Organic Controls over Biomineral Ca–Mg Carbonate Compositions and Morphologies

Abstract

Calcium carbonate minerals, such as aragonite and calcite, are widespread in biomineral skeletons, shells, exoskeletons, and more. With rapidly increasing pCO₂ levels linked to anthropogenic climate change, carbonate minerals face the threat of dissolution, especially in an acidifying ocean. Given the right conditions, Ca–Mg carbonates (especially disordered dolomite and dolomite) are alternative minerals for organisms to utilize, with the added benefit of being harder and more resistant to dissolution. Ca–Mg carbonate also holds greater potential for carbon sequestration due to both Ca and Mg cations being available to bond with the carbonate group (CO₃^2–). However, Mg-bearing carbonates are relatively rare biominerals because the high kinetic energy barrier for the dehydration of the Mg²⁺–water complex severely restricts Mg incorporation in carbonates at Earth surface conditions. This work presents the first overview of the effects of the physiochemical properties of amino acids and chitins on the mineralogy, composition, and morphology of Ca–Mg carbonates in solutions and on solid surfaces. We discovered that acidic, negatively charged, hydrophilic amino acids (aspartic and glutamic) and chitins could induce the precipitation of high-magnesium calcite (HMC) and disordered dolomite in solution and on solid surfaces with these adsorbed biosubstrates via in vitro experiments. Thus, we expect that acidic amino acids and chitins are among the controlling factors in biomineralization used in different combinations to control the mineral phases, compositions, and morphologies of Ca–Mg carbonate biomineral crystals.

Short abstract

In this study, we show how acidic amino acids and chitin promote Ca−Mg carbonate precipitation and control crystal morphologies.

Introduction

Calcium carbonate minerals are the most common mineral components in biominerals across geologic time, ranging from the Neoproterozoic to the present.¹ From algae,² foraminifera,³ coral,⁴ mollusks,⁵ and echinoderms⁶ in aquatic environments, to ants,⁷ gastropods,⁸ and humans⁹ on land, biocarbonates are observed across a wide range of environments and organisms across the tree of life. Apart from serving as biomineral shells and skeletons, carbonates also play an important role in the global carbon cycle by sequestering carbon in long-term mineral storage and regulating the carbonate chemistry and pH of the ocean.¹⁰ Finally, they preserve vital records of biological and geological evolution on Earth as fossils.¹¹

Compared to pure calcium carbonate species such as calcite and aragonite, Mg-bearing calcium carbonates such as high-magnesium calcite (HMC), proto-dolomite, and disordered dolomite (Dol) are known to be more resistant to physical weathering and are potentially more imperviable to dissolution in an acidifying ocean.¹² Mg-bearing calcium carbonate could potentially serve as a better carbon sink by utilizing Mg²⁺ cations, which are abundant in natural water. However, the occurrence of Mg-bearing carbonates in modern-day systems is incredibly rare compared to pure calcium carbonate phases and has only been occasionally reported in organisms such as corals,¹³ mollusks,^14,83 sea urchins,¹⁵ and ants.⁷ This limited occurrence stems from the reluctance of magnesium cations to dehydrate in low-temperature environments, inhibiting magnesium from being incorporated into carbonates with calcite structures.^16,17

Synthetic experiments testing the mechanisms of high-Mg calcium carbonate formation demonstrated that homogeneous nucleation in the presence of dissolved silica,¹⁸ sulfide,¹⁹ polysaccharides,²⁰ exopolymeric substances (EPS),^21,22 and amphipathic molecules such as ethanol (and likely others with similar polarity/acidic/hydrophilicity properties) dehydrates the local mineral surface to allow for Mg incorporation into Ca–Mg carbonate.¹⁶ Here, we examine how biomolecules commonly observed in biominerals, such as amino acids and chitins, may be involved in Mg incorporation in Ca–Mg carbonates. Previous works have demonstrated how various amino acids and chitin have direct effects on CaCO₃ precipitation in mollusks,⁵ chitons,^7,23 foraminifera,²⁴ and corals.²⁵ Some biomineral species even display different amino acid compositions in aragonite versus calcite regions of coral skeletons⁴ and mollusks shells.²⁶⁻²⁸ Yet, the effect of these different biosubstrates on Mg incorporation in calcium carbonates and their resulting morphologies has not been previously tested.

In this study, we explore how different amino acids and chitin may be involved in Mg carbonate precipitation in biomineralization systems. We used mineralogical and geochemical approaches to precipitate synthetic Ca–Mg carbonates via in vitro experiments in solutions with various solid substrate surfaces to examine the final Mg concentrations and morphologies of the precipitates.

