Cementing CO2 into C-S-H: A step toward concrete carbon neutrality

Abstract Addressing the existing gap between currently available mitigation strategies for greenhouse gas emissions associated with ordinary Portland cement production and the 2050 carbon neutrality goal represents a significant challenge. In order to bridge this gap, one potential option is the direct gaseous sequestration and storage of anthropogenic CO2 in concrete through forced carbonate mineralization in both the cementing minerals and their aggregates. To better clarify the potential strategic benefits of these processes, here, we apply an integrated correlative time- and space-resolved Raman microscopy and indentation approach to investigate the underlying mechanisms and chemomechanics of cement carbonation over time scales ranging from the first few hours to several days using bicarbonate-substituted alite as a model system. In these reactions, the carbonation of transient disordered calcium hydroxide particles at the hydration site leads to the formation of a series of calcium carbonate polymorphs including disordered calcium carbonate, ikaite, vaterite, and calcite, which serve as nucleation sites for the formation of a calcium carbonate/calcium-silicate-hydrate (C-S-H) composite, and the subsequent acceleration of the curing process. The results from these studies reveal that in contrast to late-stage cement carbonation processes, these early stage (precure) out-of-equilibrium carbonation reactions do not compromise the material's structural integrity, while allowing significant quantities of CO2 (up to 15 w%) to be incorporated into the cementing matrix. The out-of-equilibrium carbonation of hydrating clinker thus provides an avenue for reducing the environmental footprint of cementitious materials via the uptake and long-term storage of anthropogenic CO2.


Extended methods
Materials. For all samples, pure alite (C 3 S) (chosen for its much simpler chemistry compared to the more heterogeneous OPC) was obtained from Mineral Research Processing (Meyzieu, France), and combined with five different concentrations of sodium bicarbonate (NaHCO 3 ): 0% (control sample), 5%, 10%, 15%, and 20%. A water:solid (w/s) ratio of 0.42 was used for the 0%, 5%, and 10% NaHCO 3 samples, and 0.43 and 0.44 w/s for the 15% and 20% samples, respectively, to maintain comparable sample workability.

Micro-indentation.
Before testing, all samples were first cut to size with a water-lubricated slow-speed diamond saw (which washed away any residual Na), and wet surface-polished using sequential silicon carbide grinding papers with P2500 and P4000 grits, and dry aluminum oxide abrasive discs with sequential particle sizes of 3 μm and 1 μm. For mechanical characterization of the different samples, instrumented micro-indentation (Anton Paar Instruments) was employed. All measurements were carried out with a Berkovich tip, and anomalous force-displacement curves resulting from indentations at locations containing surface cracks were removed from the analysis. For all data sets, indentation hardness H and indentation modulus M were calculated using the Oliver-Pharr method (1). All specimens were loaded in force-control up to the limiting maximum indentation depth of 24 μm at a loading rate of 3 N/min. At peak load, the force was held constant for 10 s before the load was removed at a rate of 3 N/min, which was fast enough to avoid creep effects that can affect the elastic unloading. 15 to 30 microindentation measurements were performed at each curing age (10 h, 1 day, 2 days, and 7 days) to analyze the change in mechanical properties due to an addition of NaHCO 3 . The microindentation results are presented in terms of specific indentation modulus (M*=M/c) and specific indentation hardness (H*=H/c), where c is the mass of alite divided by the total mass of solids (e.g., c=1 for the reference samples, and c=0.8 for the samples with 20% replacement of NaHCO 3 ).

Raman spectroscopy.
A Confocal Raman Microscopy (CRM) system (Alpha 300RA; WITec, Germany) was used to obtain Raman spectra at different curing ages (1 h, 5 h, 10 h, 1 day, 2 days, and 7 days) using a Nd:YAG laser (λ = 532 nm) and a 63× Zeiss water immersion objective for underwater measurements. The excitation wavelength was calibrated using a silicon wafer standard. Raman maps of 100×100 μm scan areas (80×80 points) were acquired with a continuous laser beam and an accumulation time of 0.2-0.3 s per point. At least three maps at each hydration step were acquired in order to ensure the representativeness of the results.
For the Raman spectroscopy measurements acquired at time points up to 24 h, no sample preparation was required. After mixing, each sample (while still in a slurry-like state) was placed on a glass microscope slide and the top surface was covered with a quartz coverslip to prevent the sample from drying and to eliminate its contact with the surrounding environment. After 24 h (after which the samples had solidified), sample preparation followed the same procedure used for micro-indentation measurements, as described above.
The chemical maps obtained by CRM were analyzed using correlation functions (two-point auto-correlation (S 2 (r)) and cross-correlation (X 2 (r)) functions) to quantitatively represent the phase distributions and their spatial correlation at progressing stages of hydration. WITec Project and Matlab software were used for data analyses and visualization. More details regarding the applied methodology can be found in (2). Thermogravimetric analysis. To assess the quantity of precipitated carbonates, thermogravimetric analysis (TGA) was employed. After 3 months of curing, the reference sample and those with 10% and 20% of bicarbonate substitution were ground to a fine powder, and then heated in air at a uniform rate of 5℃/min from 25 to 1000℃. For each sample, two replicates were performed using 25-30 mg of material. Three modes of CO 2 dissociation were distinguished following the work of Thiery et al. (3). Mode I, characterized by a peak in the range of 700-750℃, corresponds to well-crystallized carbonate, i.e., calcite. Modes II (600-700℃) and III (450-600℃) are associated with amorphous and less crystalline carbonates. The quantity of CO 2 (by mass) for each mode was calculated using deconvolution (based on Gaussian distributions) of the derivative of mass change.
Theoretical CO 2 capturing capacity. The theoretical capacity of the forced carbonation process (to offset CO 2 production during the preparation of OPC) was determined based on a stoichiometric analysis of the hydration reaction in the presence of abundant carbon dioxide to mimic the effects of forced out-of-equilibrium carbonation. Traditionally, OPC is a mixture of limestone and clay that is calcined at high temperatures (1450°C) and ground to produce a multiphase clinker (primarily alite and belite) (Eq. 1): Because of the intrinsic alkaline nature of the products formed from clinker hydration (Eq. 2 for Ca/Si ratio of C-S-H equal to 1.7), the hardened cement serves as a significant CO 2 absorbing agent through natural carbonation of calcium hydroxide (CH): The hydration of alite and belite in presence of carbonate ions leads to the formation of both calcium carbonate (CaCO 3 ) and a C-S-H gel. E.g., for alite we have: Furthermore, any portlandite formed will convert over time into CaCO 3 if carbon dioxide is dissolved in excess: Dataset S1 (separate file). A Matlab script (Correlation_functions.m) was created to evaluate the correlation functions (two-point auto-correlation (S 2 (r)) and cross-correlation (X 2 (r)) functions) to quantitatively represent the phase distributions and their spatial correlation at progressing stages of hydration.