Cédric’s journey into rare earth research started with a professional bachelor’s degree in Chemistry, where he collected plenty of hands-on laboratory experience. He then continued with a Master’s in Chemistry at KU Leuven, completing his master’s thesis in Prof. Koen Binnemans’ SOLVOMET group. He is now about halfway through his PhD on the ELECTRA project (FWO project G017224N).
What is ELECTRA project about?
In short, this project is about creating an efficient design for continuous counter-current solvent extraction for separation of rare earth metals.
What do you mean by efficient design?
Rare earth solvent extraction has been around for decades and is used on a large industrial scale. The interesting thing is that even the biggest industrial plants with up to hundreds of mixer-settler stages still rely mostly on empirical models, which are built entirely from experimental data which is labor-intensive and time consuming. Those models work quite well within the experimental conditions they were developed for, but as soon as you want to change something, for example rare-earth element concentrations, temperature or volume ratios, they have difficulties predicting the result. Then you basically have to start doing additional experiments.
So what is your solution?
We will minimize the required experiments by using predictive thermodynamic modelling, based on the mixed solvent electrolyte (MSE) model. We are developing a semi-empirical thermodynamic model that uses standard and excess thermodynamic data in combination with Gibbs energy minimization (GEM) methods. The result is that our model can predict all sorts of properties and conditions that are essential for chemical process design: compositions, densities and volumes, with a minimum amount of experiments. Empirical models require you to obtain all that information experimentally for every process condition.
We use OLI Systems software, which provides a strong thermodynamic framework. It is also coupled to a flowsheet simulator. So once we have built a model for a particular separation, we will be able to simulate an industrial process and estimate process parameters, for example how many extraction stages would be needed to achieve the required separation.
If a semi-empirical model is better than an empirical one, why not go fully thermodynamic?
Because a fully thermodynamic model would require far too much computational power! Calculating the composition of a solvent extraction system is too complex for the current computational methods to converge in a reasonable amount of time. With a semi-empirical thermodynamic model, the computational time is limited and still reliable to predict liquid-liquid equilibria.
Would you say your work is more fundamental or more applied?
Difficult to say! On the one hand, I do fundamental studies to determine the composition of the chemical system, because their input is required for developing the model. On the other hand, I perform solvent extraction experiments in batch, and I will perform continuous counter-current experiments later on, so it has a very applied side as well.
Do you identify more as an experimental or a computational chemist?
I would say it is 50:50. For a semi-empirical model, you still need experimental data. Luckily, there is already a lot of data available in the literature, but you still have to verify the quality and obtain new data where it is lacking.
What is your favourite rare-earth metal?
Currently my top three are praseodymium (Pr), neodymium (Nd) and samarium (Sm). I chose them because besides studying the separation itself, you also need to analyse their concentration profile throughout the process. These three rare earths have bright colours in solution. Neodymium is pink, praseodymium is green and samarium is yellow, which makes it possible for both XRF and UV-Vis analysis and incidentally makes them a bit more fun to work with.
So how soon could your semi-empirical model be used in industry?
I have made individual models for praseodymium, neodymium and Currently I am working on combined system containing those three elements together. Expanding the model to all 14 lanthanides will still take quite a bit of time😊
The results of my work are looking promising, but we still need to validate the model in a continuous counter-current system. That is planned towards the end of the project, using mixer-settlers that we have 3D printed ourselves.
What is your biggest challenge in the ELECTRA project?
For sure it will be the validation of the model! This will require a continuous counter-current setup of at least 20 to 30 mixer-settlers to have a significant overview of the concentration profiles. To get such a large continuous setup running smoothly in the lab will be challenging.
Cédric Coomans is part of Solvomet KU Leuven. To know more about Solvomet, visit the website.
