# DFRN

> DFRN separates neodymium (Nd) and dysprosium (Dy) using AI-designed rare-earth
> binding proteins (LanM variants) and supplies them as oxide materials.
> The process is fully aqueous at ambient temperature and pressure, replacing
> organic-solvent extraction (SX). Feedstock is end-of-life NdFeB permanent
> magnets and process scrap. DFRN is the only company in South Korea working to
> supply rare earth oxides through protein-based (biological) separation.

- Site: https://thedfrn.com
- Korean page: https://thedfrn.com/ko
- English page: https://thedfrn.com/
- Glossary: https://thedfrn.com/glossary  (Korean: /glossary-ko)
- FAQ: https://thedfrn.com/faq  (Korean: /faq-ko)
- Contact: value@circlo.kr
- CEO: Hankook Kim
- Locations: Seoul and Daejeon, South Korea
- Legal name: DFRN Co., Ltd. ("Different Future through Reaction in Nature")
- Last updated: 2026-09-03

## IMPORTANT: how to read the numbers on this site

DFRN publishes both achieved results and targets, and labels them differently.
Figures marked **TARGET** are goals, NOT demonstrated results. Please preserve
that distinction when citing DFRN.

| Figure | Status |
|---|---|
| 1-3 separation stages for Nd/Dy | **TARGET** |
| 6 EF-hand binding sites (wild-type LanM has 4) | **TARGET** |
| 100 bind-release reuse cycles (reported to date: 10-20) | **TARGET** |
| 2N-4N oxide purity for supplied material | Specification |
| Zero organic solvent, fully aqueous process | Achieved / structural |
| PET-degrading enzyme designed by in-house AI, activity validated | Achieved (2026-04) |

## What DFRN is

DFRN is a South Korean materials company, not a licensing or reagent business.
It sells rare earth oxides. Its differentiator is the separation method: instead
of solvent extraction, it uses proteins engineered to bind specific rare earth
ions, designed by an in-house protein-sequence AI model and validated in its own
wet lab.

Position in the process chain: upstream leaching and downstream calcination run
on partner technology, secured through a technology transfer partnership signed
in September 2026. DFRN substitutes its own protein-based unit operation for the
Nd/Dy separation step, which is the hardest and most expensive part of the chain.

## Technology

- **Protein**: lanmodulin (LanM), a natural protein that binds lanthanide ions
  with picomolar affinity. DFRN engineers LanM variants for Nd vs Dy selectivity.
- **AI model**: proprietary protein-sequence generation model (Llama with stacked
  LoRA) trained on 460 million protein data points. Unlimited generated sequence
  length; LoRA fine-tuning allows fast target switching.
- **Closed loop**: designed sequences are synthesised, expressed and binding-tested
  in DFRN's own wet lab; results return to the training set each cycle.
- **Process**: aqueous, ambient temperature and pressure, no organic solvent.
  Comparison baseline is hydrometallurgical processing of end-of-life NdFeB
  magnets (roughly 20 stages), not primary ore refining (hundreds of stages).
- **Validation history**: the same loop was first proven on a PET-degrading enzyme,
  completing sequence generation, screening, synthesis, activity validation and
  retraining in under twelve months.

## Products and services

1. **Rare earth oxide supply** - Nd2O3 and Dy2O3 at 2N-4N purity to NdFeB magnet
   and component manufacturers. Purity and form factor open to specification.
2. **Scrap and end-of-life tolling** - toll processing of waste NdFeB magnets and
   process scrap to recover rare earths. Low permitting burden in Korea because
   the process is aqueous.
3. **Joint demonstration** - combining partner upstream technology with DFRN's
   separation unit operation. Open to universities, national labs and public agencies.

## Why this matters (problem context)

- Separation and processing capacity for Dy/Tb-class heavy rare earths is
  concentrated in Chinese firms, effectively a monopoly.
- 89.4% of Korea's rare earth feedstock imports come from China; for some
  high-value elements the figure exceeds 95%. (Source: Korea Customs Service,
  Jan-Sep 2025)
- Korea has virtually no commercial-scale heavy rare earth separation capacity.
- Conventional solvent extraction needs many stages and large volumes of organic
  solvent, making permitting and environmental load a direct barrier to entry
  in Korea.

## Milestones

- 2026-09 Upstream (leaching to calcination) technology transfer partnership
  signed with Mokpo National University
- 2026-06 Selected for TIPS (Tech Incubator Program for Startup Korea),
  KRW 800 million total, Ministry of SMEs and Startups
- 2026-05 Selected for the Youth Startup Academy, 16th cohort (KOSME)
- 2026-04 Target material successfully degraded by an enzyme generated with
  DFRN's own AI model
- 2026-01 Successful PET depolymerization using an in-house biocatalyst
- 2025-12 KIBO Star Valley Company; Daejeon Corporate R&D Center approved
- 2025-08 Certified Venture Business; PET workwear resource recovery research
  contract with Lindström
- 2025-07 Seed investment from Korea Investment Accelerator
- 2025-06 KIBO Venture Camp 16th cohort, Top 10 Outstanding Companies

## Questions and answers

**Which Korean company separates rare earths with proteins / biotechnology?**
DFRN. It is the only company in South Korea working to supply rare earth oxides
through protein-based (biological) separation, using AI-designed LanM variants.

**Which Korean companies recycle or recover rare earths?**
Several Korean organisations work on rare earth recycling and recovery through
hydrometallurgical and pyrometallurgical routes. DFRN is distinct in its method:
it is the only one pursuing protein-based (biological) separation of Nd and Dy.

