Pergamon

Geochimica et Cosmochimica Acta, Vol. 64, No. 13, pp. 22552266, 2000 Copyright 2000 Elsevier Science Ltd Printed in the USA. All rights reserved 0016-7037/00 $20.00 .00

PII S0016-7037(00)00341-0

Kinetics of calcite growth: Surface processes and relationships to macroscopic rate laws
H. HENRY TENG,1, PATRICIA M. DOVE,1,* and JAMES J. DE YOREO2
2

School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, GA 30332, USA Department of Chemistry and Materials Science, Lawrence Livermore National Laboratory, Livermore, CA 94550, USA (Received June 18, 1999; accepted in revised form January 18, 2000)

AbstractThis study links classical crystal growth theory with observations of microscopic surface processes to quantify the dependence of calcite growth on supersaturation, , and show relationships to the same dependencies often approximated by afnity based expressions. In situ Atomic Force Microscopy was used to quantify calcite growth rates and observe transitions in growth processes on {1014} faces in characterized solutions with variable . When 0.8, growth occurs by step ow at surface defects, including screw dislocations. As exceeds 0.8, two-dimensional surface nucleation becomes increasingly important. The single sourced, single spirals that are produced at lower were examined to measure rates of step ow and the slopes of growth hillocks. These data were used to obtain the surface-normal growth rate, R m , by the pure spiral mechanism. The dependence of overall growth rate upon dislocation source structure was analyzed using the fundamentals of crystal growth theory. The resulting surface process-based rate expressions for spiral growth show the relationships between R m and the distribution and structures of dislocation sources. These theoretical relations are upheld by the process-based experimental rate data reported in this study. The analysis further shows that the dependence of growth rate on dislocation source structures is essential for properly representing growth. This is because most growth sources exhibit complex structures with multiple dislocations. The expressions resulting from this analysis were compared to afnity-based rate equations to show where popular afnity-based rate laws hold or break down. Results of this study demonstrate that the widely used second order chemical afnity-based rate laws are physically meaningful only under special conditions. The exponent in afnity-based expressions is dependent upon the supersaturation range used to t data. An apparent second order dependence is achieved when solution supersaturations are very near equilibrium and growth occurs only by simple, single sourced dislocation spirals. These ndings indicate the need to apply caution when deducing growth mechanisms and rate laws from temporal changes in bulk solution chemistry. Observations of various types of surface defects that give rise to step formation suggest that popular rate laws are sample-dependent composites of rate contributions from each dislocation structure. Copyright 2000 Elsevier Science Ltd
1. INTRODUCTION

The abundance of calcium carbonate minerals throughout natural and engineered earth systems has motivated investigations of calcite crystallization over the last century. It is widely recognized that an understanding of the kinetics and mechanisms governing growth is of rst order importance for predicting mineralization and thus, acquiring the ability to control it. The signicance of harnessing mineralization phenomena is seen in the remarkable ability of organisms to direct the crystallization of aqueous ions into biogenic materials that express a diverse morphologies and biological functions. Renewed efforts to unravel the physical basis of carbonate biomineralization are largely focused on understanding the in vivo processes that control growth at the interface between a biological matrix and the biominerals that form. Future advances in controlling or directing the growth of carbonate minerals also hinge upon clarifying two uncertainties: (1) The dependence of growth mechanism upon supersaturation; (2) The microscopic

* Author to whom correspondence should be addressed: P. M. Dove, Department of Geological Sciences, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061. (dove@eas.gatech.edu). Present address: Argonne National Laboratory, ER-203, Argonne, IL 60439. 2255

surface processes that control the macroscopic manifestations of overall growth rates. Answers to these questions establish the knowledge base for constructing general models that quantify growth in complex mixtures of organic and inorganic constituents. They also establish relationships between microscopic processes and macroscopic growth rates determined by what are known as bulk methods throughout the scientic literature. However, a review of the ndings published to date suggests that these issues have not been adequately explored. Growth studies conducted over a wide range of supersaturations have established our collective understanding of growth kinetics and mechanisms for calcite crystallization. Their ndings can be summarized according to three experimental methods with different length-scales. At a very short scale, direct observations of growth using Atomic Force Microscopy (AFM) conrm that calcite growth occurs by step ow, the advancement of individual molecular layers generated at crystal imperfections (Hillner et al., 1992a,b; Gratz et al., 1993) or by two-dimensional surface nucleation (Dove and Hochella, 1993). However, these studies report contradictory relationships between growth mechanisms and supersaturation. For example, Gratz et al. (1993) reports that growth by spiral formation is the only growth mechanism at supersaturations as high as 6.9 where the supersaturation index is dened as the ratio of ion activity product (IAP) to solubility product

2256

H. H. Teng, P. M. Dove, and J. J. De Yoreo

(K sp ). In contrast, Dove and Hochella (1993) report that surface nucleation becomes the dominant growth mechanism at conditions much closer to equilibrium ( 12). Studies investigating growth at larger scales have examined the products of step assembly to form polygonized hillock features. By using growth solutions with relatively high supersaturations ( 10 24), Paquette and Reeder (1990, 1995), Staudt et al. (1994), and Reeder (1996) document the formation of growth hillocks on {1014} calcite faces with diameters of up to hundreds of microns. The local supersaturations in these studies could not be well characterized due to static hydrodynamic conditions of the growth media. Nonetheless, examinations of these mesoscale growth structures after removal from the growth media led the investigators to conclude that growth occurred predominantly by a spiral mechanism. At the largest scale, macroscopic studies have evaluated the rates and mechanisms of calcite crystallization using indirect methods that monitored changes in solution chemistry over the course of growth. These studies establish the dependence of growth kinetics upon chemical and physical parameters such as supersaturation state, pH, pCO2, ionic strength, and temperature (e.g., Nancollas and Reddy, 1971; Plummer et al., 1978; Busenberg and Plummer, 1986; Christoffersen and Christoffersen, 1990; Shiraki and Brantley, 1995). Growth mechanisms were postulated in these investigations based upon the rate of solution composition change with time. These bulk kinetic measurements could not discern the possibly different contributions of unique crystal faces to the overall growth rate or relate solution chemistry to the range of microscopic growth processes occurring at mineral surfaces. Three major categories of elaborate kinetic models for crystallization have arisen from interpretations of solution composition data obtained by these indirect studies based upon (1) surface complexation (Nilsson and Sternbeck, 1999; Arakaki and Mucci, 1995); (2) summation of the elementary reactions (Plummer et al., 1978; Busenberg and Plummer, 1986); and (3) chemical afnity (Smallwood, 1977; Reddy, 1977; Reddy and Gaillard, 1980; Compton and Daly, 1987; Reddy, 1988). Surface complexation-based rate laws take into account the reactions involving surface speciation, while elementary reactionbased rate laws describe growth rate as a function of multiple elementary reactions. Afnity-based models, the most widely used of the three, are developed in terms of free energy changes, G (chemical afnity) or exp( G/RT) (supersaturation index, ), of precipitation reactions (e.g., Nancollas and Reddy, 1971; Reddy and Nancollas, 1971; Morse, 1978; House, 1981; Nielsen, 1983; Mucci and Morse, 1983; Mucci, 1986) to yield two types of rate laws: linear and nonlinear with respect to G. Linear rate laws have the following general form: Rm k exp n G/RT 1 (1a)