Methods

Synthesis Experiments

In vitro synthesis experiments were conducted in polyethylene bottles at room temperature (20 °C). Eight different amino acids were tested to probe the effect of different amino acid properties, such as polarity, hydrophilicity, and acidity, on carbonate formation (Table 1). Amino acids (7.5 mM) and chitin (0.025 g/L) were mixed into 50 mL of deionized (DI) water and continuously stirred in polyethylene bottles for half an hour. Then, 6.25 mM calcium chloride dihydrate (CaCl₂·2H₂O), 31.25 mM magnesium chloride hexahydrate (MgCl₂·6H₂O), and 25 mM sodium bicarbonate (NaHCO₃) were dissolved into the organic substrate bearing solutions. After another 30 min of mixing, to ensure that everything had dissolved, the bottles were sealed and kept at room temperature. After a week, solutions were filtered and air-dried.

Table 1. Amino Acid and Chitin Physicochemical Properties and the Resulting Carbonate Phases Precipitated in Solution with Starting Concentrations of [Mg²⁺] = 31.25 mM, [Ca²⁺] = 6.25 mM, [CO₃^2–] = 50 mM, and 20 mM of Organic Substrate (0.1 g/L for Chitin)^a.

substrate	polarity	charge at pH = 7.4	hydropathy index	hydrophobic/hydrophilic	physiochemistry class	precipitation in solution	MgCO3 mol %
aspartic acid	polar	negative	–3.5	hydrophilic	acidic	aragonite + HMC	21.80
glutamic acid	polar	negative	–3.5	hydrophilic	acidic	aragonite + HMC	21.60
glycine	nonpolar	neutral	–0.4	hydrophilic	aliphatic	aragonite
isoleucine	nonpolar	neutral	3.8	hydrophobic	aliphatic	aragonite
leucine	nonpolar	neutral	4.5	hydrophobic	aliphatic	aragonite
lysine	polar	positive	–3.9	hydrophilic	basic	aragonite
phenylalanine	nonpolar	neutral	2.8	hydrophobic	aromatic	aragonite
serine	polar	neutral	–0.8	hydrophilic	hydroxylic	aragonite
chitin						aragonite + HMC	21.85

^aDetailed structures of amino acids and chitin can be found in Table S2.

In a separate set of experiments to test the effects of solid surfaces on Ca–Mg carbonate precipitation, aspartic acid, glutamic acid, and chitin were first individually mixed with deionized (DI) water. Natural amorphous opal-A (opal) slices from Australia, glass slides (GS), and gold-coated glass slides (GCGS) were then placed into these solutions for 24 h with closed lids to allow solutes to adsorb onto their surfaces. Next, calcium chloride dihydrate (CaCl₂·2H₂O) and magnesium chloride hexahydrate (MgCl₂·6H₂O) were added to the solutions to provide the Ca and Mg needed to precipitate Ca–Mg carbonate minerals (see Table 2 for solution concentrations). In most solutions, the initial Mg/Ca ratio was set to 5 as an analogue to the seawater Mg/Ca ratio (∼5); however, some solutions had lower Mg/Ca ratios of 3–2 (Table S1). Absolute concentrations of Mg range from 40 to 20 mM (Table S1). Polyethylene bottles containing these solutions (100 mL each) were placed into a sealed desiccator with ammonium bicarbonate ([NH₄]HCO₃) powders (20 g), allowing for NH₃ and the necessary CO₂ to diffuse into the open-lid solutions for the carbonate minerals to precipitate. After one week, the solutions were removed from the desiccators, and precipitates were filtered out with 20–25 μm pore-size filter papers. Surface materials, i.e., opal, GS, and GCGS, were picked out and rinsed with DI water for 30 seconds to remove precipitates that had been formed in the solution but settled on the surfaces, leaving behind only the minerals that precipitated directly onto the solid surfaces.

Table 2. Experimental Conditions for Solid Surface Experiments and Their Resulting Mineral Phases and mol % MgCO₃ in Calcite-Structure Carbonate^a.

substrate	surface	substrate concentration	[Mg2+] (mM)	[Ca2+] (mM)	precipitation phases	d104 (Å)	MgCO3 mol %
aspartic acid	solution	7.5 mM	31.25	6.25	aragonite + HMC	2.976	21.8
	Opal				aragonite + HMC	2.958	27.4
	GCGS				aragonite + Mg-cal	2.999	12.9
	GS				aragonite + Mg-cal	3.008	9.9
chitin	solution	0.0375 g	31.25	6.25	aragonite + HMC	2.974	21.9
	Opal				aragonite + Dol	2.931	46.3
	GCGS				aragonite + HMC	2.949	35.1
	GS				aragonite + HMC	2.953	32.3
control	solution		31.25	6.25	aragonite + calciteb
	Opal				aragonite + calciteb
	GCGS				aragonite + calciteb
	GS				aragonite + calciteb

^aMgCO₃ values were calculated based on d₁₀₄ values of the calcite -structure carbonates following procedures from Fang and Xu.³¹

^bDenotes a trace amount.

