Direct Observations of the Dissolution of Fluorite Surfaces with

Article pubs.acs.org/crystal

Direct Observations of the Dissolution of Fluorite Surfaces with Different Orientations Jose R. A. Godinho,*,†,‡ Christine V. Putnis,§ and Sandra Piazolo†,∥ †

Department of Geological Sciences, Stockholm University, 114 18 Stockholm, Sweden Chemical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, United States § Institut für Mineralogie, University of Münster, Corrensstrasse 24, 48149 Münster, Germany ∥ ARC Centre of Excellence for Core to Crust Fluid Systems, Department of Earth and Planetary Sciences, Macquarie University, NSW 2109, Australia ‡

W Web-Enhanced Feature * S Supporting Information *

ABSTRACT: Atomic force microscopy has been used to observe the surface dynamics during dissolution of polished fluorite surfaces with different orientations. These surfaces, with an initially high density of atomic scale defects, showed fast changes during the first seconds in contact with a solution. Different types of structures developed on each surface, depending on its initial orientation and solution composition. These structures dissolved slower than the main surface persisting for at least 67.5 days of continuous dissolution. A new interpretation of traditional kinetic and thermodynamic models of dissolution applied to surfaces with a high density of steps is proposed to explain the observations. The new model includes the following: (a) fast initial dissolution at defect sites, (b) formation of a fluid boundary layer at the mineral−solution interface enriched in the dissolving ions, and (c) precipitation of more stable fluorite structures nucleated at surface defects. This model highlights the importance of considering surface defects and crystal orientation for advancing our understanding of processes happening at the mineral−solution interface and for developing more accurate kinetic dissolution and crystal growth models essential in Earth and material sciences.

1. INTRODUCTION Dissolution kinetic models traditionally consider the dissolution rate as a function of a surface area that is assumed to react homogeneously.1−4 In addition, the solubility of a mineral is commonly considered a bulk property of the mineral and independent of surface defects.5 However, the density of surface defects, such as steps and kinks, are known to affect both kinetics and thermodynamics of dissolution.6−9 Furthermore, the reactions at the mineral−solution interface are not wellunderstood. The existence of a fluid boundary layer at the interface between a dissolving surface and a solution has been proposed to significantly affect or control dissolution− precipitation mechanisms.10−14 Some observations of natural and experimental dissolution and precipitation can only be explained by the formation of a fluid boundary layer enriched with some elements dissolved from the surface.15,16 This may cause a local supersaturation in the fluid boundary layer even though the bulk solution is far from equilibrium with respect to the bulk solid phase. It is a common approach to use results from surface studies on a cleaved plane to extrapolate the observations to natural environments.17,18 However, natural surfaces have a higher density of defects such as steps and kinks that define weathering pits, fractures and crystal edges, or dislocations created, for example, by the incorporation of other minor element ions during crystal growth. Consequently, if we know that the © 2013 American Chemical Society

cleavage plane is the most stable surface, any dissolution rate estimated from these theoretically perfect surfaces will be underestimating the dissolution rate of a naturally occurring imperfect material. Furthermore, powders used in dissolution experiments also have a high density of surface defects. Therefore, to correctly link laboratory and field dissolution rates, a better understanding of how surface properties (e.g., density of steps) affect dissolution kinetics is essential.19−22 This will also benefit the understanding of other processes common in rocks such as dissolution−precipitation creep23 and solventmediated phase transformations.24,25 We have used natural fluorite, CaF2, as a suitable experimental material for dissolution experiments for several reasons. (I) The reactivity of surfaces with different crystallographic orientations is related to the surface structure and is therefore expected to be similar to the different mineral phases with the same crystallographic structure.26 Thus, CaF2 is a suitable analogue for materials with the fluorite structure,27 such as UO2, ThO2, and PuO2.28 (II) CaF2 is an important natural mineral predominantly found in hydrothermal veins, especially as a gangue mineral associated with ore-bearing deposits, such as galena and sphalerite.29 (III) Synthetic fluorite is an important optical Received: July 24, 2013 Revised: November 3, 2013 Published: November 25, 2013 69