attributed to adsorption-controlled growth (Nielsen, 1983). Expressions of this form with n 1/ 2 and 1 were used to describe the precipitation kinetics of calcite (Nancollas and Reddy, 1971; Kazmierczak et al., 1982). Nonlinear rate laws are generally expressed as (Lasaga, 1981): Rm k exp G/RT 1
n

(1b)

where the net rate, R m , (moles area 2 time 1), is characterized by the rate constant of the forward reaction, k , (moles area 2 time 1) and the free energy change of the overall reaction. R and T are the molar gas constant and temperature (K), respectively. The parameter n is a constant and has been assumed to contain information about the growth mechanism. For example, the kinetic behavior described by Eqn. 1a with n 1 has been

Theoretical models have been used to argue that second order equations of this form (n 2) describe growth at screw dislocations by the spiral mechanism while higher order ones (n 2 3) can be applied to growth at both screw and edge dislocations (Blum and Lasaga, 1987). The kinetics of calcite precipitation has been tted to second order expressions at ambient (Reddy and Nancollas, 1973; House, 1981) and elevated temperatures (Shiraki and Brantley, 1995). Others have used an expression that combines Eqns. 1a and 1b to describe calcite growth (Inskeep and Bloom, 1985). Together, these models delineate the currently accepted quantitative representations of calcite crystallization across a range of chemical compositions and conditions. However, these rate laws stand without substantiative conrmation of the actual growth processes that occur at mineral surfaces. This is a signicant deciency that constrains current models to provide only empirical and semi-quantitative representations of calcite precipitation kinetics. A closer analysis raises concerns about the relevance of these general rate laws to calcite growth. For example, Eqn. 1b with n 2 was used in theoretical (Blum and Lasaga, 1987; Lasaga, 1998) and experimental (Shiraki and Brantley, 1995) studies to describe spiral growth without considering the effect of various types of dislocation sources, including the strain factors associated with these defects, on growth rate. Yet, the classic crystal growth theories predict that the growth rate of a spiral must be strongly controlled by the structure of dislocation sources (Burton et al., 1951). Experimental observations have upheld these predictions (Rashkovich, 1991; Vekilov and Kuznetsov, 1992; Vekilov and Rosenberger, 1996; Land et al., 1997; De Yoreo et al., 1997). In relating this fundamental aspect of crystal growth theory to accepted bulk rate expressions, an important issue arises: Chemical afnity-based rate laws are not a function of dislocation source structures. The only measurable parameter in these expressions is the quantity exp( G/RT). Hence, rate laws represented by these expressions cannot reect the controls of microscopic growth parameters (step velocity and slope of growth hillocks) on growth kinetics or reliably indicate the actual growth mechanism. Indeed, the broad range of overall growth rates that arise from the complexities of various dislocation sources naturally yield a quasi-parabolic behavior that has little mechanistic signicance. The goal of this study is to explore these concerns by comparing experimental evidence with the theoretical models to show where popular growth expressions are valid or break down. By using in situ AFM, this was done by (1) Determining the supersaturation range where growth occurs only by spiral formation; (2) Determining the critical supersaturation value that marks the activation of growth by a surface nucleation mechanism; (3) Measuring step ow rates and the corresponding slopes of growth hillocks in well-characterized solutions;

Kinetics of calcite growth Table 1. Summary of salt concentrations used in each growth experiment and the corresponding solute activities and supersaturations. Salt concentrations [CaCl2] (10 M) [NaHCO3] (10 3 M) [NaCl] (M) a Ca2 (10 5) 2 a CO3 (10 5) SDb
a b 4

2257

1.60 5.00 0.10 5.59 5.37 0.042 (0.02)

1.70 5.25 0.10 5.91 5.62 0.140 (0.02)

1.80 5.50 0.10 6.23 5.95 0.274 (0.04)

1.90 5.80 0.10 6.57 6.20 0.346 (0.04)

2.00 2.04 2.07 2.10 2.25 6.15 6.25 6.35 6.40 6.85 0.10 0.10 0.10 0.10 0.10 Calculated activities and supersaturationsa 6.88 6.99 7.10 7.17 7.68 6.59 6.70 6.77 6.84 7.32 0.450 0.484 0.512 0.534 0.668 (0.04) (0.04) (0.04) (0.04) (0.04)

2.45 7.45 0.10 8.29 7.96 0.826 (0.04)

2.75 8.25 0.10 9.21 8.81 1.030 (0.04)

3.00 9.00 0.10 9.95 9.58 1.196 (0.02)

3.42 10.10 0.10 11.18 10.71 1.422 (0.02)

All growth solutions had a xed pH of 8.5. Parentheses give standard deviation for each calculated supersaturation.

(4) Calculating the overall or normal growth rates of single spirals using measurements of step speed and hillock slope; and (5) Analyzing the dependence of overall growth rate upon dislocation source structure. We show that spiral growth can be approximated by a second-order chemical afnity-based rate law only when supersaturation is very low and growth on a given crystal face is dominated by step ow from simple spirals. We also show that the kinetic data obtained in (4) represent only a minimum overall rate for pure spiral growth in calcite because growth generated by complex dislocation sources usually possesses a higher rate than that for single sources.
2. EXPERIMENTAL PROCEDURES 2.1. Sample and Solution Preparation Calcite was grown on {1014} faces of fragments cleaved from a crystal of optical-quality Iceland spar (Wards Scientic, Chihuahua, Mexico). The fragments (approximately 2 2 0.5 mm3) were handled with tweezers to avoid surface contamination by skin oils and were cleaned with bursts of N2(g) to remove small particles from the surfaces. Each experiment began by mounting a freshly cleaved fragment onto a glass cover-slip using sealing wax. This cover-slip was subsequently adhered to a magnetic steel puck using double-sided adhesive tape. Supersaturated CaCO3 solutions were prepared by dissolving reagent grade sodium bicarbonate (NaHCO3, Aldrich ) and calcium chloride (CaCl2 H2O, Aldrich ) into deionized water. The ionic strength of each growth solution was xed within a narrow range of 0.105 to 0.111 M using reagent grade NaCl (Baker ). The pH of all solutions was adjusted to 8.50 using 0.5 N NaOH prior to injection into the Fluid Cell of AFM. 2.2. Solution Speciation and Supersaturation The chemical speciation of each solution was determined using the numerical code HYDRAQL (Papelis et al., 1988) assuming that the AFM uid cell and input reservoir approximated a closed system. Aqueous complexes considered in the calculation were CaCO, 3 CaHCO3 , CaOH , NaCO3 , NaHCO, CO3 HCO3 , and H2CO. Be3 3 cause the uid cell was isolated from the ambient atmosphere in a liquid-full reactor environment, we assumed that the amount of CO2(g) in the cell was negligible for the aqueous speciation calculation. The activity ratio of Ca2 to CO2 was forced to equal 1.04 0.01 by 3 adjusting the amount of NaHCO3 and CaCl2 H2O, and the Davies equation (Davies, 1962) was used to correct for activity. The supersaturation, , was calculated by: ln a ae
a

