Langmuir 2004, 20, 8587-8596
8587
Probing Crystallization of Calcium Oxalate Monohydrate and the Role of Macromolecule Additives with in Situ Atomic Force Microscopy Taesung Jung,† Xiaoxia Sheng,‡ Chang Kyun Choi,§ Woo-Sik Kim,| Jeffrey A. Wesson,*,⊥ and Michael D. Ward*,‡ Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota 55455, Department of Chemical Engineering, Seoul National University, Seoul 151-744, Korea, Department of Chemical Engineering, Kyunghee University, Yongin, Kyunkee-do 449-701, Korea, and Nephrology Division, Department of Veterans Affairs Medical Center and the Medical College of Wisconsin, Milwaukee, Wisconsin 53295 Received May 6, 2004. In Final Form: July 5, 2004 Kidney stones are crystal aggregates, most commonly containing calcium oxalate monohydrate (COM) microcrystals as the primary constituent. Macromolecules, specifically proteins rich with anionic side chains, are thought to play an important role in the regulation of COM growth, aggregation, and attachment to cells, all key processes in kidney stone formation. The microscopic events associated with crystal growth on the {010}, {121 h }, and {100} faces have been examined with in situ atomic force microscopy (AFM). Lattice images of each face reveal two-dimensional unit cells consistent with the COM crystal structure. Each face exhibits hillocks with step sites that can be assigned to specific crystal planes, enabling direct determination of growth rates along specific crystallographic directions. The rates of growth are found to depend on the degree of supersaturation of calcium oxalate in the growth medium, and the growth rates are very sensitive to the manner in which the growth solutions are prepared and introduced to the AFM cell. The addition of macromolecules with anionic side chains, specifically poly(acrylic acid), poly(aspartic acid), and poly(glutamic acid), results in inhibition of growth on the hillock step planes. The magnitude of this effect depends on the macromolecule structure, macromolecule concentration, and the identity of the step site. Poly(acrylic acid) was the most effective inhibitor of growth. Whereas poly(aspartic acid) inhibited growth on the (021) step planes of the (100) hillocks more than poly(glutamic acid), the opposite was found for the same step planes on the (010) hillocks. This suggests that growth inhibition is due to macromolecule binding to both planes of the step site or pinning of the steps due to binding to the (100) and (010) faces alone. The different profiles observed for these three macromolecules argue that local binding of anionic side chains to crystal surface sites governs growth inhibition rather than any secondary polymer structure. Growth inhibition by cationic macromolecules is negligible, further supporting an important role for proteins rich in anionic side chains in the regulation of kidney stone formation.
Introduction Kidney stone disease, which occurs in approximately 10% of the U.S. population, causes substantial suffering and occasional renal failure, but the disease mechanism is poorly understood. Kidney stones are microcrystal aggregates, most commonly containing calcium oxalate monohydrate (COM, also known as whewellite) crystals as the primary constituent.1,2 The occurrence of kidney stones derived from COM is thought to be associated with four critical processes, all which involve its crystal surfaces: (i) nucleation, (ii) growth, (iii) crystal aggregation, and (iv) attachment of crystals or aggregates to epithelial cells. In vitro studies have suggested that anionic molecules or macromolecules with substantial anionic functionalities (e.g., carboxylate)3-6 play an important role in directing †
Visiting BK-21 Fellow from Seoul National University. University of Minnesota. § Seoul National University. | Kyunghee University. ⊥ Medical College of Wisconsin. ‡
(1) Mandel, N. S.; Mandel, G. S. J. Urol. 1989, 142, 1516. (2) Prein, E. L. J. Urol. 1963, 89, 917. (3) Lieske, J. C.; Huang, E.; Toback, F. G. Am. J. Physiol. Renal Physiol. 2000, 278, F130. (4) Millan, A. Cryst. Growth Design 2001, 1, 245. (5) Guo, S.; Ward, M. D.; Wesson, J. A. Langmuir 2002, 18, 4284.
the selectivity toward crystallization of COM and calcium oxalate dihydrate (COD, or weddellite), the latter found less often in kidney stones even though it is found in the urine of asymptomatic individuals.7,8 Certain macromolecules isolated from human urine were found to inhibit COM crystal growth in vitro and favor the formation of COD over COM, suggesting these macromolecules in urine provide a natural defense against stone formation through crystallization selectivity.8-10 Anionic molecules and macromolecules also are thought to affect COM crystallization rates, the aggregation of COM crystals, and the attachment of crystals and their aggregates to cells. Synthetic anionic macromolecules such as poly(aspartic acid) and poly(glutamic acid) (polyD and polyE, respectively; Chart 1) inhibit COM crystal growth.11 Kidney stones contain measurable amounts of carboxylate-rich proteins that may serve as adhesives that promote aggregation of COM crystals. In contrast, many urinary macromolecules, such as uropontin12 and nephrocalcin,13 which are rich in carboxylate groups, are known to inhibit stone formation. (6) Lieske, J. C.; Deganello, S.; Toback, F. G. Nephron 1999, 81 Suppl 1, 8. (7) Elliot, J. S.; Rabinowitz, I. N. J. Urol. 1980, 123, 324. (8) Dyer, R.; Nordin, B. E. Nature 1967, 215, 751. (9) Wesson, J. A.; Worcester, E. Scanning Microsc. 1996, 10, 415. (10) Wesson, J. A.; Worcester, E. M.; Kleinmann, J. G. J. Urol. 2000, 163, 1343. (11) Ito, H.; Coe, F. Am. J. Physiol. 1977, 233, F455.
