- An important research challenge is understanding the interaction between organic molecules and nanoparticle reactivity and how those interactions differ in the laboratory versus in the natural world.
- Increased understanding of the role of proteins in nucleation and biomineralization processes could aid in the development of tools to direct the assembly of nanoparticles.
- The growth of biominerals provides a rich area for research at the mesoscale, and new analytical tools are changing fundamental understanding about the growth behavior of even common, well-studied materials. Additional theoretical, computational, and in situ experimental tools that can visualize molecular events could be useful in furthering this work.
- A comprehensive model that quantitatively describes the diverse pathways of mineralization is lacking, and one essential challenge for developing that model is to describe the ensemble behavior of coupled solvent–solid systems.
- Models for crystal growth that work well for describing the growth of bulk soluble crystals at low supersaturation do not work as well when applied to crystals that grow via particle-based processes.
- Bulk analysis of paleo proxies—geochemical signatures that are recorded in an archive that has been preserved and that can be measured—have proven important for understanding large-scale changes in the Earth’s geochemistry, but these proxies display unusual and unpredicted properties when examined at the mesoscale, in part as a result of large compositional heterogeneity in the materials.
The third session of the workshop explored mesoscale phenomena involved in biomineralization and geochemical processes. Pupa Gilbert, Professor of Physics at the University of Wisconsin-Madison, described two kinds of biomineral structures and discussed how knowledge of those structures proved indispensable for understanding the mesoscale biomineral formation pathways. John Spencer Evans, Professor at the Center for Skeletal Sciences at New York University, talked about the families of extracellular matrix proteins, how they guide biomineralization processes, and how that leads to interesting material features. Jim De Yoreo, Chief Scientist for Materials Synthesis and Simulation across Scales at the Pacific Northwest National Laboratory (PNNL), spoke about the ongoing advances in the temporal and spatial resolution of in situ imaging methods and how those methods are being applied to understand the emergence of order in protein matrices and the controlling factors in matrix-directed mineral formation. Alex Gagnon, Assistant Professor in the School of Oceanography at the University of Washington, discussed work from his laboratory that aims to explain small-scale heterogeneity in the geologic record and to understand the implications regarding efforts to explore how the carbon cycle has changed over the course of the Earth’s history. In the final talk, Andrew Madden, Associate Professor in the School of Geology and Geophysics
at the University of Oklahoma, presented a number of important directions for future research for mesoscale chemistry in the Earth and planetary sciences. An open discussion followed, moderated by Patricia Thiel, the John D. Corbett Professor Chemistry and Distinguished Professor of Chemistry and of Materials Science & Engineering at Iowa State University, co-chair of the workshop organizing committee, and a member of the Chemical Sciences Roundtable (CSR).
Pupa Gilbert started her presentation by showing a micrograph of single crystals of calcite that form the spines of sea urchins. These crystals have rounded structures that defy the usual understanding of crystals as structures with flat faces and sharp edges. She pointed out that these crystals co-orient with each other within 0.025 degrees. Using spectroscopic methods to explore the electronic structure of the crystals reveals that they have a sequence of mineral faces that develop as the crystals grow from amorphous precursors (Politi et al. 2008, Radha et al. 2010). During the animal’s lifetime, a sea urchin’s spines can grow as large as 30 cm long. Each spine comprises a single crystal that has holes on the order of 10 microns, and this crystal grows to length in about 10 months, whereas with current techniques in the laboratory it would take about 82 years if it could be grown that large at all, said Gilbert. Understanding the mechanisms that support such growth and reproducing it in the laboratory could have implications for the electronics and solar panel industries and for carbon sequestration.
The crystal growth mechanism that is most familiar is one where crystals grow from ions attaching one at a time to a nucleus, which grows into a bulk crystal with facets and edges. Urchin calcite crystals take a different path, one of particle-by-particle growth. The process starts with amorphous nanoparticles that can be as large as 50 to 100 nm. These nanoparticles aggregate into a three-dimensional but amorphous solid that then undergoes crystallization. This random-walk process spreads in three dimensions as the crystal grows. A question that emerges from this is whether the two methods produce similar crystals, and if not, what are the differences in the crystals that result from the different growth patterns? Those are questions that Gilbert and her collaborators have been exploring.
No differences are apparent using x-ray diffraction techniques, but other methods do highlight some variations. Gilbert showed a calcium distribution map of the triradiate sea urchin spicules imaged using x-ray absorption near-edge spectroscopy and photoelectron emission spectromicroscopy (Figure 5-1). Photoelectron emission spectromicroscopy reveals the different states of calcium carbonate present in the spicule, that is, whether it is crystalline calcite, anhydrous amorphous calcium carbonate, or hydrated amorphous calcium carbonate. These images reveal that freshly deposited material at the outer edge of the growing spicule comprises a mixture of amorphous and hydrous amorphous calcium carbonate, while the center is mostly crystalline calcite. Crystallinity propagates from the center to the outer rim of the spicule (Gong et al. 2012) in a series of phase transitions that occur as the urchin grows from about 30 hours after fertilization occurred to create the urchin embryo. By day three, the spicules are completely crystalline, which means the phase transitions in the animal are different when this process occurs in the laboratory. “So question number one is how do biological systems control these phase transitions?” Gilbert said. Energetically, this phase transition is exothermic, with calcite being the lowest energy state.
