The Use of Microfluidic Platforms with Raman Spectroscopy for Investigating the Co-Precipitation of Metals and Radionuclides in Carbonates

Abstract

In the context of long-term safety assessments of deep geological repositories for radio-active wastes, a rigorous understanding of the retention of radionuclides such as ²²⁶Ra due to co-precipitation with carbonate and sulphate minerals is important for a realistic prediction of radionuclide migration behaviour in the repository near and far field. The co-precipitation of ²²⁶Ra in sulphate minerals, in particular barite, has been studied experimentally and numerically in detail throughout the last decade to establish the thermodynamic properties and mixing behaviour of its solid solutions over a wide range of temperatures. However, so far, few studies have been dedicated to the incorporation of ²²⁶Ra into carbonates, and little is known about the mixing behaviour of ²²⁶Ra and calcium carbonate phases such as calcite, aragonite, or vaterite. The aim of the work presented here was to develop and explore innovative microfluidic experiments in combination with in situ Raman spectroscopy that can be used to investigate co-precipitation processes of radionuclides in carbonate minerals, using stable Ba as a chemical analogue for ²²⁶Ra in the first step, due to their similar ionic radii. Different microfluidic set-ups were developed to address co-precipitation in bulk solution as well as in confinement or under diffusive flow regimes. It could be shown by XRD and SEM-EDX analyses that high amounts of Ba can be incorporated into the lattice of calcite when formed via an amorphous precursor phase, suggesting that the formation of calcium carbonates can contribute to the retention of ²²⁶Ra in the repository environment, which will be verified in future studies using the presented microfluidic platforms.

1. Introduction

In various industrial applications, co-precipitation is considered as the holy grail for the removal of metal contaminants from industrial wastewater and groundwater. For example, the co-precipitation of phosphate is induced for the remediation of soil and groundwater contaminated with lead, cadmium, or zinc, while the co-precipitation of Ra-bearing sulphate minerals has proven to be an effective remediation method for the treatment of water from hydraulic fracturing in gas-bearing shales (Refs. [1,2] and references therein). In the context of radioactive waste disposal, a rigorous understanding of the retention mechanisms of safety relevant radionuclides, in particular for radium, is important for a more realistic prediction of radionuclide mobility in the repository near and far field [3]. Besides sorption processes on clay minerals [4], Ra can be immobilized or retained in the repository environment through co-precipitation processes as (i) (Ba,Ra)SO₄ formed from barium and radium released from the wastes and the intruding sulphate-bearing pore water [3,5] and (ii) as radium-bearing carbonates (in particular CaCO₃) formed at clay-cement interfaces during carbonation of cementitious materials introduced as construction or backfill material in the repository [6] or in carbonate-bearing clay rock [7].

The co-precipitation of radium in sulphates has been studied experimentally and numerically to establish the thermodynamic properties and mixing behaviour of its solid solutions over a wide range of temperatures [8,9,10,11,12,13,14]. To date, there have been limited experimental investigations focused on the incorporation of radium into carbonates, resulting in a lack of understanding regarding the mixing behaviour of ²²⁶Ra and calcium carbonate [15,16]. In most experiments, radium is usually used at tracer concentrations (<10⁻⁸ M), or barium is employed as a surrogate. Indeed, experiments involving radium are difficult because of its radioactivity; ²²⁶Ra (half-life 1600 years) decays to form gaseous ²²²Rn and a series of other short-lived α emitting isotopes. In recent safety assessments for deep geological repositories for radioactive wastes, potential upper limits for the ²²⁶Ra concentration in the repository environment are expected to be in the range of 10⁻⁵ to 10⁻⁴ M, depending on the composition of the pore water and the assumed solubility-limiting solids, with best estimates often distinctly lower (i.e., between 10⁻¹¹ and 10⁻⁹ M; cf. [17,18]). In contrast, in the sulphate-poor pore water in the Boom Clay at Mol, Belgium, the expected Ra concentration is about 7 × 10⁻⁵ M (best estimate) with a proposed upper level of 1.2 × 10⁻⁴ M [19].