Analytic Procedures

X-ray diffraction (XRD) analyses were conducted using a Rigaku Rapid II XRD system (Mo Kα radiation, λ = 0.7093 Å) equipped with a two-dimensional (2D) image-plate detector in the S.W. Bailey X-ray Diffraction Laboratory in the Department of Geoscience at the University of Wisconsin-Madison. Diffraction patterns were collected on the image-plate detector and integrated using Rigaku 2DP software. Mineral phases were identified using the MDI Jade 9.5 software package with the American Mineralogist Crystal Database (AMCSD) and the PDF-4+ database from the International Centre for Diffraction Data (ICDD). Rietveld refinements calculated unit cell parameters, crystal sizes, phase fractions, and crystallinity using the Bruker TOPAS program and crystal structures from the AMCSD. Pearson VII peak functions were used for all refinements. We used the d₁₀₄ values of calcite and dolomite structures to estimate MgCO₃ percentages in Ca–Mg carbonates based on an empirical curve from previous works.^29,30

Scanning electron microscopy (SEM), X-ray energy-dispersive spectroscopy (EDS), and electron backscatter diffraction (EBSD) were conducted using a Hitachi S3400 equipped with Oxford Aztec EDS and EBSD systems at the Department of Geoscience at the University of Wisconsin-Madison. EDS imaging, including backscatter electron (BSE) imaging and secondary electron (SE) imaging, and EDS, was conducted at 15 kV with a horizontal setting. EBSD was performed at 20 kV with a 70° tilt. All samples were coated with 1 nm thick iridium to prevent charging. The EDS results were calibrated with a dolomite standard (Delight dolomite) with the published value of 50.48 mol % MgCO₃.²⁰²⁰ Additional SEM images were collected using a Thermofisher Quattro S ESEM at the Institute of Materials Science & Engineering at the Washington University in St. Louis.

The samples were analyzed using a JEOL 7600F field-emission-gun scanning transmission electron microscope at NASA Johnson Space Center (JSC) to obtain high-resolution transmission electron microscopy (HRTEM) images. Chemical compositions were characterized using a JEOL SD60GV 60 mm² ultrathin window silicon drift detector (SDD). A combination of spot analyses and element maps was collected from each sample to understand element distributions, abundances, and ratios. The energy-dispersive X-ray (EDX) mapping analyses were obtained using a 2 nm probe to reduce beam damage and showed the spatial distribution of major and minor elements.

Raman spectra were collected on a subset of experiment samples using a Horiba LabRAM Evolution Raman system with a CCD detector at the Smithsonian National Museum of Natural History in the Department of Mineral Sciences. We used an unpolarized 532 nm green laser coupled with an 1800 groove grating to collect spectra from 50 to 4000 cm^–1 Raman shift. Each spectrum was an average of five replicates collected for five seconds each at a 10% laser power (using neutral density filters) of approximately 4.4 mW. The spot size of the laser using the 100× objective was <5 μm in diameter. The laser was calibrated using a Si wafer and laser power measurements were made with a ThorLabs S170C Microscope Slide Power Sensor system on a silicon wafer. Representative Raman spectra of dolomite (NMNH161804, Penfield Quarry, Rochester, NY), low-Mg calcite (NMNH144953-33, Iceberg claim, Copper Mountain Mining District, Taos Co., NM), and aragonite (NMNHB10095, Spania Dolina, Slovakia) were collected from mineral samples from the Smithsonian National Gem and Mineral Collection, and are referred to as “standards.”

Results

Mineralogy of Synthetic Ca–Mg Carbonate Precipitates in Solution with Organic Substrates

XRD results show that suspended precipitates formed in solution with aspartic acid (Asp) and glutamic acid (Glu) are dominated by high-magnesium calcite (HMC; d₁₀₄ = 2.9755 and 2.9756 Å corresponding to 21.8 and 21.6 mol % MgCO₃ for aspartic acid and glutamic acid) with aragonite (Figure 1 and Table 1). Other amino acids (glycine, isoleucine, leucine, lysine, phenylalanine, serine) resulted in aragonite precipitation in the solution (Figure 1 and Table 1). Similar to the aspartic acid and glutamic acid experiments, chitin experiments also resulted in precipitates composed of HMC (d_104 = 2.9754 Å corresponding to 21.9 mol % MgCO₃; Figure 1 and Table 1).

