1. The SAL MicroFab in Context

Silicon Austria Labs is Austria’s flagship research centre for electronic-based systems, with three sites — Graz, Linz, and Villach. The Villach facility hosts SAL MicroFab, a 1,400 m² purpose-built cleanroom certified to ISO 14644:2015 and organised across two zones: an ISO 5 front-end line for MEMS/M(O)EMS micro-fabrication, and an ISO 8 packaging-prototyping-test area. Substrate capability runs to 200 mm wafers with batch processing up to 25 wafers.

This is the kind of facility that EVG’s cleaning unit is being installed into — and it is also the kind of facility that took a decade of capital planning, FFU procurement, ceiling-grid layout, and ISO validation to build. When SAL signs an award notice that names both the tool and “Filter Fan Unit ISO 3 for Process and Handling Area” as options in the same lot, the EU procurement office is documenting a unified buy: process tool + cleanroom envelope.

2. The Hidden Cost of “Equipment-First” Procurement

Most semiconductor and micro-electronics facilities are still procured in two separate workstreams:

1. Tool procurement. Lithography stepper, wafer bonder, cleaning system, MOCVD reactor — sourced from the equipment OEM through a tender like the one above.
2. Cleanroom construction. ISO 5/6/7 hardwall rooms, FFU ceiling grids, raised floors, air-handling units — sourced from a cleanroom contractor as a separate capex project, often 12-24 months earlier.

This split creates three predictable problems:

Interface mismatches. The tool arrives with a footprint of 1.8 m × 1.4 m and a service clearance of 600 mm on three sides. The cleanroom was laid out for a 1.5 m × 1.2 m predecessor. Now you are re-routing FFU ductwork, re-cutting raised-floor panels, and re-validating the particle count. Every rework is a cleanroom shutdown — and every shutdown is lost wafer throughput.

Particle-budget fragmentation. An ISO 3 process area (≤1,000 particles/m³ at ≥0.1 µm, per ISO 14644-1) cannot be retrofitted around an existing tool that was specified for ISO 5 (≤100,000/m³). The FFU layout, ceiling coverage ratio, return-air path, and worker-entry gowning all have to be co-engineered with the tool’s internal airflow and exhaust — not after it.

Schedule drag. When the cleanroom and the tool run on independent Gantt charts, the tool is the bottleneck. You pay for an idle cleanroom (or worse, an unvalidated cleanroom that the tool vendor refuses to install into). The cost of the stand-by FFU power, the conditioned air, and the locked-out operator shifts compounds quickly.

3. What “FFU ISO 3” Actually Means in a Working Cleanroom

Filter Fan Units are the lungs of any hardwall cleanroom. A ceiling-grid FFU pulls ambient air through a HEPA (H14, ≥99.995% at MPPS) or ULPA (U15/U16, ≥99.9995%) filter and discharges it as a unidirectional (laminar) flow downward over the work zone. ISO 3 classification at the process level requires:

• FFU coverage ratio of ~80-100% of the critical zone ceiling area
• Air change rates of 600-700 per hour
• Face velocity of 0.36-0.54 m/s (laminar flow band)
• Return-air path designed to avoid short-circuiting
• Pre-filtration on the intake side to extend HEPA life

For a wafer-cleaning process — which is what EVG’s tool is doing at SAL — the FFU grid has to be coordinated with the tool’s exhaust. The cleaning chemistry (DI-water, megasonic, NH₄OH/H₂O₂, brush stations) generates aerosolised particles that must be carried away from the wafer, not re-deposited. The FFU ceiling pattern, the tool’s local exhaust, and the operator’s gowning protocol form one aerodynamic system.

This is why the FFU appears in the same lot as the tool in the SAL award. It is not a building fixture that the facility team will source later. It is part of the process specification.

4. The “Solution, Not Widgets” Shift in EU Semiconductor Procurement

EU procurement law (Directive 2014/24/EU) has progressively moved buyers away from spec-by-spec widget purchasing toward functional / outcome-based lots. The SAL award is a textbook example. The buyer did not say “supply an EVG301” or “supply a standalone FFU”. They said:

“Supply, install, commission, and train operators on a frame/wafer cleaning unit — with optional Filter Fan Unit ISO 3 for the Process and Handling Area, and optional post-warranty service.”

One lot. One supplier. One acceptance test. The economic operator (EVG) bears the integration risk. The buyer gets a single point of accountability.