With the ongoing technological transfer from biotechnology to geosciences [20], microfluidics is emerging as a versatile tool to investigate mineral crystallization processes [21,22]. Whittaker et al. [23,24,25], as well as Meldrum et al. [26,27], have successfully employed droplet microfluidics to generate hundreds of droplets in an immiscible phase in which each droplet is an independent batch experiment, allowing for the investigation of nucleation of carbonates or iron hydroxides in confinement. Poonoosamy et al. [28] combined microfluidic experiments with fluid dynamics modelling to derive the crystal growth kinetics of sulphates in confinement. Further investigations including in situ Raman spectroscopy were used to investigate whether oscillatory zoning of (Ba,Sr)SO₄ is driven by kinetics or diffusion [29]. The advantage of this technology is that only small volumes of fluid are required to run hundreds of experiments in parallel, providing good statistics. This technology, when harnessed, could result in a breakthrough in the field of radio-geochemistry as it will offer the opportunity to work with high concentrations of radium or other radionuclides in solution while the radionuclide inventory in the experimental setup is low, keeping the radiation doses minimal as in classical batch experiments or below.

The goal of this work is to explore and develop microfluidic experiments that can be used to investigate metal co-precipitation processes in carbonate minerals that can be later transferred for the investigation of radium incorporation into carbonates. We propose two experimental set-ups: one enabling a quick mixing of reactive solutes (droplet microfluidics) and another that enables the slow diffusion of reactive solutes (passive microfluidic setup), analogous to fluid flow in fractured crystalline rocks and clay rocks, respectively. We combine Raman spectroscopy, scanning electron microscopy (SEM) combined with energy-dispersive X-ray spectroscopy (EDX), and powder X-ray diffraction (P-XRD) analysis for the investigation of the polymorphism of carbonates, and the incorporation of Ba into the carbonate lattice, using Ba as chemical analogue for ²²⁶Ra, due to their similar ionic radii (Ra²⁺: 164 pm, Ba²⁺: 160 pm; for coordination number 12 [30,31]).

2. Materials and Methods

2.1. Chemical System and Geochemical Modelling

Calcium carbonate exhibits several anhydrous polymorphs: (i) calcite (hexagonal structure, space group R$\overline{3}$ch), the thermodynamically stable phase at ambient conditions, (ii) aragonite (orthorhombic structure, space group Pmcn), a high-pressure phase unlikely to form at room temperature in abiotic systems or unless other additives, e.g., magnesium, are added [32,33], and (iii) vaterite (hexagonal structure, space group P6₃/mmc) a meta-stable phase readily transforming into calcite or aragonite (depending on temperature) [34]. Moreover, there are three hydrated CaCO₃ phases known to exist, namely (with increasing solubility) monohydrocalcite (CaCO₃ × H₂O, trigonal, space group P3₁12), ikaite (CaCO₃ × 6H₂O, monoclinic, space group C2/c), which is stable only at low temperature (<8 °C [35]) and rapidly decomposes into vaterite and/or calcite and water at higher temperatures, and amorphous (hydrous) calcium carbonate (ACC). The ACC phase is unstable and a precursor for the formation of crystalline phases and usually has the short-range ordering of calcite or aragonite. The stable polymorphs can incorporate divalent metal cations, depending on their size: cations of a smaller ionic radius than calcium are incorporated into calcite, while larger cations, such as strontium or barium, will be incorporated preferably into the aragonite structure (e.g., [36]). However, recent work has shown that even cations larger than calcium (e.g., barium) can also be incorporated in the calcite lattice up to a maximum concentration corresponding to the chemical formula Ba_0.7Ca_0.3CO₃, if a synthesis route through the formation of ACC is involved [37]. Indeed, these results were even more surprising as witherite (BaCO₃) crystallizes in an orthorhombic (aragonite) structure.

The chemical system used in this work is the (Ca_1−xBa_x)CO₃ system. Two reactions were investigated:

$$C{a}_{\left(aq\right)}^{2+}+C{O}_{3\left(aq\right)}^{2-}\leftrightarrow CaC{O}_{3\left(s\right)}$$

(1)

$$\left(1-x\right)C{a}_{\left(aq\right)}^{2+}+xB{a}_{\left(aq\right)}^{2+}+C{O}_{3\left(aq\right)}^{2-}\leftrightarrow C{a}_{1-x}B{a}_{x}C{O}_{3\left(s\right)}$$

(2)

The mixing of calcite with witherite involves a non-isostructural mixing which can be described using Darken’s Quadratic Formulation DQF model as was done for calcite and strontianite [38]. However, the implementation of such a model was beyond the scope of this work which focused on the development of experimental methods. For simplicity, the results will be interpreted with respect to the supersaturation of the end member phases. Gibbs energy minimization (GEM) approaches implemented in the GEMS3K solver [39] were used to calculate the solution chemistry for the evaluation of the saturation indices (SI) of the solution with respect to the various Ca-carbonates and witherite (BaCO₃), using the solubility products given in

Table 1. Solubility products of various carbonate minerals.