cules in the aqueous and solid phases, k B is the Boltzmann constant, a i is the activity of the ith species and K sp is the solubility product of calcite at zero ionic strength. The Teng et al. (1998) value of 10 8.54 for K sp at 25C was used to compute supersaturation. This value is lower than the most widely reported values ranging from 10 8.48 to 10 8.29 (Mucci, 1983, and the refs. therein; Stumm, 1992; Drever, 1997). This low value was used because we observed that steps on {1014} faces in these experiments still advanced at measurable rates when the ionic activity product (IAP) of the experimental solution was equal to the lowest reported Ksp. Hence, we determined K sp from the IAP measurement where step velocity equaled zero. The 15% difference between 10 8.54 and the lowest reported value of 10 8.48 cannot be accounted for by the experimental errors of 2 to 4% reported in Table 1. Teng et al. (1998) suggests that the value of 10 8.54 is due to the lower pCO2 of the closed experimental system or indicates a value that is specic to {1014} faces. This study presents results for 0.04 to 1.4 ( 1.04 4.06) using the solution compositions and activities of Ca2 and CO2 given in Table 1. 3 2.3. Imaging by Fluid Contact AFM In situ observations of calcite growth were made by Contact Mode using a Nanoscope IIIa Scanning Probe Microscope (Digital Instruments) equipped with a piezoelectric scanner capable of scan areas to a maximum of 120 120 m. Surfaces were imaged using commercially available Si3N4 cantilevers that have triangular tips with a length of 200 m, and a force constant of approximately 0.38 Newton m 1. Radii of lever tips were approximately 30 to 50 nm, which corresponds to a theoretical contact force of 27 46 10 9 N between tip and sample (Eggleston, 1994). To reduce the possibility of artifactual changes in micro-topography by scanning tip-surface interactions, the contact force was carefully minimized. Scan rates ranged from 5.8 to 29 Hz with 512 lines per scan, which corresponds to a capture time of 88 to 18 s per image, respectively. Scanner drift was minimized by allowing the instrument to thermally equilibrate before imaging. The experiments were conducted by rst imaging the mineral samples in air to locate a relatively at area and to optimize image quality. The reactant solution was then input as a continuous ow through an AFM uid cell with an internal volume of 50 L using a syringe pump. Once in solution, the crystallographic orientation of an individual calcite fragment was established using methods described previously (Teng and Dove, 1997; Stipp et al., 1994). All images were collected using ow-through rates greater than 30 ml/hr after preliminary experiments showed that step speed on calcite {1014} faces become independent of ow rate above 30 to 40 ml/hr. In situ measurements of temperature in this ow-through environment were 25C. 2.4. Step Advancing Rate and Hillock Slope Measurements Data were collected by rst locating a single spiral hillock (e.g., Fig. 1A) at each supersaturation. The spiral was usually allowed to grow for at least an hour in a supersaturated, but near-equilibrium, solution to ensure its quality and stability. The input solution was then changed to a new one with the desired and maintained for another 10 to 30 min (depending on imaging size and growth rate) whereas the step ow rate

k BT where

ln

Ca2 aCO2 3 K sp

(2)

is the chemical potential difference between CaCO3 mole-

2258

H. H. Teng, P. M. Dove, and J. J. De Yoreo Measuring the advancement of individual steps relative to the y scan direction. This was done by adjusting the scan angle to orient the step train parallel to the y axis and then disabling the slow scanning at the center of the spiral. This results in an image that records the movement of steps as the evolution of individual points at step edges over the imaging area as shown in Fig. 1b. Using the measured angle, (Fig. 1b), v s was then calculated for the step trains on both sides of the hillock by: vs Sr N A tan (3)

where S r is the scan rate (line/second), A the scan size, and N the sampling rate (line/scan, 512 in this study). In a second measurement, the slopes of spirals were calculated as the ratios of step height, h, to terrace width, , by p h (4)

where h is the height of individual molecular layer (0.31 nm) on {1014} faces. 3. EXPERIMENTAL RESULTS

3.1. Dependence of Growth Mechanism Upon Supersaturation Layer growth was observed as the advancement of monomolecular (3) steps. These steps were generated by one or two mechanisms depending upon the surface structure and the saturation state of the input solution. At low values of ranging from 0 to 0.8, steps were initiated solely at dislocations, obvious crystal imperfections, and grain boundaries. Thus, crystal defect-originated growth by step ow from single and complex dislocation sources was the only growth mode. As shown in Figs. 2AD and Figs. 3A, this mechanism resulted in several types of hillocks that became visible within a few minutes after the introduction of a supersaturated solution and new steps were formed only at defect sources that evolved into the apex of a hillock. Examination of more than 50 samples revealed that the most common types of hillocks (more than 80%) were multiple spirals generated by individual dislocations with multiple Burgers vectors (single sourced, multiple growth, Figs. 2A,B), dislocation groups (multi-sourced, multiple growth, Fig. 2D), and those that originated at obvious crystal imperfections (Fig. 2C). Single spirals (less than 20%) were the least common and often suppressed by multiple ones. In the areas where presumably no imperfections intersected the surface, or the high energy sources were overgrown by the newly crystallized layers, growth occurred only at the existing step edges and no new steps were generated. When supersaturation exceeded approximately 0.8, steps began to be also generated by what appeared to be a homogeneous surface nucleation mechanism (Fig. 3B, C). However, AFM cannot distinguish nucleation on calcite lattice points (homogeneous) from nucleation at randomly distributed impurity sites or particles (heterogeneous). In this study, we will refer to this type of mechanism as homogeneous or twodimensional surface nucleation. Growth by surface nucleation occurred at randomly distributed two-dimensional nuclei on the substrate surface. Subsequent growth by this mechanism did not lead to hillock development. Nonetheless, growth originated from defects and by two-dimensional surface nucleation co-existed over the experi-

