Discovery of a hyperalkaline liquid condensed phase: significance toward applications in carbon dioxide sequestration

Bicarbonate ion-containing solutions such as seawater, natural brines, bovine serum and other mineralizing fluids have been found to contain hyperalkaline droplets of a separate, liquid condensed phase (LCP), that have higher concentrations of bicarbonate ion (HCO3 −) relative to the bulk solution in which they reside. The existence and unique composition of the LCP droplets have been characterized by nanoparticle tracking analysis, nuclear magnetic resonance spectroscopy, fourier transform infrared spectroscopy, dissolved inorganic carbon analysis and refractive index measurements. Carbon dioxide can be brought into solution through an aqueous reaction to form LCP droplets that can then be separated by established industrial membrane processes as a means of concentrating HCO3 −. Reaction of calcium with the LCP droplets results in calcium carbonate precipitation and mineral formation. The LCP phenomenon may bear on native mineralization reactions and has the potential to change fundamental approaches to carbon capture, sequestration and utilization.

Varian Inova 500 magnet operating at 126 MHz using a 5 mm broadband probe.All experiments were conducted at 298 Kelvin.Deuterium oxide (Aldrich, 151882) was used to obtain a lock at a volume fraction of 2.5% of the total sample.Data was processed using NUTS™ and Microsoft Excel software when deconvolution of overlapping spectral peaks was required.90° pulses were used with acquisition times of 6.35 seconds.The T2 relaxation measurements were conducted using a Carr-Purcell-Meibloom-Gill (CPMG) sequence with increasing tau (τ) times of 0.025, 0.05, 0.1, 0.2, 0.4, 0.8, 1.6, 3.2, 6.4 seconds.

Refractive Index measurement
Species in the size range of the bicarbonate-rich LCP are generally in the Rayleigh scattering regime, where a close approximation of the efficiency of light scattering is given by equation (S1) below: 2 5  3 Where: I is the measured intensity of the scattering event σ s is the scattering cross section d is the species diameter λ is the wavelength of incident light, here λ = 402 nm n is the ratio of the scattering species refractive index to the solvent refractive index Equation (S1) relates the measured intensity of a scattering event (I) to the diameter of the scattering center species (d), the wavelength of incident light (λ) and the refractive indices (RI) of the scattering species and solvent.The NTA technique measures I directly and d by means of Brownian motion.If the RI of the solvent is known, a standard with a known RI can be used to calibrate the technique to account for software and equipment measurements of the relative intensity.Equation (S1) can then be used to calculate the RI for the scattering species (2,3).
To determine the RI of bicarbonate-rich LCP, silica (SiO2) nanoparticles of 50, 80, 100 nm (hydrated diameters of 62, 92, 116 nm, respectively) were used as the standard reference material (RI = 1.51).Equal masses of the different sized SiO2 nanoparticles were placed into a solution of 100 mM Na2CO3 to the concentration of approximately 10 8 particles/ml.The standard solution was then titrated with 1 N HCl, to pH 9.0, and analyzed for light scattering events using NTA. Figure S2 displays the standard curve.This standard curve was applied to the data shown in Figure 2C to demonstrate the reliability of the technique and to verify and validate the technique.

Pitzer Modeling
We carried out thermodynamic calculations of ion activities in carbonate solutions using Geochemist's Workbench software (GWB).The plot in Figure S4-A was obtained by subtracting the ion activity predicted by Pizer's Equations from the ion activity predicted by Debye-Hückel (D-H) theory.D-H theory accurately predicts ion activities up to about 10 mMolal concentration.
Our assumption is that non-ideality at higher concentrations may be in part due to formation of LCP which removes some ions from bulk solution and thus lowers their activities.The sum of the deviations from D-H activities shown in S4-B are simply the sum of the deviations for all ions and approximate the percent of ions present as LCP (assuming it is responsible for the lowered ion activities).We then compare trends in ion activities vs. pH for our calculated percent of ions present in LCP with observations of the volume of LCP based on light scattering measurements of real carbonate solutions (Figure S4-C).S7).20 μl was pipetted onto the ATR crystal and the reaction was recorded in a time resolved fashion using a Macro applied to Omnic 9.2 software.The spectra were recorded at 0, 10, 20, and 1800 seconds.

