figure is contoured at constant values of S. The in the following way. The rate of silica dissolution or
abscissa represents S = 1 (equilibrium) where the rate precipitation in moles per square meter for a surface
of precipitation is zero. For example, a solution con- of area A (m2) from a system containing M (kg) of
taining 500 ppm silica is in equilibrium at 274°C and solution is
r;14sio4 is zero. Decreasing the temperature of this
&,sioJA = (dmu,sio,/dt)/A (57)
solution increases S, and thus the rate of quartz preci-
pitation, until at about 230°C a maximum is reached. by definition
Below this temperature the value of the rate constant
(k,) decreases faster than S increases so the net rate of mH&iOd = flH4Si04/M (58)

TEMPERATURE, “C
Fig. 8. The rate of quartz precipitation (logarithmic scale) as a function of temperature for a system
where aHZo, c+,~,, j’H,y104and (A/M) are one. Each line represents a different value of the activity product.
1694 J. D. RIMSTIDT
and H. L. BARNFS

SO geothermal reservoir and, as such, it provides a useful

example of how silica-water kinetics affects natural
(59)
systems. It is assumed that the silica concentration in
The number of moles of silica that dissolve from or a geothermal reservoir is controlled by quartz solu-
precipitate on I m2 in 1 set is bility. Occasionally some geothermal tluid leaks
through the nearly impermeable zone which caps the
(dn ,,sio,/dWA = (&,sio,)/(A/M) (60)
reservoir and finds a way to the surface where it
The rate that the surface changes in thickness can appears in thermal springs. Application of the quartz
then be found by multiplying this rate by the molar geothermometer assumes that the silica concentration
volume of quartz in the water of the thermal springs is identical to the
concentration in the reservoir. Thus. the temperature
rsr, = (V,,,)(r;l,sio,)/(A/W (61)
of the geothermal reservoir can be found by determin-
Combining (55) and (61) gives ing at what temperature the silica concentration in
the thermal spring water would be in equilibrium
with quartz (FOURNIER and ROWE. 1966).
This can be further simplified in a system of low sali- The least tenable assumption here is that the silica
nity to concentration of the water does not change during its
transit through the zone between the geothermal
raft = (~q/,,,M+)U - S) (63) reservoir and the surface. There are three possible
Note that r,,, is independent of the extent of the sys- ways by which this assumption may break down. In
tem. This means that the rate that quartz grows on free flowing, highly fractured systems. some water
the walls of a fracture is controlled only by the degree may boil away causing an increase in the silica con-
of saturation and is independent of its width. centration in the liquid phase. In such a system, the
The rates of silica-water reactions change over quartz geothermometer will indicate a subsurface
many orders of magnitude in the range of O-300°C. In temperature that is too high. The rising geothermal
order to represent reaction rates over this entire tem- fluid may also be diluted by groundwater from a
perature range it is useful to plot r;14sio4 on a logarith- cooler source to reduce the silica concentration
mic scale. Figure 8 follows this format and shows the (TRUESDELLand FOURNIER,1977) or finally, the silica
effect of changing the concentration of silica in solu- concentration may be reduced by silica precipitation
tion on the rate of quartz precipitation. Figure 9 uses as the solution cools and becomes supersaturated. In
the same format to show the reaction rates between these cases the silica geothermometer prediction of
various silica phases and a solution containing the reservoir temperature will be too low. The effects
600 ppm silica. The respective phases are in equilib- of silica-water kinetics on the amount of silica preci-
rium with the solution at the temperature where the pitation will be considered here.
lines become vertical. There are three important variables that can affect
the rate of silica precipitation during the fluid’s trip to
.iZfect of reaction rates on the quartz geothermometer
the surface. These are the geothermal gradient. the
The quartz geothermometer is one of the most (A/M) ratio and the rate at which the fluid ascends
common methods of estimating the temperature of a toward the surface (i.e. the rate of movement perpen-