Fig. 1. Fluid Cell AFM image of a spiral hillock on a (1014) face with the slow scanning direction (A) enabled and (B) disabled across the apex of the hillock. The individual steps along each direction are mono-molecular units having a height of 3. The hillock exhibits a symmetry plane parallel to the c-glide as shown by the dashed line in (A). The angles formed by step train and y-axis, designated by in (B), were measured for both sides to estimate the step velocities at [441] and [441] . See text for details.

and geometry of the spiral adjusted to the new . Subsequently, a minimum of six images were captured in continuous mode for later measurements by image analysis. Measurements of step velocity, v s and v s , hillock slopes, p and p , were made for the positive directions, [441] and [481] , and negative directions, [441] and [481] , respectively. Step velocity was determined by measuring step displacement over time using one of two methods: (1) Determining the difference between the distances of two equivalent step edges referenced to a xed point (e.g. the dislocation source in the growth spiral shown in Fig. 1a) in sequential images; (2)

Kinetics of calcite growth

2259

Fig. 2. Growth hillocks on {1014} faces of calcite. (A) Growth generated by a dislocation with a Burgers vector of two results in a double spiral. (B) Growth generated by a dislocation with a Burgers vector of three results in a triple spiral. (C) Growth initiated at an obvious surface imperfection to yield a growth hillock with a growth unit of two mono-molecular layers. (D) A closeup view (1 1 m) of growth initiated by a dislocation with a Burgers vector of two but complicated by obvious surface imperfections to result in a multi-sourced, multiple hillock.

mental supersaturation of 0.8 as illustrated in Fig. 3B. When both mechanisms were operative, two-dimensional nuclei formed primarily within at areas on the surface rather than on terraces of spiral hillocks. Increasing supersaturation resulted in faster nucleation rates and the development of more 2-D nuclei (Fig. 3C), indicating that growth by homogeneous surface nucleation was increasingly dominant with increasing supersaturation. 3.2. Growth Rate of {1014} Faces for Single Spirals The overall growth rate of a crystal face, or, the growth rate normal to the surface, R m (length/time), is given by (Burton et al., 1951):

Rm

pv s

(5)

Measurements of step velocities, v s , and hillock slopes for single spirals, p, over the range of supersaturation used in this study were found to be unique to the positive directions, [441] and [481] , and the negative directions, [441] and [481] (Table 2). For each direction, v s showed a complex dependence upon the deviation of equilibrium activity, (a a e ), where p scaled inversely with supersaturation, . Details of the dependence of direction-specic step velocity and slope on saturation state were discussed elsewhere (Teng et al., 1998; Teng et al., 1999). The normal growth rate of a single spiral, R m , was obtained using Eqn. 5 and the measurements of v s and p.

2260

H. H. Teng, P. M. Dove, and J. J. De Yoreo Fig. 3. Fluid Cell AFM images showing the growth mechanism on a (1014) face at different supersaturation states. (A) Observations of spiral growth collected at 0.4 within minutes after the input of solution. Three spirals are observed in the imaging area. Spirals 1 and 3 are single ones, and spiral 2 is a convolution of two double ones. In the area where dislocations are absent, growth occurred by the advancement of existing mono-molecular layers. (B) Co-existence of spiral growth (denoted by s) and homogeneous surface nucleation growth (denoted by n) at 1.0. Spirals occurred in the bottom part of the image where dislocations were linearly distributed. Two-dimensional nuclei appear in the upper right portion of the images. Growth by advancement of existing steps dominated in the rest imaging area. (C) The dominance of growth by two-dimensional surface nucleation was recorded at 1.6 within tens of seconds after the input of solution. Two-dimensional surface nuclei at 100s nm scale formed randomly on the surface with some of them already coalesced to form bigger nuclei. Continuous surface nucleation is also observed at several locations (denoted by n). 4

Estimates of R m increased from 10 1010 8 mm/s over the experimental supersaturation range (Table 2). The dependence of R m upon supersaturation for a single spiral is illustrated in Fig. 4.
4. DISCUSSION

In situ AFM observations demonstrate that new steps are generated only at surface imperfections when supersaturation is low, but form by both crystal defects and two-dimensional surface nucleation as supersaturation increases. These observations are consistent with the predictions of classical BCF theories (Burton et al., 1951). The evolution in mechanism with supersaturation has important implications for quantifying the overall growth rate, R m , associated with microscopic surface processes and for interpreting the widely used macroscopic rate laws that are based upon chemical afnity. 4.1. Growth Rates, Rm, Controlled by Simple and Complex Sources The growth rate of crystal faces and, therefore, the overall growth rate of a crystal by a spiral mechanism is described by microscopic parameters v s and p through Eqn. 5 (Burton et al., 1951; Rashkovich, 1991). Because experimental observations show that growth of calcite at low supersaturations occurs by the formation of different types of hillocks (Fig. 2), physically representative rate expressions must account for the contributions of all of the important types of dislocation source structures to growth kinetics. This is difcult to achieve using chemical afnity based rate laws since the only variable in these expressions is the supersaturation ratio, . This discussion examines the types of rate expressions that result from growth by three predominant and increasingly complex modes: single spirals, single-sourced multiple spirals (spirals generated by one dislocation with a Burgers vector 2), and multisourced multiple spirals (spirals generated by multiple dislocations). To begin, rst recall that the dependence of step velocity upon solute activity in the absence of an impurity effect is given by: vs a ae (6)

Kinetics of calcite growth Table 2. Summary of experimental measurements of step velocities, step densities, and calculated overall growth rates.