Dissolved inorganic carbon (DIC) analysis
The dissolved inorganic carbon (DIC) content of solution and solid carbonate samples were determined by acidometric titration and coulometric detection using a CM150 carbon analysis system (UIC, Inc.).The samples were typically titrated with 2N H2PO4 (Sigma Aldrich).To detect CO2 evolved in reactions of CaCl2 (Sigma Aldrich) with NaHCO3 (Aqua Solutions), however, the samples were not titrated with H2PO4, but rather, a solution of CaCl2 was titrated with a solution of NaHCO3 because titration with H2PO4 would result in liberation of CO2 from CaCO3.This allowed CO2 to be quantified by coulometric detection; any solid formed in the reaction was then isolated, dried and analyzed by FTIR to confirm its composition as CaCO3.All analyses using the CM150 system were completed at 40 °C.

Time-resolved pH measurement
The pH was recorded in a time resolved manner using an OrionStar A215 pH meter with an Orion 8157BNUMD Ross Ultra pH/ATC Probe.Data was logged using StarCom 1.0 sampling every 3 seconds while dosing 0.

Synthesis of Calcium Carbonates Materials with Desirable Properties
Carbonate was produced by mixing CaCl2 (aq) and NaHCO3 (aq) at a molar ratio of 1:2, similar to what is described in the Timer-resolved pH measurement section of the Material and Methods.The precipitate was then pressurized using a carver press to 20,000 lbf and allowed it to dwell for 30 minutes.The compact calcium carbonate was then placed in a humidity chamber (Fisher Scientific Isotemp Oven Model 615F, made 100% humid with water) for 7 days at 40 °C.Finally, the sample was cured for 12 days in 1M Na2CO3 solution, which was kept in a water bath to maintain the temperature at 40°C.

Life Cycle analysis
Calculations of lb CO2/yd 3 mortar were based on the assumption that an average of 2,044 lb of CO2 is emitted for every 2,205 lb of Ordinary Portland Cement (OPC) produced in the U.S., depending on fuel type, raw ingredients, and the energy efficiency of the cement plant (9) (essentially 1:1 CO2:OPC produced).Therefore, ASTM C109/C109M mortar cube mix design has 1,057 lb CO2/yd 3 mortar (extrapolated from mix design detailed in Figure 4B); this also assumes no CO2 contributions from the water and aggregate components of the mix design.When a carbonreducing LCP liquid that contains 1 wt% CO2 is used as a complete water replacement in the Ordinary mix design, the lb CO2/yd 3 mortar is reduced by 5 lb CO2.When the carbon-reducing liquid is used in combination with a 15% replacement of OPC by interground (IG) limestone, the lb CO2/yd 3 mortar is reduced to 898 lb CO2/yd 3 (from 1057 lb CO2/yd 3 ).This is due to the 1 wt% CO2 in the liquid and the 15% offset of CO2 that would have otherwise come from the manufacturing of OPC.Still-shots of the scattering projections obtained by nanoparticle tracking analysis (NTA) strongly suggest that the bicarbonate ion participates in a condensation, as reported for bicarbonate-rich LCP.A) A solution containing 100 mM NaHCO3 and 100 mM NaCl contain many scattering events presumably due to the formation of bicarbonate-rich LCP.B) A solution containing 200 mM NaCl does not display scattering events at the same conditions.This is evidence that the bicarbonate ion participates in a condensation to form bicarbonate-rich LCP even in relatively simple, undersaturated solutions.The species seen in A are a distribution of sizes centered around 50-60 nm in diameter as shown in Figure 1C of the main report.