,<5*
,<;/’
,;?
a-p transition L i

TEMPERATURE, “C
Fig. 9. The rate of reaction of various silica phases with water for a system where Q,~. jlHaSloaand (A/M)
are one and the silica concentration is 600 ppm. The dashed lines reuresent S < 1 (dissolution) and the
solid lines are for S > I (precipitation). They converge to a vertical line at the temperature where the
respective phases are in equilibrium with the solution.
 The kinetics of silicawater reactions 1695

0
near surface geothermal gradient of 138C/km. Al-
though this is somewhat lower than the geothermal
-Magmamax 1,300”lkn gradient mentioned above, the passage of hot geother-
‘Mesa 6-2. 175’1 km mal fluids through a conduit is expected to heat the
E surrounding rocks and, thus, somewhat lower the rate
Y
of change of temperature with depth. The lower the
Z’ Impermeable Zone geothermal gradient, the longer the fluid will remain
at high temperatures where the reaction rates are fas-
% test
GEOTHERMAL
RESERVOIR
Although the plumbing of geothermal systems is in
general quite complex, two extreme cases can be iso-
4 I I lated for discussion. In a system dominated by large
I
loo 200 300 fractures which allow a direct path to the surface
TEMPERATURE, “C (Fig. I l), the (A/M) ratio will be small and the ascen-
Fig. IO. Typical geothermal gradients for geothermal areas sion rate will be high. This means that the fluid will
(PALM~H, 1975 and REED. 1975). The gradient is very high reach the surface essentially unscathed by its trip and
near the surface where conduction controls the heat flux. the concentration of silica that it contains will be
However. within the geothermal reservoir convection of
nearly the same as the concentration in the reservoir
the geothermal Ruid maintains nearly isothermal
conditions.
fluid. This is the best case for the silica geotherm-
ometer. An example of this situation might be found
in Yellowstone National Park, Wyoming. On the
dicular to the isotherms). Although these variables other hand, if the ascending fluid must pass through
are, no doubt, very interdependent in real systems, a series of fine grained, unconsolidated sediments
they have been separated here to emphasize their (Fig. 12) the ascension rate will probably be low and
effects. the (A/M) ratio very high, so that the fluid may re-
The variation of temperature as a function of depth equilibrate at each point along its path. Also, this con-
in typical geothermal areas is shown schematically in figuration makes dilution by near surface water very
Fig. IO. The geothermal gradient above most geother- likely. This is the worst case for the quartz geotherm-
mal reservoirs is controlled by the conduction of heat ometer. Examples of this may be found in the geo-
to the surface and is generally very high. For example, thermal fields in the Imperial Valley of California.
near surface geothermal gradients of ZOO”C/km are The error in the quartz geothermometer tempera-
not uncommon in the Imperial Valley. On the other tures for the spectrum of possible configurations that
hand, convection of the fluid within the geothermal lie between these extremes can be estimated using
reservoir causes it to be nearly isothermal. For the silica-water kinetics. Figure 13 shows the effect of the
subsequent calculations, a reservoir with a tempera- ascension rate on a system with an (A/M) ratio of 100.
ture of 300°C is assumed to occur at a depth of 2 km. which is equivalent to a fracture 20,~m wide (see
If the surface temperature is 25°C this system has a Fig. 6). The concentration of silica remaining in sol-

::::::.::.~~.;~,::ii,,,:,:,:.:.~
‘:.‘:.::::.::.‘:..; ..; I .. . . . . ..~....

FRACTURED ZONE

IMPERMEABLE ZONE

GEOTHERMAL RESERVOIR
(QUARTZ SATURATION)

Fig. I I. Schematic illustration of the ‘best case’ for the quartz geothermometer. Here a fracture gives the
geothermal fluid a direct path to the surface. The surface expression of this situation would be a hot
spring with a high flow rate.
1696 J. D. RIMSTIDT and H. L. BARW

z UNCONSOLIDATED ZONE

+** 7-f ~ ‘f;

?_:‘** *:*:I
* *‘++*‘+*++. IMPERMEABLE ZONE
*A A***
l9
.;.f.);;*:
:i:::::.*.y; :
&.......:.:*:....‘.::. GEOTHERMAL RESERVOIR
.:.::..::.::
.,.. . . . . . .
;::;; ::‘:...> (QUARTZ SATURATION)
. .*.:..:..