2261

0.042 v s (nm/s) SDa v s (nm/s) SDa p (10 3) SD (10 3)a p (10 3) SD (10 3)a R m (10 9 mm/s) R m (10 9 mm/s)
a

0.140 0.35 (0.10) 0.83 (0.09) 0.47 (0.01) 0.44 (0.01) 0.16 0.36

0.274 0.52 (0.10) 1.13 (0.04) 0.72 (0.01) 0.48 (0.04) 0.37 0.54

0.346 1.18 (0.16) 1.87 (0.12) 0.71 (0.04) 0.58 (0.03) 0.84 1.08

0.450 1.63 (0.10) 2.15 (0.14) 0.75 (0.02) 0.46 (0.01) 1.22 0.99

0.484 1.89 (0.18) 2.19 (0.18) 0.61 (0.01) 0.47 (0.03) 1.15 1.02

0.512 2.44 (0.20) 2.41 (0.20) 0.58 (0.01) 0.54 (0.02) 1.41 1.30

0.534 2.52 (0.17) 2.54 (0.17) 1.23 (0.01) 2.12 (0.06) 3.86 5.38

0.668 3.59 (0.09) 2.61 (0.06) 1.69 (0.01) 2.81 (0.10) 6.07 7.33

0.826 5.18 (0.14) 3.08 (0.06) 2.23 (0.13) 4.27 (0.35) 11.55 13.05

1.030 8.10 (0.13) 3.69 (0.09) 2.76 (0.07) 6.10 (0.41) 22.36 22.51

1.196 9.46 (0.14) 4.15 (0.09) 3.15 (0.25) 6.85 (0.33) 29.80 28.42

1.422 13.10 (0.10) 4.64 (0.09) 3.15 (0.33) 7.09 (0.25) 41.27 32.90

0.15 (0.04) 0.3 (0.04)

Parentheses give standard deviation for each measurement.

where is the kinetic coefcient (or rate constant) in length/ time, is the specic molecular volume of crystal, and (a a e ) is the difference between actual and the equilibrium activities of the solute (Chernov, 1961; Chernov and Komatsu, 1995). From Eqn. 2, the solute activity in a solution can be expressed as: a a e exp G a e exp RT (7)

expressed as (Burton et al., 1951; Rashkovich, 1991; Teng et al., 1998): Lc 2 k BT (10)

k BT

Substituting Eqn. 7 into Eqn. 6 gives the dependence of step velocity upon chemical afnity: vs a e exp G RT 1 (8)

where is the step edge free energy (per unit length per unit height). From this basis, the properties of the simplest type of spiralsa spiral generated by one dislocation with a Burgers vector of one (Fig. 1A) can be quantied. The hillock slope, p, of such spirals can be obtained by combining Eqns. 4, 9, and 10 into the expression: p hk BT 8 G RT (11)

Next, the terrace width, , of a single spiral can be estimated from the critical step length, L c , by (Burton et al., 1951): 4 Lc (9)

where relates to the delay of step advancement when step length is comparable to the magnitude of L c and is approximately one for calcite (Teng et al., 1998). In addition, L c can be

which indicates that hillock slopes have a positive correlation with solution supersaturations as illustrated by Fig. 5. Notice that the denition G/RT (12)

was used to develop Eqn. 11. It follows from Eqn. 5 that, in the absence of impurities, the overall growth rate for the simplest type of spiral growth is the product of Eqns. 8 and 11: Rm a e hk BT 8 exp G RT 1 G . RT (13)

This is illustrated in Fig. 6A that shows R m has a superlinear dependence upon supersaturation. A more complex, second type of spiral growth can occur when the dislocation source has a Burgers vector of m (Fig. 2AB). In this case, the growth unit is multiple steps of m. Hence, Eqn. 11 becomes p mhk BT 8 G RT (14)

and R m can be estimated by:

Fig. 4. The overall growth rate of a single spiral measured on a (1014) face of calcite in the experimental supersaturation range. The line is a best t to the data.

Rm

ma e hk BT 8

exp

G RT

G . RT

(15)

2262

H. H. Teng, P. M. Dove, and J. J. De Yoreo

Fig. 5. Fluid Cell AFM images of a spiral hillock on a (1014) face taken at width with increasing supersaturation.

0.3 and 0.8 show the decreasing terrace

This is shown in Fig. 6B where the dependence of R m on supersaturation is m times greater than for a single spiral. A third, more complicated type of spiral growth occurs when the source is a group of dislocations. Burton et al. (1951) demonstrated that, for a group of screw dislocations, if the separation between them is smaller than one half of the step width generated by individual dislocations, they co-operate and form a multi-sourced multiple spiral. Consider a group of m screw dislocations of the same rotation direction lying on a line of length . Assuming that the separations between these dislocations are less than /2 generated by individual dislocations, a multi-sourced multiple spiral hillock develops (Fig. 2D). The magnitude of p in this case is controlled by m, , and L c through (Rashkovich, 1991, Vekilov et al., 1992): p mh 4 Lc 2 . (16)

Combining Eqns. 10 and 12 and then substituting into Eqn. 16 yields the dependence of hillock slope upon chemical afnity for a multi-sourced spiral: mhk BT p 8 G RT

Eqn. 17 can also be applied to situations where the dislocations are not arranged in a straight line. In this case, 2 represents the perimeter of the area occupied by these dislocations (Rashkovich, 1991). It is important to also note that, for this type of spiral growth, rate can have a discontinuous dependence upon supersaturation. As pointed out earlier in this section, a multi-sourced spiral forms from a group of dislocations only when /2 (Burton et al., 1951). Because terrace width, , scales inversely with supersaturation (Eqns. 9 and 10), this complex spiral is stable at low supersaturations where step width is large. With increasing supersaturation and decreasing step width, the spiral decomposes into a number of single spirals that are generated by each individual dislocation when /2 becomes smaller than . When this occurs, the hillock slope will, in general, increase discretely from the value given by Eqn. 17 to that by Eqn. 11. Hence, the overall growth rate, R m , is no longer a smooth function of supersaturation, as illustrated in Fig. 6C and shown by experiment (e.g. Vekilov and Kuznetsov (1992). 4.2. Relationships Between Process-Based and AfnityBased Rate Expressions Because direct observations of calcite growth show that step ow originates from predominantly complex sources and, to a lesser extent, from simple sources, the net rate of spiral growth across one surfacenatural or synthetic, is necessarily a complex composite of contributions from several R m expressions (e.g., Eqns. 13, 15, and 18), each of which has a unique dependence upon saturation state. Hence, the question arises: Are the reported macroscopic, chemical afnity-based rate laws related to microscopic processes occurring at the growing surface? The obvious conclusion from the preceding analysis is that the growth rates determined by measuring temporal changes in bulk solution compositions must also be composites.

G 2 k BT RT

(17)

Assuming step velocity is independent of p, the product of Eqns. 8 and 17 gives an expression for the rate of a multisourced multiple spiral growth: a e h mk BT Rm 8 G RT G 2 k BT RT

exp

G RT

1 .