Fig. S3
. The standardization of nanoparticle tracking analysis with silica particles for refractive index measurements.Scattering intensity vs. diameter of silica nanoparticles in water obtained from the NS500 nanoparticle tracking analyzer.To establish a standard curve and calibrate the NS500 NTA for refractive indices measurements, silica (SiO 2 ) nanoparticles (nanoComposix) of 50, and 80 nm (hydrated diameters of 62, and 92 nm, respectively) were used as the standard reference material (RI = 1.51).A curve was estimated using the Raleigh approximation (equation S1) to fit the intensity of the scattering events with the measured size of the particles.The area of the data points (blue circles) represents the relative statistical certainty of the measurement.Fourier Transform Infrared Spectra (FTIR) of a reaction 1 dump reaction at times of 0 seconds (purple), 10 seconds (green), 30 seconds (red), 30 minutes (blue) post mixing.Calcite infrared active bond vibrational modes of, v3 (1400 cm-1), v1 (1087 cm-1), v2 (877 cm-1), and v4 (714 cm-1) are seen.The asymmetrical C-O stretching of the carbonate bond, v3, is seen shifting through a bidentate, resulting in a characteristic calcite peak suggesting that calcium carbonate formation may be forming through a bicarbonate pathway similar to one proposed in nature (10).The symmetric carbonate vibrational mode, v1, relates to free carbonate available in the structure.Out of plane bending, v2, and in plane bending, v4, are identified by (877 cm-1) and (714 cm-1) respectively.(B) A FTIR spectra identifying CaCO According to FTIR spectra in Figure S5A, the structures were initially hydrated and amorphous as reported previously, showing broad peaks in the observed range (11,12).As the reaction progresses, however, gradual appearance of sharp peaks are related to the development of crystalline structure of the carbonate polymorphs as seen with the increase of 1400 cm -1 (ν 3 asymmetrical CO 3 ), 1087 cm -1 (ν 1 symmetrical CO 3 ), 877 cm -1 (ν 2 out-of-plane band of CO 3 ), and 714 cm -1 (ν 4 in-plane-band of CO 3 ) (13), indicating the formation of calcite phase (14).This particular reaction was denoted as Reaction 1 in the main report and was compared to conventional CaCO 3 precipitation pathway, Reaction 2. The products as the result of Reaction 1 and 2 are identical as shown in Figure S5B.The yield of CO 2 and CaCO 3 were 90% and 80%, respectively, confirming the stoichiometry and chemical pathway of Reaction 1. pH was also measured in a time-resolved fashion and suggests that reaction 1 occurs at a lower pH compared to the conventional Reaction 2. This is directly related to LCP-formation mechanism as Ca 2+ has the propensity to interact with HCO 3 -, enabling precipitation reaction to take place at neutral pH.In both cases, pHs in the initial stages decrease slightly due to onset of CaCO 3 precipitation.-and CO 3 2-ions charge balanced with H + and various M n+ cations, e.g., Na + , K + , Ca 2+ , etc. (B) A system containing the same constituents as (A) but now they are arranged in a two-phase system that has bicarbonate-rich LCP, illustrated by the single large droplet.Even though the systems are identical in a global sense (overall DIC and alkalinity), the variables measured only in the bulk of the two-phase system, such as pH, conductivity, and selected ion concentrations, will not reflect the contents of the LCP.This can lead to misinterpretation of the system behavior unless the two-phase system is considered.For this illustration, the pH of the two-phase system would be higher due to the known sequestration of H + in the LCP, leading to the false conclusion that the overall HCO 3 -/CO 3 2-ratio has dropped when, in a global sense, it has remained constant.In a two-phase system, such as one containing bicarbonate-rich LCP, pH and alkalinity are independent due to the presence of an extra degree of freedom.Fig. S6.An illustration of a two-phase system that can alter the interpretation of system measurements such as pH.On the left is a hypothetical one phase system which shows speciation ideally to yield a 1:10 carbonate to bicarbonate ratio.The system on the right contains the same constituents as on the left but it is arranged in a two-phase system containing bicarbonate-rich liquid condensed phase droplets within a mother liquor which is at a carbonate-to-bicarbonate ratio of 1:5.Even though the systems are identical in a global sense, the measured pH value is different between the systems which would lead to incorrect interpretations if the two-phase system is not considered.

(B)
The difference in ion activity between actual ion activities (approximated using Pitzer's equations) and that predicted by Debye-Hückel theory for a 50 mM NaHCO 3 with varying amounts of HCl added to adjust pH.Species with positive deviations (lower activities) are interpreted as strong candidates to be participating in the LCP phase.Negative deviations are interpreted to be species that are relatively enriched in the bulk phase and are excluded from the LCP.The composition of the LCP phase is then the sum of the positive deviations of the curves, which provides the amounts of species missing from bulk solution (C) The comparison of measured LCP droplet concentrations via NTA vs. the predicted percent of ions participating in LCP as predicted from part B. The NTA experimental data was collected in triplicate and is presented as an average with a standard deviation of error in either direction.The amount of LCP droplets detected by NTA decreases in a way qualitatively similar to our prediction.The drop in concentration of LCP droplets detected occurs more rapidly than the drop according to the Pitzer equations.This suggests that the LCP droplets may be smaller as well as fewer as the pH is lowered and therefore are not being detected by the NTA (which has a size detection lower limit of 40 nm).We speculate that there may be a spinodal for the LCP below pH 7.5.This would be consistent with an LCP phase that is bicarbonate-rich and slightly acidic as reported previously as the energy of interfacial formation would be reduced in this environment.