dx.doi.org/10.1021/cg401119p | Cryst. Growth Des. 2014, 14, 69−77

Crystal Growth & Design

Article

Figure 1. AFM deflection images of a cleaved (111) CaF2 surface at different dissolution times (a) 0 min, (b) 2 min in solution 1, (c) 2 min in solution 1 + 2 min in solution 2, (d) 2 min in solution 1 + 31 min in solution 2, and (e) 2 min in solution 1 + 52 min in solution 2. North England.40,41 These were chosen because of the large size of the crystals (up to 2 cm) and the known geological setting. Elementary data about chemical composition indicates the presence of rare earth elements below 274 ppm40 (see the Supporting Information). These ions are known to substitute calcium atoms in the fluorite structure.42 The samples were cut using a diamond saw, polished down to a 1 μm surface roughness using diamond paste and further polished by mechanochemical etching in commercially available colloidal silica for 5 h before being used. The mechanochemical polishing minimizes the effect of the nanometer thick crushed surface layer, which may form during polishing, ensuring an average roughness less than 0.5 nm measured from AFM height images using the AFM software over an area of 10 × 10 μm. The final surface orientations were (104), (102), (110), (115), or (334). These surfaces were chosen to represent different surface structures in terms of types of steps and main planes. Furthermore, detailed long-term dissolution studies of those surfaces have already been published.38,39 Due to natural perfect twinning of the original crystals, it was possible to analyze within the same experiment surfaces (334) and (104) and the interface between them. In order to compare experimental results to the commonly used cleaved (111) surface, a cleaved, nonpolished sample was also studied. The solutions used were (1) deionized water, with pH = 5.4 (resistivity > 18 mΩ cm−1 at 25 °C); (2) 0.05 M NaClO4 adjusted to pH = 3.6 with HClO4; (3) HCl solution with pH = 3.6; (4) HCl solution with pH = 2.0; and (5) carbonated water with pH ∼ 4, left equilibrating with a pure CO2 atmosphere. Solutions 2−5 were prepared using solution 1. NaClO4·H2O and HClO4 70% reagents p.a. grade from Sigma-Aldrich were used to prepare solution 2. The effect of all solution compositions on dissolution was studied on surface (334). The effect of surface orientation on dissolution was studied mostly using solution 4, for which surfaces (334), (110), (115), and (102) were tested. 2.2. Analytical Methods. Crystal orientations were obtained by electron backscatter diffraction analysis (EBSD) using a field emission environmental scanning electron microscope (ESEM) (Phillips XL-30) with a coupled electron backscattered detector (Nordlys, Oxford instruments) and the software HKL Channel 5 (Oxford Instruments). The analysis conditions were high vacuum and 20 kV. Angles were measured with an accuracy of 0.3°. The same ESEM was used to acquire secondary electron (SE) images and obtain elemental information of the

material with applications in lithography and laser science.30,31 For these applications, the understanding of the etching process and surface reactivity especially at defect sites is essential.32 (IV) A variety of dissolution kinetic data using different types of surfaces was previously published. The natural cleavage plane of the fluorite structure {111} was studied by scanning probe microscopy during growth2,34,35 and dissolution.32−34,36 Growth of undetermined particles on the surface was reported by Guntram34 and Sangwal.37 Dissolution of surfaces with other orientations has been studied by Godinho.38,39 It was found that under the same solution composition, surfaces with a higher density of step edges dissolve faster. Consequently, the atomically flat (111) is the slowest dissolving surface in a pH = 3.6 NaClO4 solution. In this study, we analyze topography changes on fluorite surfaces with different crystallographic orientations during the first minutes of reaction between the surface and a solution. Direct imaging during the dissolution process, using atomic force microscopy (AFM), enabled real-time in situ observations on a nanoscale. On this scale, we can test the hypothesis proposed by Godinho38,39 that dissolution rates are faster on surfaces with a higher density of steps. As the highest density of steps is present at the beginning of dissolution and continuously decreases thereafter, the initial minutes of dissolution are expected to be the most dynamic period of dissolution. Results from this study show highly dynamic surfaces during the first minutes of dissolution. The nature of the dynamics can be attributed to the stability of the surface, which is dependent on the density of steps and solution composition. As such, this study contributes to a better understanding of processes occurring at the mineral−solution interface.

2. EXPERIMENTAL METHODS 2.1. Samples and Solutions. The crystals used in this study were natural green fluorite cubes from the Rogerley mine, Durham County, 70



Article

Figure 2. AFM height images of the interface between (104) (lower side) and (334) (top side) CaF2 surfaces at different dissolution times in solution 2 (a) 0, (b) 2, (c) 6, (d) 20, and (e) 37 min. Figure 2e is a zoom out of the area of (a−d). Note the differences between scanned and unscanned areas and the accumulation of rounded features on surface (104) at the edge of the scanned area, which indicates that they might have been moved by the scanning tip. external stirring of 110 rotations per minute, at room temperature (21 ± 1 °C). Twenty milliliters of the respective solution was used and renewed every 72 h. At selected dissolution times, the surface was analyzed by CP.