These amorphous precursors, said Gilbert, explain how the resulting calcite crystals can have rounded shapes and grow much larger and 100 times faster than in the laboratory. “The morphology is determined before crystallization, at the time of aggregation,” she explained.
Gilbert then discussed the production of nacre as a second example of biomineralization at the mesoscale. Nacre, the lamellar material out of which many mollusk shells are made, is iridescent because it is layered, and the thickness of the layers is about 0.5 microns. Nacre, she said, is among the most well-studied biominerals, with over 1,000 papers published on its structure and properties, yet when Gilbert and her team used polarization-dependent imaging contrast to examine the aragonite crystals in nacre, they found that the published structures were wrong (Olson et al. 2013). Instead of the C axis of the aragonite crystals in nacre being oriented parallel to one another and perpendicular to the nacre layers, their orientation differs depending on the species
Figure 5-1 Sea urchin spicules imaged using x-ray absorption near-edge spectroscopy (left) and photoelectron emission spectromicroscopy (right). SOURCE: Gong et al. (2012). Reprinted with permission from Proceedings of the National Academy of Sciences of the United States of America.
(Figure 5-2). In one sample, for example, the crystals were staggered diagonally, in another they formed sheets, and in a third they stacked in columns.
Of the many features of nacre that Gilbert’s lab has since elucidated, one is that the aragonite crystals grow epitaxially and a second is that these crystals order gradually by a physical, not biological, process. During epitaxial growth, the crystals are almost always connected to one another, and the ones that are not oriented correctly to the other crystals appear to stop growing and die out, while oriented crystals grow faster and fill space faster. “This is a perfectly abiotic mechanism,” said Gilbert. “It’s a competition for space model and there’s beautiful math that goes along with that. [It] is precisely the same math that describes the evolution of biological organisms in computation for space, food, and other resources.”
Gilbert’s team also found that the initial layers of nacre are more disordered than in the final state, and that these initial crystals die out as the co-orientation process takes over. The speed with which co-orientation occurs is species dependent, she added. In some species, co-orientation begins within 10 to 20 microns of the initial growth, while in others it can take 100 to 200 microns for order to develop. She concluded her remarks by noting that there are 70 different polymorphs of calcium carbonate, with the ones she discussed being just two of them. “Each and every one of them can teach us something very interesting technologically,” she said.
In biological systems, cells are the architects of the mesoscale, said John Evans. Cells have the genomic blueprints that dictate how the process is going to go from initiation to its conclusion. He also noted that there are more than 60 different biominerals on Earth produced by 55 phyla in all five kingdoms. Of these, 80 percent are crystalline materials and 20 percent are amorphous, and they are made from 22 different elements. It is likely, he added, that there are common mechanisms for forming these biominerals (Veis 1990), though this hypothesis has yet to be proven.
Figure 5-2 Side view of fractured nacre from two different species of mollusk shows the different crystal arrangements when imaged using polarization-dependent imaging contrast. SOURCE: Olson et al. (2013). Reprinted with permission from the Journal of Structural Biology.
If cells are the architects of biominerals, then biomolecules are the construction workers of the mesoscale, and biomineralization can be considered a “hard–soft” assembly process in which “soft” nanocomponents, particularly proteins, assemble to form “hard” mesoscale and larger biomineral structures. Cells use proteins to generate the nanoscale components that assemble into larger-scale solids with either short- or long-range ordering, a process that involves protein-based mediation of nucleation and disorder-to-order transformation. Using nacre from mollusk shells as a model system, each nacre tablet comprises a coherent aggregation of calcium carbonate nanograins that range in size from 3 to 10 nm, and nanoscale bridges and asperites—rough, nanosized bluffs—on the tablet surfaces help interlock the tablets and keep them from being displaced during crystal growth.
Proteins play a key role in creating the tablets. Each table comprises a polysaccharide layer of beta-chitin onto which is deposited a silk protein gel matrix. Inside this matrix is a family of proteins called framework proteins that are responsible for nucleation that occurs in the gel phase. Inside the tablets is another set of proteins known as intercrystalline proteins and they exist as inclusions or voids inside the table crystals. Each of these proteins contains regions of amino acids that do not fold and do not form stable structures; these regions are labile, dynamic, and they destabilize aggregation. These proteins also contain regions of amino acids that are prone to aggregate similar to the way that beta-amyloid plaques aggregate in the brains of patients with Alzheimer’s disease. Together, these different regions simultaneously promote and destabilize aggregation, which is what is seen in atomic force micrographs (Figure 5-3). “We see evidence of aggregation but no real ordering of the structures,” said Evans, who added that these proteins can actually scuttle along the surface of the developing crystals.
