A lab-on-a-chip approach integrating in-situ characterization and reactive transport modelling diagnostics to unravel (Ba,Sr)SO4 oscillatory zoning

Abstract

The co-precipitation of sulphate minerals such as celestine and barite is widely studied because their formation is ubiquitous in natural and anthropogenic systems. Co-precipitation in porous media results in crystallization of solid solutions yielding characteristics such as oscillatory zoning that are rarely observed in bulk solution or in batch experiments. In the past, the precipitation of compositionally-zoned (Ba,Sr)SO₄ crystals was observed post-mortem in macroscopic silica gel counter-diffusion experiments. Their formation was originally explained by the difference in the solubility products of the end-members combined with diffusion-limited transport of solutes to the mineral-fluid interface, while a later study favored the idea of kinetically controlled reactions. With recent advances combining in-operando microfluidic experiments and reactive transport modelling, it is now possible to verify hypotheses on the driving forces of transport-coupled geochemical processes. We developed a "lab on a chip" experiment that enabled the systematic study of the nucleation and growth of oscillatory-zoned (Ba,Sr)SO₄ crystals in a microfluidic reactor. The compositions of the solid solutions were determined by in-situ Raman spectroscopy. Our investigation shows (1) that the composition of the nucleating phases can be approximated using classical nucleation theory, (2) that the oscillatory zoning is not solely controlled by the limited diffusional transport of solutes, and (3) that nucleation kinetics plays a major role in the switch between different stoichiometric compositions. The zoning phenomena is governed by the complex interplay between the diffusion of reactants and the crystallization kinetics as well as other factors, e.g. surface tension and lattice mismatch.

Figure 1

(a) Schematic representation of the microfluidic setup, (b) microfluidic reactor with an array of 50 growth chambers, (c) top view of microfluidic growth chamber with the two adjacent supply channels, (d) an enlargement of the growth chamber with an array of pillars constituting a barrier structure in the middle.

Figure 2

Temporal evolution of crystals in chamber 1 (a) and 4 (b) as revealed by optical microscopy. Image (c) shows hyperspectral Raman images of the ν₁ (SO₄) intensities from crystals formed after 1200 min in chamber 1 and 4 with intensities of 1000 ± 3 and 994 ± 3 cm⁻¹.

Figure 3

(a) Comparison of the average Raman spectra of the barium (f, h) and strontium (g) enriched layers of selected crystals in chamber 1 and those of commercial 99% pure BaSO₄ and SrSO₄, (b) locations where the spectral averaging was done in chamber 1 with an enlargement of the ν₁(SO₄) band, (c) comparison of the average Raman spectra of the barium (j) and strontium (i) enriched layers of selected crystals in chamber 4 and those of commercial 99% pure BaSO₄ and SrSO₄ and PDMS, (d) locations where the spectral averaging was done in chamber 4 along with corresponding Raman spectra of the ν₁(SO₄) frequency region.

Figure 4

(a–e) experimentally derived crystallization rates of crystals 1–5 in reaction chamber 4.

Figure 5

(a) Distribution map of the velocity magnitude at steady state, and (b) profile of the velocity magnitude along line y = 0 with the center of the growth chamber as origin.

Figure 6

Map of (a) BaCl₂, (b) SrCl₂ and (c) Na₂SO₄ concentrations distributions in the growth chamber and associated stoichiometric supersaturation ratio with respect to (d) Ba_0.88Sr_0.12SO₄, (e) Ba_0.5Sr_0.5SO₄ and (f) Ba_0.05Sr_0.95SO₄ with their respective plots of stoichiometric saturation ratio profiles along y = 0 in (g) for a better visualization.

Figure 7

(a) Mineral distribution profiles of the two stoichiometric compositions (Ba_0.05Sr_0.95)SO₄ and (Ba_0.5Sr_0.5)SO₄ across a simulated 1D reactor for case study 1 at 30 and 275 s and case study 2 at 275 s using a discretization of 1 µm. Points 1–5 refer to the sampling points used to calculate nucleation and crystallization kinetics in Fig. 9. (b) Comparison of simulated precipitation rates using OGS-GEM with instantaneous precipitation (case study 1) and with kinetic constraints (case study 2) using mesh discretizations of 5 µm and 1 µm against experimental results (total amount of minerals that precipitated per unit time).

Figure 8

(a) Histograms of the solid solution compositions sampled in the first 5 growth chambers, (b) stoichiometric supersaturations computed for the aqueous solution compositions at the locations where crystals 1–5 in chamber 4 start to nucleate (aSr²⁺/aBa²⁺ ~ 10) and (c) corresponding nucleation rates, (d) stoichiometric supersaturation function and e) associated nucleation rate for a hypothetical aqueous solution composition with aSr²⁺/aBa²⁺ ~ 100.

Figure 9

(a) Nucleation rates, (b) associated crystallization rates, and (c) calculated effective surface tension (primary vertical axis) and lattice mismatch (secondary vertical axis) as function of the solid solution series.