Mineral	Solubility Product	Reference
calcite	10−8.48	[36]
aragonite	10−8.336	[36]
vaterite	10−7.913	[36]
monohydrocalcite	10−7.6	[40]
ikaite	10−6.585	[41]
ACC	10−6.393	[42]
witherite	10−8.56	[43]

. The SI is defined as log₁₀ (Ω), where Ω is the ratio of the ion activity product (IAP) divided by the solubility product of the respective solid (K_sp):

$$SI=\log \Omega =\log \frac{IAP}{K_{\mathrm{sp}}}$$

(3)

2.2. Batch Experiments

Micronized “batch” experiments were carried out to establish Raman measurements of CaCO₃ phases in the presence of Ba as a reference for subsequent analysis and for comparison with the results of conventional bulk experiments. The batch experiments comprised the mixing of two reacting solutions and the collection of the precipitates for analysis.

A solution of 10 mM of calcium chloride (CaCl₂) and a solution of 10 mM of sodium carbonate (Na₂CO₃) was prepared and filtered using MF 187MilliporeTM Membrane Filters (Merck KGaA, Darmstadt, Germany) of 0.22 μm pore size. Then, 10 µL of the Na₂CO₃ solution followed by 10 µL of the CaCl₂ solutions was pipetted in a Petri dish with 0.17 µm glass bottom (ref 150,680 from Thermo Scientific, Waltham, MA, USA), and 300 µL of oil was added to cover the aqueous solution and thus prevent evaporation (cf. Figure 1). This experiment was labelled A1. The mixed solution in the Petri dish was monitored using time-lapse microscopy between 5 and 90 min using an inverted Microscope Eclipse Ti2 (NIKON, Tokyo, Japan) with a CFI Plan Fluor DL 10× objective (numerical aperture 0.3; working distance 0.16, Nikon, Tokyo). Raman spectroscopy was performed within the first hour after mixing using a Witec alpha300 Ri inverted confocal microscope with a Nikon CF plan 100× objective of numerical aperture 0.95. The instrument is equipped with a Nd:YAG laser (λ = 532 nm), a thermoelectrically cooled CCD Camera, and a Ultra-High-Throughput Spectrometer UHTS300. Raman measurements were collected using a grating of 1800 grooves per mm at a laser power of 20 mW collecting 100 measurements each lasting 0.5 s for amorphous phases and 50 measurements each lasting 0.5 s for the crystalline phases. A second experiment (A2) was conducted with 10 µL of a mixed solution of 9 mM CalCl₂ and 1 mM barium chloride (BaCl₂) and 10 µL of 10 mM Na₂CO₃ solution. The Petri dishes were closed and sealed with parafilm and kept for 48 h for further analysis.

The crystals recovered were collected using glass fibres and allowed to dry before being measured using P-XRD. For these measurements, considering the small amount of material, a Supernova Single Crystal Diffractometer (Rigaku Oxford Diffraction) equipped with a microfocus Cu Ka (k = 1.54060 Å) radiation source and an Eos™ CCD detector was used to measure samples from experiments A1 and A2. Obtained patterns were analysed using the Rietveld method using the program GSAS-II [44]. The background was estimated using a 6 to 12 term shifted Chebyschev function, and peak shapes were modelled using a pseudo-Voigt function. The scale factor, detector zero-point, and lattice parameters, as well as peak profile parameters, were refined together with phase amounts quantitatively. Additionally, SEM-EDX investigations of the precipitates were performed with a FEI Quanta 200F SEM with a field emission cathode and an Apollo X Silicon Drift Detector (SDD, EDAX).

