Short-lived calcium carbonate precursors observed in situ via Bullet-dynamic nuclear polarization

The study addresses how short-lived calcium carbonate prenucleation species (PNS), existing under high oversaturation and lifetimes well below 5 seconds, can be detected and characterized. Classical nucleation theory is challenged by observations of metastable prenucleation clusters (PNC) that precede solid formation in many systems, including calcium phosphates and carbonates. Prior experimental work has largely focused on mild oversaturation where PNC/PNS are longer-lived and experimentally accessible. High-oversaturation regimes, where early precursors are short-lived and dynamic, remain underexplored due to methodological limitations. Dissolution dynamic nuclear polarization (DDNP) can provide >10,000-fold NMR signal enhancements and, combined with rapid mixing, access sub-second time windows; it has been demonstrated for calcium phosphate. However, for calcium carbonate near neutral pH, the pneumatic transfer in DDNP purges carbonate as CO2, rendering conventional DDNP unsuitable. The purpose of this work is to overcome these limitations using Bullet-DNP, which transfers the frozen hyperpolarized pellet and dissolves it directly in the NMR tube, preserving carbonate. The study aims to reveal and characterize the earliest CaCO3 precursors forming immediately after Ca2+ and carbonate encounter at high oversaturation, test whether non-classical pathways operate under these conditions, and quantify their kinetics and spectral signatures.

Metastable prenucleation clusters have been reported across diverse materials, including calcium phosphates and carbonates, prompting reevaluation of classical nucleation-and-growth paradigms. Reports indicate coalescence and dehydration of PNCs leading to solids, and debates persist on reconciling such observations with classical nucleation theory. Most prior studies probed mild oversaturation where PNC/PNS are more stable and long-lived. Non-classical pathways for CaCO3 include formation of ion pairs and subsequent assembly into larger aggregates or condensed phases. DDNP-enhanced NMR has enabled detection of short-lived precursors in calcium phosphate systems at lower oversaturation. For calcium carbonate, DDNP observations are feasible only at elevated pH ≥10 due to CO2 outgassing during pressurized transfer, as shown by Balodis et al.; near-neutral pH conditions purge carbonate under traditional DDNP. Bullet-DNP has been developed to transfer frozen samples, preventing degassing and enabling the study of pressure/temperature-sensitive systems. The present work leverages these advances to access the high-oversaturation, short-lifetime regime of CaCO3 precursors.

Bullet-DNP setup and transfer: A 50 µL aliquot of a 500 mM 13C-enriched sodium carbonate solution (in glycerol-d3/D2O/HEPES) was hyperpolarized at TDNP = 1.5 K using OX063 radical. The frozen sample was transferred as a pellet (“bullet”) to a 9.4 T NMR spectrometer at 298 K using a pneumatic transfer. A custom injector caught the bullet in a brass stopper; the frozen pellet was ejected into an NMR tube and dissolved upon impact into 500 µL of a 20 mM CaCl2 solution in 50 mM HEPES at pH 6, corresponding to >200-fold oversaturation. NMR acquisition was triggered immediately after dissolution/mixing. The system minimized thermal and pressure perturbations during transfer, preserving carbonate content. Experiments were also performed at pH 7 and 8. Sample preparation: 0.1 M HEPES was prepared in 40 mL water; pH adjusted to 6, 7, or 8. Solutions were sonicated under vacuum for 15 min to remove dissolved oxygen. CaCl2 solution was prepared by dissolving 4.6 mg CaCl2 in 1 mL buffer (0.041 M). For Bullet-DNP measurements, 600 µL of the buffer solution plus 50 µL D2O were placed in an NMR tube to yield 0.038 M CaCl2. For DNP, 0.5 M 13C-labeled Na2CO3 was dissolved in glycerol-d3/D2O/HEPES buffer at a 50:40:10 volumetric ratio; OX063 radical was included at 0.015 M. 50 µL aliquots were loaded into bullets, frozen in liquid nitrogen, and inserted into the DNP system (6.7 T, 1.5 K) with continuous microwave irradiation at 187.650 GHz until polarization build-up saturation. During transfer, pulsed solenoids provided a 70 mT guiding field. Dissolution in the NMR tube produced a 700 µL final volume with approximately 35.7 mM carbonate and 35.1 mM CaCl2. Detection used no decoupling; probe temperature 298 K. Data acquisition and processing: Hyperpolarized 13C NMR spectra were recorded time-resolved immediately after mixing. Solvent (glycerol) 13C signals were used to correct for convection effects that initially reduce apparent signal intensity due to sample movement. All data were zero-filled to twice the FID length and apodized with an exponential window adding 5 Hz line broadening before Fourier transformation, followed by baseline correction. Peaks were fitted with MATLAB 2022 function 'fitnlorentzian.m'; amplitudes were corrected for convection using solvent signals. Apparent signal enhancement factors were estimated by comparing hyperpolarized spectra (e.g., 1–2 s after mixing) to thermal equilibrium references acquired after completion of precipitation when feasible. Powder X-ray diffraction (post-precipitation) assessed the crystalline phases formed.