Figure 1.

Mineralogical analyses of precipitates from biosubstrate-rich solutions. (A) X-ray powder diffraction patterns (Mo Kα) of precipitates formed with different amino acids and chitin solutions. Patterns in bold lines show the presence of high-Mg calcite (HMC) phases with (104) peak positions located between the calcite and dolomite standards. (B) Transmission electron microscopy (TEM) image of HMC formed on a chitin-coated opal surface. (C) Selected-area electron diffraction (SAED) patterns of (B) along [100] zone axis showing distinct “a”-reflections indicating large coherent domains and no “b”-reflection suggesting no Ca–Mg cation ordering. (D) TEM image of HMC formed in an aspartic acid solution. (E) SAED pattern of (D) along the [182] zone axis showing distinct a-reflections indicating large coherent domains and no b-reflection suggesting no Ca–Mg cation ordering.

TEM results confirm the presence of HMC in aspartic acid and chitin experiments and show that HMC crystals from synthesis experiments lack cation ordering with the absence of b-reflections typically observed in natural dolomite (Figure 1B–E; 30). b-Reflections originate from the reduced symmetry of the R3̅ space group compared to the R3̅c space group in calcite with only a-reflections present. HMC crystals in chitin experiments are rhombohedral in shape with well-developed {104} crystal facies (Figure 1B), whereas precipitates from aspartic acid experiments show as bundles of elongated crystals with less regular shaped crystals (Figure 1D).

Mineralogy of Synthetic Biocarbonate Precipitates on Solid Surfaces with Organic Substrates

Upon introducing solid surfaces (opal, glass slides, and gold-coated glass slides) for the organic substrates to attach to, we observed an additional set of carbonate precipitation behaviors. XRD and SEM–EDS revealed a significant increase in the Mg content in calcium carbonate precipitates on opal surfaces compared to in solution. Precipitates from aspartic acid experiments increased from 21.8 mol % MgCO₃ in solution to 27.4 mol % MgCO₃ on opal surfaces, and chitin experiments increased from 21.9 mol % MgCO₃ in solution to 46.3 mol % MgCO₃ (disordered dolomite) on opal surfaces (Figure 2 and Table 2). Compared to opal surfaces, experiments on glass slide (GS) and gold-coated glass slide (GCGS) surfaces showed different trends between aspartic acid and chitin, where the Mg content increased in chitin experiments (32.3 mol % MgCO₃ on GS and 35.1 mol % MgCO₃ on GCGS) but decreased in aspartic acid experiments (9.9 mol % MgCO₃ on GS and 12.9 mol % MgCO₃ on GCGS) compared to precipitates in solution (Table 2). Aragonite was present in all solid precipitation experiments.

Figure 2.

SEM images of precipitates on opal, glass slide (GS), and gold-coated glass slide (GCGS) surfaces in control, aspartic acid, and chitin experiments showing various morphologies of Ca–Mg carbonates, such as high-Mg calcite (HMC) with the {104} and {100} crystallographic forms and disordered dolomite (Dol) and aragonite (Arg). MgCO₃ values are obtained from both XRD and SEM–EDS.

SEM images demonstrated that precipitates have different crystal sizes and morphologies on opal, GS, and GCGS surfaces (Figure 2). In aspartic acid experiments, HMC forms as elongated spherulites on opal surfaces form (∼10 μm) and as euhedral crystals on GS (∼5 μm) and GCGS (∼2 μm; Figure 2). In chitin experiments, disordered dolomite precipitates as 1–3 μm (smaller based on TEM image) sized euhedral crystals with the {104} crystallographic form on opal surfaces while HMC forms as slightly more elongated euhedral crystals with both the {104} and {100} forms on GS and GCGS surfaces (Figure 2). Aragonite spherulites precipitated on all three surfaces with a slower growth rate on the opal surface (Figure 2). Electron backscatter diffraction (EBSD) revealed that precipitates on GCGS surfaces displayed a strong preferred crystallographic orientation with the c-axes perpendicular to the surface, while the preferred orientation was diminished on GS surfaces (Figure 3). Twinning is only observed in precipitates from solution in aspartic acid experiments (HMC with (012) twins) but not in chitin experiments (Figure S1).

Figure 3.

EBSD pole figures showing a strong preferred crystallographic orientation of high-Mg calcite (HMC) on gold-coated glass slide (GCGS) surfaces and a lesser degree of preferred orientation on the glass slide (GS) surfaces from aspartic acid and chitin experiments.