10.1021/la0488755 CCC: $27.50 © 2004 American Chemical Society Published on Web 08/25/2004
8588
Langmuir, Vol. 20, No. 20, 2004 Chart 1
Stones and COM crystals often are found attached to epithelial cells at the tip of renal papilla,14,15 and COM crystals adhere readily to renal tubule cells grown in culture.16 Despite the obvious importance of macromolecule binding to COM crystal surfaces during growth, aggregation, and attachment to cells, a microscopic understanding of these processes has yet to be developed. During the past decade, advances in real time in situ atomic force microscopy (AFM) have made possible direct imaging of nucleation and crystal growth events at the microscopic, and even near-molecular, level. These studies have enabled elucidation of crystal growth mechanisms, at the microscopic level, including examination of the role of molecular additives on growth rates and crystal surface topography.17-22 Our laboratory demonstrated that AFM could be employed to examine growth in the (100) plane of COM single crystals and the influence of certain macromolecules, specifically polyD and polyE,5 which were chosen for these studies because all natural proteins known to affect calcium oxalate crystallization contain 10-40% aspartate and glutamate residues.23 Due to the absence of well-defined steps on the (100) surface, which prevented direct measurement of step velocities, the growth rates along specific crystallographic directions were deduced from the time-dependent filling of pits created in the (100) surface by a preliminary etch of the surface in undersaturated aqueous solutions. These studies revealed that polyD and polyE were more effective at inhibiting growth than their corresponding monomers. Furthermore, polyD suppressed growth along the c-axis but polyE suppressed growth along the b-axis, reflecting preferred binding affinities of these molecules for certain step surfaces exposed at the pit walls. In this case, polyD bound preferentially to {h21} planes whereas polyE preferred to bind to {010} planes. The ability of polyD to suppress growth along the c-axis, the fast growth direction of COM, was consistent with the pronounced shift in selectivity from COM to COD in the presence of polyD. These results prompted us to perform a more comprehensive examination of the effect of macromolecules on COM growth, extending our previous AFM studies to additional COM crystal faces and other macromolecules. (12) Worcester, E. M.; Blumenthal, S. S.; Beshensky, A. M.; Lewand, D. L. J. Bone Miner. Res. 1992, 7, 1029. (13) Deganello, S. Calcif. Tissue Int. 1991, 48, 421. (14) Wesson, J. A. Unpublished results. (15) Khan, S. R., Finlayson, B., Hackett, R. L. Lab. Invest. 1979, 41, 504. (16) Wesson, J. A.; Worcester, E. M.; Wiessner, J. H.; Mandel, N. S.; Kleinman, J. G. Kidney Int. 1998, 53, 952. (17) Teng, H. H.; Dove, P. M.; Orme, C. A.; De Yoreo, J. J. Science 1998, 282, 724. (18) Hillier, A. C.; Ward, M. D. Science 1994, 263, 1261. (19) Yip, C. M.; Ward, M. D. Biophys. J. 1996, 71, 1071. (20) Yip, C. M.; Brader, M. L.; DeFelippis, M. R.; Ward, M. D. Biophys. J. 1998, 74, 2199. (21) Davis, K. J.; Dove, P. M.; De Yoreo, J. J. Science 2000, 290, 1134. (22) Mao, G.; Lobo, L.; Scaringe, R.; Ward, M. D. Chem. Mater. 1997, 9, 773. (23) Shiraga, H.; Min, W.; VanDusen, W. J.; Clayman, M. D.; Miner, D.; Terrell, C. H.; Sherbotie, J. R.; Foreman, J. W.; Przysiecki, C.; Neilson, E. G.; Hoyer, J. R. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 426.
Jung et al.