• Immediately after mixing, three main 13C resonances were observed: free carbonate at δ = 160.4 ppm; a second peak at δ ≈ 161.4–162.3 ppm; and a broad peak centered at δ ≈ 168.9 ppm. A weak CO2 resonance at 125.9 ppm was also detected.
• Linewidths: free carbonate 9 ± 1 Hz; the ~161.4 ppm species 18 ± 1 Hz; the broad ~168.9 ppm species 318 ± 21 Hz. The broad line indicates slow tumbling consistent with larger ionic assemblies (PNS). The two additional peaks suggest two distinct CaCO3 precursor species with different sizes and environments.
• Kinetics: Effective decay rates (Reff) derived from time-dependent intensities at pH 6 were 0.08 ± 0.01 s⁻1 for free carbonate; 0.37 ± 0.03 s⁻1 and 0.50 ± 0.03 s⁻1 for the two Ca2+-bound PNS, respectively.
• pH dependence: Ratios of broad-to-narrow PNS signal integrals 2 s after acquisition start were 0.62 ± 0.2 (pH 6), 0.87 ± 0.3 (pH 7), and 0.96 ± 0.4 (pH 8), indicating faster conversion toward larger aggregates/condensed phases at higher pH. The broad species largely disappeared by 5 s after mixing in all cases.
• Apparent hyperpolarized signal enhancement reached ε ≈ 52,700 ± 500 (comparison between hyperpolarized spectrum and thermal equilibrium for free carbonate and narrow PNS at pH 6).
• Post-precipitation powder XRD showed pH-dependent mixtures of vaterite and calcite, indicating that observed PNS did not lead uniquely to one crystalline polymorph.
• The chemical shift separation (>6 ppm) between the two PNS indicates distinct local structures and environments (e.g., coordination, hydration, pH microenvironments, exchange kinetics).

The findings demonstrate that Bullet-DNP enables direct observation of very short-lived CaCO3 prenucleation species under high oversaturation, addressing a previously inaccessible regime due to carbonate loss in traditional DDNP. The simultaneous detection of free carbonate and two Ca2+-bound species with markedly different linewidths and decay rates supports a non-classical pathway involving multiple precursors, consistent with models from mild oversaturation conditions. The narrower PNS near 161–162 ppm is consistent with smaller ionic clusters (e.g., ion pairs), while the broad resonance near 169 ppm indicates larger assemblies or condensed phases with slower tumbling. pH-dependent evolution of the broad-to-narrow PNS ratios suggests accelerated conversion into larger aggregates at higher pH, aligning with known carbonate chemistry and non-classical nucleation behavior. While alternative pathways cannot be excluded (e.g., direct transformation of free carbonate to one PNS with the other being off-pathway), the observations are compatible with multiple precursor-driven routes to solid formation. The rapid disappearance of the broad species within seconds underscores fast kinetics at high oversaturation; some steps may occur outside the accessible time window. Nevertheless, the data reveal early-stage kinetics and structural differences sufficient to infer precursor roles in initiating precipitation and guiding phase outcomes (vaterite/calcite mixtures), reinforcing the significance of PNS in CaCO3 formation across oversaturation regimes.

Two major conclusions are drawn. (1) Bullet-DNP enables liquid-state NMR spectra with enhancements over four orders of magnitude for pressure- or temperature-sensitive samples, overcoming traditional DDNP limitations (e.g., carbonate purging) and expanding the scope of ex-situ hyperpolarization. For hyperpolarized carbonate, this grants access to the first seconds of Ca2+-bound precipitation under high oversaturation. (2) Calcium carbonate prenucleation species can be identified and characterized immediately after mixing, revealing coexisting precursors consistent with non-classical pathways operative even at high oversaturation. Combined with traditional DDNP for other ions, these approaches broaden the toolkit for studying prenucleation phenomena in important oxyanion systems. The methodology paves the way for structural and kinetic studies (e.g., ultrafast exchange and diffusion spectroscopy) and integration with molecular dynamics simulations at relevant concentrations and timescales. It opens new regions of precursor space for designing CaCO3-based functional materials, potentially enabling control over solid morphologies and properties via targeted PNS selection along solidification pathways in high-oversaturation regimes.

• Traditional DDNP with pressurized dissolution and transfer purges carbonate at near-neutral pH, precluding observations; Bullet-DNP circumvents this but highlights method-specific constraints.
• Rapid precipitation complicated determination of solution-state enhancement factors using thermal equilibrium references; only partial referencing was feasible (free carbonate and narrow PNS at pH 6).
• Strong convection immediately after mixing necessitated correction using solvent signals; as a result, polarization could not be reliably inferred directly from SNR, and correction increases noise.
• The solution likely remains inhomogeneous for several seconds after dissolution, introducing spatial concentration gradients that may bias kinetic parameters and affect apparent reaction rates and oversaturation.
• Gas bubbles can broaden lines in D-DNP; mitigated by degassing and adequate fill volume, but residual effects cannot be fully excluded.
• Presence of glycerol (from hyperpolarization matrix) may influence CaCO3 formation kinetics and microenvironments.
• Extremely fast steps at high oversaturation may occur outside the Bullet-DNP detection window, limiting pathway resolution and mechanistic discrimination (e.g., whether the ~161 ppm species is on-pathway).