A series of Raman spectra and optical images reveal a transformation from magnesium-rich amorphous calcium carbonate (Mg ACC) to HMC precipitates in the glass slide aspartic acid experiments. The Mg ACC phases were still present approximately six months after being precipitated (Figure 4A). We assume that these are Mg ACC spectra due to the very wide full width at half-maximum (FWHM) of the ν₁ carbonate group symmetric stretching vibrational mode peak, which signals increased disorder and also resembles the shape of the lower-Mg ACC standard (Figure 4A,C). It is important to note that wider FWHMs can also be a function of smaller particle sizes, but this is not likely the case here because we still observe wide FWHM in spectra collected off of single crystals. As the precipitates transformed from Mg ACC to become more like HMC, the ν₁ mode peaks became thinner and sharper (Figure 4C). We also observed that the Raman water envelope (∼3500 cm^–1) for Mg ACC (Figure 4A, spectra E–G) resembled the reference low-Mg ACC standard and began to disappear in the crystal, which appeared to be transforming to a more anhydrous HMC-like structure (Figure 4A, spectrum H). The high Mg content in these precipitates is observable via Raman by how the ν₁ vibrational mode is shifted toward higher frequencies compared to a low-Mg natural calcite sample, but not as high as a natural dolomite sample (Figure 4C).³²

Figure 4.

Raman spectroscopy measurements of Mg-rich amorphous calcium carbonate (Mg ACC) to high-Mg calcite (HMC) transformations in precipitates from the aspartic acid on glass slide experiments, compared to carbonate standards. (A) Full spectra from 20 to 4000 cm^–1. Zoomed-in spectra featuring (B) low-wavenumber internal modes, such as translational and librational modes between 50 and 400 cm^–1, and (C) v₁ symmetric stretching mode between 1050 and 1130 cm^–1. Peak positions for the major carbonate modes in this sample fall between peak positions for calcite (light blue) and dolomite (dark blue) standards and have FWHMs that fall between the range for the low-Mg ACC example and the crystalline carbonate mineral standards. Photomicrographs in reflected light show where spectra were collected from (D) an Mg ACC example grown on a gold-coated glass slide aspartic acid experiment sample, and (E–H) ACC-to-HMC transformations in the glass slide aspartic acid sample.

Discussion

Despite extensive works on the effects of amino acids on the calcium carbonate system,³³⁻³⁶ the effects of amino acids and chitins on Ca–Mg carbonate crystallization are poorly constrained. This work provides an overview of the effects of amino acids and chitins on Ca–Mg carbonate mineralogy, composition, and morphology in both solutions and on solid surfaces via in vitro experiments.

Acidic Amino Acids and Chitin Enable Mg Incorporation into Calcium Carbonate

Previous studies have demonstrated that acidic proteins promote calcium carbonate nucleation when on solid templates and inhibit crystal growth in solution.^26,37−40 In nature, acidic proteins are commonly associated with the biomineralization of carbonates that make up shells, skeletons, and spicules.^5,25,41 Aspartic acids in particular were shown to modify calcite crystal morphology by altering the surface energy at certain crystallographic facies.^36,42−44 As a consequence, aspartic acid exerts a strong influence on crystal morphology.⁴⁵ Chitin also wields a significant influence on ACC and calcium carbonate growth.^46,47 Chitin is known to be widespread in biocalcifiers and likely contributes to the formation of calcium carbonates in coralline algae,^48,49 crustaceans,⁵⁰ ostracods,⁵¹ and mollusks.⁵² Laboratory experiments have also demonstrated that chitin has the ability to promote heterogeneous nucleation of calcite at room temperature without the presence of magnesium.^53,54

Mg incorporation in calcite-structure carbonates is difficult at Earth surface conditions due to the high kinetic barrier from strong Mg–water complexing that prevents Mg from bonding with CO₃^2–.^16,17,55 This constitutes part of the famous “dolomite problem”, where the formation of massive deposits of dolomite in the geologic past cannot be explained by modern surface geological processes. Previous experiments demonstrated that catalysts such as dioxane,⁵⁶ carboxylated polystyrene,⁵⁷ dissolved silica,¹⁸ ethanol,¹⁶ hydrogen sulfide,¹⁹ polysaccharide,²⁰ and exopolymeric substances (EPS; 21) could promote the Mg–water complex to dehydrate at ambient conditions and allow for Mg to incorporate into the calcite structure. This study demonstrates that only acidic, hydrophilic, and negatively charged polar amino acids, such as Asp and Glu, are able to catalyze Mg–water dehydration and promote anhydrous Mg-bearing carbonate precipitation, while glycine, serine, phenylalanine, leucine, and isoleucine solutions only precipitated aragonite (Figure 1 and Table 1). Previous synthesis experiments also established that aspartic acid could destabilize ACC and crystallize to calcite in the presence of magnesium in the solution.⁵⁸