The inclusion of other crystal faces provides an opportunity to characterize crystal growth modes more fully and elucidate the role of surface structure and composition on macromolecule binding, particularly when those crystal faces have well-defined steps that allow direct measurements of step motion in the absence and presence of additives. Since our initial findings, two reports have appeared that describe the dynamic behavior of steps under growth conditions and the effect of certain additives (osteopontin, citrate, Al3+, Fe3+) on step growth due to their adsorption on the COM crystal surfaces.24,25 In addition to probing the influence of these species on growth itself, such information can identify the crystal faces that most likely bind macromolecules, which is crucial to understanding crystal aggregation and attachment to epithelial cells. We report herein direct AFM visualization of the surface morphology and step growth on the (100), (121 h ), and (010) faces of COM and the inhibiting effect of polyD, polyE, and poly(acrylic acid) (polyAA) on step growth, which is synonymous with the attachment of CaOx units to specific crystal planes. The steps defining the periphery of growth hillocks on these crystal faces do not always mimic the polygonal habit anticipated from the bulk crystals, revealing kinetic morphologies under the conditions used here. Growth inhibition was found to depend on the macromolecule structure, macromolecule concentration, and the identity of the step site. The effect on step growth by cationic macromolecules was negligible, underscoring the importance of anionic groups to regulating stone formation processes. The kinetic profiles suggest that growth inhibition is associated with macromolecule binding to both crystal planes of the step site or with step pinning by macromolecules adsorbed on the hillock terraces, which are important considerations when developing models that explain the role of additives on crystal growth. Results and Discussion Characterization of COM Crystal Surfaces. Previously reported studies of COM crystal growth in our laboratories relied on the use of COM single crystals that had been prepared by rapid mixing of 10 mM aqueous solutions of CaCl2 and sodium oxalate (Na2Ox) at neutral pH (pH ) 7.5) and high ionic strength (150 mM NaCl).26 This procedure produced crystals with predominant {100} bounded by smaller {010}, {021}, and {121 h } faces. The assignment of these faces was based on the single-crystal structure reported by Tazzoli and Domeneghetti27 (P21/c, a ) 6.290 Å, b ) 14.5803 Å, c ) 10.116 Å, β ) 109.46°). This index setting was used recently by Millan in a comparison of theoretical and experimental crystal habits.4 This space group setting and accompanying unit cell choice differs from those proposed by Deganello and Piro28 (P21/ n, a ) 9.976 Å, b ) 14.588 Å, c ) 6.291 Å, β ) 107.05°), which has been used by others during related investigations of calcium oxalate crystallization. These crystal structures, both collected at room temperature, are identical. The Tazzoli setting is used here for convenience and ease of description (more of the relevant crystal planes (24) Qiu, S. R.; Wierzbicki, A.; Orme, C. A.; Cody, A. M.; Hoyer, J. R.; Nancollas, G. H.; Zepeda, S.; De Yoreo, J. J. Proc. Natl. Acad. Sci. 2004, 101, 1811. (25) Gvozdev, N. V.; Petrova, E. V.; Chernevich, T. G.; Shustin, O. A.; Rashkovich, L. N. J. Cryst. Growth 2004, 261, 539. (26) Sheng, X.; Ward, M. D.; Wesson, J. A. J. Am. Chem. Soc. 2003, 125, 2854-2855. (27) Tazzoli, V.; Domeneghetti, C. Am. Miner. 1980, 65, 327. (28) Deganello, S.; Piro, O. N. Jb. Miner. Mh. 1981, 2, 81.
Crystallization of Calcium Oxalate
Figure 1. SEM images of COM crystals with their (010), (12h 1 h ), and (100) faces oriented upward (top) and their schematic representations generated by the modeling program SHAPE (bottom). The shaded areas represent the shape of the growth hillocks typically observed on the faces. The crystal face assignments were confirmed by analysis of the angles between the face edges and comparison with the crystallographic values. The angles are denoted by letters that correspond to the entries in Table 1. The calcium ion surface densities differ for the three faces: (010) ) 0.033 Ca2+/Å2; (12 h1 h ) ) 0.041 Ca2+/Å2; (100) ) 0.054 Ca2+/Å2.
and directions are principal ones). Millan has cataloged the indices for both unit cell choices and the equivalent planes for each.29 The COM crystals used in our previous AFM investigations proved to be of limited use because the relatively small areas of faces other than (100) prohibited reliable positioning of these faces under the AFM tip. Consequently, two alternative methods, adapted from previously reported procedures,30,31 were employed to grow COM crystals (see Experimental Section for complete details): Method 1. Double jet precipitation at high temperature and low ionic strength produced crystals with large (010) faces that were suitable for AFM studies (Figure 1). The (010) face and all its adjoining faces were assigned Miller indices by comparison of the measured angles between edges of the adjoining faces and those expected from the crystal structure. The angles between the relevant edges, denoted by letters in Figure 1 and Table 1, all are within 0.5° of the expected values, confirming their assignment. This analysis is aided by crystal habit models using the crystal modeling program SHAPE,32 which generates models of the external habit of a particular crystal structure and displays the crystal model from the perspective of the viewer. The upper face also was confirmed as (010) by its real-space AFM lattice images of the faces and the corresponding Fourier transform (see below). The size of the (010) face generated by method 1 ranged from 10 to 30 µm, which was sufficiently large for AFM studies. X-ray diffraction verified that these crystals were COM, uncontaminated by its dihydrate or trihydrate forms (COD and COT, respectively). As with the crystals grown at (29) COM undergoes a reversible phase transition between 38° and 45° to a high-temperature structure (the so-called “basic” structure), which crystallizes in the I2/m space group. We note that in ref 37 (Millan) the crystal structure depicted in Figure 1 for the (100) plane of the Tazzoli notation is the (1 h 01) plane of the I2/m phase, not the identical room temperature P21/n phase. Consequently, the view of the (100) plane depicted in our Figure 1 differs somewhat from that in ref 37. (30) Cody, A. M.; Horner, H. T. Scanning Electron Microsc. 1984, 3, 1451-1460. (31) Millan, A. Personal communication. (32) SHAPE, Shape software, Kingsport, TN.
