Artificial neural network modeling of rare earth element solvent extraction based on pH and extractant concentration
Dilan S. Udawattha , Shafiq Alam
Green and Smart Mining Engineering ›› 2026, Vol. 3 ›› Issue (2) : 145 -155.
Rare earth elements (REEs), which comprise 15 lanthanides together with scandium and yttrium, are commonly separated by solvent extraction, in which the equilibrium distribution coefficients (lg D) depend on the extractant type, solution pH, and extractant concentration. Owing to the nonlinear interactions among these variables, the quantitative interpretation of REE extraction data remains challenging. A curated dataset of experimentally reported REE solvent-extraction equilibrium data encompassing diverse extractants and operating conditions was compiled from the literature. Two artificial neural network (ANN) models were employed as data-driven tools to simulate and analyze the behavior of lg D based on the solution pH and log-scaled extractant concentration. The extractant identity was represented using one-hot encoding, and the lanthanide atomic number was included as an auxiliary descriptor to represent systematic trends across the rare-earth series. The ANN results showed strong agreement with experimental data across multiple systems, thus demonstrating the internal consistency and analytical value of the compiled dataset for data-centric studies regarding REE solvent-extraction equilibria.
Solvent extraction / Rare earth elements / Artificial neural network / Distribution coefficient / PH / Extractant concentration
| [1] |
|
| [2] |
|
| [3] |
|
| [4] |
|
| [5] |
|
| [6] |
|
| [7] |
|
| [8] |
|
| [9] |
|
| [10] |
|
| [11] |
|
| [12] |
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
|
| [21] |
|
| [22] |
|
| [23] |
|
| [24] |
|
| [25] |
|
| [26] |
|
| [27] |
|
| [28] |
|
| [29] |
|
| [30] |
|
| [31] |
|
| [32] |
|
| [33] |
|
| [34] |
|
| [35] |
|
| [36] |
|
| [37] |
|
| [38] |
|
| [39] |
|
| [40] |
|
/
| 〈 |
|
〉 |