2.3. Microfluidic Droplet Experiment

Experimental setups consisted of a droplet generator chip made out of TOPAS^® COC (cyclic olefin copolymer) and connected to a storage chip made out of PMMA (polymethyl methacrylate), where the droplets were monitored after fabrication (Figure 2). Two inlets were dedicated to the injection of the oil phase (Novec™ 7500 fluorinated oil, Fluigent Le Kremlin-Bicetre, France) (indicated by yellow circles in Figure 2b), and two for the injection of the dispersed (aqueous) phase (indicated by red and green circles in Figure 2b), which was a mixture of 10 mM of calcium chloride (CaCl₂) and a solution of 10 mM of Na₂CO₃, hereafter referred to as experiment B1. The flow rates used for the experiment were 1 µL min⁻¹ for both CaCl₂ and Na₂CO₃ and 5 µL min⁻¹ for the oil phase. The storage chips had cylindrical pillars (Figure 2c) to trap the droplets. When the storage chip was filled with droplets, it was disconnected from the droplet generator chip, closed at both ends with microfluidic plugs, and the mineralogical transformations were monitored using time-lapse microscopy. Raman spectra were collected to determine the polymorphism of the newly formed precipitates. A second experiment (experiment B2) was conducted, replacing the solution of 10 mM CaCl₂ with a mixed solution of 1 mM BaCl₂ and 9 mM CaCl₂. To determine whether the barium was incorporated into the carbonate lattice, droplets were additionally collected in an Eppendorf tube for further analysis using SEM-EDX.

2.4. Micronized Counter-Diffusion Experiments

The classical U-tube counter diffusion experiment, which has been extensively used to investigate the crystallization of solid solutions [45,46,47,48,49,50], was micronized as shown in Figure 3. This constitutes a passive microfluidic setup (no pumps required) to investigate diffusion-induced crystallization of carbonates. The chip made out of TOPAS^® consisted of a microfluidic channel of 1.8 cm in length and a cross-sectional area of 100 × 100 µm² with Luer locks at each end, to which mini tanks each of 500 µL volume were connected. An agarose solution (1 wt.% agarose in deionized water) was prepared and injected into the channel using a syringe. The solution was allowed to set for 1 h at room temperature to solidify into a gel. The tanks were filled with a 10 mM solution of CaCl₂ and 10 mM Na₂CO₃, respectively, and fixed to the channel. The gel ensures the diffusive flux of reacting solutes through the gel and no advective mixing. The effective diffusivity of the agarose gel is ~1 × 10⁻¹⁰ m²s⁻¹ ([51]). The reactions in the channel were monitored using time-lapse microscopy. A second experiment (experiment C2) was conducted with a mixed solution of 9 mM CaCl₂ and 1 mM BaCl₂ and a solution of 10 mM Na₂CO₃. Both experiments C1 and C2 were triplicated to verify reproducibility.

3. Results

3.1. Solution Chemistry

The speciation of the aqueous solution was calculated using GEM to determine the SI with respect to carbonate minerals given in

Table 2. Saturation indices (SI) of reacting solutions with respect to carbonate minerals in a solution of pH 10.3 in equilibrium with atmospheric CO₂.

Mineral	10 mM CaCl2+ 10 mM Na2CO3	9 mM CaCl2+ 1 mM BaCl2+ 10 mM Na2CO3
calcite	3.0	3.0
aragonite	2.8	2.8
vaterite	2.4	2.4
monohydrocalcite	2.1	2.1
ikaite	1.1	1.1
ACC	0.9	0.9
witherite		2.3

, indicating that the solutions in all experiments were strongly supersaturated with respect to the carbonates, and calcite being the most supersaturated phase. N.B., with microfluidics, it is difficult to sample the aqueous solution for chemical analysis of the reactants. Thus, the data provided on the aqueous composition hereafter are based on theoretical calculations, though it can be assumed that fast mixing of reactants occurred in the small volumes of solutions in the batch experiments and the droplet microfluidics. Moreover, the solute concentrations in the microchannel in the diffusion experiment vary with space and time (before a steady state is reached) and are thus accessible only via transport modelling.

