Particle Risk & Semiconductor Hardware

And how PECM's unique superfinishing capabilities can mitigate significant damage and risk across some of the world's most important equipment infrastructure.

---

When it comes to building the high-purity vacuum and gas delivery systems used in semiconductor fabrication, the exact machining methods used to make these systems don't make that much of a difference, as at the end of the day, the same underlying issues around burrs, EDM recast, laser dross, and embedded abrasive media remain the same.

While it may be a reasonable stance to blame the part designs for these faults, manufacturers should instead consider shifting part of the blame for manufacturing shortcomings on the material removal methods themselves. Whether it's gas distribution plates or vacuum manifolds, remnants from processes like EDM, laser, or abrasion can still leave their marks in the underlying internal features of these critical parts, resulting in potential future damage. Voxel's PECM methodology, however, sidesteps these issues and may act as a uniquely beneficial process for the internal geometries found in certain UHP semiconductor hardware infrastructure.

Schedule 30-minute Introductory Conversation

Key Takeaways

• Semiconductor fabs take sub-micron particle contamination as a major concern
  

• Burrs are arguably the most common, but not the only, source of particulate contamination: EDM recast, laser dross, and embedded abrasive media can all be sources as well
  

• Equipment has a variety of non-line-of-sight, difficult-to-access features that require superfinishing: from cross-drilled manifolds to intersecting holes

Why is Semicon Hardware So Sensitive to Particulate and Residue?

Consider an issue of micro-particulate contamination within an automobile's fuel delivery systems. Miniature particulate contamination in the engine of a Toyota Camry is not unexpected, and the disruption this would cause to the combustion efforts of the engine would likely be minimal, if not felt at all.

However, in the world of semicon fabrication, a singular sub-micron particle that makes its way to a wafer can be the difference between a working die and total scrap, notably made worse by ever-decreasing sizes in process nodes. A single particle less than the size of a grain of salt has the power to stall multi-million-dollar equipment.

Etched wafer. Image sourced from Wikimedia commons.

Consider standards around semiconductor cleanrooms (graded under ISO 14644-1). An ISO class 5 environment (a typical standardization for wafer fab bays) allows no more than 3,520 particles ≥0.5 µm per cubic meter of air, and 100,000 particles ≥0.1 µm per cubic meter. Even more, the ISO Class 1 (the most sensitive of process areas) allows just 10 particles ≥0.1 µm per cubic meter. Swagelok, for instance, states that its stainless steel UHP components are assembled and tested in ISO Class 5 primary cleanrooms) 352 particles ≥0.5 µm per cubic meter) with surface qualities on these UHP surfaces anywhere from 0.13 µm Ra to 0.20 µm Ra.

To put those numbers in perspective: the average particulate count (with each particle being about 1/140th the width of a human hair) in an office space (what ISO would call a 'Class 9') runs about 35,200,000 particles ≥0.5 µm per cubic meter, meaning the cleanrooms in semiconductor environments have roughly 10,000 times fewer particles than the air in a typical office room. Meanwhile, the surface quality standards cited are orders-of-magnitude smoother than a typically machined surface: ordinary milling or turning on a stainless steel component may run about 3.2µm Ra, and Swagelok's UHP specs are about 25 times smoother than that.

So why is a single particle such a problem? If a particle contaminates an area during the photolithography, the light patterning the surface can become blocked, even on a micro-scale, affecting the feature underneath. If the particle is put on during the deposition or etch, it can shadow the surface, making material build up unevenly.

Where Do Particles Come From?

Most of the particulate contamination in gas hardware, fundamentally, originals in how the part or feature was cut, drilled or finished. In this section, we can identify four distinct, but crucial, artifacts of conventional machining that leave these parts.

This image was enabled with generative AI.

1. Burrs. Burrs form from cutting tools that do not shear the material cleanly all the way to the exit edge of a hole or cut. As the tool approaches that edge, the remaining material may deform plastically instead of separating...folding over into a raised lip. Generally, these are considered byproducts of the manufacturing process, not at the specific fault of an operator or machine.
   

2. Recast Layers. During the EDM process (and other heat-based processes...), melted material may sometimes re-solidify itself on the surface faster than the dielectric fluid (in the case of EDM) may flush it away. This resolidified layer is often porous, and generally poorly bonded to the base material, causing it to flake free later as debris.
   

3. Laser Dross. Laser dross forms in a similar way to recast layers, except it is when the beam melts material at the cut edge and some of that melt re-deposits (and re-solidifies) at the kerf rather than fully ejecting. Sputtering is a similar common issue with this process.
   

4. Embedded media. This, ironically, can come from deburring methodologies meant to fix issues found in the other three. Abrasive blasting and media-based deburring removes burrs through mechanical impact, but that same impact can actually drive fragments of the abrasive itself into the metal surface. Notably, this embedded media is often invisible under standard visual or white-metal inspections, so a part can pass routine checks and still carry hidden particle sources.

PECM: An Unconventional Fix

Consider how each of the aforementioned 4 issues traces back to the same root causes: The process either made direct contact with, or mechanically melted/deformed the material to remove it. Pulsed electrochemical machining (PECM) removes that problem entirely as it removes the material through a controlled anodic dissolution as opposed to tool contact or thermal melting. This means PECM completely avoids:

• Burrs, as there is no shearing taking place, just uniform dissolvement

• Recast, as the temperature of the electrolytic fluid is nowhere near hot enough to melt material (not to mention, it changes the chemical composition of the metal to a hydroxide)

• Dross or sputter, for similar aforementioned reasons

• Embedded media, as the electrolytic fluid gently flushes away particulate waste material, providing no opportunities to hide material in microcracks

Additionally, PECM and OPECM are able to reach non-line-of-sight geometries often found in cross-drilled manifolds or intersecting hole junctions. With custom tooling (including some in-situ capabilities) PECM can remove material and superfinish components (Down to 0.005-0.4um Ra) without any further processing steps required.

Finally, PECM is also able to machine a wide range of tough-to-machine materials other processes may struggle with, as material hardness or toughness does not correlate with its ability to be machined via PECM (conductivity).

Fit still depends on specifics: the manifold's internal flow paths, how electrolyte can be routed through them, the material's conductivity, and the purity spec a given part has to hit. Not every geometry is a clean fit, and 99% positivity an application will work isn't 100% which is exactly why this is worth a real look rather than a blanket assumption either way. If particle risk in your gas or vacuum hardware traces back to a deburring or finishing step, send the drawing, internal geometry, material, and spec to Mike and Kirk at info@voxelinnovations.com for a fit review.