As far as what these protein phases, as he called the aggregates, do in the nucleation process, Evans said that a nonclassic nucleation model postulates that nucleation starts from supersaturated conditions when prenucleation clusters form. These
Figure 5-3 Atomic force micrographs of framework protein N16.3 (left) and the intracrystalline protein AP7 (right) showing regions of aggregation but no ordering. SOURCE: Evans (2014). Reprinted with permission from John S. Evans.
prenucleation clusters are small ion chains that are no more than 1 to 3 nm across and that can co-assemble into an amorphous aggregate of about 100 nm in size. This amorphous precursor has one of two fates: it can dissolve and precipitate again to form a crystalline phase or it can persist in an amorphous phase if there is something present to stabilize that state. Evans explained that some interesting mesoscale effects are seen during this process, and the role of some proteins in the nucleation process is beginning to be understood. Focusing on a pair of proteins, Evans noted that the intercrystalline protein AP7 (Figure 5-3, right) slows down but supports the formation of the prenucleation clusters by a factor of 2, while the N16.3 (Figure 5-3, left) framework protein affects a different part of the process and destabilizes those prenuclear clusters.
Using flow cell technology with electron microscopy to observe the nucleation process, Evans and his collaborators have shown that with AP7 intercrystalline protein present in a saturated solution of calcium carbonate, the mineralization reaction forms clusters of ring-like or donut-like structures that eventually produce branching mineralized structures (Figure 5-4), whereas when AP7 is absent, amorphous calcium carbonate forms rapidly in a chaotic manner (Chang et al. 2014b). As the AP7mediated process continues over a period of minutes, calcite crystals grow and become coated with protein, which directs nanoparticles of calcite to attach to the growing crystal to create a three-dimensional, orthogonally arranged, ordered structure that again has protein on its surface (Figure 5-5). “There is a continual deposition of a protein phase on top of an existing crystal, causing nucleation and the assembly of nanoparticles on the surface,” said Evans. The framework protein N16.3, on the other hand, appears to be directing the emergence of new crystal growth directions and creating nanotexturing along the sides where emerging crystals are growing (Figure 5-6) (Chang et al. 2014a). Slicing through crystals grown in the presence of AP7 reveals that they are porous and that there are micro-, meso-, and macrosized pores all within the same crystal. “So not only do we modify the surface, we modify the interior as the crystal grows,” said Evans.
Evans and his colleagues have also studied what happens when there is more than one intercrystalline protein present and found that the kinetics of crystal formation can change significantly depending on the ratios of the different proteins. In one set of experiments, they paired AP7 with PFMG-1, a protein found in Japanese oyster pearl nacre and that by itself forms clusters of crystals that resemble pine cones. When the two proteins are mixed in equal mole amounts, the result is a pine cone structure with nanoparticles growing on top of it.
Figure 5-4 AP7 protein phases assemble and organize mineral nanoparticles, starting with ring-like structures in a premineralization step (A) and proceeding through early (B) and late (C) mineralization steps. SOURCE: Perovic et al. (2014b). Reprinted with permission from Biochemistry.
Figure 5-5 Intercrystalline protein AP7 induces the growth of ordered single-crystal nanocalcite coatings. SOURCE: Chang et al. (2014b). Reprinted with permission from Biochemistry.
Measurements of amorphous calcium carbonate solubility as a function of time showed that nucleation events are slightly destabilized when the AP7:PFMG-1 ratio was 10:1, a little more destabilized at a 1:10 ratio, and strongly destabilized at a 1:1 ratio, though on a longer time frame. At equal molar amounts, explained Evans, “it takes longer for amorphous calcium carbonate to form and it is highly unstable relative to what either protein does on its own.” In addition, there is a second, smaller nucleation event that occurs later when the two proteins are present in equal amounts. Further experiments showed that nucleation can even stop with the result that a stable phase forms in which nucleation is essentially suspended. This is called a polymer-induced liquid-phase precursor, something that was observed a decade ago with polymer systems that form lipid phases with amorphous precursors. He noted that this interesting phenomenon needs further study. He also predicted, in closing, that these biomineral assembly proteins could be used to direct the assembly of nanoparticles made of other minerals.
Figure 5-6 Framework protein N16.3 directs crystal growth and nanotexturing. SOURCE: Chang et al. (2014b). Reproduced with permission from CrystEngComm.
“Mineral formation is a manifestation of the intimate link that exists between biology and the environment,” said Jim De Yoreo. The process of biomineralization, he added, occurs on such a vast scale that it affects the planet’s chemistry and geology. The White Cliffs of Dover, for example, are made up almost exclusively of the deposits of a single organism and they record the interaction between the planet and its biota stretching back through the Cambrian Period. He also noted that the nucleation of ice in clouds is driven by mineral aerosols that originate in soils, often the result of biological activity. “The point is, you cannot separate the near-surface mineralogy of the Earth from the action of biology,” said De Yoreo.