Fig. 12. Schematic illustration of the ‘worst cast’ for the quartz gcothermometer. The fluid must travel a
tortous path and thus. it will ascend slowiq and come in contact with large surface areas on which silica
can precipitate. The surface expression of this situation Lcould he a spring with a relatively low Row rate.

ution at a given depth has been calculated using reported as I km;yr which is 3 x IO- ’ m/see (Cusica-
QTZTHERM, a FORTRAN computer program (see nour et al., 1967). If we accept this as an average
RIMSTIDT.1979 for a description of QTZTHERM). value, the ascension rate of fluid in fractures is prob-
This concentration can be converted to a quartz ably much higher so the quartz geothermometer will
geothermometer temperature by referring to the solu- generally give accurate results only for fracture-
bility of quartz as a function of temperature. The tem- dominated systems.
perature at a given depth is determined by the geo- The effect of the (A/M) ratio on the quartz geo-
thermal gradient. So we see here, that if a geothermal thermometer is similar to the effect of the ascension rate
fluid initially at 300°C were to ascend through the (Fig. 14). A 1 cm-wide fracture has an (A/M) ratio of
system at 10-4m,!sec, its silica concentration at the 0.2 while fine-grained sediments would have values of
surface would indicate a reservoir temperature of only 100 to 10,000. Here again, the quartz geothermometer
225°C. The extreme possibilities here are shown by a gives the most accurate results in a highly fractured
vertical line which represents no silica precipitation system. Also notice that most of the silica precipita-
and the dashed line which shows total equilibration tion occurs in the temperature interval from 300 to
with quartz during the ascension of the fluid to the 200°C.
surface. For reference, the rate of fluid movement in The effect of temperature on the rate of silica preci-
the El Tatio, Chili geothermal system has been pitation is illustrated in Fig. 15. Here the rate is 0 at

QUARTZ GEOTHERMOMETER, “C
100 200

5
200 400 600
SILICA CONCENTRATION, ppm
Fig. 13. The effect of the ascension rate on the quartz geothermometer temperature. The geothermal
gradient is 138 Vkm and the (A ‘M) ratlo is 100.
 The kinetics of silica water reactions 1697

QUARTZ GEOTHERMOMETER, OC
100 200
I I I 1

SILICA CONCENTRATION, ppm

Fig. 14. The effect of the (NM) ratio on the quart7 geothermometer temperature. The geothermal
gradient is 138 C/km and the ascension rate is IO 4 m sec.

300°C because the solution is in equilibrium with the tial rate equation has been derived to express the
surrounding rocks but, as it moves upward into rates of dissolution and precipitation of silica phases
cooler regions, it becomes more and more supersatu- in terms of the degree of saturation of the system [eqn
rated and the rate of precipitation increases. Soon, (36)]. In this derivation, the reaction rate is directly
however, the strong temperature dependence of the proportional to the interfacial area (A) between the
rate constants override the effect of supersaturation solid and aqueous phases and inversely proportional
and limit the precipitation rate. So, as the tempera- to the mass of water (M) into which the silica is dis-
ture decreases further, so does the rate, until below solvmg. The simplicity of this rate equation allows it
200°C the rates become negligible in most geologi- to be integrated to a form which relates the degree of
cally realizable systems. This justifies the use of the saturation of the system to the elapsed reaction time
quartz geothermometer for systems which have initial [eqn (41)]. These equations describe reaction rates in
temperatures below 200°C. On the other hand, higher both supersaturated (S > 1) and undersaturated
temperature systems will probably give results which (S < 1) systems (see Fig. 3. for example).
cluster around 200°C so the true reservoir tempera- By fitting experimental data to the integrated rate
ture will have to be determined by other means. equation, rate constants for precipitation (k-) have
been found at a number of temperatures between 0
and 300°C. These rate constants have been correlated
by the Arrhenius equation (Fig. 4) and expressed in
CONCLUSIONS
the form log k- = a + c/T (Table 1). Realizing that
By assuming that the reaction mechanism is di- k, = (K)(kk) allowed the temperature functions for
rectly related to the stoichiometry, a simple differen- log K to be added to that for log k- to obtain log k,