Increasing the aspartic acid concentration further enhances Mg incorporation into carbonate, similar to the behavior of other catalysts (Figure 5A; 16, 18–20). Higher initial Mg concentration in solutions will also result in higher Mg incorporation (Figure 5B). Our findings demonstrate that Asp, Glu, and chitin, in solution, have similar effects as all other catalysts in promoting Mg–water dehydration, increasing Mg incorporation into the calcite-structure carbonate, and inhibiting aragonite precipitation.

Figure 5.

(A) Mole percent of MgCO₃ in carbonates precipitated from the solution increases with increasing aspartic acid concentrations with [Ca²⁺] = 10 mM, [Mg²⁺] = 20 mM, and [CO₃^2–] = 40 mM. (B) Mole percent of MgCO₃ in carbonates precipitated from solutions and on solid surfaces (opal, GCGS, and GS). Results are compared for aspartic acid (14 mM, red symbols), chitin (0.0375 g/L, blue symbols), and control (no added biosubstrates, black symbols) experiments. Experiments with aspartic acid and chitin positively correlated with increasing [Mg²⁺] in the initial solution ([Ca²⁺] = 10 mM, [CO₃^2–] = 40 mM).

The ability to catalyze Ca–Mg carbonate precipitation with these organic molecules might be comparable to absorbed ethanol. It has been shown previously that adsorbed ethanol on carbonate surfaces produces a hydrophobic layer, retarding the calcium carbonate precipitation rate,⁵⁹ disrupting the Mg hydration shell, and promoting Ca–Mg carbonate formation.¹⁶ It has also been postulated that acidic amino acids can regulate calcium ions through carboxylate side chains, which may also serve as a way to feed Mg into the calcite structure.^60,61 Aspartic acid is suggested to have a stronger inhibiting effect than glutamic acid in solution on calcium carbonate crystallization.⁶¹ This difference in capacity in regulating calcium carbonate precipitation stems from the slight variance in geometry and larger space that allows the carboxylate group in aspartic acids to bind with multiple calcium ions, while glutamic acid can only bind to fewer cations.⁶¹ However, we do not observe a clear difference in the Mg contents of Ca–Mg carbonate precipitates between aspartic acid and glutamic acid experiments (Table 1), which is likely a result of magnesium being a much smaller cation compared to calcium and resulting in no preferential binding at the carboxylate group between the two acidic amino acids. This nonpreferential binding at the carboxylate group was also observed in prior work on amorphous mineral transformation kinetics.⁵⁸

Ca–Mg carbonate crystals precipitated with aspartic acid and chitin on solid surfaces have significantly different behaviors in both composition and morphology, suggesting a different crystallization mechanism when organic catalysts are absorbed onto solid surfaces. Previous works in pure calcium carbonate systems propose that acidic amino acids⁶² and chitin⁵⁴ would promote heterogeneous nucleation and inhibit homogeneous nucleation in solution. We observe elevated Mg incorporation in Ca–Mg carbonate precipitates in both solution and solid surface experiments with aspartic acid and chitin, although Mg tends to be more elevated in solid surface experiments (Table 2). The exception is the reduced Mg content in Asp experiments on GS and GCGS surfaces. The strongest effect we observed was chitin on opal surfaces, resulting in disordered dolomite nucleation and growth, suggesting a stronger catalytic effect than chitin in solution or amino acid induced heterogeneous nucleation (Figures 1 and 2). It is possible that the high surface area and regular distribution of the nanospheres on the freshly broken opal-A surface might have contributed to Ca–Mg carbonate nucleation in a manner similar to clay mineral surfaces.^63,64

Being a polysaccharide could also explain chitin’s ability to promote HMC and disordered dolomite precipitation, since some polysaccharides (carboxymethyl cellulose and agar) were shown to catalyze disordered dolomite formation in the laboratory.²⁰

Raman spectroscopy measurements on solid surface precipitates reveal the presence of amorphous calcium carbonate (ACC), known to stay preserved in this form in abiogenic settings with increased Mg content.⁶⁵ This may indicate that ACC precursor phases play a role in the crystallization mechanism of the heterogeneous precipitation experiments. Numerous investigations have found that chitin induces ACC formation;⁶⁶⁻⁶⁸ however, the transformation of Mg ACC to crystalline carbonate was not observed in the chitin solid substrate experiments in this study, at least on longer timescales (months after the experiment). We were still able to measure ACC and Mg ACC in the aspartic acid experiments after six months. This likely suggests either a different crystallization pathway between aspartic acid and chitin biosubstrates, or ACC in the chitin experiments had a substantially shorter residence time. Previous studies have demonstrated that aspartate can also stabilize CaCO₃ prenucleation clusters and thus potentially stabilize Mg ACC.^35,69,70

