Keeping legacy semiconductor tools running when parts go obsolete

A process tool built twenty or thirty years ago can still do good work. Chambers, vacuum systems and wafer stages are robust, and the physics of a sputter source or an RIE chamber has not changed. What usually ends a legacy tool's life is not the process hardware but the parts around it: a control computer that no longer boots, a discontinued interface board, an RF generator with no repair route, or a manual that left with an engineer years ago.

Keeping legacy semiconductor tools running is mostly a matter of finding those weak points before they fail and deciding, for each one, whether to stock, repair or replace. This post sets out how to do that in a research or pilot-line setting.

Where legacy tools actually fail

Across etch, deposition and implant tools of the 1980s to 2000s, the same groups of components tend to cause the long outages, because they combine electronics that ageing affects with designs that only one maker ever produced:

• Control computers. Industrial PCs running DOS, early Windows or proprietary operating systems, with ISA or PCI cards, floppy or early hard drives, and CRT displays.
• Interface and I/O boards. Custom boards that connect the computer to valves, gauges and power supplies. Often unique to the platform and long out of production.
• RF generators and matching networks. Older generators and their control electronics, especially models whose makers have withdrawn support.
• Vacuum components. Turbomolecular pump controllers, cryopump compressors and gauges whose controllers have been superseded.
• Mass flow controllers. Analogue MFCs with obsolete connectors or communication protocols.
• Wafer handling. Robots and their controllers, which combine precise mechanics with platform-specific electronics.

Mechanical parts, such as O-rings, valves, bellows and chamber parts, can usually be sourced or made, if necessary from a drawing or a measured sample. Electronics and software are where the real risk sits, because a single failed board or corrupted disk can stop the whole tool.

Start with an audit

The first step is a written audit of each tool, listing every subsystem and answering three questions:

1. Is it still supported by its maker? Check for current part numbers, repair services and documentation.
2. Is there a current equivalent? Many gauges, MFCs and pump controllers have drop-in or near drop-in modern replacements.
3. What happens if it fails tomorrow? Estimate downtime, and whether a failure would stop the tool entirely or only degrade it.

The answers sort components into three groups: those you can buy new, those you need to stock because no replacement exists, and those you should plan to retrofit. Keep the audit with the tool's maintenance log and review it each year.

Back up software and settings now

Software loss is the failure most likely to be permanent. Old hard drives fail without warning, and a control system without its software is often unrecoverable.

Do these while the tool still works:

• Image the control computer's disks with a sector-level copy, and store copies off the machine.
• Export recipes, calibration tables and configuration files in whatever format the software allows. Print the critical ones.
• Photograph settings screens, DIP switch and jumper positions on boards, and the wiring at key connectors.
• Record licence keys, dongles and passwords. Some legacy software is tied to a hardware key; losing it can be as bad as losing the disk.
• Collect every manual and drawing you can find, scanned and filed with the tool record.

If the computer uses floppy disks or old drive interfaces, solid-state emulators can often replace the drive while keeping the original software unchanged. This removes a common mechanical failure without touching the control logic.

Build a sensible spares holding

Stocking every part is neither affordable nor necessary. Focus on the components that are both obsolete and critical:

• Interface boards and power supplies unique to the platform.
• Spare control computer hardware matching the original, configured and tested with your disk image.
• Sensors and gauges whose replacements would need control changes.
• Consumables with long lead times, such as specific quartz parts, electrodes and seals.

Sources for obsolete parts include donor tools from decommissioned lines, independent repair houses and the second-hand market. Buy only tested parts where you can, and test spares on arrival rather than when you need them. Our spare parts service sources and tests parts for older platforms, including from donor tools.

Repair rather than replace

Many failures on legacy electronics are repairable at component level. Electrolytic capacitors dry out, relays wear, connectors corrode and power supplies fail in predictable ways. A board that the maker no longer offers can often be diagnosed and repaired by an electronics technician with the schematic, or even without it.

RF generators, turbo controllers and cryopump compressors can often be refurbished by independent specialists. Ask for a test report on each repaired unit, not just a working sticker.

Preventive replacement also helps. Recapping power supplies on a tool that has run for decades, replacing cooling fans and cleaning boards can prevent the failures that cause the longest outages.

When a retrofit is the better choice

At some point, maintaining the original control system costs more in risk than replacing it. Signs that a retrofit is due include repeated failures of the same subsystem, no remaining spares for a critical board, software that cannot run on any available hardware, or a need for features the original cannot provide, such as data logging or network access.

A control system retrofit replaces the computer, I/O and software with current industrial hardware while keeping the chamber, vacuum system and process hardware. Done well, it gives you supported parts, modern recipe management and process data logging. Points to plan for:

• Recipe transfer. Existing recipes need translating and validating on the new controller.
• Interlocks. Every safety interlock must be mapped, reimplemented and tested.
• Requalification. Run your reference processes before and after, and compare.
• Documentation. The new system needs its own drawings, I/O list and user manual.

Retrofits are also a natural time to replace other obsolete items in the same outage, such as gauges, MFCs or a turbo controller.

Plan the budget over years, not failures

Legacy tools are cheap to run until something fails. Spreading the work over a few years, with backups first, critical spares next, and retrofits when the risk justifies them, keeps costs predictable and avoids emergency purchases at any price. The cost drivers are the number of unique obsolete components, the availability of donor parts and the effort to requalify processes after each change.

What we can do

We audit older tools, back up their software and settings, source and repair obsolete parts, and retrofit control systems when the original reaches the end of its life. See our control system retrofit and repair and troubleshooting services for how we approach legacy platforms.