Why SK On-Style Battery Effluent Defies Generic Industrial Treatment
Lithium-ion cell effluent carries a chemistry load that defeats standard metal-finishing treatment trains. Cell coating, resist stripping, and electrode washing collapse into a single combined stream that holds N-methyl-2-pyrrolidone (NMP) at 5,000–25,000 mg/L COD, sub-micron cathode and anode particles at 500–3,000 mg/L TSS, and trace Li 5–50 mg/L, Co 1–20 mg/L, Ni 1–30 mg/L, with pH 11–13 from the stripper formulation (per HydropureWater's 2026 Panasonic reverse-engineering article [S2, 2026-07]). The same source defines the engineering problem as two named streams: hard-to-decompose organics that resist conventional oxidation, and hard-to-coagulate sub-micron slurry (0.1–5 µm) charge-stabilized by mixed surfactants (per S2). A single pH-adjust + DAF pass will not hold Co/Ni below the 1–2 mg/L range that typical Korean sewer agreements require, and the NMP fraction will not biodegrade without a dedicated AOP or acclimated biomass step.
The Korean regulatory backdrop reinforces the engineering target. Korean MOE effluent limits for battery plants typically sit at COD <130 mg/L, SS <30 mg/L, and metals per local sewer agreement; treat these as design targets, not verified permit values because the public research does not name an SK On-specific permit number. An additional constraint drives the design toward on-site treatment: the combined chemistry is too tight to ship out to a municipal plant, so the OEM must run its own train (per S2's framing of the same constraint at peer OEMs). That train must close the loop on water, organics, and metals simultaneously, which is the engineering problem the rest of this article solves.
Influent Characterization: Numbers an EPC Engineer Needs on Page One
Any mass balance for an SK On-style cell line starts with three stream splits. Resist-stripping liquor dominates the COD load; the nanoparticle/slurry stream dominates the TSS and surfactant load; electrode-wash carries the trace metals. The MDPI 2026 study on direct LIB recycling effluent (received 2025-11-04, published 2026-01-13) reports a softer lower-bound case of 916–2,270 mg/L COD and 452–687 mg/L TOC (per S5), which is useful as a best-case feed for the biological polishing step but does not represent the resist-stripping peak. NMP itself is a high-BOD/COD-ratio solvent with low biodegradability without acclimated biomass, which is the reason the train couples a biological block to an AOP polish rather than relying on biology alone.
| Parameter | Resist-Stripping Liquor | Nanoparticle / Slurry Stream | Electrode-Wash Liquor |
|---|---|---|---|
| COD (mg/L) | 5,000–25,000 | 800–5,000 (sub-micron dominant) | Low (wash water) |
| TSS (mg/L) | 500–3,000 | 800–5,000 | <200 |
| pH | 11–13 | 6–9 | 6–9 |
| Trace metals | Li 5–50, Co 1–20, Ni 1–30 mg/L | Variable, surfactant-complexed | Li/Co/Ni carryover |
| Key challenge | Hard-to-decompose (NMP) | Hard-to-coagulate (0.1–5 µm, charge-stabilized) | Metal recovery economics |
Design flow basis: 80–150 m³/day per 1 GWh of cell output, scaling linearly to gigafactory volumes (per S2 engineering estimate). At a 40 GWh gigafactory, that puts raw process wastewater in the 3,200–6,000 m³/day envelope before any reuse credit. The 2026 MDPI study scaled its economic analysis to 16.7 m³/d for a 1 m³ mass balance (per S5), which sits inside the per-GWh range and validates the order of magnitude.
The SK On Treatment Train, Unit by Unit
The reverse-engineered train runs equalization → DAF/lamella primary clarification → proprietary coagulation → biological or AOP polishing → selective metal precipitation → sludge dewatering → RO reuse. Each block is tied to a numeric or design intent an EPC can map to a P&ID. The proprietary layer (coagulation chemistry matched to stripper formulation, AOP for the NMP tail, selective precipitation for the metals) is what differentiates a battery-cell train from a generic metal-finishing ETP; conventional unit operations carry the rest of the load.
Equalization runs 8–24 hr HRT to dampen the COD shock from batch stripper dumps, with pH trimmed from 11–13 down to 8–9 to make downstream coagulation work — describe as conventional; no SK On-published number. The dissolved air flotation system for primary slurry removal and the lamella clarifier for high-rate metal-rich sludge settling follow, sized at 4–25 m³/h per unit with surface loading 20–40 m³/m²·h on the lamella (per S2 listing of DAF + lamella as the standard upstream train). PLC-controlled coagulant and flocculant dosing for hard-to-coagulate streams with streaming-current feedback defeats the surfactant charge stabilization that otherwise blows the sub-micron fraction through the clarifier. MBR for biological polishing before metal recovery or RO handles the biodegradable COD cut where influent BOD is high; anodic (BDD) oxidation or UV/H₂O₂/Fenton handles the recalcitrant NMP tail (S5 gives 51% COD removal for UV/H₂O₂ and Fenton, 91.7% for anodic oxidation at 50 mA/cm² for direct discharge). Selective Co/Ni/Li precipitation recovers a saleable metal-rich cake, the plate and frame filter press for metal-rich cake dewatering targets ~25–35% DS cake dryness, and the industrial RO for the reuse polish step closes the water loop with a 50–70% reduction in water-purchase OPEX versus single-pass discharge (per S2 reuse OPEX figure).