25

0 20 40 60
PRECIPITATION RATE , mol/sec X10”

Fig. 15. The precipitation rate for various (A/M) ratios as a function of temperature and depth. The
geothermal gradient is IX-C’km and the ascension rate ih 10 mJ m,‘sec.
1698 J. D. RIMSTIIJTand H. L. BARNES

as a function of temperature. The coefficients for all of Dr W. F. DOWNS for his help in building and operating the
these equations are listed in Table 4. circulating hydrothermal system in which many of these
experiments were carried out. G. SOLOMON performed the
The activation energy for these reactions range
arg0nB.E.T. surface area measurements on the run ma-
between49 and 77 kJ mol- ’ (11-19 kcal/mol) (Table 4). terial which was furnished by Corning Glass Works. State
These high values indicate that the reaction rates College, Pennsylvania. Dr E. USDOWSKI, Professor of Sedi-
are limited by the breaking of the strong Si(0 mentary Petrography, University of Gottingen provided
bonds. for use in this analysis his excellent experimental data
which allowed the rate model to be tested before embark-
Rewriting the rate equations produces expressions
ing on the experiments, This research was supported by the
for rates in some important, simple cases. For U.S. Department of the Interior, Bureau of Mines Electro-
example, the time constant of a system, defined as metallurgy Research Group under Contract number GO
t, = I/k’, is a fundamental expression of how rapidly 155140 and also. in part, through Grant number AER
it approaches equilibrium. Because a reacting system 74-08473 under the Geothermal Program of the Division
of Advanced Energy Research and Technology of the
has effectively achieved equilibrium when t = 5t,, the National Science Foundation.
larger the time constant of a system, the longer it will
take to equilibrate. Also, the rate of precipitation of
silica at a given temperature and (A/M) can be found
from eqn (56). This rate can be transformed to the REFERENCES