3.2. Batch Experiments

The precipitation processes in experiments A1 and A2 were monitored using optical time-lapse microscopy. In experiment A1 (Figure 4a), a fine precipitate of amorphous calcium carbonate (ACC) was observed, which dissolved in favour of more thermodynamically stable crystalline phases within 20 min. Crystals of two distinct morphologies, i.e., cuboidal-shaped and rounded-shaped crystals, were observed. The polymorphs of calcium carbonate can be distinguished using Raman spectroscopy. According to literature (cf. Table 1 in Wehrmeister et al. [52]), calcite is characterized by a ν4 in-plane bending of CO₃²⁻ at 712 cm⁻¹, a ν1 symmetric stretching of CO₃²⁻ at 1085 cm⁻¹, and the ν3 antisymmetric stretching of CO₃²⁻ at 1436 cm⁻¹. Vaterite can be easily distinguished from calcite by the doubling of the ν1 band, giving bands at 1074 and 1090 cm⁻¹. The distinction between calcite and aragonite is made by the ν4 and ν3 bands, which are positioned at 703 and 1464 cm⁻¹, respectively, for the latter.

The Raman spectrum of aqueous Na₂CO₃ prior to reaction is reported in Figure 4c(i) and the formation of ACC is characterized by the Raman spectrum 4c (ii). The cuboidal-shaped crystals in experiment A1 were calcite, while the rounded-shaped crystals were vaterite, characterized by the doubling of the ν1 band (1075 and 1090 cm⁻¹). After 48 h, only euhedral calcite crystals were observed, as shown by SEM images in Figure 5a, and confirmed by P-XRD analysis (Appendix A; Figure A1), suggesting a complete conversion of vaterite to calcite.

In the case of experiment A2, a fine precipitate of an amorphous calcium barium carbonate (ACBC) was observed, which dissolved completely in favour of a crystalline phase after 90 min. In experiment A2, no cuboidal-shaped crystals were observed, but rather bow tie-shaped crystals (Figure 4d(ii)). In experiment A2, the Raman spectra of the bow tie-shaped crystals were like those of pure calcite, except for the broadening of the ν1 band at 1085 cm⁻¹ (instead of 1091 cm⁻¹), which suggests the incorporation of barium into the calcite lattice. SEM images obtained 48 h after reaction (Figure 5a) also showed the presence of vaterite (spherical-shaped crystals) and two distinguishable types of barium containing calcite with different morphologies (Figure 5b(i),(ii)). SEM-EDX measurements showed average compositions of Ca_0.31Ba_0.69CO₃ and Ca_0.84Ba_0.16CO₃ (10 measurements each, relative standard deviation about 5%) in (i) and (ii), respectively, while barium incorporation into vaterite crystals can be excluded based on the SEM-EDX measurements.

P-XRD analysis via the Rietveld method for crystal samples recovered from experiment A2 indicated the occurrence of 3 separate phases that could be modelled against hexagonal (space group (SG) = R$\overline{3}$ch), cubic (SG = Fm$\overline{3}$m) and hexagonal (SG = P6₃/mmc) structures, respectively, consistent with calcite, halite (NaCl; a residual salt resulting from the drying of the reacting aqueous solutions), and vaterite, which agrees with SEM measurements. Quantitative phase analysis indicated their phase assemblage to be 66.2 (2), 26.3 (1) and 7.45 (3) wt.% for calcite, halite, and vaterite, respectively. Similar P-XRD analysis was performed for crystal samples recovered from experiment A1 which identified calcite and halite with 41.4 (4) and 58.6 (4) wt.% proportions, respectively, again consistent with SEM measurements. Interestingly, it was found that the calcite phase from experiment A2 had a larger lattice volume than that of experiment A1 (372.42 (5) vs. 368.65 (10) ų, respectively), where the latter has a unit cell volume consistent with previously measured calcite [53]. This unit cell expansion is consistent with the observations from SEM measurements, namely that Ba has been incorporated into the calcite lattice, as the larger Ba⁺² cation substituting for the smaller Ca⁺² cation, leading to unit cell expansion [54]. The vaterite identified in experiment A2 was found to have a unit cell volume of 124.99 (22) ų, which is relatively typical for pure vaterite [55] and does not support inclusion of Ba into the crystal lattice, again consistent with SEM measurements.

3.3. Droplet Experiments

In experiment B1 (Figure 6), the reaction pathway was similar to that of experiment A1 with the formation of an ACC phase, which was found to transform into vaterite or calcite within an hour, revealing the same shapes as in experiment A1. In the 200 µm diameter droplets, mainly calcite crystals were formed; in droplets with 70 µm diameter, predominantly vaterite precipitated.