Nucleation, he said, occurs through unstable density fluctuations that overcome a free energy barrier, a theory put forth by Josiah Williard Gibbs in two papers published in 1876 and 1878 as a means of explaining how water droplets form but that was subsequently applied to crystals. De Yoreo explained that the rate of nucleation can be written in terms of two exponential Boltzmann factors, one of which describes the atomistic processes involved in crystallization, including the desolvation of ions, and the other describing the free energy barrier that Gibbs postulated. He also explained that the free energy expressions have a dependence on volume that is stabilizing and a dependence on area that is destabilizing. The volume effect arises because precipitation into the bulk drops the free energy of the system to zero, while the surface effect comes with a free energy penalty. Together, these competing energy terms create the free energy barrier.
As Evans noted in his presentation, recent data and simulations argue for nonclassical mechanisms of nucleation in which multi-ion clusters first aggregate to form amorphous particles that then transform to crystals. In this model, the classical nucleation barrier is eliminated by the aggregates of these clusters, said De Yoreo. However, there are also data regarding the distribution of smaller clusters of calcium carbonate that are seen in cryoelectron micrographs that he said argue for
classical mechanisms of nucleation. “If you look at the distribution, they look essentially classical,” said De Yoreo, who added that this debate needs further study to resolve, he said.
Another possible mechanism posits that these systems crystallize by first phase-separating into two liquids, one of which is a dense liquid phase that then dehydrates and turns into a crystalline phase. Molecular dynamics simulations suggest that the free energy of clustering is a downhill drop with no barrier, and that the diffusivity of the ions within these clusters is essentially that of liquids and orders of magnitude larger than what would be expected for solids, he explained (Bewernitz et al. 2012, Wallace et al. 2013).
There has also been a change in thinking in recent years about what takes place during crystal growth after nucleation. Classical crystal growth happens by the addition of monomers to step edges that can be generated by two-dimensional nucleation or via a dislocation. This classical view works well for bulk soluble crystals growing at low supersaturation. Many crystals, however, grow instead through particle-based processes, both of biogenic and nonbiogenic origin, such as those that Gilbert and Evans discussed and that other investigators have described. What is needed, said De Yoreo, is a comprehensive model that quantitatively describes the diverse pathways of mineralization (Figure 5-7). The essential challenge, he said, is to describe the ensemble behavior of a system that represents a coupled solvent–solid system. “That coupling happens across scales, and I don’t just mean length scales, I mean time scales and energy scales. And we want to describe the ensemble process, but we want it to be predictive, and that means we have to maintain molecular fidelity where it’s needed,” said De Yoreo.
The challenge, then, is to create theoretic frameworks and computational tools to describe ensemble phenomena with molecular fidelity and in situ experimental tools that can visualize molecular events. In terms of key mesoscale science challenges, De Yoreo listed the need to understand solution structure and fluctuations at solution–solid interfaces and confined interparticle regions; the evolution of fields and forces during assembly and translation into motion; the size dependence of free energy of formation, solvation energy, and phase stability; and the existence of prenucleation clusters and dense liquid states in electrolyte solutions.
His group is tackling some of these challenges by using in situ methods to watch how nucleation and growth occur. In one set of experiments, his group is using in situ atomic force microscopy to examine how microbial membranes form two-dimensional crystals with a well-ordered lattice structure. These crystals form with the aid of proteins that start out in a disordered form but over time become crystalline. Once they reach the crystalline stage, tetramers form along the growing crystals’ edges. De Yoreo explained that because these tetramers never form before crystallization starts, it is “apparently not possible for the protein to find its way to the folded tetrameric state without first passing through this liquid-like state.” Once they do form, however, they never leave and the clusters never disappear, so this is not a nucleation process because the solid does not communicate reversibly with the reservoir of proteins in solution. Taken together, these observations suggest that structural fluctuations occur on a longer timescale than density fluctuations and that conformational entropy drives a two-step pathway via an irreversible formation of liquid-like precursors. The conformational entropy drives the system through this two-step pathway to irreversibly form liquid-like droplets in which the proteins are in close enough contact for a long enough time to find the specific bonds they need to form an ordered structure. In this system, De Yoreo said, the emergence of order catalyzes assembly (Chung et al. 2010).
Mineral systems are no less complex, said De Yoreo. Studies on calcium carbonate nucleation demonstrated that multiple processes can occur simultaneously. In one process, amorphous calcium carbonate grows in size until a new nucleation event, which occurs on the surface of the particle, transforms amorphous calcium carbonate into crystalline aragonite. What these experiments showed was that all of the phases of the calcium carbonate system are stable relative to the solution and metastable relative to calcite. Moreover, the level of supersaturation needed stabilize amorphous calcium carbonate makes all of the pathways available and which one dominates in the transition to calcite is unpredictable.
