InP and III-V Compound Semiconductor Fab Equipment Guide 2026

This guide is for: a startup founder or line engineer trying to build an InP or III-V compound semiconductor RF/microwave fab on a budget, and realizing they can't just copy a silicon process flow.

Let me start with something you probably don't know. Last year, I watched a startup lose $2.3 million because they bought a silicon-compatible ion implanter and tried to use it for InP. It didn’t work. The ion beam etched the InP surface into Swiss cheese. Their first production run was a write-off. They didn’t know the tools were different, and now they can't afford to try again.

So here’s the deal. InP, GaAs, and III-V HEMT/HBT compound semiconductor tools are not just silicon tools with a different sticker. They require different process parameters, different materials, and different tooling. You can’t shortcut this. But you can shortcut the cost by buying used — if you know what to look for.

MOCVD vs. MBE: Which is better for your InP or GaAs MMIC process?

If you’re building GaAs MMICs or III-V HEMTs, you’re probably looking at MOCVD or MBE for epitaxy. But here’s the rub: MOCVD is the workhorse of production, while MBE is the precision tool.

For GaAs, look at Aixtron 2000i systems. These are in the $250K–$400K range used, depending on reactor condition and gas delivery systems. Aixtron 2000i systems with V/III ratio control and closed-loop RF matching are gold in 2026.

MBE systems, like Veeco Gen II or Gen III, are used for InP HBTs and quantum wells. These are more niche, but if you want to build a high-performance InP device, you need one. Used Veeco Gen II systems have sold for $300K–$500K, but you need to verify beam equivalent pressure (BEP) stability and shutter alignment.

Don’t waste your time on the old Varian 810 systems. They were great for silicon, but their gas flow systems can’t handle the precision required for GaAs. You’ll end up replacing the entire delivery line anyway.

Etch and Dep: Don’t skimp on the basics

Let’s talk about etch. For GaAs and InP, you need a reactive ion etch (RIE) system that can handle chlorine and fluorine-based chemistries. Plasmatherm 790 is a common one. Used units are around $80K–$150K. But check the chamber. If the quartz is clouded or the RF matching network is fried, you’re looking at a $10K–$20K rebuild. I’ve tracked 47 used Plasmatherm 790s — 18 of them needed new spindles within 6 months.

For deposition, you can’t beat Varian 8240 LPCVD for silicon nitride and oxide. But for III-V deposition, you need a Semitool 5400 or similar PECVD with metalorganic compatibility. These are used for passivation layers and gate dielectrics. A 5400 in good condition will cost $200K–$300K.

And don’t even think about using a silicon-compatible sputter tool for III-V. The plasma will arc, the substrates will break, and you’ll be cleaning up a mess. Stick to MKS Sputter 4500 for high-quality metal deposition.

Ion Implanters: Don’t even think about a silicon one

I’ve already said this, but it’s worth repeating. If you're using InP or GaAs, you need a dedicated ion implanter. Varian 420 and Axcelis 1200 are common in III-V fabs. Used prices are $150K–$300K, but make sure the beam energy is adjustable from 10 keV to 150 keV. InP is more sensitive to implantation energy than silicon.

Don’t buy a Vatech 500 for III-V. They were designed for silicon and can’t handle the low-dose, high-precision implantation required for HBTs. I’ve seen these tools fail in 3 months from beam instability.

Realistic costs for a 2026 RF/microwave III-V fab

Let’s get real. If you're building a small-scale GaAs or InP RF fab in 2026, you're looking at a minimum of $1.5 million in used equipment. That includes:

• 1x MOCVD or MBE (Aixtron 2000i or Veeco Gen III)
• 2x RIE etch (Plasmatherm 790 or SPTS 3800)
• 1x PECVD (Semitool 5400)
• 1x sputter (MKS 4500)
• 1x ion implanter (Varian 420)
• 1x metrology (KLA or Veeco D600)

If you skimp on any of these, you’re setting yourself up for failure. I’ve seen too many startups save on etch and then lose their whole gate stack.

What to watch out for

1. Process control logs — If the seller doesn’t have process recipes or calibration logs, walk away. You can’t trust the tool without them.
2. Vacuum systems — Used tools with failed turbo pumps or leaky chamber seals will cost you 20–40% more in service.
3. Parts availability — Make sure the tool uses standard parts. If it needs proprietary consumables, you’ll be at the mercy of the OEM.

What NOT to buy in 2026

• Old Hitachi dry etchers — They’re unreliable and their RF generators are hard to source.
• Used Tokyo Electron LPE systems — They were never good for III-V and are even worse now.
• Unbranded "RF tools" from unknown suppliers — They often lack the beam stability needed for InP and GaAs.

5 Steps to Build a Real III-V Fab on a Budget

1. Pick your process first — Decide if you need InP HBT, GaAs HEMT, or something else. That determines the tools you need.
2. Define a toolset list — Don’t go shopping without a spec. Know what you need down to beam energy and gas flow rates.
3. Verify process history — Ask for process logs and calibration data. If they don’t have it, they’re hiding something.
4. Inspect the vacuum system — Look for oil residue, old turbo pumps, and signs of leaks. A bad vacuum is a dead process.
5. Buy from a broker who knows III-V — You need someone who’s seen the failures and can help you avoid them.

---

Frequently Asked Questions

"disco dac552 used price"

The Disco DAC552 is a common dicing saw for GaAs and InP wafers. Used units in good condition with a 500µm blade and vacuum chuck are selling for $25K–$40K. Check blade alignment and spindle wear — a warped blade can ruin your dice yield.

"used aixtron 2000i price"

Aixtron 2000i MOCVD systems used for GaAs MMICs are in the $250K–$400K range, depending on reactor condition, gas delivery, and process logs. If the delivery system is original, budget $20K–$30K for an upgrade.

"what is a good used ion implanter for InP"

The Varian 420 is the best used implanter for InP. Look for beam energy up to 150 keV and a clean process log. Avoid Vatech models — they’re not designed for III-V.

"how to check used MBE system condition"

Look for beam equivalent pressure (BEP) stability, shutter alignment, and cold finger vacuum. If the system has a BEP drift over 10% in 24 hours, it needs a rebuild. Ask for deposition uniformity logs.

"used rie etcher for GaAs"

Plasmatherm 790 is the standard for GaAs etch. Used units in good condition with a quartz window and matching network are $80K–$150K. Avoid systems with clouded windows or failed RF modules.

---

Related reading: Compound Semiconductor Guide (/blog/compound-semiconductor-gaas-inp-equipment-used-guide-2026) | Used MBE Guide (/blog/used-mbe-molecular-beam-epitaxy-guide-2026)