Key Takeaways
  • Introduction: Why GEM200 vs GEM300 Matters
  • What Are GEM200 and GEM300?
  • Protocol Stack: SECS-I vs HSMS
  • Carrier Management: E84 and E87
  • Process Management: E40 and E94

Introduction: Why GEM200 vs GEM300 Matters

GEM200 and GEM300 describe two distinct levels of SEMI equipment communication compliance. GEM200 serves 200mm fabs; GEM300 encompasses the expanded standards mandated by 300mm automated factories. A tool that only supports GEM200 cannot deploy in a modern 300mm fab without significant rework. See our SECS/GEM Protocol Guide for foundational context.

What Are GEM200 and GEM300?

Both build on SEMI E30 GEM. GEM200 = E4 + E5 + E30 + E37 (classic stack). GEM300 adds E39 (Object Services), E40/E94 (Process/Control Jobs), E84/E87 (Carrier Management), E90 (Substrate Tracking), E42 (Recipe Management), and E116/E134/E157 enhancements. The philosophical shift: the host moves from observer to controller.

Protocol Stack: SECS-I vs HSMS

SECS-I (E4): Serial Legacy

RS-232 serial at 9600 baud, point-to-point, block-based transfer. Adequate for 200mm data volumes but a bandwidth bottleneck for modern tools.

HSMS-SS (E37): TCP/IP Standard

TCP/IP networking with 100Mbps+ bandwidth, no block overhead, standard Ethernet infrastructure. Mandatory for GEM300. Supports concurrent message handling needed for full automation.

Carrier Management: E84 and E87

E84 defines the parallel I/O handshake between AMHS (OHT) and load ports. Not needed for manual 200mm loading; non-negotiable for 300mm FOUP automation.

E87 (COSEM) manages the carrier lifecycle: ID verification (BCR/RFID), carrier state model, slot map reading, access mode control. The most complex GEM300 standard to implement. Absent from GEM200.

Process Management: E40 and E94

E40: Process Jobs define host-directed processing — specific wafers, specific recipe, specific chamber. State model: Queued → Setting Up → Processing → Complete.

E94: Control Jobs group Process Jobs, orchestrate carriers and execution sequence. Enables full host-directed automation. No equivalent in GEM200.

Recipe Management: E42

E42 adds namespaces, versioning, pre-execution validation, and recipe locking. GEM200 only has basic S7 upload/download. Critical for deterministic recipe management in high-volume 300mm manufacturing.

Substrate Tracking: E90

Mandatory in GEM300. Equipment reports every substrate movement through internal stations. Enables wafer-level traceability, WIP tracking, and per-wafer process data correlation for APC/FDC. GEM200 typically provides only lot-level tracking.

GEM200 vs GEM300: Complete Comparison Table

Feature GEM200 GEM300
Transport SECS-I or HSMS HSMS-SS mandatory
Object Services (E39) Not required Required
Process Jobs (E40) Not supported Mandatory
Recipe Mgmt (E42) Basic S7 Full E42 RMS
Load Port (E84) Not required Required for AMHS
Carrier Mgmt (E87) Not supported Mandatory
Substrate Track (E90) Not supported Mandatory
Control Jobs (E94) Not supported Required
Automation Level Operator-centric Host-centric, fully automated

When Should You Support GEM300?

Must: Any tool for 300mm fabs, tools handling FOUPs/AMHS, tools requiring wafer-level traceability. GEM200 sufficient: Standalone 200mm legacy tools, offline metrology, R&D prototypes. Both: Tools serving 200mm and 300mm markets.

Migration Path: GEM200 to GEM300

Five phases: ①Transport (SECS-I→HSMS) → ②E39 Object Model → ③E84+E87 Carrier Management → ④E40+E94 Process/Control Jobs → ⑤E90+E42 Substrate Tracking + Recipes. MST’s NeuroBox platform natively supports both GEM200 and GEM300, providing pre-built implementations that cut development from years to months. See our open-source SECS/GEM driver for a starting point.

SECS/GEM integration still taking months?

We open-sourced our SECS/GEM Python driver. 3 lines of code to connect. 15,000+ downloads, 40+ contributors.

View Open-Source Driver →
MST
MST Technical Team
Written by the engineering team at Moore Solution Technology (MST), a Singapore-headquartered AI infrastructure company. Our team includes semiconductor process engineers, AI/ML researchers, and equipment automation specialists with 50+ years of combined fab experience across Singapore, China, Taiwan, and the US.