Living systems, however, cannot tolerate this type of flexible system, and they use proteins and
Figure 5-7 The many diverse pathways of mineralization. SOURCE: De Yoreo (2014). Reproduced with permission from Jim De Yoreo and Pupa Gilbert.
other biological polymers to direct transformation to specific crystalline forms. In nature, surfaces are covered in proteins and other organics, which impacts both nucleation rate and the mineral formation pathway. De Yoreo explained that when a nucleus forms on a surface, this creates an interfacial energy between the crystal and the substrate, but it also eliminates an interface between the substrate and liquid that had existed. The result is that the new interfacial energy is a composite term that is equal to the old one times a correction factor.
It turns out, said De Yoreo, that adding organic polymers to a supersaturated solution of calcium carbonate can redirect nucleation pathways. For example, polystyrene sulfonate, a polyelectrolyte polymer, binds large percentages of the calcium ions in solutions and produces globules that are 20 to 80 nm in size. Adding carbonate to the solution results in nucleation only occurring in these globules and the nuclei that form are amorphous. In another set of experiments with iron oxide, De Yoreo and his colleagues observed that ferrihydrate forms dumbbell-shaped particles that engage in long-range interactions over 5 to 10 nm with other particles that causes them to jump to each other over a 5- to 10-angstrom distance (Nielsen et al. 2014). Plots of translational and angular velocity versus distance show a sharp increase occurring when two particles come within about 1 nm of each other (Li et al. 2012). De Yoreo said that calculations suggest that van der Waals and ion correlation forces can act at long range to bring the particles together, but that electrostatic forces dominate in this case.
He concluded his presentation with his view of some of the key mesoscale science challenges regarding understanding nucleation and growth. “We don’t know what the solution structure and its fluctuations are at solution–solid interfaces, and we know even less about it in the region between particles. We don’t know how the fields and forces evolve during assembly, and how those fields and forces translate into motion. We have to understand what the size dependence of free energies, solvation energies, and phase stabilities are and those impact
whether or not prenucleation clusters really exist and take part in the formation process or whether there are dense liquid states within the system,” said De Yoreo. Above all, he added, there is a need for theoretical frameworks and computational tools that can describe the ensemble phenomenon in these systems while keeping molecular fidelity where it is needed, and experimental tools that can visualize molecular-scale events in real time.
Paleo proxies are one of the fundamental tools used to understand Earth’s history and climate, explained Alex Gagnon. The chemical composition, or geochemical signature, of that archive reflects how mass and energy has moved around the planet. While measurements of these paleo proxies have been incredibly important for conducting bulk analysis, they display unusual and unpredictable properties when used at the mesoscale. “If you look at a really small scale, what we typically see, especially in biominerals but in a large number of systems, is small-scale but large-magnitude compositional heterogeneity that presents both a challenge in how we interpret how we think proxies work and an opportunity in that there is rich information encoded in this heterogeneity that could potentially help us understand and reconstruct climate signals in new and improved ways,” said Gagnon.
One biomineral of particular interest in this regard is produced by a free-living, single-cell organism called foraminifera that produces a calcium carbonate shell. When this organism is building its calcium carbonate shell, it is incorporating the O16/O18 ratio that exists in the ocean at that particular time. This ratio is influenced by the amount of water that is deposited in ice—the process of evaporating water and making snow fractionates stable isotopes, and when done over a long enough period of time, such as during an ice age, the composition of seawater will change and that is reflected in this O16/O18 ratio. As these organisms die, they fall to the bottom of the ocean and build up large deposits of carbonates that can be sampled by taking cores. Analyzing those cores for their O16/O18 ratio provided the first insights into the fact that there had been regular ice ages in the past and that analysis, conducted in the 1960s, showed both the time and pace of these ice ages.
Another important paleo proxy, one used widely to reconstruct sea surface temperatures, also uses foraminifera and it is based on the empirical observation that its skeleton is not pure calcium carbonate but includes a number of impurities, particularly magnesium. The empirical observation is that if these organisms are cultured at different temperatures, the magnesium/calcium ratio of a collected group of these organisms changes as a function of temperature, and that an analysis of sediment cores for that ratio provides a proxy for past sea surface temperatures. At the bulk scale, the correlation between temperature and magnesium/calcium ratio is robust within a particular species of foraminifera. Again, the signals from this paleo proxy show alternating periods of warm conditions and ice age conditions, and they are reproduced in sediments from around the world. One interesting feature of this paleo proxy is that it provides tropical and high-latitude temperatures and the differences between them are what drive heat and power weather systems around the planet.