Acidic Amino Acids and Chitin Controlling Carbonate Crystal Morphology on Solid Substrate Surfaces

It is known that organisms can control the morphology of biominerals through various approaches, such as using ACC precursor phases in echinoderms,⁷¹ or an organic matrix framework in mollusk nacre.⁷² Here, aspartic acid and chitin on solid substrate surfaces show a strong control on crystal morphology through heterogeneous nucleation (Figure 2). In the pure calcium system, it is established that chitin on exoskeleton surfaces could control calcium carbonate formation⁶⁷ and orientation.^67,73 Our study demonstrates that chitin adsorbed onto solid surfaces assumes the same role in controlling crystal orientation for HMC (Figure 3).

Since there is no evidence suggesting the presence of ACC in our chitin in vitro experiments, and oriented growth of Mg carbonate is only known to happen on clay, and other mineral surfaces through classical crystallization pathways,^63,64 it is likely that chitin adsorbed onto solid surfaces provides heterogeneous nucleation sites that lower the crystallization energy and Mg–water dehydration energy, resulting in oriented growth of HMC and disordered dolomite (Figures 2, 3, and 6). Although aspartic acid adsorbed surfaces show a similar trend of Mg calcite crystal-oriented growth, it is less coherent compared to chitin counterparts, indicating a weaker heterogeneous nucleation process (Figures 2 and 3). This may help to explain why some organisms utilize both acidic amino acids and chitins to control the morphology and crystal orientation of precipitated Ca–Mg carbonates during biomineralization, similar to cases for pure calcium bicarbonates.⁷² In this study, we demonstrate that Asp and chitin experiments display different effects on crystal texture during heterogeneous nucleation (Figure 2). The difference in carbonate morphology on different solid substrate surfaces, i.e., opal, GS, and GCGS, could be a result of different degrees of the short-range ordering of silica tetrahedral. While the GS and GCGS are amorphous without short-range ordering, opal-A surface was shown to exhibit local six-member rings of silica tetrahedral and short-range ordering.⁷⁴ These six-member rings of silica on the surface could exert a stronger surface effect and promote heterogeneous nucleation on them.^18,63,75 In Figure 6, we outline a model of how different Ca–Mg carbonate crystal morphologies were achieved by following various homogeneous and heterogeneous nucleation pathways set up by our experiments combining solutions with organic substrates and solid surfaces. Surface-controlled (heterogeneous nucleation) pathways would result in euhedral crystals via relatively slow growth rates with higher Mg incorporation, while solution-controlled pathways (homogeneous nucleation) would result in spherulite formation, indicating rapid growth rates.

Figure 6.

Schematic showing various factors controlling morphology and growth rate of carbonate minerals. For more detail on crystal morphologies, see Figures S2 and S3. Here we connect spherulitic growth associated with faster growth rates and lower Mg concentrations with solution-controlled systems that depend primarily on solution chemistry, whereas the euhedral crystal morphologies (that are also more Mg-rich) align best with surface controlled conditions where organic substrates and solid surfaces play a larger role.

Ca–Mg Carbonate Growth in Biomineral Systems

Despite dolomite precipitation being thermodynamically favorable, in seawater, aragonite is the predominant carbonate mineral to precipitate due to high Mg/Ca ratios and the high kinetic barrier of the Mg–water complex.¹⁶ Thus, aragonite is a common biomineral phase used by many marine organisms, such as corals and mollusks. Still, other organisms continue to form calcite and other carbonate phases, likely with the aid of acidic proteins and chitins that have been shown to promote the precipitation of calcite-structure minerals in coccoliths,⁷⁶ mollusks,⁷⁷ and sea urchins.⁷⁸ In addition to their acidic properties, proteins made up of negatively charged amino acids, such as aspartic acid, also appear to significantly influence calcium carbonate formation.^35,45 The ability to drive calcite-structure mineral precipitation over aragonite via amino acid substrates is likely similar to other catalysts that drive Mg incorporation into calcite-structure carbonates, such as dissolved silica.^18,79 In this study, HMC grown on GCGS surfaces with chitin substrates has similar euhedral morphologies, preferred crystal orientations, and Mg compositions to Ca–Mg carbonates formed on the chitin-rich exoskeletons of leaf-cutter ants⁷ (Figure S2). This suggests that a combination of organic substrates and solid surface properties may control calcite structure formation and Mg incorporation via biomineralization in nature. The appearance of (012) twinning in the HMC in aspartic acid experiments is also commonly observed in modern Ca–Mg carbonates in natural abiotic settings.^30,75

