Myriad Mapping of nanoscale minerals reveals calcium carbonate hemihydrate in forming nacre and coral biominerals

Marine carbonate biominerals (e.g., mollusk shells, coral skeletons) are central to the global carbon cycle and form via complex pathways. Prior work emphasized amorphous calcium carbonate (ACC) precursors, but mechanisms remain debated because they influence resilience to ocean warming/acidification and interpretations of geochemical proxies. This study asks whether crystalline hydrated calcium carbonate phases, specifically calcium carbonate hemihydrate (CCHH) and monohydrocalcite (MHC), occur naturally as transient precursors during biomineral formation, and how they are spatially distributed relative to amorphous precursors and mature phases. The authors introduce Myriad Mapping (MM), a quantitative imaging method to map more than three mineral phases at nanoscale resolution, to test the occurrence and role of multiple transient phases in forming coral skeletons, nacre, and sea urchin spines.

Multiple studies have shown amorphous calcium carbonate (ACC; hydrated and anhydrous) as a transient precursor in diverse biominerals (sea urchins, mollusks, corals). However, crystalline precursors have been less frequently identified; vaterite has been reported as a precursor to calcite in certain foraminifera. Calcium carbonate hemihydrate (CCHH) has been characterized only synthetically. Analytical approaches for mapping phases were previously limited to three components (RGB/CMY), potentially obscuring additional phases. Ikaite spectra are difficult to obtain due to its instability at room temperature, making its biogenic role uncertain. This context motivates the search for crystalline hydrated intermediates (CCHH, MHC) in aragonitic and calcitic biominerals using improved spectral mapping and corroborative infrared methods.

• Sample types: Forming coral skeletons (multiple species including Stylophora pistillata), nacre from Haliotis rufescens, and regenerating sea urchin spines (Strongylocentrotus purpuratus, Paracentrotus lividus). Total of 86 areas (20–65 µm field), each acquired in duplicate or triplicate, yielding 180 acquisitions and analyzing on the order of 200 million Ca-containing pixels; >1.95×10⁴ spectra summarized in detailed datasets.
• Preparation: Fixation and dehydration protocols designed to preserve metastable phases (cacodylate buffers in carbonate solution; ethanol dehydration; embedding in EpoFix; cutting/polishing using 22 g/L Na₂CO₃ coolant; thin Pt coating for conductivity). Steps aimed to minimize dissolution/recrystallization artifacts.
• X-ray spectromicroscopy (PEEM-3, ALS beamline 11.0.1.1): Calcium L-edge (340–360 eV) XANES image stacks (121 images per stack; 0.1 eV steps between 345–355 eV; typical pixel size 60 nm; lateral resolution down to 20 nm; detection depth ~3 nm). Circular polarization to minimize orientation effects; regions were virgin (no prior exposure). Each pixel contains a Ca L-edge spectrum of unknown composition.
• Component spectra and fitting: Component library (Cni16) derived from synthetic standards for ACCH₂O (hydrated ACC), ACC (anhydrous), CCHH, MHC, aragonite (or calcite for spines). All component spectra peak-fitted with consistent backgrounds/arctangents; normalized so peak 1 = 10 at 352.6 eV. Unknown single-pixel spectra were fit by nonlinear least-squares linear combinations of components over 340–355 eV with allowance for small rigid energy shifts. Goodness-of-fit assessed by reduced χ².
• Myriad Mapping (MM): For each component, proportion maps (pMaps) were generated (0–1 proportion per pixel). Duotone images were created per phase (0% black to 100% assigned color), then only pixels with >50% proportion for a given phase were retained and overlaid as layers, ensuring non-overlapping phase assignments. This allows display of more than three phases without ambiguity (mixed pixels <50% for any phase are not shown). Masking by Ca difference maps and χ² thresholds minimized noisy assignments.
• Model selection/statistics: Compared 3-, 4-, and 5-component fits. Inclusion of both CCHH and MHC (5 components) significantly improved fits over 3-component analysis (e.g., p = 3.0×10⁻²⁰⁷ for CCHH pixels). Adding vaterite (6 components) did not significantly improve fits (e.g., p = 0.63 for vaterite-assigned pixels), so vaterite was excluded from MMs.
• Infrared validation (SINS-FTIR, ALS BL 2.4 and 5.4): Surface-sensitive nano-FTIR (voxel ~20 nm³; probing depth ~20 nm) measured ν₃ and ν₂ carbonate bands on fresh coral surfaces. Synthetic references (CCHH, MHC, ACC, ACCH₂O, aragonite) were measured; CCHH shows ν₃ splitting into three peaks; MHC into two. Shifts in biogenic spectra relative to synthetic were analyzed to assess phase identity despite matrix effects. XRD confirmed phase identity of synthetic standards.
• Additional details: For nacre, interfaces between forming tablets were examined. Repeat acquisitions assessed temporal transformations (beam exposure/time). Extensive supplemental datasets include peak parameters, χ² comparisons, and area-wise summaries.