The problem with these proxies arises when looking at individual foraminifera, and the reason is that temperature is not the only factor involved in influencing the magnesium/calcium ratio, and a number of different anomalies hinted at the problem, including the fact that simple precipitation experiments produced much different results than what is seen in biominerals. The most conclusive experiments were done on individual foraminifera that had been grown in the laboratory at constant temperature. Accurate spot analysis of individual shells showed that the magnesium/calcium ratio could vary by as much as a factor of 2 to 4 in a single organism cultured at constant temperature. Measurements using a variety of microanalytical techniques—including nanoscale time-of-flight secondary ion mass spectrometry (TOF-SIMS), which provides compositional isotopic maps at the submicron scale on spots the size of a few hundred nanometers—revealed regular banding of magnesium and calcium levels even with the shells of individual foraminifera. This finding raises the question, then, of what modulates this banding beyond temperature.
Figure 5-8 Micrographs showing the primary organic membrane (POM) on which foraminifera and corals grow their mineralized structures. SOURCES: (A) Spero (1988). Reprinted with permission from Marine Biology. (B) Eggins, Sadekov, and De Deckker (2004). Reprinted with permission from Earth and Planetary Science Letters.
One possibility is that pH could be playing a role, and indeed, when Gagnon and his collaborators grew foraminifera in the lab in a range of pHs, they saw the submicron scale of magnesium/calcium banding changed as a function of pH. Foraminifera cultured at low pH showed high-amplitude banding, while those grown at high pH showed much lower-amplitude banding. Additional analyses done on impurities other than magnesium, such as sodium, identified similar patterns of variation at the mesoscale. His group is now collecting data for a number of different ions using nano-TOF-SIMS to construct detailed compositional maps of foraminifera grown under different conditions. He noted that in every foraminifera examined so far there is a high-sodium region associated with a known organic membrane that is laminated to the calcite skeleton. He also said that his team has found these variations at the mesoscale in corals and other biominerals.
Taken together, these data raise questions about how these mesoscale processes of biomineralization affect the interpretation of records of Earth’s climate history. What is needed, said Gagnon, is a set of tools to help map this heterogeneity and to explain this heterogeneity at a chemical scale. He said that he believes this heterogeneity can provide important clues about the past climate history of the Earth that are presently missing. In addition, a better understanding of these processes could provide inspiration for biomimetic approaches for controlling the composition of minerals and materials at very small scales.
Turning to the issue of identifying mechanisms that might be producing these banding patterns, Gagnon first reviewed some possible mechanistic forces that could be operating, including diurnal forcing of the foraminifera microenvironment, the dynamics and chemistry at the site of calcification as influenced by organismal ion-specific pumping, and the mineral growth kinetics and thermodynamics as affected by organic mineral interactions. Foraminifera, it turns out, are a good model for studying how organic–mineral interactions might affect composition, in part because they have a compositional variability that is structured and because they lay down a very clear organic layer, known as the primary organic membrane, upon which they grow their skeletons (Figure 5-8). Examining the detailed composition of the minerals extending out from this membrane could provide information on how the organic mineral interface influences composition.
Gagnon and his team have used atom probe tomography, a method for reconstructing the atom-by-atom composition of a sample, to explore this interface in detail. This field emission technique is typically used with conductive materials because it requires applying a large charge to the sample in order to ionize individual atoms from a prepared section of the sample, but with some creative and brute-force approaches and by taking advantage of the fact that carbonates are insulators, his team was able to analyze the atomic composition at the organic mineral interface from a single foraminifera skeleton. One observation from these experiments was that the sodium concentration is increased dramatically on the organic side of the interface, a finding that he said can help distinguish how
Figure 5-9 Sodium intensity image obtained using nanoscale TOF-SIMS. SOURCE: Gagnon (2014). Reproduced with permission from E. Bonnin and Andrew Gagnon.
organics and the organic–mineral interface influences composition.
“There is a compositional signature, and the location can tell us about mechanism,” he said,adding that these data are consistent with the signature of the large sodium signature detected with other methods, such as time-of-flight secondary ion mass spectrometry (Figure 5-9), that are used to make measurements at larger scales. “What we’ve now shown is that we can scale up from nanometer-scale measurement that tells us about the chemistry of these interactions to microscale maps of composition and then move from these micron-scale maps of composition to be able to tell us about ensemble behavior and global climate,” said Gagnon.
In the final presentation of this session, Andrew Madden illustrated the wide applicability of mesoscale chemistry to the study of Earth and other planetary systems. To do so, he presented three unrelated vignettes. The first vignette explored how to investigate size-dependent reactivity in the natural world. One approach would be to do laboratory experiments and study processes that might be important for the environment, make observations of the natural world, and see if the two sets of data correlate with each other. “But what we actually find in the natural environment is that nanoparticles aren’t really isolated,” said Madden, suggesting that it might be possible to use observations from the natural environment to inform new laboratory experiments. “In that way, we could say whether or not size-dependent reactivity and nanoscale size effects were important for the natural world.”