By comparing the mineralogy of biocarbonates and the composition of amino acids reported in the literature, we observe that organisms that utilize both aragonite and calcite in their hard parts (skeletons, shells, etc.) have higher Asp and Glu content in the calcitic regions and lower acidic amino acid contents in the aragonitic regions^{4,26,48,72,73} (Figure 7). This supports our hypothesis that many biomineralizing organisms likely control their mineralogy, crystal chemistry, and crystal morphology by mixing different organic substrate catalysts to precipitate specific calcium carbonate minerals. Understanding these combinations and subsequent precipitates could have implications for forming synthetic biomimetic materials in the future, especially ones with higher Mg contents and improved material properties.

Figure 7.

Compiled data showing aspartic and glutamic acid proportions in regions with calcite and with aragonite in bimineralic organisms. Crassostrea irredescenes,²⁶Mytilus californiaus,⁸¹Atrina rigida,^52,80 and Stylophora pistillate.⁴

Compared to those that produce pure Ca carbonates, biomineralizing organisms that produce Ca–Mg carbonates may hold added benefits, such as lower dissolution rates when the Mg content reaches at least ∼40 mol %, higher Mohs hardness, and a greater potential for carbon sequestration. Leaf-cutter ants, which were recently shown to form Ca–Mg carbonate on their chitin-rich exoskeletons, may take advantage of this by using harder carbonates as armor,⁷ as well as a robust mineral-based carbon sink to reduce the extremely high pCO₂ (up to 3 vol %) in their colonies.⁸² Using a closed box mass balance model, we roughly estimate that a colony with 8 million ants forming a 1 μm thick layer of Ca–Mg carbonates on their exoskeletons could sequester up to 5–10% of the carbon dioxide in the colony in this solid form (Supporting Information). In contrast to a 50:50 Ca/Mg carbonate phase, using the same amount of Ca²⁺ available on the ant exoskeletons, precipitation of a pure calcium end-member would provide half the carbon sequestration potential and face more significant dissolution from the low pH environment in the ant colony. This demonstrates that Ca–Mg carbonate biominerals may serve as a much more practical carbon sequestration agent in the biosphere.

Conclusions

Our results demonstrate that acidic, negatively charged, hydrophilic amino acids and chitin polysaccharides can act as catalysts to dehydrate local mineral surfaces to allow Ca–Mg carbonate precipitation. Different combinations of organic substrates and solid surfaces create various crystal morphologies in experimental biomineral analogues, some of which resemble natural HMC/disordered dolomite crystals on ant exoskeletons. The ability of Mg-bearing calcium carbonates to sequester carbon, resist ocean acidification, and form harder mineral parts in biominerals offers a largely untapped reservoir of information on substrate-controlled biomineralization mechanisms, biomimetic materials synthesis, and carbon dioxide and ocean acidification mitigation responses.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.cgd.3c00102.

• (A) TEM image of (012) twinning of high-magnesium calcites (HMC) close to the [100] zone axis; (B) selected-area electron diffraction (SAED) pattern of (A) showing two sets of diffraction patterns, which are labeled in yellow and white; (C) crystal form models from of (012) twinning of HMC close to the [100] zone axis with same traces labeled as (A); (D) crystal form models from of (012) twinning of HMC along the [100] zone axis (Figure S1); (A) backscattered (BSE) image of synthetic quasi-rhombohedral HMC on a gold-coated glass slide surface in the chitin experiment; (B) BSE image of rhombohedral-shaped HMC on a leaf-cutter ant exoskeleton;⁷ (C) BSE image of a synthetic HMC spherulite from a chitin solution experiment; (D) euhedral shape of HMC with {104} and {100} crystallographic forms; (E) ideal shape of rhombohedral HMC with only {104} faces developed (Figure S2); (A) cross-polarized light image of Mg-calcite spherulites growing from solution in chitin experiments; (B) BSE image of elongated HMC spherulites growing on an opal surface in the aspartic acid experiment; (C) euhedral HMC growing on a GCGS surface in the chitin experiment (Figure S3); box model for estimating CO₂ sequestration via dolomite precipitation (CaMg(CO₃)₂) in an ant colony (Figure S4); and results of in vitro experiments with aspartic acid, chitin, and no organic substrates (Table S1); *denotes a minor amount (PDF)

The authors declare no competing financial interest.