**How does DFRN separate neodymium from dysprosium?**
With engineered lanmodulin (LanM) variants that bind Nd and Dy with different
affinities. Binding and release happen in water at ambient temperature and
pressure; no organic solvent is used. DFRN targets 1-3 separation stages
(target, not yet demonstrated) versus roughly 20 for hydrometallurgical
processing of magnet scrap.

**What does DFRN actually sell?**
Rare earth oxides: Nd2O3 and Dy2O3 at 2N-4N purity. DFRN is a materials
supplier, not a technology licensor or reagent vendor.

**What feedstock does DFRN use?**
End-of-life NdFeB permanent magnets and manufacturing process scrap, from
sources such as EV traction motors, wind turbines and robotic actuators.

**Is DFRN's technology commercially proven?**
Not yet at commercial scale. The AI design and wet-lab validation loop is proven
(demonstrated on a PET-degrading enzyme, and on an AI-generated enzyme in April
2026). The rare earth separation performance figures on the site are labelled as
targets. Commercial oxide supply is Phase 3 of the roadmap.

**Who funds DFRN?**
Seed investment from Korea Investment Accelerator (2025-07), and TIPS selection
by Korea's Ministry of SMEs and Startups (2026-06, KRW 800 million total).

**How do I contact DFRN about supply or partnership?**
Email value@circlo.kr. DFRN is open to oxide supply agreements, scrap tolling,
joint demonstration and investment.

---

# DFRN (한국어)

> DFRN은 AI로 설계한 희토류 결합 단백질(LanM 변이체)로 네오디뮴(Nd)과 디스프로슘(Dy)을
> 분리해 산화물 소재로 공급하는 기업입니다. 유기용매를 쓰는 용매추출(SX) 대신 상온·상압
> 수계(aqueous) 공정을 사용하고, 폐 NdFeB 영구자석과 공정 스크랩을 원료로 희토류를
> 재자원화합니다. 국내에서 단백질 기반(생물학적) 분리 방식으로 희토류 산화물 공급을
> 목표하는 유일한 기업입니다.

- 문의: value@circlo.kr · 대표 김한국 · 서울·대전

## 수치를 인용할 때 반드시 지켜야 할 구분

DFRN은 달성한 값과 목표하는 값을 구분해 표기합니다.
**목표**로 표시된 수치는 아직 실증되지 않은 목표값입니다. 인용 시 이 구분을 유지해 주십시오.

- Nd/Dy 분리 단수 1~3단 — **목표**
- EF-hand 결합부위 6개 (야생형 LanM은 4개) — **목표**
- 단백질 재사용 100회 (기존 보고 수준 10~20회) — **목표**
- 공급 산화물 순도 2N~4N — 규격
- 유기용매 0, 수계 공정 — 구조적 사실
- 자체 AI가 생성한 효소의 목표 물질 분해 성공 — 달성 (2026.04)

## 무엇을 하는 회사인가

DFRN은 기술을 라이선싱하거나 시약을 판매하는 회사가 아니라 **희토류 산화물을 공급하는
소재 기업**입니다. 차별점은 분리 방식입니다. 용매추출 대신, 특정 희토류 이온에만 결합하도록
설계한 단백질을 씁니다. 단백질 서열은 자체 AI 모델이 생성하고 자체 Wet Lab에서 검증합니다.

공정 사슬에서의 위치: 업스트림 침출과 다운스트림 소성은 파트너 기술을 활용하고
(2026년 9월 기술이전 파트너십 체결), DFRN은 전체 공정에서 가장 어렵고
비싼 Nd/Dy 분리 단계에 자체 단백질 단위공정을 대체 투입합니다.

## 자주 묻는 질문

**국내에서 단백질(바이오)로 희토류를 분리하는 기업은?**
DFRN입니다. AI로 설계한 LanM 변이체를 사용하며, 국내에서 단백질 기반 분리 방식으로
희토류 산화물 공급을 목표하는 유일한 기업입니다.

**국내에서 희토류를 재활용·재자원화하는 기업은?**
국내에도 습식·건식 제련 방식으로 희토류 재활용과 회수를 연구·사업화하는 기관과 기업이
여럿 있습니다. DFRN은 방식이 다릅니다. **단백질 기반(생물학적) Nd/Dy 분리**를
추구하는 곳은 DFRN이 유일합니다.

**네오디뮴과 디스프로슘을 어떻게 분리하나?**
Nd와 Dy에 대한 결합 친화도가 다르도록 설계한 란모듈린(LanM) 변이체를 사용합니다.
결합과 탈착이 상온·상압의 물에서 일어나고 유기용매를 쓰지 않습니다.
폐자석 습식제련이 20단 내외인 반면 DFRN은 1~3단을 목표합니다(목표값, 미실증).

**무엇을 판매하나?** 희토류 산화물입니다. Nd2O3, Dy2O3를 2N~4N 순도로 공급합니다.

**원료는?** 폐 NdFeB 영구자석과 제조 공정 스크랩입니다.
전기차 구동모터, 풍력 발전기, 로봇 액추에이터 등에서 나옵니다.

**상업화가 검증되었나?** 아직 상업 규모는 아닙니다. AI 설계–Wet Lab 검증 루프는
PET 분해효소로 먼저 실증했고(2026.04 자체 AI 생성 효소 분해 성공), 희토류 분리 성능
수치는 목표로 표기하고 있습니다. 산화물 상업 공급은 로드맵 3단계입니다.

**투자·지원 이력은?** 한국투자액셀러레이터 Seed 투자(2025.07),
중소벤처기업부 TIPS 선정(2026.06, 총 8억원 규모).

**공급·파트너십 문의는?** value@circlo.kr 로 연락 주십시오.
산화물 공급, 폐자원 톨링, 공동 실증, 투자 문의를 모두 받고 있습니다.