Langmuir, Vol. 20, No. 20, 2004 8589
high ionic strength, crystals generated by method 1 are characterized by the absence of {001} faces, consistent with fast growth along the 〈001〉 direction. The (010) face of crystals grown by this method is bounded by six faces. We note that COM crystals grown at high ionic strength also display (010) faces flanked by six planes, but some crystals appear to be bounded by only four planes owing to small {021} faces. Overall, the area of the observable faces decreased in the order {010} > {100} > {121 h} > {021}. These crystals are often accompanied by twinned crystals, which were easily distinguished and separable from the untwinned version. Method 2. Slow evaporation of aqueous saturated CaOx solutions under acidic conditions (pH ) 1.5) produced COM crystals with the xiphoid (knifelike) habit depicted in Figure 1C. The xiphoid crystals were 20-50 µm long and 10-20 µm wide, with a prominent five-sided face that was assigned as the plane of COM using analysis described above. The (121h ) face and the adjoining crystallographically equivalent (12 h1 h ) face have similar areas, as expected by the mirror symmetry about the b-axis (this is more apparent in Figure 1F than in the perspective view of Figure 1D). Method 2 also produced crystals with reasonably sized rhombus-shaped (100) faces. Unlike the crystals grown at neutral pH or in gels,30 the (010) faces did not adjoin (100). Instead, the (100) face was bounded by (121 h ), (12 h1 h ), (021), and (02 h 1). AFM Characterization of COM Crystal Faces. COM crystals prepared by either method 1 or 2 were harvested by filtration of the crystallization media and then transferred to the specimen disk of an AFM cell designed for image acquisition in liquid media. The crystals were anchored to the specimen holder by pressing them gently onto a partially cured UV-curable polymer film previously cast onto the specimen disk, followed by a final curing of the polymer. The crystals typically assumed orientations with their largest face beneath the AFM tip. The face beneath the AFM tip was identified using an optical microscope above the AFM cell and the crystal morphology analysis employed with the SEM images described above. This also allowed unambiguous determination of the azimuthal (i.e., in-plane) orientation of the crystal with respect to the AFM image frame, which could be used to corroborate the lattice images of the upper-facing plane and the assignment of step features on growth hillocks grown under supersaturated conditions (see below). If the AFM cell was filled with an aqueous solution saturated (or slightly supersaturated) in calcium oxalate (saturation concentration ) 0.11 mM) the (010), (121h ), and (100) faces were sufficiently stable (i.e., negligible growth or dissolution) for the acquisition of lattice images (Figure 2). The real-space images and Fourier transforms (not shown) obtained for the (010) face of crystals grown by method 1 clearly revealed a two-dimensional periodic motif with an oblique unit cell having lattice parameters of 6.4 Å, 10.5 Å, and 106.0°. These values are in reasonable agreement with the lattice constants for the (010) plane (a ) 6.290 Å, c ) 10.116 Å, β ) 109.46°). Scans acquired over larger areas of the (010) face under these conditions revealed micrometer-sized terraces separated by steps with heights of 14.7 Å, in agreement with d010 ) 14.580 Å ()b), which corresponds to four calcium oxalate layers (Figure 3). Images of the (121h ) face clearly revealed a twodimensional periodic motif corresponding to an oblique unit cell, with lattice parameters of 17.6 and 25.1 Å along [2 h 10] and [012], in reasonable agreement with the values of 19.26 and 24.94 Å expected from the bulk COM crystal structure. The difference between the measured and
8590
Langmuir, Vol. 20, No. 20, 2004
Jung et al.
