How a 19‑Year‑Old Built a 1200‑Transistor CPU in His Garage

Sam Zeloof’s DIY CPU Journey

In 2018, 19‑year‑old American college student Sam Zeloof completed a fully functional 1200‑transistor CPU, named Z2, built entirely by hand in his garage over three years.

The Z2 uses a 10 µm polysilicon gate process, mirroring the technology of Intel’s 1970s 4004 processor, and represents an evolution from his earlier 6‑transistor Z1 chip made at age 17.

Motivation and Background

Zeloof was inspired by Jeri Ellsworth’s YouTube tutorials on low‑cost silicon wafer cutting and transistor fabrication, prompting him to create his own integrated circuits.

He purchased parts and chemicals from eBay, assembling a makeshift semiconductor lab without a cleanroom, using improvised tools and non‑pure chemicals.

Design and Layout

The chip layout began with a simple 10 × 10 transistor array designed in Photoshop, where each column of ten transistors shares a common gate and rows are series‑connected, sharing source/drain regions.

Images illustrate the transistor array and the individual 10 µm NMOS devices, showing misaligned metal layers, polysilicon gates (red outlines), and source/drain regions (blue).

Fabrication Process

Zeloof first employed a metal‑gate process, which suffered from high threshold voltages (>10 V) due to large work‑function differences between aluminum gates and silicon channels.

Switching to a polysilicon gate process eliminated the high threshold voltage, reduced power consumption, and allowed compatibility with 2.5 V and 3.3 V logic levels.

Key steps included:

Designing the photomask and cutting the wafer.

Etching the polysilicon gate.

Self‑aligned processing and high‑temperature diffusion for doping, avoiding ion implantation.

Despite the low‑cost approach, the lack of pure chemicals and a cleanroom resulted in a low yield (under 80%).

Results and Impact

Zeloof reported fabricating 15 chips (≈1500 transistors), with at least one fully functional and two partially functional, demonstrating the feasibility of garage‑scale CPU production.

The project attracted attention from veteran engineers who offered advice on cloning the 4004 architecture, and it sparked online discussions about “breaking Moore’s Law” in a hobbyist setting.

Legacy

Zeloof’s work showcases how curiosity and resourcefulness can enable complex semiconductor experimentation outside traditional industrial facilities, inspiring a new generation of hardware hackers.