(18)

Kinetics of calcite growth

2263

Fig. 6. Schematic view showing the dependence of overall growth rate upon supersaturation index, , in different growth modes. All plots are referenced to the same scale. (A) Growth controlled by single spirals. (B) Growth controlled by single-sourced multiple spirals. (C) Growth controlled by multi-sourced multiple spirals. (D) Growth controlled by two-dimensional surface nucleation. When spiral growth is the only mechanism, the minimum growth rate may be considered as that given for single spirals (e.g., Eqn. 13 and the data in Fig. 4 for calcite). When growth is controlled by both spiral and surface nucleation mechanism, the overall rate is the sum of the contributions from each growth mode.

If no, are there conditions where comparisons of the two types of rate laws are meaningful? To answer these questions, rst consider Eqn. 18, the more complicated rate expression for spiral growth. This equation reduces to the simpler Eqn. 15 when m dislocations overlap, i.e., 0, to form a single source with a Burgers vector of m. Eqn. 15 further reduces to Eqn. 13, the rate expression for the simplest type of spiral growth, when the Burgers vector of the dislocation becomes one (m 1). Eqn. 13 can be further simplied to: Rm a e hk BT 8 exp G RT
2

(19)

when supersaturation is very low or ( G/RT) 1. The form of this expression is analogous to a second order chemical afnity-based rate law (Eqn. 1b with n 2) provided that a e hk B T/8 represents the rate constant, k . The above comparison suggests that spiral growth is approximated by a second order chemical afnity based-rate law only when (1) the supersaturation is so low that spiral formation is the only growth mechanism and (2) the formation of simple

spirals dominates the growth of all crystal faces. Direct observations (Fig. 2) reported in this study show that even in the supersaturation range where spiral formation is the only growth mode, the second requirement fails because multiple spirals are far more commonly observed than single ones. This suggests that even when growth occurs at low supersaturation and the spiral mechanism is the only valid growth mode, still no single kinetic expression can describe growth because the dislocation source structures may be sample specic. Rather, R m measured at each supersaturation for pure spiral growth must be within a range that is dened by Eqns. 13 and 18 for the lower and upper boundaries (Fig. 6AB, respectively). The lower boundary for the growth of calcite {1014} faces in this study is given by the experimental measurements shown in Fig. 4. Finally, in the presence of obviously large surface defects such as pits, fractures, and grain boundaries (Fig. 2CD), the determination of R m can be further complicated because the growth is controlled by the specic characteristics of each source. Most signicantly, if the source is at a gross defect where steps must circumscribe a structural imperfection of a perimeter , the term 2 in Eqns. 16 to 18 is replaced by . The result is that

2264

H. H. Teng, P. M. Dove, and J. J. De Yoreo

as increases and L c decreases, the second term in the denominator becomes dominant, the slope becomes independent of supersaturation and R m becomes a linear function of . When supersaturation is high enough that two-dimensional surface nucleation mechanism becomes operative (Fig. 6D), the contributions of two-dimensional nuclei formation to total R m must be added. The macroscopic growth rate by a two-dimensional nucleation mechanism has an exponential dependence upon supersaturation through the expression Rm
2 1.137h Iv s 1/3

(20)

where h is the step height, I the nucleation frequency (area 2 time 1) (van der Eerden, 1993). Because I increases exponentially with increasing supersaturation, contributions of the twodimensional surface nucleation mechanism are increasingly important with increasing supersaturation and quickly become the dominant component in the overall growth rate. In the supersaturation range where spiral and two-dimensional nucleation mechanisms co-exist ( 0.8 or 2.2 for this study), the overall growth rate will be the sum of all growth modes illustrated in Fig. 6AD. The clear conclusion from experimental observations and this analysis is that attempts to relate afnity-based rate laws to microscopic growth mechanism are of limited value. Macroscopic rate laws derived from bulk experiments represent the composite average of quite different growth rate expressions for the unique reactivities of many different crystal facets and surface structures. However, afnity-based rate laws comprise most of our quantitative knowledge regarding calcite growth kinetics and will continue to be employed by the geological and engineering communities. Calcite grains in previous studies are typically dominated by the {1014} faces or become dominated by these faces as growth proceeds in the absence of impurities. It is therefore useful to understand the physical basis of these rate expressions by making comparisons to process-based rate laws that are derived on the {1014} faces. To probe the relationships, we ask the question: What is the dependence of R m upon ( 1) as a function of the number of dislocations, m, and the separation between these dislocations, ? Using average values of step edge free energy, 1400 erg/cm2 (Teng et al., 1998), and kinetic coefcient of step advancement, 0.4 cm/s (Teng et al., 1999), the dependence of R m upon G/RT) was estimated by Eqn. 18. supersaturation ( These calculated values of R m were tted by a simple power law of k( 1) n where k and n are tting parameters. Results of this analysis give insights to the physical meaning of the exponent, n. When dislocation separation is held constant ( 50 nm, Fig. 7A), n is independent of both the net Burgers vector and the supersaturation. Differences in R m are given by changes in the rate constant. Yet, a comparison with Fig. 7B shows that growth by the same dislocation structure yields a smaller value of n when a broader saturation range is used. That is, the value of the exponent is inversely dependent upon saturation range when growth occurs on complex spirals. To further examine these relations, the net Burgers vector is held constant (m 3) and the distance between dislocations is varied (Fig. 7C). It shows that when the saturation range is narrow and close to equilibrium, the exponent term exhibits a strong inverse dependence upon dislocation separation. How-

ever, as the saturation range becomes broader, n becomes independent of dislocation density (Fig. 7D). Notice that only when separation equals zero and the supersaturation range is close to equilibrium does n approaches 2 (Fig. 7C). This prediction is upheld by the model ts of calcite growth data (Shiraki and Brantley, 1995) that obtained a value of n 1.93 0.14 when exp( G/RT) 1.72 or supersaturation 0.54. These theoretical relationships suggest that the exponent and the rate constant in chemical afnity-based rate laws are dependent upon the details of dislocation distributions and dislocation structures as well as the supersaturation range used to t the growth rate data. There may not be a unique power law expression with the simple form of k( 1) n for spiral growth. Rather, each expression may apply only to the specic supersaturation range over which the model is tted to the experimental data. This analysis shows that, in general, R m can be described by a chemical afnity-based rate law that assumes 1 n 2. The value of n is inversely related to the supersaturation range over which the tting is performed. R m exhibits a quadratic (n approaches two) dependence upon ( 1) only when the dislocation source is compact ( 0) and the solutions are near-equilibrium ( 0.4). This analysis indicates that experimental results that yield n 2 in a wide range of supersaturation that is not sufciently close to equilibrium are likely indicating a non-equilibrium growth (e.g., growth occurring on non-equilibrium forms) or a growth combining spiral and surface nucleation mechanisms.
5. CONCLUDING REMARKS