Table 1. Dihedral Angles between the Edges of Crystal Faces Bounding the Three Crystal Faces Depicted in Figure 1 (12 h1 h ) face
(010) face
(100) face
code
edges
measd (deg)
calcd (deg)
code
edges
measd (deg)
calcd (deg)
code
edges
measd (deg)
calcd (deg)
a
(100) (021) (021) (1 h 21) (1 h 21) (1 h 00) (1 h 00) (021 h) (021 h) (121 h) (12 h 1) (100)
109
109.46
g
141
141.34
l
108.43
107.02
h
88
87.94
m
72
71.57
144
143.52
i
131
130.71
n
108
108.43
109
109.46
j
49
49.29
o
(021) (121 h) (121h ) (12 h1 h) (12h 1 h) (02 h 1) (02h 1) (021)
108
107
72
71.57
108
107.02
k
(01 h 0) (02 h 1) (02 h 1) (100) (100) (121 h) (12 h 1) (02 h1 h) (02 h1 h) (01 h 0)
131
130.71
143
143.52
b c d e f
expected values for the first lattice parameter is somewhat larger than the typical error associated with AFM measurements of unit cell dimensions (generally not more than 5%), suggesting some slight reconstruction of the exposed {121 h } plane compared with its bulk structure. The measured acute angle (56°) of the oblique unit cell, however, is very similar to the expected value (58.2°), suggesting that this reconstruction is relatively minor. Scans over large areas revealed terraces separated by steps with heights measuring 4.6 Å, essentially identical to d121h (4.39 Å), which is normal to this surface. Due to the roughness of the (100) face, two-dimensional periodicity was difficult to discern in its real-space lattice images. Fourier transforms of the AFM data, however, revealed one strong frequency component attributable to a force corrugation along a direction perpendicular to the edges of the (121 h ) or (12 h1 h ) faces, which adjoin the (100)
Figure 2. AFM lattice images of the (010), (121h ), and (100) faces of COM single crystals (upper panels) and corresponding images of larger areas (lower panels). The images were acquired in aqueous solutions containing 0.11 mM calcium oxalate in deflection mode. The upper and lower halves of the lattice images correspond to raw and Fourier-filtered data, respectively. (A, B) The lattice image of the (010) face reveals a two-dimensional oblique unit cell with lattice parameters of a ) 6.4 Å, b ) 10.5 Å, and β ) 106.0°, in good agreement with the bulk values. The average step height in the large area scans is 14.7 Å, essentially identical to b in the bulk crystal. (C, D) The AFM image of the (121 h ) face reveals an oblique unit cell with [2h 10] ) 17.6 Å and [012] ) 25.1 Å, and an acute angle of 56.0°. The average step height in the large area scans is 4.6 Å, essentially identical to d121 in the bulk crystal. (E, F) The AFM image of the (100) face exhibits a one-dimensional corrugation coinciding with the [021] direction, which is defined by chains of calcium and oxalate ions forming an angle of 36° with respect to the c-axis (see Figure 3). The average step height in the large area scans is 6.3 Å, identical to d010 in the bulk crystal. (Note: the bottom panels are not aligned in the same direction as their respective upper partners.)
Figure 3. Views of the COM crystal structure illustrating the d-spacings corresponding to the step heights observed on the (010), (121h ), and (100) faces. The dashed line in the left panel represents the [021] direction that coincides with the corrugations observed in the AFM image of the (100) face (see Figure 2).
face. Fourier-filtered images based on this component revealed this corrugation as one-dimensional features oriented along the [021] direction, at an angle of 35.2° with respect the c-axis. Comparison with the structure of the (100) plane reveals that this force corrugation coincides with one-dimensional chains of oxalate ions that run parallel to the (121 h ) and (12 h1 h ) edges and are inclined by 36° with respect to the c-axis. Larger scan areas revealed narrow terraces aligned along the c-axis, separated by steps with heights of 6.3 Å, essentially identical to d100 ) 5.93 Å and in agreement with our previous report.26 These lattice images and step height measurements corroborate the assignment of the (010), (121h ), and (100) faces deduced from the crystal habit analysis, providing the unambiguous determination of the crystal orientation that is necessary for reliable characterization of crystal growth modes. Growth Hillocks and Crystallographic Assignment of Surface Features. Collectively, the three crystal habits described above provide three crystal faces that are amenable to dynamic AFM visualization of crystal growth events under supersaturated conditions. Quantitative measurement of these events, typically deduced from step velocities associated with growth on specific crystal planes, requires surfaces with well-characterized topographic features. Exposure of the crystals with the (010), (121 h ), and (100) faces to highly supersaturated CaOx solutions ([CaOx] ) 0.20-0.25 mM; pH ) 7.0) produced dramatic changes in the surface topography of each face. The (010) surface formed four-sided parallelogram hillocks with well-defined step edges (Figure 4), similar in appearance to hillocks recently observed on this face by
Crystallization of Calcium Oxalate
Langmuir, Vol. 20, No. 20, 2004 8591
Figure 4. AFM images of growth hillocks and etch pits on various faces of COM crystals. Panels A, C, and E: Growth hillocks on the (01 h 0), (121 h ), and (100) faces observed in supersaturated CaOx solution. Note that (01 h 0) ) (010) and (121 h ) ) (12 h1 h ) by the COM monoclinic symmetry. Panels B, D, and F: AFM images of etch pits on the three faces, acquired in undersaturated CaOx solutions. The COM crystal in (B) depicts the surface of a twinned COM crystal in the (010) projection, and its (121h ) edge is visible at the upper left. An optical micrograph of the crystal, with the twin sectors labeled, is depicted in the inset. Panels G and H depict well-defined triangular (100) hillocks observed in air for a crystal retrieved from the crystallization medium used to prepare COM crystals. The hillocks emanate from point sources at the intersections of the (121 h ) and (12 h1 h ) planes. The triangular habit is indicative of the mirror symmetry about the b axis and a prominent (001) face. Although the (001) face is not observed on the bulk crystals grown under these conditions (pH ) 1.5), this face has been observed occasionally on crystals grown at neutral pH. The AFM image in (H) also reveals triangular pits oriented in a direction opposite to the hillocks. The height of the hillocks and depth of the pits are 6.3 Å, corresponding to d100, a single unit of translation along a*. This suggests that the (001) face is prominent during growth but (001 h ) is prominent during etching.