surface by X-ray energy dispersive spectroscopy (EDS) using the AZtec acquisition software (Oxford Instruments). Raman spectroscopy was used to identify the surface phases. The Raman spectra were collected with a high-resolution Jobin Yvon HR800 Raman spectrometer. Surface topography was studied with a Sensofar PLu2300 confocal profilometer (CP) and a Digital Instruments Nanoscope III Multimode AFM (Bruker). The CP has confocal lenses with 10, 50, and 150 times magnification (fields of view 1273 × 955 μm, 253 × 190 μm, and 84 × 63 μm) and numerical aperture of 0.3, 0.8, and 0.95, respectively. The system has a lateral resolution of 111 nm and a vertical resolution of 1 nm with the 150× magnification lens. Slopes up to 71° can be analyzed. The data were acquired using the program SensoScan 2.45 and represented as height-dependent color-coded images. The AFM was equipped with a fluid cell from Bruker with a total volume of 50 μL. Images were collected and analyzed using the Nanoscope software version 5.31r1 at room temperature 21 ± 1 °C, using Si3N4 tips, model NP-S20 with spring constants 0.12 and 0.58 N/m from Bruker. Both deflection and height AFM images were obtained. Deflection images record information from the variation of the cantilever position caused by the near-surface repulsive forces while scanning the surface. Height images record information from the vertical variation of the cantilever caused by differences of topography. Height images allow a quantification of the Z-direction variation, while deflection images often allow a clearer visualization of the surface morphology. 2.3. Experimental Procedure. Dissolution experiments were performed in the AFM fluid cell, allowing the direct observation of the surface dynamics during dissolution. The solution volume in the cell with the sample inside was approximately 38 μL. Images were taken at a scan rate of 3 Hz in the contact mode, resulting in an acquisition time of approximately 90 s per image. Given dissolution times correspond to the time at the end of the scanned area. Injections of 2 mL of fresh solution were made between scans. No solution was injected while scanning to avoid interference during image collection. Images of the initially dry surfaces were obtained followed by the surface in contact with the solution. After being studied in the fluid cell, selected samples were dried and subjected to further dissolution in a 30 mL polypropylene reactor with

3. RESULTS All fluorite surfaces exhibit dynamic changes within the first seconds in contact with the different solutions used. The nature of these changes is dependent on the crystallographic orientation of the surface as well as the solution composition. With the exception of surface (111) all surfaces develop new structures. High-resolution EDS analysis of these structures detect only calcium and fluorine (see the Supporting Information). Raman spectroscopy detects only the presence of a fluorite phase (see the Supporting Information). 3.1. Effect of Surface Orientation on Surface Dynamics. Figure 1 shows the dissolution of a cleaved surface, presenting several steps separating terraces along the (111) surface. Dissolution is visible at the step edges after 1 min in solution 1. After 2 min, the fluid cell was flushed with solution 2. During the first 2 min of contact with solution 2, dissolution is faster than in deionized water, occurring mainly at the steps edges. After 31 min, the smaller steps are no longer visible and rounded features protrude out from the edge of larger steps. This is visible in Figure 1e that shows higher regions with more developed rounded shapes, aligned where previously step edges existed. The rounded features increase in size with time, constituting the main units of the overall topography. The triangular-shaped etch pits characteristic of these surfaces32,36 were not visible during this experiment. The polished surfaces exhibit faster changes than the unpolished cleaved (111) surface. Figure 2 shows the different surface changes on (104) and (334) after contact with solution 2 71



Article

Figure 3. AFM height images of different CaF2 surfaces after the initial dissolution minutes in solution 4 (a) (110), 2 min; (b) (115), 6 min; and (c) (102), 3 min. Note the development of the surface structures from the top of (a) to the bottom (later scanning time). Horizontal lines are possibly caused by the AFM tip moving loose particles on the surface.

at different dissolution times. The roughening of the surface is visible during the first scan (Figure 2, panels a and b) and continues during the first 37 min of dissolution (Figure 2e). During the first 2 min of reaction, small pits with no specific dominant shape open on surface (334), while round structures develop on surface (104) (Figure 2b). After 20 min of continuous scanning over the same area, the topography on surface (334) is characterized by aggregates of rounded structures separated by deeper areas (Figure 2d). A zoom out of the area continuously scanned during 37 min shows a faster dissolution in the scanned areas [higher etch pits on surface (334) and lesser rounded structures on surface (104), Figure 2e]. A smooth layer covers some areas of the unscanned (334) surface. At this stage both planes are covered by similar rounded features. However, it should be noted that the rounded structures on surface (334) are formed by the dissolution of the surface around the structures, and on surface (104), the structures protrude out from the surface. Figure 3 shows the differences in topography on different surfaces after the initial minutes of dissolution in solution 4. On (110) and (115), a surface layer similar to the layer developed on surface (334) forms. On (115), loose particles on the surface are also visible. The layer developed on surface (102) is restricted to smaller areas (Figure 3c). In figure 3a it can be seen that the topography changes within the scanning duration as the surface structures are more developed at later scanning times (bottom of the image). 3.2. Effect of Solution on Surface Dynamics. Surfaces (104) and (334) exhibit different surface dynamics when exposed to different solutions (Figure 4). The rounded features developed on surface (104) and the roughness developed on surface (334) are less pronounced when dissolution occurs in solutions 3 and 5 than when solution 2 is used (compare Figures 2c and 4). Also significant are the differences observed after dissolution on surface (334) in solution 1, showing the development of loose particles on the surface (Figure 5a) and in solution 4 showing a well-defined layer formed over the dissolving surface (Figure 5b). 3.3. Evolution of Surface Features with Time. When a surface layer is developed during the initial minutes of contact between surface and solution, dissolution of the layer occurs faster in the scanned areas than in the unscanned areas. Details of the layer’s dissolution are exemplified in Figure 6 (see Movie 1) that shows the formation of characteristic triangular-shaped etch pits that spread laterally. Once this surface layer dissolved, the surface underneath dissolves faster, even in the areas not continuously scanned (Figure 6d). Figure 6e shows terraces that form during dissolution of the layer.