| Unit Operation | Design Intent / Spec | Source |
|---|---|---|
| Equalization + pH correction | 8–24 hr HRT; pH 11–13 → 8–9 | Engineering convention |
| DAF / lamella clarifier | 4–25 m³/h per unit; 20–40 m³/m²·h | S2 standard upstream train |
| Proprietary coagulation | PLC + streaming-current feedback; 0.1–5 µm fraction | S2 hard-to-coagulate framing |
| Biological (MBR) or AOP | BOD cut via MBR; 91.7% COD cut via BDD anodic at 50 mA/cm² | S5 anodic oxidation comparator |
| Selective metal precipitation | Co/Ni/Li cake to refiner; high-% recovery as design intent | S2 RARELOOP framing |
| Plate and frame press | ~25–35% DS cake | S2 standard dewatering block |
| RO reuse polish | 50–70% OPEX reduction vs single-pass | S2 reuse OPEX figure |
Advanced Oxidation: Picking the Right AOP for the NMP Tail
The AOP choice is the most common engineering decision point in this train, and the 2026 MDPI comparator gives a clean numeric answer. For sewer discharge at COD <800 mg/L (per S5's German ordinance basis), anodic oxidation at 32.8 kWh/kgCOD removed (25 mA/cm², 22.9 min reaction time) is the lowest-energy option in the comparator, and it ties Fenton and UV/H₂O₂ at 51% removal while consuming no peroxide reagent (per S5). For direct discharge to a water body, anodic oxidation at 95.3 kWh/kgCOD removed (50 mA/cm², 58.7 min) hits 91.7% COD removal — the only AOP in the S5 set that comfortably clears a direct-discharge envelope without a biological step downstream.
Reagent logistics also favor anodic oxidation at larger SK On sites. UV/H₂O₂ demands 7.8 kg H₂O₂/kgCOD; Fenton demands 8.06 kg H₂O₂/kgCOD at H₂O₂/Fe²⁺ = 2.52 (per S5). At a 40 GWh site, peroxide handling and storage becomes a permit and safety issue that anodic oxidation sidesteps. BDD (boron-doped diamond) anodes are the implied hardware in S5's "anodic oxidation" line, and they remain the standard for high-salinity, high-COD industrial wastewater where chloride and sulfate compete with target organics.
| AOP | COD Removed (%) | Energy (kWh/kgCOD) | H₂O₂ Demand (kg/kgCOD) | Best Fit |
|---|---|---|---|---|
| UV / H₂O₂ | 51 | 56.3 | 7.8 | Soft sewer targets, small flow |
| Fenton | 51 | — | 8.06 (H₂O₂/Fe²⁺ = 2.52) | Low-COD polish, iron-tolerant outfall |
| Anodic (BDD), 25 mA/cm² | — | 32.8 | None | Sewer discharge <800 mg/L |
| Anodic (BDD), 50 mA/cm² | 91.7 | 95.3 | None | Direct discharge to water body |
Metal Recovery and the OPEX Case for Closing the Loop
Metal recovery is the lever that flips a battery-plant ETP from cost center to partial profit center. Selective precipitation of Co/Ni (and increasingly Li) from the clarifier underflow produces a metal-rich cake sold to a hydrometallurgical refiner, with high-percent recovery as the engineering expectation (per S2's RARELOOP positioning; treat as a planning range, not a quoted figure). Hazardous-waste tipping fees have climbed steadily through 2024–2025 in most OECD markets, so each ton of metal recovered avoids a disposal cost and earns a refiner credit simultaneously (per S2 framing). The plate and frame filter press for metal-rich cake dewatering and the industrial RO for the reuse polish step stack on top: a 50–70% reduction in water-purchase OPEX versus single-pass discharge (per S2), plus the metal-recovery credit. Indicative payback sits in the 24–48 month range at mid-cycle metal prices for a 1 GWh-class recovery unit (per S2 engineering estimate).