AFM.24,25 The height of the steps on these hillocks, corresponded to either b/2 (ca. 7 Å) or b (ca. 14 Å), with the b/2 steps exhibiting bunching to create the 14 Å steps. Notably, these hillocks were enclosed by four step planes, whereas the (010) face of the bulk crystal was bounded by six faces. The (121h ) face under these conditions typically exhibited two-dimensional islands with shapes that mimicked the xiphoid habit of the bulk crystal, with growth hillocks observed only occasionally. The islands were separated by 4.6 Å high steps, identical to the step heights in Figure 2D. The (100) face exhibited large tear-shaped hillocks emanating from a single point source at one end of the bulk crystal, where the (121 h ) and (12 h1 h ) planes intersect. These tear-shaped hillocks were observed recently by AFM.24 Reliable measurement of step velocities requires unambiguous assignment of the steps that define the hillocks on each face. These assignments were performed here using a combination of optical microscopy and AFM. In Figure 5, for example, optical microscopy provides a view of a COM crystal normal to its (010) face, which is bounded by six faces as described above. Two of the faces on the periphery of the crystal, however, are not visible because they are perpendicular to the (010) face. The monoclinic symmetry of COM warrants that these invisible faces be assigned to (100) and (1h 00), as these faces form an angle of 90° with the (010) face, as illustrated in the SHAPE representation in Figure 1.33 The remaining four faces, which clearly are not perpendicular to the (010) face, must then be assigned as (021), (021h ), (1h 21), and (121h ), consistent with the assignments made based on crystal habit above (the latter two faces in this image appear dark because (33) An identical view also is illustrated in: Millan, A. J. Mater. Sci.: Mater. Med. 1997, 8, 247.
Figure 5. (top left) Photograph of (010)-oriented COM crystal on the AFM stage. The {100} faces form 90° angles with the upper (010) face and therefore are not visible in this view. The (021), (021 h ), (1 h 21), and (121 h ) faces extend outward and can be seen under the microscope. (top right) Same crystals in the AFM cell with the cantilever approaching from the right. (bottom, left to right) SHAPE representation of the crystal in the photograph, with Miller indices, and AFM images of the hillocks and pits (arrows) on this crystal, in the same orientation as the optical image The hillocks and pits have four edges whereas the bulk crystal has six, reflecting the absence of {100} step planes.
of the difference in focal distance). The AFM cantilever was then positioned on top of the crystal, and images of the hillocks were acquired with the crystal in the same orientation, thereby allowing unambiguous assignment of the step planes that enclose the hillocks. Interestingly,
8592
Langmuir, Vol. 20, No. 20, 2004
the growth hillocks exhibited only four step planes, whereas six step planes might have been anticipated from the bulk crystal habit. On the basis of the orientation of the hillocks and the bulk crystal, the (010) hillock step planes must be (021), (021 h ), (1 h 21), and (121 h ). The step edges on the (010) hillocks form obtuse and acute angles of 108° and 72°, respectively, which are identical to the angles of the {121 h } and {021} faces in the (010) projection (expected crystallographic values: 107.02 ° and 72.98°; see shaded area in Figure 1). Clearly, the hillocks did not exhibit {100} faces, which are evident for the bulk crystal. This argues that the hillocks reflect a kinetic (i.e., nonequilibrium) morphology, with {100} faces growing out of existence due to rapid kink advancement along the {121h } and {021} step planes. Although predictive models for the shape evolution of crystal facets in two dimensions are not adequately developed, habit changes during growth and the appearance and disappearance of crystal facets during growth have been documented.34 Crystal habit and shape evolution depend on lattice geometry, solute-crystal interactions, solute-solvent interactions, solvent-crystal interactions, and interplanar spacings. Models have been proposed to describe the evolution of “virtual” faces, like the missing (100) faces on the hillocks, to morphologically significant ones.35 The unique behavior exhibited by the COM (010) hillocks suggests an interesting opportunity to examine these phenomena in more detail. We also note these step plane assignments, which are unambiguous and have been verified numerous times on different COM crystals, differ from those reported recently for parallelogram-shaped (010) hillocks on COM crystals grown from gel, where two planes of the hillocks appear to have been assigned to (100) planes (in the Tazzoli convention).24 We have no explanation for this apparent discrepancy. The {100} faces were also absent in pits that appeared on the (010) face when the AFM cell was filled with undersaturated 0.05 mM CaOx solution. These pits adopted an identical shape and orientation as the hillocks, indicating that the internal step edges also corresponded to (021), (021h ), (1h 21), and (121h ). The facile formation of the {121 h } and {021} step planes also was evident in AFM images of pits on the (010) surfaces of twinned COM crystals. These twinned crystals were characterized previously by SEM to be enclosed by (021), (021 h ), (1 h 21), and (121 h ) step planes.36 The