rate that the layer of silica on the surfaces in the BARNES H. L. (1971) Investigations in hydrothermal sulfide
system increases in thickness as given by eqn (61). systems. In Research Techniques for High Pressure and
The effect of silica--water reaction rates on the vali- High Temperature. (ed. G. C. Ulmer). pp. 317-356.
dity of the quartz geothermometer is a good illus- Springer-Verlag.
BECKWITH R. S. and REEVE R. (1969) Dissolution and de-
tration of the application of this work to natural sys-
position of monosilic acid in suspensions of ground
tems This analysis showed that the quartz geotherm- quartz. Geochim. Cosmochim. Acta 33, 745~ 750.
ometer is generally accurate in areas that are highly CHRISTAIN J. W. (1975) The Theory oj Transformations in
fractured so that the reservoir fluid can pass quickly Metals and Alloys. Part 1. Equilibrium and General Kin-
etic Theory, 2nd edn. Pergamon Press.
to the surface without lingering contact with large
CUSICANOUR H. MAHO~’ W. A. J. and ELLIS A. J. (1967) The
areas of mineral surfaces at high temperatures. Fortu- geochemistry of the El Tatio geothermal held. northern
nately, most geothermal areas are tectonically active Chili. 2nd U. N. Symp., Deu. Use Geothermal Resow.
so faulting and fracturing are common. Also, because Vol. 1, pp. 70337 11.
of the rapid increase in the rate of quartz precipita- DEVINE J. C. and SHUR N. H. (1977) Determination of
silicon in water samples, At. Ahsorpt. Newsl. 16, 39-41.
tion with temperature, the quartz geothermometer is
ELMER T. M. and NORDBERG M. E. (1958) Solubility of
most accurate for lower temperature systems. Systems silica in nitric acid. J. Am. Ceram. Sot. 41, 517-520.
giving a geothermometer value of 200°C or above EYRING H. (1935) The activated complex m chemical reac-
should be investigated by other means since this value tions. J. Chem. Phys. 3, 107.
may be much lower than the actual reservoir tem- FOUR~ZIERR. 0. and ROWE J. J. (1966) Estimation of under-
ground temperature from the silica content of water
perature. Finally, in a given geothermal region, the
from hot springs and wet-stream wells. Am J. Sci. 264,
springs that have the highest flow rates and are the 6855697.
hottest are most likely fed by the largest fractures. GOVETT G. J. S. (1961) Critical factors in the calorimetric
Thus. because they have the highest ascension rate determination of silica. Anul. Chim. Acfa 25, 69-80.
and the lowest (A/M) ratios, they will give the most GRATON L. C. and FRASER H. J. (1935) Systematic packing
of spheres. J. Grol. 43, 7855909.
accurate values from the quartz geothermometer. HAAS -J. L. JR (1976) Physical properties of coexisting
The effect of reaction rates on the quartz geotherm- phases and the thermochemical properties of the H,O
ometer illustrates that the major problem in applying component in boiling NaCl solutions. L’.S. Grol. Surr.
kinetics to natural systems is identifying and quantify- Bull. 142lA, AIO.
LEAMNSON R. N. THOMAS J. JR and EHRLINC;ER H. P. III
ing the variables involved. This problem was circum-
(1969) A study of the surface areas of particulate micro-
vented here by choosing values for the important crystalline silica and silica sand. 111. Sturr Grol Surr.
variables from a reasonable range and assuming that Circ. 444.
they were constant over time. However, many other PALMER T. D. (1975) Characteristics of geothermal wells
located in the Salton Sea geothermal field, Imperial Val-
systems, for example, interpretation of the time and
ley, California. Lawrence Livermore Laboratory Rept.
fluid quantities involved in the formation of quartz UCRL-51976.
veins. will require a knowledge of how the rate and REED M. J. (1975) Geology and hydrothermal metamor-
solubility controlling variables change over time. In phism in the Cerro Prieto geothermal field. Mexico, 2nd
such cases, a comprehensive model of the thermal and II. N. Symp. Dev. Use Grorhermal Resow Vol. I, pp.
539-547.
hydrodynamic history of the overall system will be
RIMSTI~T J. D. (1979) The kinetics of silica-water reactions.
necessary. Ph.D. Thesis, The Pennsylvania State Univ.
SEARS G. W. JR (1956) Determination of the specific surface
area of colloidal silica by titration with sodium hydrox-
ide. Antrl. Chem. 28, 19XI- 1983.
AL.kl~Ow/edgr,nenfs~We are grateful to Drs L. DARKEN, SOLOMON G. C. (1978) Kinetics of interaction between bio-
A. LASAGA and D. LANGMC’IR for reviewing parts of this tite adamellite, granitic gneiss and aqueous chloride
manuscript and providing many useful suggestions and to solutions. M.S. Thesis. The Pennsylvama State Univ.
 The kinetics. of silica-water reactions 1699

STOBEH W. (1967) Formation of silicic acid in aqueous sus- WADE W. H., COLE H. D., MEYER D. E. and HACKERMAX
pensions of different silica modifications. Adv. Chem. Ser. N. (1961) Adsorptive behavior of fused quartz powders.
67. 161 18’. Adv. Chem. Ser. 33, 3541.
TKUSIXLI A. H. and FOUKNIER R. 0. (1977) Procedure for WERT C. A. and ZENER C. (1949) Interstitial atomic diffu-
estimating the temperature of a hot-water component in sion coefficients. Phys. Rec. 76, 1169-l 175.
a mixed water by usin? a plot of dissolved silica versus WHALEN J. W. (1961) Heats of immersion in silica-water
enthalpy. J Res. U.S. Guol. Sum. 5, 49-52. systems. Ada. Chem. Ser. 33. 28lL290.