Figure 4. AFM height images of surfaces at the interface between (104) and (334) CaF2 at the initial stages of dissolution in different solutions: (a) 11 min in solution 3, (b) 11 min in solution 5, (c) 36 min in solution 3, and (d) 36 min in solution 5.

Figure 5. AFM height images of the (334) CaF2 surface at the initial stages of dissolution in different solutions: (a) 7 min in solution 1 and (b) 2 min in solution 4. The scan lines in (a) are possibly caused by the interference of nanoparticles that are dragged by the scanning tip. Note the accumulation of particles on the right side of the scanned area.

Figure 7 shows the persistence of the smooth surface layer on some areas of surface (334) for at least 67.5 h of dissolution. Note that dissolution of this surface occurred in a batch reactor without 72



Article

Figure 6. AFM deflection images of the (334) CaF2 surface after contact with solution 4 during (a) 60 min without scanning, (b) 60 min without scanning +15 min of continuous scanning, (c) 60 min without scanning +35 min of continuous scanning, (d) 60 min without scanning +37 min of continuous scanning, and (e) 130 min without scanning. Note the triangular shape of the etch pits on the surface layer. A movie (see Movie 1) can be viewed where dissolution of a (334) CaF2 surface during 50 min of continuous scanning in solution 4. The initial surface was developed after 60 min in contact with solution 4 without scanning. The animation is composed by AFM deflection images separated by approximately 2 min frames.

the theoretical structure of planes being studied is represented. The transected planes show the regular presence of step edges constituted by atoms deficiently bonded relative to the atoms on the stable (111) surface.38,39 Furthermore, scratches caused by the polishing procedure may induce an additional instability to the polished surfaces. 4.1. Effect of Surface Orientation and Solution Composition. Dissolution on the cleaved (111) surface occurred mainly at the steps between terraces (Figure 1). A consequence of the preferential dissolution at step edges was the merging of smaller terraces. The development of roundedshaped features, especially at the edges, indicates the formation of more stable structures that subsequently dissolved slower (Figure 1e). The fast changes observed at the step edges highlight their high reactivity and importance to the overall dissolution rate of a surface. This has also been observed by other authors.38,39,44,45 The topography developed on polished surfaces is higher than on cleaved surfaces. The character of the developing and changing topography is dependent on the crystallographic orientation of the surface (Figures 2 and 3) and type of solution (Figures 4, 5, and 6). Depending on these conditions, three types of structures can be developed on the surface during the initial minutes of dissolution (Table 2): Structure I, loose (not firmly attached) particles on the surface (e.g., Figures 3b and 5a); Structure II, isolated rounded features protruding out from the surface but attached [e.g., (104) in Figures 2 and 4]; Structure III, a smooth and stable layer attached to the surface (e.g., Figures 5b and 6). The formation of different structures can be caused by the specific conditions at the mineral−solution interface for each system studied. These specific conditions could possibly form different types of complexes on the surface (e.g., pH-dependent

surface disturbances caused by the AFM scanning. Dissolution occurs by lateral shrinking of the layer area and not from the top (Z-direction). The distance between the dissolving surface and the surface layer increases at an approximately constant rate (see the Supporting Information). Figure 8 illustrates the effect of the initially formed surface structures on the development of topography during dissolution. Figure 8a shows the persistence of the layer on surface (334) (red areas) in contrast with a layer-free (104) surface. The layer on (334) forms at the alignment of a pre-existing scratch visible on surface (104). Figure 8b shows how the rounded structures initially formed on the surface are attached to the surface edges and persist for at least 10 days of dissolution on (102). Higher topography develops at the edges with the rounded structures attached, while areas characterized by the absence of these rounded structures dissolve.

4. DISCUSSION OF RESULTS Rapid changes in topography were visible on all surfaces studied within the initial seconds of dissolution, even in the presence of deionized water. This is in contrast with the lower retreat rates measured using CP by Godinho38 for synthetic fluorite surfaces with a similar pretreatment and a NaClO4 solution with pH 3.6 (Table 1). In accordance with those previously measured retreat rates and the fact that the dissolution rate is known to decrease with an increase in pH,36,43 no dissolution features were expected within the first hours of dissolution. This suggests that during the initial minutes of dissolution, the dissolution rate is possibly more than an order of magnitude faster than the dissolution rates averaged over 19.5 days of dissolution. We suggest that the discrepancies observed are related to the initial higher density of reactive sites on the surfaces studied. For example, in Figure 9, 73



Article

Figure 8. Evidence of surface structures formed during dissolution of CaF2. (a) CP image of the interface between surfaces (334) (bottom surface) and (104) (top surface) after 150 h of dissolution in solution 2. Note the alignment of the red surface structures in surface (334) with the scratches on surface (104). (b) SE image of surface (102) after 10 days of dissolution in solution 2.

