This video documents an ambitious DIY project to build a photolithography system capable of producing 1-micron features, speedrunning decades of semiconductor manufacturing technology in a home shop. The project begins with the construction of a maskless lithography machine using a Digital Mirror Device (DMD) from a projector dev kit, a high-precision Zaber XY stage, and a custom sheet-metal frame. After encountering significant challenges with mechanical inaccuracies and the shallow depth-of-field of the optics, which limited consistent results to around 2 microns, the strategy pivots. A second machine, a 10:1 reduction stepper using a vintage Zeiss lithography lens, is built. The first machine is repurposed to create larger masks, which are then projected and reduced by the second machine. This two-system approach shows promising results but is halted when the UV LEDs in the first machine fail, requiring a complete and difficult disassembly to repair. The project is concluded as a "qualified win," having successfully created the tooling to achieve feature sizes on par with 1980s technology.
A DIY Quest for 1-Micron Photolithography
0:00 Project Goal: The creator sets out to build a photolithography machine in his workshop to replicate the 1-micron feature size capability of a $2 million commercial machine from 1978, using modern technology to accelerate the process.
3:06 Machine 1 Design (Maskless Stepper): The first machine is designed in CAD, featuring a sheet-metal frame, an enclosure with UV-filtering acrylic, and a Texas Instruments DMD (Digital Mirror Device) projector dev kit as its core for creating patterns digitally.
11:22 High-Precision Motion: The system is built around a donated Zaber XY stage with 1-nanometer encoders and 500-nanometer repeatability, providing an extremely accurate foundation for positioning the wafer.
12:41 Mechanical Challenges: The creator builds a custom Z-axis and rotation stage to hold the wafer. He immediately runs into design flaws and mechanical inaccuracies, such as hole misalignments and significant runout (wobble) in the rotation stage, which he measures with a dial indicator.
23:23 Initial Results & Problems: The first machine successfully produces features down to 2 microns. However, inconsistent focus across the wafer and poor photoresist adhesion prevent reliable, high-quality results.
24:25 Existential Crisis & Strategy Pivot: Realizing the mechanical precision required to overcome the shallow depth of field is too demanding, the creator decides to change his approach. He will now build a second machine—a reduction stepper—and use the first machine to create larger, less-demanding masks.
28:18 Machine 2 Design (Reduction Stepper): A second machine is designed and built, this time centered around a vintage 10:1 Zeiss reduction lens made specifically for photolithography in the 1970s.
30:10 Innovative UV Light Source: To illuminate the mask, a novel quasi-collimated light source is constructed from a stack of optical films, including diffusers, brightness enhancers, and privacy films, which helps direct the UV light in a straight path.
37:11 First Full Test: The creator successfully uses the first machine to create a flawed test mask on copper-coated glass. He then uses that mask in the second machine to project a reduced, 2.5-micron pattern onto a silicon wafer, proving the two-system concept works.
40:41 A Major Setback: While attempting to create an improved mask, the creator discovers the UV LEDs in the first machine have failed and shorted. Replacing them requires a complete teardown of the complex optical engine, halting the project.
42:55 A Qualified Win: Despite not reaching the 1-micron goal and the final hardware failure, the project is considered a success. Two complex, functional machines were built, establishing a process capable of producing chips with features on par with technology from the 1980s.
Summary of YouTube Comments
Community Reaction and Expert Insights
The comment section for this video is overwhelmingly positive, filled with a mix of awe, relatable frustration, and a remarkable amount of high-level technical advice from industry professionals. Viewers are broadly impressed by the ambition and execution of the project, with many calling it "incredible" and "insane" that such complex technology could be recreated in a home shop. A strong sense of community is evident, as many creators and engineers relate to the design mistakes, debugging frustrations, and the "existential dread" of having to disassemble a complex machine.
A significant theme is the wealth of expert feedback from viewers working in the semiconductor, optics, and precision engineering fields. Key suggestions include:
Mechanical Stability: Professionals from companies like ASML and Intel, along with machinists, strongly advised moving away from sheet metal to more thermally stable materials like granite surface plates or cast iron to minimize alignment errors. The direct-drive stepper motor on the rotation stage was a common point of criticism due to its inherent imprecision.
Optical and Process Advice: Optical scientists warned about the challenges of using components designed for visible light (like the DMD) with UV, citing potential degradation, focus shift from chromatic aberration, and inefficient coatings. Numerous suggestions were made for improving photoresist adhesion, most commonly by using a primer like HMDS (hexamethyldisilazane).
LED Failure Diagnosis: The most frequent diagnosis for the failed UV LEDs was excessive drive current or inadequate thermal management. Commenters suggested a detailed check of the driver circuitry, improving the heatsink, and using a lower drive current to ensure longevity.
Beyond the technical advice, the comments are peppered with humor, particularly regarding the creator's "shorted open" description of the failed diodes, the nonchalant conversion to "furlongs," and the discovery of a microscopic Rick Astley in the test patterns. Many viewers, from students to retired CPU designers, expressed that the video was deeply inspirational, demonstrating that with determination, even the most complex technologies can be explored and understood at a fundamental level.