In experiment B2, an amorphous ACBC was formed which was found to transform also into crystalline phases. The stability of the ACBC phase is significantly increased in confinement, with a complete transformation of ACBC to crystalline phases observed after 10 h compared to ninety minutes in the bulk experiment. The crystals in the droplets were observed to be very similar in size (~10 µm) and were either vaterite or calcite after 16 h. Further SEM analyses conducted 30 h after the reaction, as shown in Figure 7c, revealed signs of dissolution and recrystallization processes. SEM-EDX measurements conducted on the crystals indicated the incorporation of Ba in the calcite as (Ca_1−xBa_x)CO₃ with x = 0.055 ± 0.005. Additional sampling with Raman spot measurements after 30 h revealed only Ba-calcite (Figure 7b(iv)), indicating that all the precipitates transformed into a Ba-bearing calcite in the longer term, thus suggesting a dynamic dissolution and precipitation process to achieve equilibrium.

From the Raman analysis, it is difficult to discern whether the formed vaterite had any barium in its lattice but based on the SEM and P-XRD analysis of the bulk experiment, barium incorporation into the vaterite can be excluded. This suggests that calcite formed from vaterite incorporates Ba during the phase transformation/recrystallization to calcite.

3.4. Counter Diffusion Experiments

Figure 8a,b depict the precipitates in the microfluidic channels in experiments C1 and C2, respectively, after one week of reaction. The concentration of solutes in the microfluidic channel, i.e., total aqueous Ca and total aqueous carbonate, with time can be estimated using an analytical solution of Fick’s second law (Equation (A1) in Appendix C), assuming constant concentrations at the reservoirs, similar diffusion coefficients of the solutes in the gel, and diffusion without chemical reactions through the microfluidic channel of 0.018 m length [56]. In experiment C1, the concentration builds up gradually across the channel as indicated by the dashed lines and the solid lines, respectively, in Figure 7c. After 40 h, the theoretical pore solution (cf. Appendix C) in the centre of the microfluidic channel at x = 0.009 m (Figure 7c(i)) is slightly supersaturated w.r.t calcite (SI~1.02) and vaterite (SI~0.45). In the case of experiment C2, the diffusion of aqueous metal and carbonate ions, from their respective inlet reservoirs, is slightly asymmetric. After 40 h, the simulated pore solution was supersaturated with respect to the carbonate phases, i.e., SI~1.45 for witherite at x = 0.006 m and SI~0.98 for calcite at x = 0.009 m.

As expected, the onset of precipitation was significantly delayed in the counter-diffusion experiments. Precipitation of solids was observed after 40 h only. In the case of experiment C1, only single crystals were formed in the middle of the diffusion channels, while in the case of experiment C2, many smaller crystals (<5 µm) were formed. After 7 days, fine precipitates were also observed in experiment C1 (Figure 7a). The crystals in experiment C2 exhibit very fine needle shapes. Thus, the slow diffusive solute transport process seems to significantly influence the morphology of the precipitating carbonates. Raman analysis conducted on the minerals in experiment C1 indicated calcite (see Appendix B), while in experiment C2, the quality of the measurements was hampered by the relatively high fluorescence of the agarose gel, thus not allowing for an unequivocal identification of the solids. Experiments C1 and C2 were repeated three times and leading to consistent results.

4. Discussion

This work aimed to evaluate the feasibility of using microfluidic platforms for the investigation of the coprecipitation of radioactive metals in carbonate minerals. For this purpose, we tested the use of droplet microfluidics combined with in situ Raman spectroscopy and ex situ SEM investigations. In situ Raman spectroscopy can be used to monitor the transformation of the calcium carbonate polymorphs as the system achieves equilibrium, while ex situ analysis, for example, using SEM-EDX or P-XRD analyses, should be used to quantify the incorporation of metal ions in the carbonates.