Figure S2 .
Figure S2.Bicarbonate ions form bicarbonate-rich liquid condensed phase.Still-shots of the scattering projections obtained by nanoparticle tracking analysis (NTA) strongly suggest that the bicarbonate ion participates in a condensation, as reported for bicarbonate-rich LCP.A) A solution containing 100 mM NaHCO3 and 100 mM NaCl contain many scattering events presumably due to the formation of bicarbonate-rich LCP.B) A solution containing 200 mM NaCl does not display scattering events at the same conditions.This is evidence that the bicarbonate ion participates in a condensation to form bicarbonate-rich LCP even in relatively simple, undersaturated solutions.The species seen in A are a distribution of sizes centered around 50-60 nm in diameter as shown in Figure1Cof the main report.

Fig. S4 .
Fig. S4.Two pathways to calcium carbonate formation; a high pH pathway and a low pH pathway.0.25 M CaCl 2 was added to equal volumes of either 0.5 M NaHCO 3 or 0.5 M Na 2 CO 3 in a dump reaction manner and were analyzed immediately post mixing.The results suggest that there are two distinct pathways toward calcium carbonate formation; a familiar one designated in the main text as reaction 2 (CaCl 2 (aq) and Na 2 CO 3 (aq) at high pH, carbonate pathway) and another pathway designated in the main text as reaction 1 (CaCl 2 (aq) into NaHCO 3 (aq) at neutral pH, bicarbonate pathway).(A) A Time Resolved 3 (calcite) formed by LCP Reaction 1, and Reaction 2. The end product of both pathways appears to be identical.(C) A nanoparticle tracking analysis (NTA) still-shot image of 0.25M NaHCO 3 .Bicarbonate-rich liquid condensed phase droplets can be seen.(D) A NTA still-shot image of a reaction 1immediately post mixing provides a visualization of what is measured in time-resolve fashion in part A. (E) The chemical pathway of LCPdriven low pH reaction (Reaction 1) vs. conventional high pH reaction (Reaction 2)., (F) The measured yields of reaction 1 vs. reaction 2, with respect to CaCO 3 and CO 2 , as determined by DIC analysis.The results reinforce the difference between reaction 1 and reaction 2 pathways due to differences in evolved CO2 (expected for reaction 1).(G) The time-resolved pH response of reaction 1 dump reaction shows an initial drop in pH, presumably due to removal of bicarbonate.(H) The time-resolved pH response of reaction 2 dump reaction shows little pH drop suggesting that carbonates are being consumed during mineral formation and are buffered by bicarbonates.During the reaction of carbonate formation, liquid condensed phases (LCP) evolve in the presence of calcium ion and nucleating to form CaCO 3 .As CaCO 3 precipitation proceeds, dehydration of the reaction product occurs as seen by the drop of δ O-H vibrational peak.

Fig. S5 .
Fig. S5.Illustration depicting how the two-phase bicarbonate-rich LCP system might alter the interpretation of system measurements such as pH.(A) Hypothetical one-phase system, at relatively neutral pH, that consists of HCO 3-and CO 3 2-ions charge balanced with H + and various M n+ cations, e.g., Na + , K + , Ca 2+ , etc. (B) A system containing the same constituents as (A) but now they are arranged in a two-phase system that has bicarbonate-rich LCP, illustrated by the single large droplet.Even though the systems are identical in a global sense (overall DIC and alkalinity), the variables measured only in the bulk of the two-phase system, such as pH, conductivity, and selected ion concentrations, will not reflect the contents of the LCP.This can lead to misinterpretation of the system behavior unless the two-phase system is considered.For this illustration, the pH of the two-phase system would be higher due to the known sequestration of H + in the LCP, leading to the false conclusion that the overall HCO 3 -/CO 3 2-ratio has dropped when, in a global sense, it has remained constant.In a two-phase system, such as one containing bicarbonate-rich LCP, pH and alkalinity are independent due to the presence of an extra degree of freedom.