In fact, what Madden and his colleagues did was to do experiments in both the lab and nature simultaneously by taking three different sizes of magnetite nanoparticles and putting them in groundwater in nature and in synthetic groundwater in the laboratory. What they found from comparing the results from these two environments is that the nanoparticles were far more reactive in the laboratory than in the field. Moreover, in the lab the smallest particles were the most reactive, but in the field it was the largest particles that displayed any reactivity at all (Swindle et al. 2014). What was happening, explained Madden, was that large organic molecules accumulate on the nanoparticles in natural groundwater, which has the effect of passivating the ability of the nanoparticles to react (Figure 5-10).
Figure 5-10 An organic film on a magnetite nanoparticle suppresses field reactivity. The black bar represents 10 nm. SOURCE: Swindle et al. (2014). Reprinted with permission from Environmental Science & Technology.
Another project, this one involving a bacterium that oxidized magnesium, found that organics were required for optimal manganese oxidation and crystal formation (Learman et al. 2011). In these experiments, manganese oxidation under abiotic conditions produced disordered, poorly crystalline manganese oxide as the end product, but when conducted with cell-free extracts prepared from a bacterium that oxidized manganese, the result was initially the same, but over time much larger, and ordered manganese oxide crystals formed. “So one of the research challenges we could think about in applying nanoscale size effects to the natural world is how can we sort out when organics passivate nanoparticle reactivity and when they actually promote nanoparticle reactivity, and what’s the balance between those in different environments,” said Madden.
The second vignette examined whether nanomaterials and mesomaterials could be involved in the processes that trigger earthquakes. Following the 1999 Chi-Chi earthquake in Taiwan, researchers collected core rock material from the fault zone that was active during that earthquake. While this might seem to be a stretch, a careful examination of the material collected showed that it was made out of nanoscale to mesoscale particles (Ma et al. 2006). Using an instrument that can simulate the type of processes occurring in fault zones, Madden and University of Oklahoma colleague and collaborator Ze’ev Reches found the same type of nanoparticulate materials in the gouged rocks. It turns out that these nanoscale and mesoscale particles that are gouged out of the rock as a fault begins to slip may be coming together to create what are essentially roller bearings. These bearings then reduce the friction at the fault line and allow the fault to continue to slip, producing an earthquake (Figure 5-11) (Reches and Lockner 2010).
Madden explained that the drop in friction between the two plates in Reches’s instrument that simulate a moving fault correlates with the number and diameter of the rollers present between the moving plates. He noted, too, that there is a fundamental aspect of how the inner particle forces in the gouge particles are creating strength enough to hold those
Figure 5-11 Rollers composed of nanoscale mineral grains with composition similar to bulk rock are formed under conditions re-created to mimic those that occur during a fault-slip earthquake. SOURCE: Madden (2014). Reproduced with permission from Xiaofeng Chen, Andrew Elwood Madden, and Ze’ev Reches.
rollers together until they reach a critical size, at which point they fall apart. Understanding the forces involved will require further study, he said. Vignette number three dealt with efforts to trace the history of water on Mars and to understand how long rocks on Mars were in contact with water and how that contact produced the geological signatures seen today. The key mineral of interest in this study is called jarosite, a hydrous sulfate of potassium and iron that only forms a very narrow set of chemical conditions. Outside of those conditions, jarosite breaks down with continued contact with water to produce simpler iron oxides. “The fact that jarosite is found on Mars tells us that the duration of water–rock interaction was not that long, and we can actually try to time it,” explained Madden.
Madden, working with the principal investigator for this project, Megan Elwood Madden, measured the dissolution kinetics of jarosite under various conditions and saw what thousands of previous geochemical experiments have seen—solute release into an aqueous solution rises quickly and then plateaus. However, further experiments using saturated brines found that when jarosite is in contact with these saturated brines, the dissolution reaction suddenly accelerates at some point as a result of a feedback process involving the dissolution products. Further experiments found that when jarosite dissolves, it releases iron, which precipitates, and potassium sulfate, which does not. The sulfate ions combine with calcium to form gypsum, which now disrupts the charge balance in the system and the residual chlorine then attacks the jarosite, hastening its dissolution. These results are of interest because mesoscale structures known as shales are now being tapped as sources of natural as and oil. The shales contain pores that can be filled with the briny solutions being injected into shale deposits. “I think that once we start interrogating these rocks in different ways, we’re going to find some unique mesoscale effects that will change the types of mineral–water interactions that might be important from what we thought otherwise,” said Madden in concluding his presentation.
Session moderator Patricia Thiel started the discussion by asking if the panelists saw any connections between their work and the work on catalysis that was discussed in the workshop’s first panel presentation. Gilbert responded first by noting that there is one big difference between catalysis and biomineralization processes: catalysis is about accelerating reactions, whereas biomineralization retards both the nucleation and crystal growth process that proceed rapidly under abiotic conditions. Evans elaborated on Gilbert’s remarks by noting that in biomineralization there is a “pause button” where biomolecules come together and regulate nucleation and subsequent steps in a way that allows many other components to come into play at different times in the nucleation and growth phases to further regulate and control the process to
produce biominerals with the necessary physical properties.