assignments of the step planes in these pits were confirmed here by comparison of their orientation with the (121h ) edge of the bulk crystal, which was visible in the same AFM image frame. Similar assignment procedures were applied to the hillocks on the (121 h ) face (and on its equivalent (12 h1 h ) face), which were observed occasionally when the CaOx concentration was reduced to 0.13 mM, just above the equilibrium concentration. The step edges of the hillocks were not well defined, but they appear to be bounded predominantly by {100}, {010}, and {121 h } step planes, with smaller and (021) planes evident upon closer inspection (Figure 4C). Similar shapes were observed for etch pits in this face. The step planes bounding the tear-shaped (100) hillocks were not as well defined as the faces of the bulk crystal. (34) (a) Thomas, L. A.; Wooster, N.; Wooster, W. A. Discuss. Faraday Soc. 1949, 5, 343. (b) Johnsen, A. Sitzber. preuss. Akad. Wiss. physik.math. Klasse 1923, 208. (c) Winn, D.; Doherty, M. F. AIChE J. 1998, 44, 2501. (d) Winn, D.; Doherty, M. F. AIChE J. 2000, 46, 1348. (e) Winn, D.; Doherty, M. F. Chem. Eng. Sci. 2002, 57, 1805. (35) (a) Borgstrom, L. H. Z. Krist. 1925, 62, 1. (b) Kozlovskii, M. Kristallografiya 1957, 2, 760. (b) Szurgot, M., J. Prywer, Cryst. Res. Technol. 1991, 26, 147. (36) Cody, J. Cryst. Growth 1987, 83, 485.
Jung et al.
Nevertheless, comparison with the bulk crystal orientation indicates that the narrow and broad ends of each hillock coincided with the c-axis (the [001] direction). Their shape was indicative of the crystallographic (010) mirror plane perpendicular to the b-axis. Although the step edges were irregular, the shape of the hillock does suggest the presence of (021) and (02 h 1) at the broad end of the hillock and (12 h1 h) and (121 h ) step planes at the narrow end. Interestingly, AFM images of the (100) face acquired in air for crystals harvested from the crystallization medium occasionally revealed well-defined triangular growth hillocks emanating from the point source at the intersection of the (12h 1 h) and (121h ) step planes (Figure 4G). This triangular shape signifies the presence of a large (h01 h ) step plane at the broad end of the hillocks, which was surprising given the absence of this face in the bulk crystals of COM. When the crystal was immersed in the supersaturated CaOx solution (0.20 mM) growth along the c-axis was sufficiently fast that the well-defined {h01} boundaries of the hillocks vanished (Figure 4E) and the hillocks assumed the tear shape. When the concentration was lowered to 0.10 mM CaOx, just below the equilibrium concentration, the {100} face dissolved and well-defined triangular pits were formed on the crystal surface (Figure 4G). The triangular etch pits and growth hillocks on (100) were oriented in opposite directions along the c-axis. Step Velocities and Growth Rates. The assignment of the hillock step planes permits analysis of growth along specific crystallographic directions by measurement of the velocity of step advancement. These step velocities were measured from consecutive image frames acquired during a 1-h period, at 20 s per frame. As described above, the step planes enclosing the growth hillocks on the (010) surface are assignable to (021), (021 h ), (121 h ), and (1 h 21). The rate of attachment of CaOx growth units to these steps can be deduced from measurements of their advancement velocities, which were limited here to measurements of v021 and v121h . Unlike (010), the steps on the (100) face are not well-defined. Therefore, growth on this face was characterized by measuring the advancement of the hillocks along the [001] direction, expressed as v[001]. The complex nature of the {121 h } faces prohibited reliable measurements of step velocities. During the course of crystal growth measurements the CaOx concentration will decline owing to growth on the crystal under examination as well as on other crystals (out of the field of view) in the AFM cell. Reliable measurement of step velocities also requires avoiding stagnant layers at the crystallizing interfaces, as this effectively reduces the concentration of the solute at the growth interface. These needs can be addressed by continual replenishment of the AFM cell with a constant composition CaOx solution administered to the cell at a constant flow rate. The importance of this procedure can be demonstrated by comparing step velocities under three distinct conditions: (i) a quiescent growth medium introduced by a one-time injection of a CaOx solution to the AFM cell, (ii) continual introduction of a premixed CaOx solution at a constant flow rate, and (iii) continual introduction of a CaOx solution at constant flow rate, prepared by “in-line mixing” to minimize the aging of the solution. Others have reported that continual flow of the crystallization solution through the AFM cell is important for accurate measurements of step motion on the surfaces of calcium-containing biominerals,17,21,25,37 but to our knowledge a direct comparison of conditions i-iii has not appeared. The data in Figure 6 clearly illustrate that the step velocitiessas measured by advancement along the [001]
Crystallization of Calcium Oxalate
Langmuir, Vol. 20, No. 20, 2004 8593
Figure 6. The effects of flow and mixing behavior on the step velocity along the [001] direction of the (100) hillocks (left) and the velocity of the (021) step on the (010) hillocks (right).