Despite the solutions being highly supersaturated with respect to calcite (SI = 3.0), the crystalline carbonate minerals were observed to form via the amorphous precursor ACC and ACBC, respectively. The maximum barium incorporation into calcite formed via an amorphous precursor in the bulk experiment A2 is in the same order of magnitude as the one found by Saito et al. [25]. In experiment B1, the formation of both calcite and vaterite was observed, which is consistent with the droplet experiments conducted by Yashina et al. [26]. It was observed that the stability of ACC is strongly influenced by the size of the droplets. Specifically, when the droplet size was decreased from 200 µm to 70 µm in diameter, ACC was found to be present in 70% of the droplets even after 24 h, and a transformation primarily to vaterite was observed. These observations are not discussed in detail here, as they require the development of automated tools for complete quantitative assessment. However, these findings are consistent with previous research by Cavanaugh et al. [25] which suggested an increase in the stability of ACC in confinement.

In the droplet experiments, the formation of (Ca,Ba)CO₃ occurred through the formation of precursors of ACBC, and Ba was incorporated into the calcite lattice. The presence of barium increased the (temporal) stability of the amorphous calcium carbonate phase. Despite the solution being supersaturated with respect to witherite and calcite (see Table 2), no pure end-member phases were formed. The relatively smaller amounts of Ba that incorporated the crystals in the droplet experiments may be attributed to a confinement effect, where slower kinetics of crystallization are expected to occur [57].

The counter diffusion experiments revealed the distinct consequences of the transport regime on the precipitation process. In experiment C1, the crystals are observed where the saturation indices w.r.t calcite is highest, i.e., in the middle of the reactor. For experiment C2, in contrast to the batch and droplet experiments, where fast mixing of reactants occurs, under diffusive conditions, due to the stabilization of the amorphous phase in presence barium, the newly formed aggregates are instantaneously immobilized in the porous medium, leading to a fine heterogeneous spatial distribution of (Ba,Ca)CO₃ precipitates. The formation of the calcium carbonate precipitates does not correlate with the thermodynamics of the pore water chemistry where precipitation would occur at regions with the highest SI. Instead, the precipitation patterns in C2 are influenced by the limited diffusion of solutes and the kinetic crystallization pathway. These findings also suggest that in porous media where solute transport is controlled by diffusion, the presence of metal ions, or other additives that are known to stabilize amorphous calcium carbonate, can lead to an unexpected precipitation pattern in contrast to the absence of additives where precipitation will occur within regions with the highest SI, ultimately clogging the main solute transport pathway.

5. Conclusions

Based on the experimental observations in experiments A2 and B2, it can be inferred that cations larger than calcium, such as barium, can be incorporated into the calcite structure either through the formation of an amorphous precursor phase (ACBC) followed by recrystallization into Ba-calcite, or from the dissolution of vaterite in an aqueous, barium-bearing solution, followed by recrystallization into Ba-calcite. It is thus expected that ²²⁶Ra, which has a similar ionic radius than barium, can also be incorporated into calcite via similar crystallization pathways. Future studies utilizing the presented microfluidic set-ups will be conducted to verify this hypothesis. This could reveal calcium carbonates, present in the multi-barrier system of nuclear waste repositories (e.g., in bentonite buffers and clay host rocks or as newly formed phase during waste canister corrosion or due to cement-clay interactions), as relevant phases for the retention of ²²⁶Ra in the repository environment. The distinctively lower Ra concentration to be expected in the repository environment compared to the Ba concentration used in our experiment does not compromise the findings of this paper, since if it is possible to coprecipitate large amounts of an alkaline earth metal such as Ba within the structurally incompatible calcite structure, then the uptake of much smaller amounts of the similar sized Ra²⁺ ion by calcite should also be achievable via the same pathway (analogous to the well-known coprecipitation of Ra present in trace concentrations within radio-baryte).

The advantage of droplet microfluidics lies in its ability to synthesize crystals of consistent sizes and generate robust statistics from a large number of parallel experiments, via the concomitant use of only small radionuclide inventories in the experiments. Combining droplet microfluidics with atomistic simulation can be used to determine the thermodynamic properties of solid solutions, which are important parameters for geochemical models used to estimate the concentration of mobile radionuclides (e.g., Ra (aq)) under various geological or engineered conditions [9,10]. On the other hand, miniaturized u-tube experiments provide valuable insights into the diffusion-induced precipitation processes in porous media. When combined with reactive transport modelling [58], this approach can be used to develop conceptual frameworks for describing radionuclide immobilization through co-precipitation processes in porous media [21,29].