Experimental constraints on nonskeletal CaCO₃ precipitation from Proterozoic seawater

Before the advent of skeletal biomineralization, CaCO₃ production is thought to have been strongly influenced by inorganic processes. For example, distinctive sedimentary and early diagenetic fabrics documented in Precambrian rocks record secular variations in marine carbonate chemistry (Grotzinger and James, 2000); however, the specific chemical controls on abiotic CaCO₃ production are poorly constrained.

A particularly enigmatic product in this regard is synsedimentary calcite microspar cement, a widespread and abundant primary pore-filling component of Mesoproterozoic to Neoproterozoic (Tonian) carbonates (Fig. 1; James et al., 1998). This cement comprises mosaics of equant 5–15 µm calcite crystals that share identical cathodoluminescence characteristics and preserve evidence for initial spheroidal particle precipitation with polygonal overgrowths (Fairchild and Spiro, 1987; Pollock et al., 2006). Sedimentological evidence indicates that these cements crystallized rapidly relative to surrounding sediment (e.g., James et al., 1998), while geochemical data and mass balance constraints indicate an origin from fluids dominated by contemporaneous seawater (Frank and Lyons, 1998; Bishop and Sumner, 2006). Unlike other late Proterozoic carbonate components that clearly preserve evidence of former aragonite (consistent with relatively high Mg/Ca in Tonian seawater; Spear et al., 2014), the primary mineralogy of microspar is unconstrained.

The formation of calcite microspar cement has recently been suggested to have required CaCO₃ supersaturation (Ω_cal) greatly exceeding that of the modern ocean (Strauss and Tosca, 2020). The maintenance of CaCO₃ supersaturation through much of the Precambrian, by definition, would have required the inhibition and/or modification of CaCO₃ precipitation (i.e., Sumner and Grotzinger, 1996). This, in turn, suggests that before skeletons evolved, the rates and pathways of CaCO₃ production may have had a strong kinetic control.

Although several compounds are known to inhibit or modify the precipitation of CaCO₃ (including Fe²⁺, Mg²⁺, Mn²⁺, SO₄^2–, PO₄^3–, and organic acids; Burton and Walter, 1990; Sumner and Grotzinger, 1996), relatively few studies have elucidated the sedimentologic consequences of kinetic inhibition. For example, PO₄ dramatically influences nucleation and growth kinetics at μmol/L (µM) concentrations (i.e., Burton and Walter, 1990), and although observational data and theoretical models indicate both spatial and temporal variations in Precambrian PO₄ cycling (e.g., Laakso et al., 2020), the consequences for nonskeletal CaCO₃ production are virtually unknown. Here, we examined the influence of common seawater ions on the dynamics of CaCO₃ precipitation from Precambrian seawater, with a specific focus on dissolved PO₄.

We examined CaCO₃ nucleation in the presence of PO₄ with two types of experiments conducted in synthetic Tonian seawater at 22 ± 1.5 °C and stirred at 250 rpm: (1) degassing experiments, and (2) constant composition experiments (Fig. S1 in the Supplemental Material¹). Synthetic Tonian seawater (based on fluid inclusion constraints; Spear et al., 2014) was prepared from reagent-grade salts, stock solutions, and deionized water (see Table S1). Total dissolved PO₄ (hereafter [PO_4tot]) ranged from 0 to 100 µM, the higher values of which exceed most modern carbonate sediment pore waters (which can reach >30 µM; Morse, 1985). In degassing experiments, nucleation thresholds were approached by saturating the solution with 10% CO₂ gas and fixing total alkalinity (TA). CO₂ was then degassed by N₂ purging, which increased Ω_cal until nucleation occurred. Continuous monitoring of pH, with known [Ca²⁺] and TA, constrained the carbonate chemistry at the point of nucleation. In constant composition experiments, the desired Ω_cal, dissolved inorganic carbon (DIC), ionic strength, and TA were kept constant via autotitration of CaCl₂ and Na₂CO₃-NaHCO₃-NaOH titrants until CaCO₃ nucleation occurred (Fig. S1). Continuous monitoring of pH and [Ca²⁺], with ion-selective and AgCl reference electrodes, constrained carbonate chemistry at the point of nucleation. Reactor headspace was effectively eliminated in order to minimize CO₂ exchange between the solutions and atmosphere.

During both types of experiments, solution samples were extracted, syringe-filtered, and acidified to measure Ca, Mg, P, and Si concentrations via a PerkinElmer NexION 350D inductively coupled plasma–mass spectrometer (ICP-MS). [PO_4tot] was determined using the ascorbic acid method (see the Supplemental Material). Immediately upon nucleation, solids were vacuum-filtered, rinsed with isopropanol, and dried in a vacuum-desiccator at 25 °C for 24 h. Solids were analyzed with a PANalytical Empyrean X-ray diffractometer (XRD) with a Co-K∝ source (40 kV and 40 mA). The Mg content of the solids was estimated by Rietveld refinement (Tables S2A and S2B).