Gagnon said that there are analogies between the two fields because both are trying to understand how different components of a complex system interact at the mesoscale to produce a desired function. Just as model systems of enzymes that only recapitulate the inorganic metal centers are informative but do not capture the full behavior of the native enzyme, abiotic models of biomineralization provide useful insights but do not capture the full complexity of the system in nature. De Yoreo added that one similarity both fields have is that they are trying to understand the energy landscape that drives or inhibits reactions, and in both catalysis and biomineralization there are proteins that are manipulating energy landscapes, in the former to accelerate a chemical reaction and in the latter to control the kinetics of crystal formation.
Anne Chaka, from PNNL, asked the panel what role water positioning and transport plays in biomineralization processes. Gilbert replied that one of the keys to biomineralization reactions in general is to exclude water and that biomineralization always occurs behind closed membranes from which water is excluded. She added that going through an amorphous precursor, which she described in her presentation as playing an important role in the formation of calcite, is a “smart trick because if you calculate the volume of a supersaturated solution that would be needed otherwise, it would be orders of magnitude larger than the entire organism.” In short, she said, there is no water involved in biomineralization. De Yoreo tempered Gilbert’s comments by saying that the biomineral world is vast. “The number of cases that have been studied enough to understand whether or not particular aspects such as the level of hydration or the utility of water in transformations in general is small, and we really can’t say,” said De Yoreo. “We will just have to see over time how that develops.” Gagnon added, “There’s quite a wide range of strategies that different organisms appear to use about what the water-to-mineral ratio is during biomineralization.”
Wendy Shaw, from PNNL, asked De Yoreo and Evans if they had any insights on what drives iron oxide particles to make a 5- to 10-angstrom jump from one crystal to another during biomineralization or why some minerals when they’re forming start out having a donut shape with a void. De Yoreo’s short answer was that the mechanism is unknown, and that there are many potential forces at work. “The only thing we can say for certain is we know what the accelerations are and calculate the force, and we know what the change in potential energy is during that process,” said De Yoreo. Calculation using Coulomb’s Law can provide a reasonable answer that suggests that electrostatics plays a role, which would be consistent with the length of the jump, but interfaces are complex and there are many potential interactions that could be responsible. “This is one of the open questions about particle-based growth processes and oriented attachment in particular,” said De Yoreo. Evans added his conjecture, which was that prior to becoming visible, there are actually clusters already forming that are not visible yet simply because they haven’t reached a critical density that enables them to be visualized. De Yoreo said, however, that there are data that argue against that view of assembly. Evans said that another possible model is that there are oligomer gels that are not visible that are attracting ions and creating localized supersaturation that triggers nucleation on the gel filaments.
Donna Blackmond asked the panel if there are chiral biominerals, and both De Yoreo and Evans said that they were not aware of any chiral biomineral crystals. Andrew Stack, from Oak Ridge National Laboratory, then asked if the field was advanced enough to be able to predict how a given biomineral will grow into a given shape, such as a mollusk shell. The unanimous answer from the panelists was no, though Gagnon noted that it is starting to be possible to modify existing processes to change morphology or composition, but he characterized that as a baby step toward total control and prediction. De Yoreo categorized the state of the field as being at the stage of having intuitive phenomenological approaches but not able to make predictions. Gilbert summarized the state of the field by saying, “We don’t know the laws of nature at the mesoscale and because we don’t know the fundamental laws, we cannot make predictions.”
Benjamin Wilhite asked Evans about the generalizability of some of the processes he described, that is whether instead of making carbonates these processes could be adapted to work with nitrate precursors, or if instead of working with calcium they could be applied to work with other transition metals. In particular, he was wondering if it would be possible to adapt the biological systems to make columnar structures and oriented crystals for use in energy applications, for example. Evans
replied that the “devil is in the details,” and that the idea of borrowing from biology is “tantalizing.” One question that needs answering with regard to this question, said Evans, is how these proteins would behave under solution conditions, such as pH or ion concentrations, that differ from the conditions they see in nature.
Workshop organizing committee and CSR member Miguel Garcia-Garibay, from the University of California, Los Angeles, asked if the products of biomineralization processes are in a stable state, and De Yoreo said that in most cases the answer is yes. The aragonite crystals in nacre that Gilbert talked about are an exception because aragonite is not the stable state of calcium carbonate at room temperature. “That system is caught in a metastable state,” said De Yoreo. Garcia-Garibay noted that vertebrate bone, a biomineral, is in a homeostatic state, not a stable state. Peter Stair followed that comment by asking the panelists if they had any ideas on what factors are involved in creating stable structures in the mesoscale realm. Gilbert replied that the answer is no, but that proteins are likely involved. She noted that aragonite only forms when there is an excess of magnesium present along with calcium carbonate even though there is no magnesium found in the aragonite crystals. It may be that magnesium is influencing the behavior of proteins that are involved in regulating aragonite production.