pH, which is associated with a higher dissolution rate, seemed to promote the formation of a surface layer (Figure 10). More challenging is the interpretation of the different structures formed on surfaces with different crystallographic orientations when dissolving in the same solution (Figures 2 and 3). As we propose that the dissolution rates at the initial stage of dissolution are faster from those previously reported in the literature for longer dissolution times, we cannot compare the formed structures to a known rate of dissolution. However, we can expect a specific reactivity for each surface caused by differences in their theoretical initial surface structure (Figure 9). These differences can cause a different epitaxial relation between the dissolving surface and any precipitating layer. In general, surfaces that are initially formed mainly by (001) surfaces, such as (115), (104), and (102) (Figure 9), have less tendency to develop type III structures [e.g., (104) in Figures 2 and 4, panels b and c). Surfaces (334) and to a lesser extent (110) are formed mainly by (111) surfaces (Figure 9), which apparently helps to promote the formation of larger areas covered by a surface layer. 4.2. Development of Surface Structures. The characterization of the new surface structures developed on the surface is limited due to their fast formation within the first seconds of contact between the solution and the surface, their limited thickness, and their irregular distribution on the surface. Nevertheless, the following characteristics of the developing surface structures can be given: (a) less than 5 nm thickness; (b) smooth top surface, which is at a low angle to the main dissolving surface (e.g., Figure 5b); (c) presence of triangular pits that spread symmetrically on the scanned areas with dissolution time (Figure 6 and Movie 1); (d) relationship between the type of structure developed and the initial orientation of the surface and solution composition (Table 2); (e) absence of any other elements other than calcium and fluorine and of any other phase detectable using Raman spectroscopy (Supporting Information); and a (f) lower dissolution rate than the surface underneath, especially in areas undisturbed by the AFM scanning (Figures 7 and 8). On the basis of our observations, we present three hypotheses for the formation of the surface structures: (1) an amorphous layer formed during polishing, (2) precipitation of a different mineral phase, and (3) precipitation of fluorite with a different surface structure than the original surface. The formation of an amorphous layer on the surface could be caused by the stress induced by the mechanical polishing.46 However, those layers would be unstable and thus would dissolve first. This contradicts our observations; first, because they persisted for at least 67.5

Figure 7. CP image of the (334) CaF2 surface after (a) 140 h, (b) 25 days, and (c) 67.5 days. Solution 2 was used. The variation of the distance between the surface layer and the dissolving surface is approximately linear (see the Supporting Information).

species), or different dissolution rates could cause different degrees of local supersaturation at a possible fluid boundary layer forming close to the surface. Although difficult to quantify, the degree of local supersaturation could vary according to the solution pH and would depend on the initial density of step edges of each surface orientation, that is, a higher step density would result in a higher dissolution rate. For example, dissolution of surface (334) formed type I structures when deionized water was used (Figure 5a), type II structures when solutions with pH 3.6 were used (Figures 2 and 4a), or type III structures when a HCl solution with pH 2 was used (Figure 5b). Therefore, a lower 74



Article

Table 1. Average Retreat Rates of the Studied Planes Calculated for 19.5 Days in NaClO4 Solution at pH 3.6. Data from Godinho.38 surface retreat rate (nm h−1)

(111) 0.11 ± 0.07

(102) 2.8 ± 0.5

(110) 2.5 ± 0.5

Table 2. Summary of the Type of Structures Obtained for Each Surface/Solution Systema solution

structure type

cleaved 334

1+2 1 2 3 4 5 2 3 5 4 4 4

pits I pits + III pits + III III pits II II II II I + II II + III

104

110 115 102

(334) 1.0 ± 0.4

(115) 1.1 ± 0.1

minimize the formation of these layers. The second and third hypotheses involve the deposition of a mineral phase from the solution. In turn, precipitation requires the supersaturation relative to some solid phase in the solution close to the surface. As the surface structures were formed in different solution compositions, it is not expected that they would be constituted by the electrolyte ions. Alternatively, CaCO3 could be formed due to the presence of carbonate absorbed from contact with air and a possible local enrichment of calcium close to the surface. This possibility was excluded because no extra precipitation occurred when a carbonated solution was used (Figure 4d). Raman and EDS analyses suggest that the surface structures are constituted by CaF2 (see the Supporting Information). Traditionally, this would be considered thermodynamically unlikely, if we consider that the bulk solution is undersaturated relative to fluorite. However, during the initial dissolution period, during the first seconds of contact between surface and solution, the fast dissolution of a surface with high density of steps could cause a localized enrichment of calcium and fluoride ions within the mineral−solution boundary layer. For example, Figure 8a shows that the surface layer is more likely to develop at scratched areas (i.e., unstable areas that dissolve faster). However, this is relative to the stability of the surface, as the layer only developed on (334) and not on (104). If the dissolution of the dissolved ions is fast enough relative to their diffusion to the bulk of the solution, a local supersaturation close to the surface could occur.12 A similar explanation has been used to explain the growth of surface features during calcite dissolution43 and to explain the remineralization of UO2 pellets from undersaturated solutions.10 Other studies suggest the possibility of rearrangements of surfaces with high density of steps in order to minimize surface energy.47−49 Furthermore, from solid state thermodynamics, it is expected that the dissolving crystal, containing a relatively high concentration of impurities (see the Supporting Information) and defects introduced during crystal growth, would be less stable than a possible purer and more ordered phase precipitating from solution. Classically, the solubility of CaF2 is determined as a bulk property of the solid. However, due to the nature of the starting surface, with a high density of defect sites, the initial surface is unlikely to obey the same thermodynamic constraints as a cleaved plane.5 As it is not possible to calculate with confidence the concentration of ions nor the thickness of a possible enriched layer forming at the mineral−solution interface, it is not feasible to determine the exact mechanism that would form the new surface structures. However, we can draw on our observations. In Figure 8b, it is shown that the developed structures are associated with steps. Following this possibility, we expect that a local lower concentration of calcium and fluoride ions relative to the bulk saturation concentrations would be needed to grow a more stable crystal phase on the defective sites of the surface. In this way, the observed changes on the surface could be explained by a process of interface-coupled dissolution and precipitation at defective sites. This process controlled by the minimization of the surface energy would form surface structures more stable than the initial surface. Furthermore, the structures formed at the edges in Figure 8b clearly inhibit the edges from dissolving further, which results in the development of a higher topography. The