Solids were also analyzed using a Perkin-Elmer Frontier Fourier transform infrared (FT-IR) spectrometer. KBr powder was milled with vacuum-dried samples (∼3 mg) and pressed into optically transparent discs. Scanning electron microscope (SEM) and energy dispersive X-ray spectroscopic (EDS) analyses on Au-coated solids were performed using an FEI Quanta 650 field emission gun operated at 3.5–5 kV under high vacuum. In situ Raman spectra were collected using a Tornado HyperFlux PRO Plus Raman spectrometer in constant composition experiments with 12 scans averaged over 120 s intervals.

Degassing and constant composition experiments showed that the Ω_cal threshold at which CaCO₃ nucleation occurs is strongly influenced by dissolved PO₄ (Fig. 2; Tables S1C and S1D). Below 6 µM PO_4tot, spontaneous CaCO₃ nucleation occurred at Ω_cal between 20 and 30 (Fig. 2); under the conditions examined here, this range was largely independent of DIC and TA. In contrast, at [PO_4tot] greater than ∼12 µM, CaCO₃ nucleation was strongly inhibited, and the minimum threshold for CaCO₃ nucleation corresponded to Ω_cal of ∼45 (Figs. 2 and 3). No detectable CaCO₃ nucleation occurred below this threshold, even over extended experiment durations (19 d). For experiments where precipitation was observed, CaCO₃ nucleation induction time (or the time elapsed between the establishment of supersaturation and nucleation) increased strongly with increased [PO_4tot] from 25 to 100 µM (Fig. 3).

The experiments also showed that dissolved PO₄ strongly controlled the resulting CaCO₃ polymorph precipitated from seawater solutions (Fig. 2). Specifically, below 6 µM [PO_4tot], aragonite was consistently produced. Above 6 µM PO_4tot, no aragonite was observed, and precipitates were dominated by monohydrocalcite and/or Mg-calcite. Our data show that in the presence of PO₄, monohydrocalcite is produced across a broad range of Mg/Ca and CO₃/Ca ratios; however, lower Mg/Ca and CO₃/Ca ratios promote Mg-calcite formation over monohydrocalcite (Fig. 2). XRD, in situ Raman spectroscopy, and ex-situ FT-IR analyses showed that in the presence of PO₄, amorphous Ca-Mg-carbonate (ACMC) initially nucleated from solution and recrystallized to either monohydrocalcite and/or Mg-calcite (Figs. S2–S3) in minutes to hours (Fig. 4). Amorphous calcium phosphate (ACP) nucleated in experiments where PO_4tot approached 100 µM (along with ACMC), which recrystallized to octacalcium phosphate over 48 h (Fig. S3).

Our results indicate that [PO_4tot] above ∼12 µM suppresses the spontaneous nucleation of aragonite and/or calcite in synthetic Tonian seawater, most likely through adsorption onto nascent particle nuclei and subsequent inactivation of growth sites (Burton and Walter, 1990). This permits the formation of ACMC under suitably saturated conditions, which explains several experimental observations. First, the minimum Ω_cal nucleation thresholds observed in our experiments correspond closely with ACMC solubility estimates (Fig. 2; Purgstaller et al., 2019). ACMC solubility is a function of its Mg content, which is influenced by solution Mg/Ca, [CO₃^2–], and pH (Blue and Dove, 2015). Vibrational spectroscopy (Raman v₁ values) indicates that the ACMC formed in our experiments varied between 30 and 50 mol% Mg (i.e., Purgstaller et al., 2019; see Fig. 4).

The initial nucleation of ACMC in Tonian seawater (when [PO_4tot] > 12 µM) also reconciles experimental observations of CaCO₃ polymorph selection. ACMC is metastable and rapidly transforms to Mg-bearing calcite or monohydrocalcite (Blue et al., 2017). The resulting crystalline phase is dependent on the Mg content of ACMC, solution Mg/Ca, CO₃/Ca, pH, and stirring rate (Blue et al., 2017; Purgstaller et al., 2019). Consistent with this, our observations indicate that lower Mg/Ca ratios promote calcite (Fig. 2), whereas higher Mg/Ca ratios promote monohydrocalcite. Although monohydrocalcite may also recrystallize to Mg-calcite (Fukushi and Matsumiya, 2018; Purgstaller et al., 2019), this pathway was not observed under the conditions examined here.

ACMC nucleation provides a simple explanation for an apparent CaCO₃ nucleation threshold in seawater when [PO_4tot] > 12 µM. Spontaneous CaCO₃ nucleation induction time has been shown to continuously increase with decreasing Ω_cal in modern seawater, as predicted by classical nucleation theory (CNT; Pokrovsky, 1998). Conversely, our experiments show that in seawater with elevated [PO_4tot], no nucleation occurs below an Ω_cal threshold of 45 (Fig. 3), even over time scales where nucleation should have occurred according to CNT. This result is consistent with molecular dynamic calculations of supersaturated CaCO₃ solutions (Wallace et al., 2013), which show that once a key Ω_cal threshold is crossed, amorphous CaCO₃ can form through the rapid production of dense liquid droplets. This liquid-liquid separation process occurs in a compositional regime that may only be accessed if the precipitation of crystalline polymorphs is suppressed.

