An experimental and modeling study of CaCO3 nucleation and inhibition by PAPEMP under a regime of increasing oversaturation with implications for crystallization and scale formation

Abstract

Rising saturation causes nucleation in natural and engineered environments. As the saturation increases, nucleation begins, forming a scale that is detrimental to energy production, desalination, and other industrial procedures. Inhibitor addition for scale prevention is a common practice with significant economic and environmental costs. Traditional experiments to determine induction times (t_ind) and evaluate inhibitor efficiency are performed under constant oversaturation. Similarly, constant oversaturation is used in both the empirical and the classical nucleation theory modeling schemes used for scale prediction. Subsequently, experiments and models do not address the dynamic nature of oversaturation increase during energy production.

We developed an experimental system for quantitative investigation of nucleation kinetics under a regime of dynamic oversaturation and a simple algorithm for determining t_ind from laser measurements. Using our system, we studied the precipitation kinetics of CaCO₃ minerals at a pH of ∼6.76, ionic strength of I = 1 m, temperature range of 50–90 °C, and varying rates of oversaturation increase (i.e., characteristic times). We quantified the effect of a potent inhibitor (Polyamino Polyether Methylene Phosphonate; PAPEMP) on the t_ind and the forming solid phase. Finally, we show that the characteristic time controls t_ind in systems with rising supersaturation and developed a numerical model that explicitly accounts for this key parameter.

Here, we present our experimental system, results, and modeling scheme. We show that for a given set of conditions, calcite induction occurs at a similar oversaturation, regardless of the rate at which oversaturation increases. Moreover, we show that PAPEMP retards CaCO₃ nucleation at below ppm levels and that it has a temperature-dependent effect on polymorphism. Lastly, we suggest that expanding existing models such that:

$t_{ind}=f\left(\Delta SI\right)\ast f\left(t_{ind,constant}\right)$

where f(ΔSI) is a function of oversaturation with time and f(t_{ind, constant}) are existing modeling schemes, adequately describe the dynamic nature of oversaturation and show a form of f(ΔSI) that provides an excellent fit with measured t_ind.