Figure 9. Model of the theoretical initial structure of the CaF2 surfaces studied (a) (334), (b) (102), (c) (104), (d) (110), and (e) (115). Edges are marked with an arrow, and the constituting planes are marked with a dashed line for plane {111} or a continuous line for plane {100}. The configuration with atoms at the highest coordination numbers is represented considering the fluorite-type structure.

surface

(104) 0.7 ± 0.1

a

The solutions used were (1) deionized water; (2) NaClO4, pH = 3.6; (3) HCl, pH = 3.6; (4) HCl, pH = 2.0; and (5) carbonated water. The structure types are classified by pits) if no precipitated structure is observed: (I) loose particles, (II) round structures, and (III) surface layer. See text for full description of solution composition and structure types.

Figure 10. Schematic diagram illustrating the effect of pH on the type of structures developed on surface (334).

days of dissolution, and also because they dissolve by forming crystallographically controlled etch pits. Furthermore, we expect that the mechanochemical treatment in colloidal silica would 75



Article

by the surface preparation or crystal growth (step 1). During the first seconds in contact with a solution, fast dissolution occurring at surface defects causes the enrichment of the fluid boundary layer near the surface (step 2). It is difficult to predict the concentration of ions or complexes formed in this layer. However, we can assume that the diffusion rate of those species from the surface to the bulk of the solution is relatively slower when compared to the faster overall dissolution rate of the initial surface. As the dissolution rates are dependent on the initial orientation of the surface and solution composition, the concentrations of species in the enriched layer may differ for different experimental conditions, causing different processes to occur. Furthermore, the surface orientation and solution composition can also affect the type of complexes formed at the interface. These complexes cause the development of stable surface structures, either in the form of a surface layer (step 3a) or in the form of protruding structures (step 3b). Although the exact mechanisms that lead to the development of these structures could not be determined, we suggest that it is associated with the presence of high-energy sites on the surface. These can function as nucleation sites, lowering the concentrations necessary for growth to occur, even in a hypothetically undersaturated bulk solution. Finally, we suggest that the developed structures are more stable than the original surface and thus dissolve slower and protect the surface from further dissolution (step 4). The observed surface alterations are expected to affect both the short- and long-term bulk dissolution rates. As discussed above, a faster dissolution rate is generally expected when more reactive surfaces and lower pH solutions are used. However, our observations suggest that over a critical dissolution rate, for which precipitation occurs, the dissolution rate decreases. If the precipitate coats the surface with a layer more stable than the initial surface, then the overall dissolution rate decreases due to the decrease of reactive surface area. Therefore, the overall dissolution rate of materials with low solubility may not be continuously proportional to the expected reactivity of a specific surface−solution system. This principle could be used for advanced surface-engineered materials, for which a fluidmediated treatment could increase the overall stability of a surface. As other materials with the fluorite structure are expected to have a similar relation between surface orientation and dissolution rates,10,26 such treatment would be beneficial to the highly defective surfaces of spent nuclear fuel in order to minimize the overall dissolution rates of fuel pellets.

characteristic triangular shape of the etch pits and the high stability of the layer suggest that the layer may have grown to expose (111) surfaces. It can also explain why the layer does not grow thicker than a few nanometers, as once a (111) layer is formed, a higher concentration of calcium and fluoride ions is needed for further precipitation to occur. In addition, the lower dissolution rate of the surface layer would make the mineral− solution boundary layer less concentrated in ions. In other words, by inhibiting the dissolution of unstable sites of a surface, the precipitates stabilize the surface and effectively reduce the dissolution rate, which consequently reduces the concentration of precipitating ions at the mineral−solution boundary layer. Furthermore, surfaces initially formed mainly by (111) surfaces developed larger areas covered by the layer, which may have been facilitated by easier nucleation due to a favorable epitaxial relationship between dissolving surface and developing surface structure. 4.3. Proposed Dissolution Model. A new generalized interpretation of kinetic and thermodynamic theories, specifically applied to dissolution of surfaces with high densities of defects, is necessary to explain our observations. A dynamic kinetic model that accounts for a faster initial dissolution rate caused by the initial high density of more reactive sites is suggested (Figure 11). As well, the low stability of the initial