Figure 7. (left) The hillocks on the (010) face of COM. The arrows represent the directions along which the step velocities were measured. The data at the right depict the inhibiting effect of polyAA, polyD, and polyE on the (021) and (121 h ) step velocities.
direction on the (100) hillocks (v[001]) and advancement of the (021) steps (v021) on the (010) hillockssdepend on these conditions. If the solution in the AFM cell is quiescent (no flow), the step velocities rapidly decline following introduction of the CaOx solution (0.2 mM), reaching a value that is approximately one-fourth of their respective original velocities after 5 min. If this same CaOx solution is pumped through the AFM cell at a constant rate (0.72 mL/min), the step velocity declined more slowly, with v[001] and v021 decreasing to roughly three-quarters and onehalf of their original values, respectively. This is attributed to a continual decrease of the CaOx concentration owing to nucleation and growth in the supply solution. When this solution was used, AFM revealed the accumulation of particles, presumably CaOx, on the surface of the COM crystal. In contrast, if a 0.2 mM CaOx solution was prepared just before introduction to the AFM cell at the same flow rate (i.e., by in-line mixing of 0.4 mM CaCl2 and 0.4 mM Na2Ox at 0.36 mL/min each for combined flow rate of 0.72 mL/min), the step velocities remained essentially unchanged throughout the entire measurement. Therefore, all data were collected under flow with the CaOx solutions generated by in-line mixing. We note that this flow rate exceeds that used for related AFM studies of CaOx crystal growth, ensuring the independence of step velocity on flow rate.25 Under these conditions, the step velocity on the (100) face is noticeably larger than either v021 or v121h on the (010) face. Some, but not all, of this difference can be attributed to the larger step height of (37) (a) Teng, H. H.; Dove, P. M.; De Yoreo, J. J. Geochim. Cosmochim. Acta 2000, 64, 2255. (b) Booth, N. A.; Chernov, A. A.; Vekilov, P. G. Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys. 2004, 011604/1 (KH2PO4).
the (010) hillocks (d010 ) 14.58 Å vs d100 ) 5.93 Å), which is correlated with the number of CaOx growth units required to advance the step by unit increments. The effect of macromolecule additives was examined by adding small concentrations (0-2 µg/mL) of polyAA, polyD, and polyE, which were chosen because they are rich in carboxylate side chains, as are most natural proteins known to affect CaOx crystallization. Data acquired on the (010) face clearly revealed that polyAA was the most potent inhibitor of growth on the (021) and (121h ) planes (Figure 7). The data also reveal that polyE was somewhat more effective than polyD as a growth inhibitor for both (021) and (121 h ). The effect of the corresponding monomers (not shown) on step motion was negligible, even if their concentrations were increased 10-fold. This demonstrates the importance of cooperative binding of the macromolecule carboxylate groups to calcium sites on the crystal surfaces of the step, as deduced from our earlier studies.5 The effect of macromolecule binding to the (010) hillocks is even more apparent from the complete loss of hillock features when the macromolecule concentrations exceeded that required for complete suppression. For example, the (010) hillocks became unrecognizable immediately following addition of polyD at 2.5 µg/mL (Figure 8). PolyAA was a strong inhibitor of growth on steps of the (100) hillocks as well, with growth completely suppressed at concentrations as low as 0.05 µg/mL, measurably lower than concentration required on the (010) hillocks (Figure 9). Interestingly, the effectiveness of polyE and polyD was inverted compared with the (010) hillocks, with polyD inhibiting step motion more strongly. Furthermore, growth was completely suppressed at substantially lower concentrations of polyD than polyE on this surface. Although
8594
Langmuir, Vol. 20, No. 20, 2004
Figure 8. AFM images of the (010) hillocks: (A) before addition of polyD and (B) immediately (
