Synchrotron XRD study of calcium carbonate precipitation using a microfluidic device


A versatile and re-usable microfluidic platform for synchrotron X-ray studies of dynamic processes has been developed. Its use was demonstrated for the crystallisation of calcium carbonate and quantitative information including induction times and crystallographic parameters were obtained, with a sensitivity to crystalline material at ppm levels.

  • Share

Crystallisation underpins a vast array of processes ranging from the production of functional materials and pharmaceuticals to the formation of biominerals and the prevention of scale. Identification of the mechanisms that govern crystallisation therefore promises the ability to generate crystals with specific structures, morphologies and sizes, to inhibit or promote crystallisation as desired and to determine when and where crystals form. However, this can be extremely challenging as crystallisation is often rapid, and, in bulk solution, is affected by variables including the presence of impurities and the rate of stirring.

Here, a new strategy for studying crystallisation reactions under highly reproducible reaction conditions is reported.  A microfluidic device is used to create large numbers of flowing water-in-oil droplets, where each droplet is its own reaction environment. Crystallisation is then studied in situ within the droplets using synchrotron powder X-ray diffraction (PXRD). The device employed comprises a polymer insert sealed between two X-ray transparent windows. The crystallising solution is mixed at the inlet junction before forming droplets that flow along a serpentine channel (diameter 0.3 mm) in the oil carrier phase (Figure 1). The device can be used with different temperatures and solvents and offers residence times of 5-10 minutes.  

Sketch of the the microfluidic device

Figure 1. Principle of the microfluidic device showing the T-junction and the 36 viewing positions within the X-ray accessible window. The inset is a zoomed-in optical micrograph of the T-junction (black circle) showing the continuous (oil) phase and three reagent inlets.

The assembled device was mounted on beamline ID13 where it could be moved with respect to the X-ray beam.  Different positions along the flow channel correspond to specific reaction times and PXRD patterns are acquired from droplets flowing past each position.  Excellent time resolution is achieved and the short screening time of individual droplets ensures that the effect of the high-energy X-ray beam on the reaction is minimised. Crucially, the measurements are carried out using a detector frame rate greater than the frequency of passing droplets such that frames taken of the oil phase can be discarded. The remaining frames corresponding to the aqueous droplets are background subtracted, and all of the diffraction data recorded at a specific location are combined into a single diffraction pattern.

This approach was used to identify effective nucleating agents (nucleants) that accelerate the formation of calcium carbonate crystals (Figure 2). While the ability to control nucleation is desirable for numerous applications, effective nucleants are as yet only known for a small number of systems, and many questions remain concerning the mechanisms by which they operate. Non-porous and porous bioactive glasses, unfunctionalised and carboxylate-functionalised controlled pore glasses, and the minerals kaolinite, NX illite, amazonite and montmorillonite were investigated as potential nucleants, where these exhibit contrasting surface chemistries and porosities. Droplet microfluidic PXRD was used to determine the induction times (tind) – that is the time at which crystals are first detected – and results were compared with nucleant-free conditions and droplets containing calcite nanoparticles; the latter provide ideal nucleants. Induction times varied from ≤ 4.23 sec for experiments with calcite seeds to over 30 minutes (determined by halting flow and incubating droplets on-chip) for control conditions. Notably, the porous bioactive glass was almost as effective as the calcite seeds (tind ≤ 12.15 sec), while NX illite (tind ≤ 16.00 sec) and the non-porous bioactive glass (tind ≤ 40.77 sec) were also highly active. 

Illustration of the experimental set-up and  PXRD measurements for experiments with porous bioactive glass

Figure 2. a) Illustration of the experimental set-up and (b) PXRD measurements for experiments with porous bioactive glass. Each channel position corresponds to a specific reaction time.

Effective nucleants have been proposed to operate via a range of mechanisms including: (i) offering a low crystal/nucleant interfacial energy; (ii) adsorbing ions/molecules from the solution, thus locally increasing supersaturation; (iii) aligning solute molecules; (iv) promoting the formation of different polymorphs and (v) exhibiting surface defects that promote nucleation. Bioactive glasses are known to promote the formation of hydroxyapatite (calcium phosphate), and mesoporous varieties are effective nucleants for proteins, where the pores appear to promote nucleation. 

These experiments showed that both bioactive glasses are effective nucleants for calcium carbonate while controlled pore glasses with comparable pore sizes are not. Further, after normalising for surface area, the porous and non-porous bioactive glasses exhibit comparable activities. The activity of the bioactive glass can therefore be attributed to its surface chemistry rather than porosity, where local dissolution of the surface leads to the formation of an amorphous, calcium- and carbonate-rich layer, which facilitates calcite nucleation. 


Principal publication and authors
Droplet Microfluidics XRD Identifies Effective Nucleating Agents for Calcium Carbonate, M.A. Levenstein (a,b), C. Anduix-Canto (b), Y.-Y. Kim (b), M. A. Holden (b), C. González-Niño (a), D.C. Green (b), S.E. Foster (b), A. Kulak (b), L. Govada (c), N.E. Chayen (c), S.J. Day (d), C.T. Tang (d), B. Weinhausen (e), M. Burghammer (e), N. Kapur (a) and F.C. Meldrum (b), Adv. Funct. Mater. 29, 1808172 (2019); doi: 10.1002/adfm.201808172.
(a) School of Mechanical Engineering, University of Leeds (UK)
(b) School of Chemistry, University of Leeds (UK)
(c) Computational and Systems Medicine, Imperial College London (UK)
(d) Diamond Light Source (UK)
(e) ESRF