From a cleanroom-contractor perspective, this is the most important trend in semiconductor facility procurement in the last five years. The buyers who used to write three separate tenders (cleanroom shell, FFU ceiling grid, process tool) are now writing one tender that bundles the room and the tool. The suppliers who can only deliver one of the two are increasingly being designed out of the bid list.

5. GCC Cleanroom: From FFU Modules to ISO 3 Turnkey Envelopes

At GCC Cleanroom (Guangzhou, China), we have spent fifteen years working on the unglamorous half of this equation — the controlled environment envelope that lets a tool like the EVG unit perform to its specification. Our portfolio covers the layers that wrap around the process tool:

5.1 Modular hardwall cleanrooms — ISO 5 to ISO 8

Pre-engineered, factory-fabricated hardwall panels (powder-coated steel or aluminium honeycomb core) with flush glazing, integrated cable/service penetrations, and pre-cut FFU ceiling-grid apertures. A typical 200 mm wafer process line can be installed and validated in 8-12 weeks — roughly half the time of a traditional bricks-and-mortar build — because the room is a kit of parts, not a construction site.

5.2 FFU ceiling systems — ISO 3 to ISO 6 process zones

Direct-drive backward-curved impeller FFUs (sizes 2’×2′, 2’×3′, 2’×4′, 4’×4′) with H14 HEPA or U15/U16 ULPA, individually controlled face-velocity (0.36-0.54 m/s), and low-vibration (<0.5 mm/s RMS) frames suitable for sub-micron lithography and metrology adjacency. Ceiling-grid layouts are pre-engineered against the customer’s process tool footprint and the room’s return-air architecture — not adapted after the fact.

5.3 Air-handling and make-up units

Temperature (20-24 °C ± 0.5 °C) and relative humidity (40-50% ± 2%) control to within the tight bands that EVG-class cleaning tools and adjacent metrology require. DX chilled-water or direct-expansion coils, steam or electric humidification, and pre-filter banks (G4 + F7 + F9) sized for the FFU array’s static pressure budget.

5.4 Process-tool integration services

This is the layer that most cleanroom contractors do not offer — and the layer that is increasingly the differentiator. We will take the OEM’s tool installation drawing, the FFU layout, the exhaust-duct routing, the chemical-drain connection, and the operator-entry path, and produce a single integrated IFC (issued-for-construction) drawing. The cleanroom does not arrive on site and “meet” the tool. It arrives ready for the tool.

5.5 ISO 14644 validation and ongoing monitoring

At-rest and in-operation particle counting, airflow and pressure-cascade testing, recovery time, and containment-leak testing — to the revision of ISO 14644-3 in force at the time of acceptance. We also supply the ongoing particle-monitoring hardware (remote particle counters, environmental sensors) so that the room’s classification is verifiable day-after-day, not just at handover.

6. Five Facility Use Cases Where the Environment is the Project

The cases below are drawn from the kind of work GCC Cleanroom is asked to scope on a weekly basis. They illustrate why the tool vendor and the cleanroom vendor increasingly need to be in the same conversation from day zero.

6.1 Wafer-level packaging and hybrid bonding

Tools: EVG320 D2W, EVG Gemini, Besi die-bonders. Process sensitivity: particle-addition rate <0.01/cm² at 200 mm, ISO 3 local zone around the bonding chuck. Cleanroom requirement: ISO 5 background with ISO 3 FFU tunnel over the bond chamber; temperature stability ±0.3 °C; vibration isolation <VC-E.

6.2 MEMS and M(O)EMS fabrication

Tools: Deep-reactive-ion etchers (DRIE), wafer-bond aligners, sputter coaters. Process sensitivity: stiction and particle contamination on released structures. Cleanroom requirement: ISO 5 process zone with low-outgassing materials, dedicated wet-chemistry bench exhausting, and back-to-back FFU redundancy so that filter replacement does not interrupt a process run.

6.3 Power-device epitaxy (SiC, GaN)

Tools: Aixtron G5, Veeco Propel, LPE PE1O6. Process sensitivity: metal-organic precursor vapour; toxicity; high-temperature (>1,600 °C) MOCVD reactors. Cleanroom requirement: ISO 6/7 hardwall with local scrubbed exhaust, gas-cabinet rooms separated by a pressure cascade, and a slab floor load-rated for ~1,500 kg/m² reactor footprints.