• Discovery of crystalline hydrated precursors: Calcium carbonate hemihydrate (CCHH) and monohydrocalcite (MHC) were detected as transient crystalline phases localized at forming surfaces of aragonitic biominerals (coral skeletons and nacre). They were not present in bulk, where aragonite dominates.
• Spatial organization: Precursors cluster near surfaces with characteristic layering. In corals, CCHH and ACC form discontinuous layers 0–800 nm thick below the surface; ACCH₂O and MHC occur in a thinner outermost layer (<100 nm). In nacre, CCHH and ACC form discontinuous ~100 nm layers at interfaces between forming tablets; absent between mature tablets.
• Occurrence by phylum and phase fractions (proportion of precursor pixels): • Nacre: CCHH 48%, ACC 36%, MHC 9%, ACCH₂O ~5%. • Coral skeletons: CCHH 46%, ACC 39%, MHC 12%, ACCH₂O ~5%. • Sea urchin spines (calcitic): ACC 90%, ACCH₂O 5%, CCHH 3%, MHC 2% (i.e., only traces of CCHH/MHC).
• Infrared corroboration: SINS-FTIR on coral surfaces showed ν₃ splitting consistent with CCHH (3 peaks) and MHC (2 peaks); ν₃ peak positions were shifted relative to synthetic standards but retained diagnostic splitting patterns. ACC/ACCH₂O signatures were also detected at surfaces.
• Model selection/statistical validation: Five-component fits (including CCHH and MHC) significantly outperformed three-component analyses (e.g., p = 3.0×10⁻²⁰⁷ for CCHH pixels); inclusion of vaterite did not improve fits (p = 0.63 for vaterite pixels). Peak position/intensity patterns at Ca L-edge (notably peaks 2 and 4) distinguished phases.
• Temporal and spatial transitions: Repeat acquisitions revealed thermodynamically downhill transformations (e.g., ACCH₂O → ACC/aragonite; CCHH/MHC → aragonite). Spatial gradients suggested sequences such as ACCH₂O → CCHH → aragonite.
• Energy landscape: Proposed multi-pathway landscape for aragonitic biominerals (ACCH₂O → aragonite; ACCH₂O → ACC → aragonite; ACCH₂O → CCHH → aragonite; ACCH₂O → MHC → aragonite). Calcitic biominerals follow simpler pathways (ACCH₂O → calcite; ACCH₂O → ACC → calcite). Synthetic evidence and biogenic distributions support that CCHH/MHC are not significant precursors to calcite.
• Analytical advance: Myriad Mapping (MM) enables quantitative visualization of more than three phases at nanoscale resolution, revealing phase diversity previously obscured by RGB/CMY limits.

The findings demonstrate that biomineralization in aragonitic systems involves not only amorphous precursors (ACCH₂O, ACC) but also crystalline hydrated intermediates (CCHH, MHC) localized at forming surfaces. Their prevalence in aragonitic biominerals (coral, nacre) and near absence in calcitic sea urchin spines support a polymorph-specific role: CCHH and MHC are transient intermediates en route to aragonite but not calcite. This resolves aspects of prior debates limited by three-component analyses and suggests a more intricate energy landscape with multiple feasible pathways. The detection of these crystalline precursors has implications for trace element and isotope incorporation, potentially affecting paleoclimate reconstructions by introducing additional exchange or fractionation steps during multi-stage crystallization. Spatial and temporal observations indicate predominantly thermodynamically downhill transitions and suggest both solid-state transformations and, in some contexts, dissolution–reprecipitation could contribute, varying by system and stage. The MM approach, complemented by SINS-FTIR and synthetic standards, provides robust phase identification and mapping, expanding the toolkit for studying transient phases in natural and bio-inspired materials.

This work reports the first biogenic identification and spatial mapping of calcium carbonate hemihydrate (CCHH) and monohydrocalcite (MHC) as transient crystalline precursors during formation of aragonitic biominerals (coral skeletons, nacre). Introducing Myriad Mapping (MM) enabled quantitative visualization of multiple coexisting phases beyond traditional three-component limits, revealing clustered, surface-localized precursor layers and polymorph-specific pathways. A new multi-path energy landscape is proposed for aragonite formation, while calcite follows simpler routes. These insights refine understanding of biomineralization mechanisms and suggest that multi-step pathways may modulate elemental and isotopic signatures used for paleoenvironmental reconstructions. Future work should test co-occurring aragonite and calcite within the same organism (e.g., prismatic/calcitic and nacreous/aragonitic shell regions), and leverage accessible techniques such as EBSD, nanoSIMS, and PiFM to detect crystalline precursors with high spatial resolution and surface sensitivity.

• Potential artifacts: Some crystallization could occur post-mortem during sample preparation or during x-ray exposure, potentially accelerating thermodynamically downhill transitions (dehydration/crystallization). Although precautions were taken (carbonate buffers, ethanol stabilization, Na₂CO₃ coolant), complete exclusion of preparation-induced transitions is not possible.
• Mechanistic ambiguity: The study was not designed to definitively distinguish solid-state transformation from dissolution–reprecipitation; both may occur to varying extents depending on system, location, and stage.
• Surface sensitivity differences: PEEM (∼3 nm depth) is more surface-sensitive than SINS-FTIR (~20 nm), which may bias detectability of ultrathin/discontinuous surface layers.
• Phase coverage: Ikaite was not evaluated due to instability at room temperature; vaterite did not improve fits and was excluded, but rare occurrences could be below detection or confounded by other phases.
• Temporal resolution: Repeat acquisitions show transformations, but in vivo time-resolved sequences were not directly observed; the exact chronology in living organisms remains unresolved.