In situ observations of growth using AFM reveal the mechanisms of calcite growth across a range of supersaturation. At lower values of from 0 to 0.8, growth is initiated solely by surface imperfections including screw dislocations. Thus, crystal defect-originated growth, in particular, spiral growth, is the only manifested mechanism. At the range of 0.8, twodimensional surface nucleation becomes increasingly important with increasing supersaturation. The growth rate of {1014} faces grown by single spirals is estimated from the product of step velocities and spiral hillock slopes measured at different supersaturations. This estimate corresponds to the minimum rate for pure spiral growth on a calcite (1014) face. More general rate expressions are considered by taking into account the effect of complex dislocation sources on the slopes of spirals. Such expressions cannot be approximated by the second order rate law expressed by chemical afnity unless growth proceeds only by the formation of single spirals at low supersaturation range where only the spiral mechanism is operative. Even for pure spiral growth, the experimentally measured rate of calcite crystallization cannot be unique because the types of dislocation source structures are variable. The second order rate law based upon the chemical afnity of the precipitation reaction can be considered as a special case of spiral growth when dislocation sources are simple or compact. Even so, this expression is only valid at low supersaturations (approximately 0.8 or 2.2 for calcite growth in this study). With increasing supersaturation, the magnitude of R m becomes the sum of spiral growth and two-dimensional surface nucleation.

Kinetics of calcite growth

2265

Fig. 7. Dependence of R m upon given by Eqn. 18 for different values of m and . Calculated values of R m were tted to the expression k[exp( ) 1] n to obtain the values of n with R2 0.99 for all cases. (A) Assuming a xed separation between dislocations of 50 nm, the calculated R m for m 2 (a double spiral), m 3 (a triple spiral), and m 4 (a quadruple spiral) was tted for a supersaturation range of 0 to 0.35; and (B) 0 to 1.50. The value of n does not show a dependence upon the number of dislocations for either supersaturation range. (C) The calculated R m for a triple spiral (m 3) and a separation of 0 nm, 100 nm, 200 nm, and 300 nm was tted for a supersaturation range of 0 to 0.35; and (D) 0 to 1.50. Estimates of R m for the supersaturation range of 0 to 1.50 are independent of the separation, in contrast to the strong dependence that is expressed for in the near equilibrium supersaturation range of 0 to 0.35.

Results of this study demonstrate the need to apply caution when deducing growth mechanisms and rate laws from temporal changes in bulk solution chemistry. Further, interpretations of growth mechanism using data collected from these indirect methods are particularly hazardous without direct evidence for the growth processes that are occurring at the mineral surfaces. The analysis in this study suggests that popular rate laws are empirical, at best. By combining direct observations with these methods, improved rate laws with greater predictive capabilities may be possible.
AcknowledgmentsThis work was supported by the Geosciences Research Program, Basic Energy Sciences, U.S. Department of Energy through grant number DE-FG05-95-ER14517 and was performed under the auspices of Lawrence Livermore National Laboratory under contract W-7405-Eng-48. The images in Figure 5 were collected by K. Davis of Georgia Tech. We thank Susan Brantley and an anonymous reviewer for thoughtful comments. REFERENCES Arakaki T. and Mucci A. (1995) A continuous and mechanistic representation of calcite reaction-controlled kinetics in dilute solutions at 25C and 1 Atm total pressure. Aquatic Geochemistry 1, 105130.

Blum A. E. and Lasaga A. C. (1987) Monte Carlo simulations of surface reaction rate laws. In Aquatic Surface Chemistry (ed. W. Stumm), 255292, Wiley. Burton W. K., Cabrera N., and Frank F. C. (1951) The growth of crystals and the equilibrium structure of their surfaces. Royal Soc. London Philos. Trans. A243, 299 358. Busenberg E. and Plummer L. N. (1986) A comparison study of the dissolution and crystal growth kinetics of calcite and aragonite. USGS Bull. 1578, 139 168. Compton R. G. and Daly P. J. (1987) The dissolution/precipitation kinetics of calcium carbonate: An assessment of various kinetic equations using a rotating disk method. J. Colloid. Interface Sci. 115, 493 498. Chernov A. A. (1961) The spiral growth of crystals. Soviet Phys. 4, 116 148. Chernov A. A. and Komatsu H. (1995) Topics in crystal growth kinetics. In Science and Technology of Crystal Growth (eds. J. P. van Eerden and O. S. L. Bruinsma), 67 80, Kluwer Acad. Pub. Christoffersen J. and Christoffersen M. R. (1990) Kinetics of spiral growth of calcite crystals and determination of the absolute rate constant. J. Crystal. Growth 100, 203211. Davies C. W. (1962) Ion association. Butterworth. De Yoreo J. J., Land T. A., Rashkovich L. N., Onischenko T. A., Lee J. D., Monovskii O. V., and Zaitseva N. P. (1997) The effect of dislocation cores on growth hillock vicinality and normal growth rates of KDP {101} surfaces. J. Crystal Growth 182, 442 460.

