Separation

Rare earth separation is the step that splits a mixed rare earth feed into individual rare earth products, generally oxides. The IEA calls it the technical core of processing. In 2024, China accounted for 60% of global mined production of magnet rare earths but 91% of global refined output, according to the IEA.

Where rare earth separation sits in the supply chain

The U.S. Department of Energy groups separation with metal refining as processed materials production, steps it says are often associated with the midstream of the magnet supply chain (DOE, February 2022). Mining comes before it; metal refining, alloying and magnet manufacturing come after (IEA, 2026).

Why rare earths are hard to separate

Every rare earth deposit holds a mix of rare earths, and because their chemistry is so similar, separating them is particularly challenging, the DOE says. Neodymium and praseodymium are especially hard to split, so they often go into magnets unseparated as didymium (NdPr). Separating neodymium from praseodymium alone can take 30 mixer-settlers.

How solvent extraction works

Solvent extraction is the primary separation process, and each circuit is engineered for its own feed (DOE). Bastnaesite and monazite concentrates first need high-temperature cracking with sulfuric acid or other reagents so they will dissolve, and cerium is generally removed by oxidation before the extraction trains.

A train can use hundreds of mixer-settlers. In each unit, a solvent is mixed with the feed solution in one chamber, and the liquids separate by gravity in a settling chamber. The widely used organic phase is kerosene with di-2-ethylhexyl phosphoric acid or 2-ethylhexyl phosphonic acid mono-ethylhexyl ester (DOE). The process first splits light rare earths such as neodymium and praseodymium from heavy rare earths such as dysprosium and terbium, then makes further separations.

Inputs, outputs and wastes

Lynas cracks and leaches its Mount Weld concentrate at Gebeng, Malaysia, and at Kalgoorlie in Western Australia, then runs solvent extraction and product finishing (drying and calcination) in Malaysia (Lynas 2026 Annual Report). Kalgoorlie ships a mixed rare earths carbonate (MREC) to Malaysia for separation (Lynas, August 2025). At Mountain Pass, MP Materials roasts, leaches and removes impurities before separation and extraction, then product finishing (MP Materials 10-K, February 2026).

The output is separated oxide, either a single element or a combined product such as NdPr oxide. Solvent extraction consumes large amounts of acid, caustic and water, and wastewater treatment and solvent purchases are key cost drivers. Small amounts of uranium and thorium must also be removed from the rare earth streams, which means handling radioactive material (DOE).

Where rare earth separation happens (as of October 2026)

Counting existing and announced projects outside China, the IEA expects refining and separation capacity for magnet rare earths to reach less than 40,000 metric tons by 2035, against more than 50,000 metric tons of mining capacity. Activity is concentrated in Malaysia and the U.S., followed by Australia, Vietnam, Japan, the United Kingdom, France and Estonia (IEA). The DOE estimated that in 2020 China controlled almost 90% of rare earth separation and near 100% of heavy rare earth separation (DOE).

• Lynas, Malaysia. Lynas reported 7,260 metric tons of ready-for-sale NdPr in the year to June 2026. Lynas Malaysia made its first samarium oxide in March 2026, adding to the dysprosium and terbium oxides first produced there in fiscal 2025 (Lynas).
• MP Materials, Mountain Pass, California. MP, which began producing separated rare earths in 2023 (10-K), produced 840 metric tons of NdPr oxide in the second quarter of 2026. In August it said it expects to begin refining heavy rare earths in the second half of 2026, starting with terbium and dysprosium (MP Materials 10-Q).
• Solvay, La Rochelle, France. Solvay opened a rare earths line for permanent magnets at La Rochelle in April 2025 (Solvay) and said in July 2026 that it would begin industrial-scale dysprosium and terbium separation there in fall 2026 (Solvay, July 2026).
• Neo Performance Materials, Sillamäe, Estonia. A small-scale heavy rare earth line commissioned at Neo's Silmet plant in 2026 proved the chemistry and processes needed to separate dysprosium and terbium, Neo says. Neo says its Silmet 2.0 plan, named an EU Strategic Project in October 2026, is expected to triple separation capacity, subject to a final investment decision (Neo, October 13, 2026).
• Energy Fuels, White Mesa Mill, Utah. Energy Fuels says the White Mesa Mill can currently produce up to 1,000 metric tons a year of separated NdPr oxide. In July 2026 it began construction on an expansion and expects to add terbium and dysprosium circuits by the end of 2027, depending on market needs (Energy Fuels, July 29, 2026).

Newer separation methods: columns and membranes

Ucore says its RapidSX system applies solvent extraction chemistry in compact contactor columns and phase separators (Ucore), an approach covered under advanced column separation. REEx cautioned in August 2026 that Ucore's latest milestone was engineering validation, with construction, commissioning and customer qualification still ahead (REEx).

A patent originally assigned to UT-Battelle, which manages Oak Ridge National Laboratory, covers a membrane separation method that holds the organic phase inside the pores of hollow fibers. The acidic feed flows on one side and the strip solution on the other, so extraction and stripping run at the same time (U.S. Patent 9,968,887).

Why rare earth separation is the bottleneck

The IEA lists stainless steel cells for separation among key equipment bottlenecks, with very few suppliers outside China and lead times that can span several years (IEA). REEx argues that the next rare earth shortage is unlikely to begin at the mine and that downstream capabilities such as separation remain the source of China's leverage (REEx, July 2026).

Frequently asked questions