Kinetics of Magnesium Extraction from Activated Serpentine By Carbonic Acid

Engineered ex situ mineral carbonation depends
on the ability of some metals to form stable carbonates. The abundant serpentine
minerals contain sufficient magnesium to fill this role but the magnesium must
be made more available in order to enable process development. In the Albany
NETL process, serpentine is thermally activated before being exposed to water
and CO₂ in the presence of other salts at 423 K and 100 bar.  Under
these conditions, it is believed that magnesium (as MgO in the activated rock)
is leached by the protons associated with the carbonic acid equilibria

In the NETL process, carbonate is
precipitated in parallel with leaching. While precipitation of  MgCO₃
is governed ultimately by the equilibrium

this is known to be a complex process under
hydrothermal conditions. If such precipitation occurs in the porous substrate, diffusive
transport of protons and Mg²⁺ is likely to be inhibited. In this
work, we have studied the kinetics of magnesium extraction from activated
serpentine by CO₂/H₂O with a view to identifying the
underlying mechanisms.

Separately metered streams of liquid CO₂
and H₂O (flow rates of the order of 1 ml min^-1) were
mixed and preheated in a temperature-controlled oven before being introduced
into the reactor which was also located in the oven. The reactor, designed to
operate as a fluidised bed, consisted of stainless steel 3/8 ″ Swagelok
fitting into which a weighed sample was charged prior to the experiment. In
these experiments, the sample was South West Oregon Lizardite (SWOL) +45-63 µm
which had been calcined in an argon-fluidised bed for 2 hours at 900 K. Mesh (5
µm) at the reactor inlet and outlet mesh prevented loss of particles while
allowing the CO₂/H₂O mixture (3.7 molal) to pass through
with a residence time around 15 s. The reactor conditions were temperature 423
K and pressure 100 bar. The reactor effluent was cooled and discharged through
a back-pressure regulator for sampling of the aqueous phase. Samples were
collected for one minute at intervals chosen to capture the overall kinetics of
the reaction ? the samples were analysed for magnesium by quantitative
inductively-coupled plasma optical emission spectrometry (ICP-OES).

If F is the mass flow rate of the aqueous
effluent which contains a mass fraction y_Mgthen the mass of Mg in the reactor (assumed all to be in the mineral) is
changing as   

From a separate determination of the initial magnesium content of
the sample

we can write the rate of change of the
fractional extent of extraction Χ_Mg as

Numerical integration of the time-series y_Mg
(t)then allows determination of Χ_Mg (t). The magnesium mass balance (amount
extracted + amount remaining in solid vs amount initially in solid) at the end
of the experiment closes within 4%.

Figure 1 shows typical results obtained
under identical flow and reactor conditions for the concentration of Mg in the
aqueous effluent (expressed in molar concentration units in order to facilitate
comparison with the solubility of MgCO₃) and the corresponding
evolution of the degree of conversion, Χ_Mg. Increasing the solids loading in the reactor gives rise to higher
concentrations of Mg in the effluent, as expected, but extraction is clearly
slower with higher loadings, as evidenced by the curves for Χ_Mg. Since the curves for Χ_Mg should be coincident
under identical conditions, it is clear that the loading itself has influenced
the reaction conditions.

The equilibrium pH calculated (OLI Analyser
Studio 9.1, OLI Systems Inc.) for the MgO-H₂O-CO₂ system
(ie for the reaction set A, B and C above and including a large number other Mg
species) under the conditions of the experiment is strongly affected by the
amount of Mg that has been extracted, essentially as a result of reaction B
above. The pH increases sharply from pH = 3.40 in the absence of Mg until the
Mg loading reaches that for MgCO₃(s) precipitation (molality of Mg =
0.02) after which the equilibrium pH remains constant at pH= 5.32. It is
apparent that the carbonic-acid extraction of Mg is therefore self-inhibiting. Comparing
with the data in Figure 1(a), it is apparent that most of the data at high
loading (500 mg) are obtained under saturation conditions, while the opposite
is true for the data at low loading ? this observation may explain the
suppression in rate of extraction with increased particulate loading.

 We report detailed analyses of the
kinetics over a wide range of conditions, including a discussion of the kinetic
rate laws governing the process.