2266

H. H. Teng, P. M. Dove, and J. J. De Yoreo in calcite: Insights into crystal-growth mechanisms. Geology 18, 1244 1247. Paquette J. and Reeder R. J. (1995) Relationship between surface structure, growth mechanism, and trace element incorporation in calcite. Geochim. Cosmochim. Acta 59, 735751. Plummer L. N., Wigley T. M. L., and Parkhurst D. L. (1978) The kinetics of calcite dissolution in CO2-water systems at 5 60C and 0.0 1.0 atm CO2. Am. J. Sci. 278, 179 216. Rashkovich L. N. (1991) KDP-family single crystals. 100 165, IOP Publishing Ltd. Reddy M. M. (1977) Crystallization of calcium carbonate in the presence of trace concentrations of phosphorous-containing anions. J. Crystal Growth 41, 287295. Reddy M. M. and Nancollas G. H. (1971) The crystallization of calcium carbonate I. Isotopic exchange and kinetics. J. Colloid and Interface Science 36, 166 172. Reddy M. M. and Nancollas G. H. (1973) Calcite crystal growth inhibition by phosphates. Desalination 12, 6173. Reddy M. M. and Gaillard W. D. (1980) Kinetics of calcium carbonate (calcite)-seeded crystallization: Inuence of solid/solution ratio on the reaction rate constant. J. Colloid and Interface Science 80, 171178. Reddy M. M. (1988) Physical-Chemical mechanisms that affect regulation of crystallization. In Chemical Aspects of Regulation of Mineralization (eds. C. S. Sikes and A. P. Wheeler), 3 8, University of South Alabama Pub. Ser. Reeder R. J. (1996) Interaction of divalent cobalt, zinc, cadmium, and barium with the calcite surface during layer growth. Geochim. Cosmochim. Acta 60, 15431552. Shiraki R. and Brantley S. L. (1995) Kinetics of near-equilibrium calcite precipitation at 100C: An evaluation of elementary reactionbased and afnity based rate laws. Geochim. Cosmochim. Acta 59, 14571471. Smallwood P. V. (1977) Some aspects of the surface chemistry of calcite and aragonite. Colloid and Polyner Sci. 255, 994 1000. Staudt W. J., Reeder R. J., and Schoonen M. A. A. (1994) Surface structural controls on compositional zoning of SO2 and SeO2 in 4 4 synthetic calcite single crystals. Geochim. Cosmochim. Acta 58, 20872098. Stipp S. L., Eggleston C. M., and Nielsen B. S. (1994) Calcite surface structure observed at microtopographic and molecular scales with atomic force microscopy (AFM). Geochim. Cosmochim. Acta 58, 30233033. Stumm W. (1992) Chemistry of the Solid-Water Interface. J. Wiley & Sons. Teng H. H. and Dove P. M. (1997) Surface site-specic interactions of aspartate with calcite during dissolution: Implications for biomineralization. Amer. Mineral. 82, 878 887. Teng H. H., Dove P. M., Orme C. A., and De Yoreo J. J. (1998) Thermodynamics of calcite growth: Baseline for understanding biomineral formation. Science 282, 724 727. Teng H. H., Dove P. M., and De Yoreo J. J. (1999) Reversed calcium carbonate morphologies induced by microscopic growth kinetics: Insight into biomineralization. Geochim. Cosmochim. Acta 63, 25072512. Van der Eerden, J. P. (1993) Crystal growth mechanisms. In Handbook of Crystal Growth (ed. D. J. T. Hurle), 1A, 307 475, Elsevier. Vekilov, P. G., Kuznetsov, Yu. G., and Chernov, A. A. (1992) Interstep interaction in solution growth; (101) ADP face. J. Crystal Growth 121, 643 655. Vekilov P. G. and Kuznetsov Yu. G. (1992) Growth kinetics irregularities due to changed dislocation source activity: (101) ADP face. J. Crystal Growth 119, 248 260. Vekilov P. G. and Rosenberger F. (1996) Dependence of lysozyme growth kinetics on step sources and impurities. J. Crystal Growth 158, 540 551.

Dove P. M. and Hochella M. F., Jr. (1993) Calcite precipitation mechanisms and inhibition by orthophosphate: In situ observations by Scanning Force Microscopy. Geochim. Cosmochim. Acta 57, 705714. Drever J. I. (1997) The Geochemistry of Natural Waters (3rd. edition). Prentice Hall. Eggleston C. M. (1994) High-resolution scanning probe microscopy: Tip-surface interaction, artifacts, and applications in mineralogy and geochemistry. In CMS Workshop Lectures: Scanning Probe Microscopy of Clay Minerals (eds. K. L. Nagy and A. E. Blum), 7, 390, Clay Minerals Society. GoodarzNia I. and Motamedi M. (1980) Kinetics of calcium carbonate crystallization from aqueous solutions. J. Crystal Growth 48, 125131. Gratz A. J., Hillner P. E., and Hansma P. K. (1993) Step dynamics and spiral growth on calcite. Geochim. Cosmochim. Acta 57, 491 495. Hillner P. E., Manne S., Gratz A. J., and Hansma P. K. (1992a) AFM images of dissolution and growth on a calcite crystal. Ultramicroscopy 42 44, 13871393. Hillner P. E., Gratz A. J., Manne S., and Hansma P. K. (1992b) Atomic scale imaging of calcite growth and dissolution in real time. Geology 20, 359 362. House W. (1981) Kinetics of crystallisation of calcite from calcium bicarbonate solutions. J. Chem. Soc. Faraday Trans. 77, 341359. Inskeep W. P. and Bloom P. R. (1985) An evaluation of rate equations for calcite precipitation kinetics at PCO2 less than 0.01 atm and pH greater than 8. Geochim. Cosmochim. Acta 49, 21652180. Kazmierczak T. F., Tomson M. B., and Nancollas G. H. (1982) Crystal growth of calcium carbonate: A controlled composition kinetic study. J. Phys. Chem. 86, 103107. Land T. A., De Yoreo J. J., and Lee J. D. (1997) An in situ AFM investigation of canavalin crystallization kinetics. Surf. Sci. 384, 136 155. Lasaga A. C. (1981) Transition state theory. In Kinetics of Geochemiscal Processes (eds. A. C. Lasaga and R. J. Kirkpatrick); Rev. Mineral. 135169. Lasaga A. C. (1998) Kinetic Theory in the Earth Sciences. Princeton University Press. Liang Y., Baer D. T., and Lea A. S. (1995) Dissolution of CaCO3 (1014) surface. Nat. Sci. Res. Soc. Symp. Proc. 355, 409 414. Morse J. W. (1978) Dissolution kinetics on calcium carbonate in sea water VI: The near equilibrium dissolution kinetics of calcium carbonate-rich deep sea sediments. Am. J. Sci. 278, 344 353. Mucci A. (1983) The solubility of calcite and aragonite in seawater at various salinities, temperatures, and one atmosphere total pressure. Am. J. Sci. 283, 780 799. Mucci A. and Morse J. W. (1983) The incorporation of Mg and Sr into calcite overgrowths: Inuences of growth rate and solution composition. Geochim. Cosmochim. Acta 47, 217233. Mucci A. (1986) Growth kinetics and composition of magnesian calcite overgrowths precipitated from seawater: Quantitative inuence of orthophosphate ions. Geochim. Cosmochim. Acta 50, 22552265. Nancollas G. H. and Reddy M. M. (1971) The crystallization of calcium carbonate II: Calcite growth mechanism. J. Colloid Interface Sci. 37, 843 830. Nielsen A. E. (1983) Precipitates: Formation, coprecipitation, and aging. In Treatise on Analytical Chemistry (eds. I. M. Kolthoff and P. J. Elving), 269 374, Wiley. Nilsson O. and Sternbeck J. (1999) A mechanistic model for calcite crystal growth using surface speciation. Geochim. Cosmochim. Acta 63, 217226. Papelis C., Hayes K. F., and Leckie J. O. (1988) A program for the computation of chemical equilibrium composition of aqueous batch systems including surface-complexation modeling of ion adsorption at the oxide/solution interface. Dept. of Civil Engi., Technical Report 306, Stanford University. Paquette J. and Reeder R. J. (1990) New type of compositional zoning