ACMC-mediated precipitation pathways may, in turn, strongly influence the loci of CaCO₃ nucleation in marine systems where elevated PO₄ is present. In the modern ocean, inorganic CaCO₃ nucleation is often linked to the presence of microbial substrates (Robbins and Blackwelder, 1992). This is because nucleating calcite on a microbial substrate requires less energy than the direct nucleation of calcite from aqueous solution (e.g., Obst et al., 2009). However, in the presence of PO₄ (or other compounds that inhibit crystalline CaCO₃ nucleation), the energy barrier to ACMC nucleation is lower than the barrier to CaCO₃ nucleation on a foreign substrate (Wallace et al., 2013; De Yoreo et al., 2015). This in turn implies that foreign surfaces are unlikely to influence CaCO₃ nucleation thresholds in supersaturated systems where calcite and aragonite nucleation is inhibited, consistent with experimental and natural systems (Reddy and Hoch, 2012; Fukushi and Matsumiya, 2018); thus, ACMC would be expected to nucleate where variations in temperature, pressure, and/or fluid composition exceed its solubility in natural systems.

Our experimental observations provide new insight into Precambrian nonskeletal carbonate sedimentation. Specifically, these experimental data begin to constrain the origin of fabrics associated with rapid CaCO₃ nucleation, such as synsedimentary calcite microspar cement (James et al., 1998). We hypothesize that the lack of petrographic evidence for former aragonite in late Proterozoic microspar (Fairchild and Spiro, 1987) is most consistent with ACMC nucleation and subsequent transformation to calcite (Bishop and Sumner, 2006). Spheroidal cores and crystal size distributions observed in microspar cement are also consistent with Ostwald ripening of an ACMC precursor formed at high Ω_Cal values (greater than 69–100; Kile and Eberl, 2003), as is enhanced Sr incorporation into the resulting calcite because ACMC favors trace-element uptake (Littlewood et al., 2017). Together, these observations support the hypothesis that microspar cement reflects precipitation pathways enabled only under high Ω_Cal (Strauss and Tosca, 2020).

Crucially, it is the suppression of aragonite and calcite in marine systems that enables ACMC formation, which in turn opens subsequent recrystallization pathways to CaCO₃. Although Fe²⁺, Mg²⁺, Mn²⁺, SO₄^2–, or organic acids may modify CaCO₃ growth rates, they are not known to suppress aragonite nucleation at high Mg/Ca (Bots et al., 2011). Thus, among known inhibitors, PO₄ is likely to have influenced shallow-water CaCO₃ production through much of the Proterozoic. In fact, elevated [PO_4tot] has been shown to maintain high Ω_cal and lead to the production of metastable precursors in several synthetic and natural systems (Gallagher et al., 2013; Lin et al., 2018), including many alkaline lakes (Bischoff et al., 1993; Fukushi and Matsumiya, 2018). These observations, in combination with our experimental data, suggest that the stratigraphic distribution of depositional fabrics reflecting high Ω_cal, such as microspar, and possibly cap carbonates and giant ooids (Grotzinger and James, 2000; Trower, 2020), could together reflect kinetic controls on CaCO₃ precipitation through the effects of elevated pore and/or bottom water [PO₄]. Although shale-hosted geochemical records indicate relatively minor perturbations to net P flux through much of the Precambrian (including intervals of global phosphogenesis; Laakso et al., 2020), pore/bottom water [PO_4tot] may have been most strongly influenced by the degree of internal PO₄ recycling (Ingall et al., 1993) and/or the apatite burial efficiency controlled by seawater [Ca²⁺] (e.g., Zhao et al., 2020). Although the details of PO₄ cycling and burial in Precambrian CaCO₃ depositional environments are poorly constrained, Tonian Ca-phosphate biomineralizing organisms (Cohen et al., 2017) record enhanced P availability in the water column at a time when synsedimentary microspar peaked in its stratigraphic distribution (James et al., 1998).

Our data show that kinetic interactions among common seawater ions may generate nontraditional mineralization pathways that offer new explanations for enigmatic Precambrian carbonate fabrics such as synsedimentary calcite microspar cement; thus, future interpretations of isotopic and trace-elemental variations in Precambrian carbonates should take explicit account of these CaCO₃ mineralization pathways and their associated expressions. More broadly, our data show that because [PO_4tot] strongly increases nucleation thresholds, while effectively arresting the growth rates of preexisting CaCO₃ (Mucci, 1986), periods of enhanced PO₄ recycling, whether driven by ecological factors (Lenton and Daines, 2018) or ocean-atmosphere redox (Laakso et al., 2020), may have fundamentally altered the dynamics of CaCO₃ precipitation throughout much of the Proterozoic Eon.

We thank the University of Oxford, the Natural Environment Research Council (NERC) Centre for Doctoral Training (CDT) in Oil and Gas, and the donors of the American Chemical Society Petroleum Research Fund (grant 58780-DNI8 awarded to Strauss) for support of this research, and Kat Clayton and Phil Holdship for assistance in sample analysis. We thank Peir Pufahl, Patricia Dove, Malcolm Wallace, and editor William Clyde for valuable input and feedback that significantly improved this manuscript.