6.4 Compound-semiconductor R&D and pilot lines

Tools: MBE chambers, plasma-etch tools, mask aligners. Process sensitivity: mixed III-V materials, frequent recipe changes, multi-user shared facility. Cleanroom requirement: modular and reconfigurable — sub-rooms that can be re-classified in days, not months, as a research programme pivots from GaN to InP to GaAs.

6.5 Photonics, quantum, and cryogenic packaging

Tools: Die-bonders with sub-µm alignment, cryostat integration, optical-test stations. Process sensitivity: vibration, acoustic noise, particulate accumulation on optical surfaces. Cleanroom requirement: ISO 5 with acoustic enclosures, optical tables on vibration-isolating piers, and a separate ISO 7 laser-test annex with wavelength-specific blackout.

7. The Build Sequence that Saves the Schedule

Whether you are a research centre like SAL, a commercial foundry, a university spin-out, or a power-device start-up, the same sequence discipline determines whether the project lands on time or slips by 6-12 months:

1. Process definition first. What wafer size, what throughput, what technology node, what is the killer particle spec, what chemicals, what exhaust loads. Write the process-of-record before anyone picks a tool or a wall panel.
2. Tool pre-selection in parallel with cleanroom concept. A shortlist of two or three tools — not a final PO — is enough to lock down the footprint, the service envelope, the exhaust class, and the utility connections. The cleanroom concept is then drawn around this envelope, not as a separate exercise.
3. FFU layout co-engineered with tool placement. The ceiling grid is the most expensive component to retrofit. Get the FFU apertures, the return-air path, and the tool’s local exhaust into the same 3D model before fabrication.
4. Phased installation. The hardwall shell and the FFU ceiling go in first; the air-handling and BMS come second; the tools land last into a validated envelope. The process-of-record acceptance test (ISO 14644-3) happens before the tool arrives.
5. Tool commissioning inside a qualified room. The OEM’s installation engineer walks into a particle-qualified, temperature-stable, humidity-stable, exhaust-ready room. The tool’s own commissioning time drops; the rework loop between tool vendor and cleanroom vendor closes.

This sequence is what an EU procurement officer writes down when they list the FFU option in the same lot as the cleaning tool. They are forcing the suppliers to deliver against this sequence — and the suppliers who can do it will be the ones on the next award notice.

8. The Honest Trade-Offs

No solution is free. The honest trade-offs of the bundled approach:

Single-supplier lock-in. Bundling the room and the tool under one PO concentrates interface risk in one supplier. If that supplier under-performs, you have one throat to choke — but only one. The mitigation is rigorous pre-award reference checks and a clearly written acceptance protocol that survives the supplier change if needed.

Higher front-end design effort. The integrated IFC package takes longer to produce than a cleanroom-only design or a tool-only design. The savings come at installation and qualification, not at the design stage. Plan 4-6 extra front-end weeks.

Modular cleanroom is not always the answer. For 300 mm foundry-class mega-fabs (TSMC, Intel, Samsung), traditional bricks-and-mortar still wins on absolute cost-per-square-metre at scale, and the OEM equipment vendors take the cleanroom interface directly. The modular approach wins in the sub-foundry band: 200 mm and below, R&D and pilot lines, university cleanrooms, MEMS/photonic/quantum fabs, and capacity-expansion projects where speed-to-validate matters more than absolute capex.

These are real trade-offs. They should be priced into the procurement decision, not discovered during installation.

9. What to Ask Your Cleanroom Supplier Before You Sign

If you are evaluating a cleanroom partner for a semiconductor, MEMS, or photonics project, the questions below separate the equipment-aware contractors from the rest:

1. Have you previously co-engineered the FFU ceiling layout with a specific OEM tool installation drawing? Which tool families?
2. Can you deliver an IFC package that includes the tool footprint, the exhaust duct routing, the chemical drain, and the operator-entry path — or do you only deliver the room shell?
3. What is your measured FFU face-velocity uniformity across a 4 × 4 grid at 0.45 m/s setpoint? (Target: ±10%.)
4. What is your typical recovery time after a 100-person contamination event? (Target: ISO 5 class restored in <15 minutes for ISO 5 background, <20 minutes for ISO 6.)
5. Do you supply the ongoing particle-monitoring hardware and the ISO 14644-3 re-validation service, or does that go to a third party?
6. Can your hardwall system be reconfigured — moved, extended, re-classified — within a 2-4 week window if our process roadmap changes?

A “yes” to all six means you are talking to the right partner.