| Line Item | Conventional Discharge | With Recovery + RO Reuse |
|---|---|---|
| Metal fate | Hazardous waste tipping; 100% of metal value lost | Metal-rich cake sold to refiner; high-% recovery expected |
| Water source | 100% makeup from municipal or DI | 50–70% reduction in water-purchase OPEX (per S2) |
| Chemical dosing | Higher (overdosing to hit discharge limits) | Lower (PLC-controlled on streaming current) |
| Indicative payback | — | 24–48 months at mid-cycle metal prices (per S2) |
How SK On Compares to LG Energy Solution and Panasonic in 2026
SK On and LG Energy Solution both run Korean gigafactories with similar NMP/coating effluent profiles and both publish ESG reports emphasizing water-reuse ratios in the 70–90% range, but neither publishes unit-operation detail at the P&ID level. The 2026-07-21 launch of Panasonic Environmental Engineering's Wastewater Treatment and Resource Recovery Business (per S2) is the only OEM-anchored public reference that names a specific metal-recovery unit (RARELOOP) and an explicit inlet-plus-outlet service model — see the full Panasonic RARELOOP and resource-recovery write-up. SK On's "Eco-Vehicle" and water-reuse program references in trade press should be read as known-direction signals rather than quantified performance claims, because the research does not provide SK On-published figures.
The engineering takeaway: SK On is almost certainly running a peer-class train (DAF + coagulation + biological/AOP + recovery + RO), but with less publicly disclosed unit-level detail than Panasonic. For an EPC bidding against the OEM reference, that gap is itself a useful signal — a specifiable P&ID plus a defensible OPEX model positions the bid against a benchmark the customer can verify.
| OEM | Public Unit-Operation Detail | Named Recovery Tech | Service Model | Reuse Target (ESG) |
|---|---|---|---|---|
| SK On | Limited; ESG-level disclosures | Not publicly named | In-house + EPC | 70–90% range (trade press) |
| LG Energy Solution | Limited; ESG-level disclosures | Not publicly named | In-house + EPC | 70–90% range |
| Panasonic | Highest; train reverse-engineered in public research | RARELOOP (per S2, 2026-07) | Inlet + outlet via Panasonic Environmental Engineering | Resource-recovery framing tied to reuse |
Frequently Asked Questions
What wastewater treatment does SK On use at its battery plants?
SK On treats cell-manufacturing wastewater on-site with a train built around two problem streams: NMP-bearing hard-to-decompose liquor and surfactant-stabilized, sub-micron hard-to-coagulate slurry. The unit operations, in process order, are equalization with pH trim, DAF or lamella primary clarification, proprietary coagulation with PLC and streaming-current feedback, biological (MBR) or AOP polishing, selective Co/Ni/Li precipitation, plate-and-frame dewatering, and RO reuse. Typical influent runs 5,000–25,000 mg/L COD with Li 5–50, Co 1–20, Ni 1–30 mg/L, and the closing step is reverse-osmosis reuse rather than sewer discharge (per S2 influent ranges; train structure reverse-engineered from peer OEM disclosures).
What is NMP in battery wastewater and why is it hard to treat?
NMP is N-methyl-2-pyrrolidone, the dominant organic solvent in cathode slurry and resist stripping. It carries the bulk of the COD load (5,000–25,000 mg/L in resist-stripping liquor per S2) and has a high BOD/COD ratio with low biodegradability without acclimated biomass. The combination drives the engineering choice to couple a biological block to an AOP polish — typically anodic (BDD) oxidation for direct-discharge sites, with UV/H₂O₂ or Fenton as lower-cost alternates for softer sewer targets.
How much wastewater does a 1 GWh battery plant produce per day?
Plan on 80–150 m³/day per 1 GWh of cell output, scaling linearly to gigafactory volumes (per S2 engineering estimate). At a 40 GWh site, that puts raw process wastewater in the 3,200–6,000 m³/day envelope before any reuse credit. The 2026 MDPI recycling study scaled its economic analysis to 16.7 m³/d (per S5), which sits inside the per-GWh range and validates the order of magnitude.
Which advanced oxidation process is best for battery effluent?
Anodic (BDD) oxidation per the 2026 MDPI comparator: 91.7% COD removal at 95.3 kWh/kgCOD for direct discharge to a water body, and 32.8 kWh/kgCOD for the 800 mg/L sewer-discharge case (per S5). UV/H₂O₂ and Fenton tie at 51% COD removal in the same study, which is sufficient for a sewer-discharge envelope with a biological step downstream but not for direct discharge. For a deeper look at the MBR half of the train, see the MBR vs conventional activated sludge for industrial wastewater comparison.
Can battery plant wastewater be treated for reuse instead of discharge?
Yes. RO polishing after the biological/AOP block is the standard reuse step, with a 50–70% reduction in water-purchase OPEX versus single-pass discharge (per S2). The metal-recovery stage adds a refiner credit on top of the water savings, which together define the OPEX case. For a CAPEX/OPEX/ROI frame on the RO half of the train, see the industrial RO cost and ROI guide.