Figure 11. Schematic model of the initial 4 stages of dissolution: (1) surface before dissolution, (2) initial seconds of dissolution with formation of a boundary layer at the surface−solution interface rich in ions or complexes removed from the surface, (3) surface covered with fluorite precipitates either as a surface layer (3a) or as isolated precipitates (3b), (4) the precipitates inhibited the dissolution of the surface at the most reactive edges. Areas not covered by the precipitates dissolved faster.

5. CONCLUSIONS Our results show the importance of considering the surface orientation and atomic scale structure, such as type and amount of steps, in order to develop accurate dissolution and crystal growth kinetic models. Using information from cleaved surfaces, that have lower densities of steps than natural surfaces, may lead to an underestimation of dissolution and crystal growth rates. When comparing these rates based on the direct analysis of a cleaved surface, results may present significant differences from surface area normalized rates obtained from solution analyses using natural or powdered samples with a higher density of defect, step, and edge sites. Our direct observations highlight the need to carry out further in situ experiments in order to fully understand dissolution, crystal growth and mineral−solution interfacial processes.

surface is expected to cause a fast release of Ca and F and thereby rapidly causing a local supersaturation at the mineral−solution interface necessary for precipitation of a more stable CaF2 phase to occur in order to decrease the initial high surface energy. Figure 11 schematically represents the dissolution process in a cross section of a surface. Initially, the surface is composed by a high density of defects, either natural atomic scale steps, dependent on the surface orientation, or other defects caused 76



■

Article

(18) Luttge, A.; Arvidson, R. S. J. Am. Ceram. Soc. 2010, 93, 3519− 3530. (19) Casey, W. H.; Banfield, J. F.; Westrich, H. R.; McLaughlin, L. Chem. Geol. 1993, 105, 1−15. (20) Turner, B. F.; Stallard, R. F.; Brantley, S. L. Chem. Geol. 2003, 202, 313−341. (21) Ganor, J.; Roueff, E.; Erel, Y.; Blum, J. D. Geochim. Cosmochim. Acta 2003, 69, 607−621. (22) Moore, J.; Lichtner, P. C.; White, A. F.; Brantley, S. L. Geochim. Cosmochim. Acta 2012, 93, 235−261. (23) Rutter, E. H. J. Geol. Soc. (London, U.K.) 1983, 140, 725−740. (24) Putnis, A. Rev. Min. Geochem. 2009, 70, 87−124. (25) Putnis, A.; Austrheim, H. Geofluids 2010, 10, 254−269. (26) Maldonado, P.; Godinho, J. R. A.; Evins, L. Z.; Oppeneer, P. M. J. Phys. Chem. C 2013, 117, 6639−6650. (27) Godinho, J. R. A.; Piazolo, S.; Stennett, M. C.; Hyatt, N. C. J. Nucl. Mater. 2011, 419, 46. (28) Tasker, P. Journal de Physique Colloques 1980, 41, C6−488. (29) McLemore, V. T.; Giordano, T. H.; Lueth, V. W.; Witcher, J. C. In New Mexico Geological Guidebook II, 49th Field Conference, Las Cruses, New Mexico, 1998; New Mexico Geological Society: Socorro, NM, 1998; pp 251−264. (30) Rothschild, M. Mater. Today 2005, 8, 18−24. (31) Klein, C. A. J. Appl. Phys. 2006, 100, 083101. (32) Motzer, C.; Reichling, M. J. Appl. Phys. 2009, 105, 064309. (33) Hillner, P.; Manne, S.; Hansma, P. Faraday Discuss. 1993, 95, 191−197. (34) Guntram, J.; Werner, R. Surf. Sci. 1997, 371, 371−380. (35) Schick, M.; Dabringhaus, H.; Wandelt, K. J. Phys.: Condens. Matter 2004, 16, 33−37. (36) Cama, J.; Zhang, L.; Soler, J. M.; De Giudici, G.; Arvidson, R. S.; Lüttge, A. Geochim. Cosmochim. Acta 2010, 74, 4298−4311. (37) Sangwal, K.; Desai, C. C.; John, V. Krist. Tech. 1979, 14, 63. (38) Godinho, J. R. A.; Piazolo, S.; Evins, L. Z. Geochim. Cosmochim. Acta 2012, 86, 392. (39) Godinho, J. R. A.; Piazolo, S.; Balic-Zunic, T. Geochim. Cosmochim. Acta 2013, in press, http://dx.doi.org/10.1016/j.gca. 2013.11.017. (40) Falster, A. U.; Fisher, J. E.; Simmons, W. B. Rocks Miner. 2000, 76, 253−254. (41) Fisher, J. E. U.K. Journal of Mines and Minerals 2006, 27, 25−28. (42) Schwinn, G.; Markl, G. Chem. Geol. 2005, 216, 225−248. (43) Zhang, R.; Hu, S.; Zhang, X. Journal: Aquatic Geochemistry 2006, 12, 123−159. (44) Bisschop, J.; Dysthe, D. K.; Putnis, C. V.; Jamtveit, B. Geochim. Cosmochim. Acta 2006, 70, 1728−1738. (45) Schott, J.; Brantley, S.; Crerar, D.; Guy, C.; Borcsik, M.; Willaime, C. Geochim. Cosmochim. Acta 1989, 53, 373−382. (46) Vlasov, A. D.; Rez, J. S.; Fil’chenkov, M. L. Cryst. Res. Technol. 1988, 23, 1093−1101. (47) Müller, J.; Grant, M. Phys. Rev. Lett. 1999, 82, 1736−1739. (48) Haataja, M.; Müller, J.; Rutenberg, A. D.; Grant, M. Phys. Rev. B 2002, 65, 035401. (49) Aguenaou, K.; Müller, J.; Grant, M. Philosophical Magazine, Part B 1998, 78, 103−109.