Fig. S7 .
Fig. S7.The concentration of bicarbonate-rich liquid LCP droplets (as measured by nanoparticle tracking analysis) qualitatively matches the activity drop of ions predicted from our thermodynamic analysis.(A)Pitzer equations model bicarbonate solutions empirically which considers all of the known contributions to activity loss.Debye-Hückel, models only consider columbic interactions.By subtracting the activities predicted by Pizter Equations from those predicted by D-H theory, we isolate the activity loss of ions due to bicarbonate-rich LCP.Ion pairing and prenucleation clustering is negligible due to the neutral pH, lack of divalent ions, and the weak ion associations known for all three ions (Na + , Cl -, HCO 3 -).(B) The difference in ion activity between actual ion activities (approximated using Pitzer's equations) and that predicted by Debye-Hückel theory for a 50 mM NaHCO 3 with varying amounts of HCl added to adjust pH.Species with positive deviations (lower activities) are interpreted as strong candidates to be participating in the LCP phase.Negative deviations are interpreted to be species that are relatively enriched in the bulk phase and are excluded from the LCP.The composition of the LCP phase is then the sum of the positive deviations of the curves, which provides the amounts of species missing from bulk solution (C) The comparison of measured LCP droplet concentrations via NTA vs. the predicted percent of ions participating in LCP as predicted from part B. The NTA experimental data was collected in triplicate and is presented as an average with a standard deviation of error in either direction.The amount of LCP droplets detected by NTA decreases in a way qualitatively similar to our prediction.The drop in concentration of LCP droplets detected occurs more rapidly than the drop according to the Pitzer equations.This suggests that the LCP droplets may be smaller as well as fewer as the pH is lowered and therefore are not being detected by the NTA (which has a size detection lower limit of 40 nm).We speculate that there may be a spinodal for the LCP below pH 7.5.This would be consistent with an LCP phase that is bicarbonate-rich and slightly acidic as reported previously as the energy of interfacial formation would be reduced in this environment.

Fig. S8 .
Fig. S8.A flow process diagram illustrating the proposed carbon sequestration mechanism allowed by the discovery of bicarbonate-rich LCP.The four stages of the process are: (I) Pre-Capture: the creation of the CO 2 capture solution, (II) Capture: the capture solution is contacted with flue gas, (III) Concentrate: dilute solutions of LCP are concentrated by membrane dewatering, and (IV) Precipitate: a hard water source is combined with the concentrated LCP solution to precipitate synthetic limestone (calcium carbonate, CaCO 3 ).

Fig. S9 .
Fig. S9.Models for the LCP Technology in place at a 500 MW power plant.(A) Coal-fired power plant.(B) Natural gas-fired power plant.Each case assumes the solution developed at the Pre-Capture stage will have 200 mmol alkalinity available to capture CO 2 , and that the CaCO 3 precipitation occurs by reaction of one equivalent of Ca 2+ with two equivalents of HCO 3 -.
Nicolet IS-10 by Thermo-Fisher with a HeNe laser and a fast recovery deuterated triglycerine sulfate (DTGS) detector.Scans were collected on a Germanium ATR crystal at resolution of 16 and at optical velocity of 0.4747.FTIR samples were prepared by adding 0.25 M CaCl2 (Sigma, Lot#BCBL2738 & Deionized Water) to 0.5M NaHCO3 (Aqua Solutions, Lot #319302 & Deionized Water) (Figure (5)ponents: Ottowa silica sand (ELE International), 16-200 mesh glass grade limestone sand (Blue Mountain Mineral, Colombia, CA), Basalite Type II-V cement, 15% interground (IG) cement, water, and liquid condensed phase (LCP) liquid.The 15% interground cement was prepared by blending 325 μm-sized natural limestone (Blue Mountain Minerals, Columbia, CA) with the Basalite cement and was then milled for 16 hours (PTA-02 Ball Mill with 10 L Jar).Mortar components were blended in a Hobart Mixer (Hobart Inc, Troy OH) according to ASTM C305(5).Mortar was transferred to 50 mm × 50 mm cubic brass molds, specified by ASTM C109/C109M.Samples set in 100% relative humidity (RH), 73°F, were demolded at 1 day and further stored at 100% RH, 73°F, until mechanically loaded.Samples were mechanically loaded at 200 lb/s on an ELE load frame (Accu-Tek Touch 250 Series) in accordance with ASTM C109/C109M.FTIR analyses and sample preparation FTIR spectra were recorded using a