ASSOCIATED CONTENT

S Supporting Information *

Chemical composition of fluorites from the Rogerley mine,40 EDS spectrum of the dissolving surface and formed surface layer, Raman spectra of a surface before and after dissolution, and variation of the distance from the top layer to the average dissolving surface with time (in Figure 7). This material is available free of charge via the Internet at http://pubs.acs.org. W Web-Enhanced Feature *

A movie of the dissolution of a (334) CaF2 surface during 50 min of continuous scanning in solution 4 is available in the HTML version.

■

AUTHOR INFORMATION

Corresponding Author

*Address: 1 Bethel valley Road, Oak Ridge, Tennessee 37831, United States. E-mail: [email protected]. Tel: 1-865-3353377. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work has been financially supported by the EU Initial Training Network Delta-Min (Mechanisms of Mineral Replacement Reactions) Grant PITN-GA-2008-215360 and the Swedish Nuclear Fuel and Waste Management Company (S.K.B.). The Knut och Alice Wallenberg stiftelse is acknowledged for funding the ESEM setup. S.P. acknowledges the financial support by the Australian Research Council through DP120102060 and FT1101100070. This is contribution 376 from the ARC Centre of Excellence for Core to Crust Fluid Systems (http://www.ccfs. mq.edu.au) and 919 in the GEMOC Key Centre (http://www. gemoc.mq.edu.au).

■

REFERENCES

(1) Lasaga, A. Kinetic Theory in the Earth Sciences. Princeton series in Geochemistry; Princeton University Press: Princeton, NJ, 1999. (2) White, A. F.; Brantley, S. L. Chem. Geol. 2003, 202, 479−506. (3) Brantley, S. L.; Kubicki, J.; White, A. F. Kinetics of Water-Rock Interaction; Springer: New York, 2008. (4) Oelkers, E. H.; Schott, J. Rev. Mineral. Geochem. 2009, 70, 207−258. (5) Fan, C.; Chen, J.; Chen, Y.; Ji, J.; Teng, H. H. Geochim. Cosmochim. Acta 2006, 70, 3820−3829. (6) Blum, A. E.; Yund, R. A.; Lasaga, A. C. Geochim. Cosmochim. Acta 1990, 54, 283−297. (7) den Brok, S. W. J. Geophys. Res. Lett. 2001, 28, 603−606. (8) Lasaga, A.; Lüttge, A. Science 2001, 291, 2400−2404. (9) Lüttge, A. Am. Mineral. 2005, 90, 1776−1783. (10) Römer, J.; Plaschke, M.; Beuchle, G.; Kim, J. J. Nucl. Mater. 2003, 322, 80−86. (11) Putnis, A.; Putnis, C. V. J. Solid State Chem. 2007, 180, 1783− 1786. (12) Ruiz-Agudo, E.; Putnis, C. V.; Rodriguez-Navarro, C.; Putnis, A. Geology 2012, 40, 947−950. (13) Peruffo, M.; Mbogoro, M. M.; Edwards, M. A.; Unwin, P. R. Phys. Chem. Chem. Phys. 2013, 15, 1956. (14) Putnis, C. V.; Ruiz-Agudo, E. Elements 2013, 9, 177−182. (15) Hövelmann, J.; Putnis, C. V.; Ruiz-Agudo, E.; Austrheim, H. Environ. Sci. Technol. 2012, 46, 5253−5260. (16) Hellmann, R.; Wirth, R.; Daval, D.; Barnes, J. P.; Penisson, J. M.; Tisserand, D.; Epicier, T.; Florin, B.; Hervig, R. L. Chem. Geol. 2012, 294−295, 203−216. (17) Renard, F.; Montes-Hernadez, G.; Ruiz-Agudo, E.; Putnis, C. V. Chem. Geol. 2013, 340, 151−161. 77


Direct Observations of the Dissolution of Fluorite Surfaces with

Recommend Documents