Why Is mSAP the End of Subtractive Etching for High-End PCBs?

For decades, subtractive etching defined how the industry built circuits.

The logic was simple:

• start with copper foil
• remove what you don't need
• keep the rest

This approach worked well when:

• line widths were wide
• frequencies were low
• signal margins were forgiving
• board density was moderate

But today's electronics landscape has fundamentally changed.

We are now designing:

• HDI PCB with ultra-fine geometries (<30 μm, moving toward <15 μm)
• High-Speed PCB supporting 56G / 112G PAM4 channels
• mmWave RF structures at 28 GHz, 39 GHz, 77 GHz
• edge-AI hardware with dense interconnect and power coupling
• advanced packaging interfaces that blur the boundary between PCB and substrate

In this environment, the question is no longer: Can subtractive etching be optimized further?

It is: Has subtractive etching reached its physical and electrical limit?

And the answer, from an engineering standpoint, is yes.

mSAP (modified Semi-Additive Process) is not just a process upgrade.

It represents a fundamental shift in how conductors are formed, controlled, and optimized for electrical performance.

1. Subtractive Etching Was Designed for a Different Era of PCB Physics

Subtractive etching was developed in a time when PCB requirements were fundamentally different.

Historically:

• trace widths were large (100 μm–200 μm or more)
• frequency effects were secondary
• signal rise times were slow
• copper surface characteristics had minimal electrical impact

Under those conditions:

• small geometric variation was acceptable
• undercut did not significantly affect performance
• copper roughness was beneficial for adhesion and not harmful electrically

So subtractive etching was not "bad."

It was perfectly matched to the requirements of its time.

The problem today is not that subtractive etching has degraded.

It is that PCB physics has changed faster than the process can evolve

Modern designs now demand:

• tight impedance control within narrow tolerance windows
• extremely consistent conductor geometry
• low-loss transmission at multi-GHz frequencies
• minimal variation across panels and production lots

Subtractive etching was never designed for this level of control.

2. The Core Limitation: Etching Is Isotropic, Not Geometrically Controlled

At the heart of subtractive etching is a chemical reaction:

• etchant attacks exposed copper
• copper is dissolved away

But this reaction is not directional.

It removes copper:

• vertically (desired)
• laterally (undesired)

This is called isotropic etching behavior.

Even with process optimization:

• spray pressure
• chemistry concentration
• temperature control

you cannot eliminate lateral etching

This means:

• the final trace shape is not defined purely by the mask
• it is defined by the interaction between mask and chemistry

In other words: you are not fully controlling geometry—you are negotiating with chemistry

3. Undercut Is Not a Defect—It Is an Inevitable Outcome of the Process

Undercut is often treated as a process issue to be minimized.

But fundamentally: undercut is unavoidable in subtractive etching

When copper is etched:

• the etchant penetrates beneath the resist
• removes material laterally
• creates a trapezoidal profile

As line widths shrink:

• the ratio of undercut to total width increases

For example:

• at 100 μm lines → undercut may be acceptable
• at 30 μm lines → undercut becomes dominant
• at 15 μm lines → geometry collapses

This leads to:

• unpredictable line width
• inconsistent cross-section
• variation across panel and batch

So the real insight is: Undercut is not a defect—it is a built-in limitation that scales poorly with miniaturization

4. Electrical Consequences: Why Geometry Error Becomes Signal Error

At low frequencies, small geometry variation is tolerable.

At high speeds, it is not.

In High-Speed PCB:

• impedance depends on trace width, thickness, and dielectric geometry
• even small variation shifts impedance

With trapezoidal profiles:

• top width ≠ bottom width
• effective impedance becomes ambiguous

This causes:

• reflection (return loss degradation)
• insertion loss variation
• timing skew across channels
• channel-to-channel mismatch

At 112G PAM4: margin is already extremely tight

So even small geometric inconsistency can push the channel outside acceptable performance.

This is why: geometric precision becomes electrical performance

5. Copper Surface Morphology: The Hidden Driver of High-Frequency Loss

At high frequencies, current flows along the conductor surface (skin effect).

This makes surface condition critical.

Subtractive etching typically uses:

• roughened copper foil (for adhesion)

This creates:

• micro-scale peaks and valleys

Electrically:

• increases path length
• increases resistance
• increases insertion loss

At mmWave frequencies: copper roughness becomes a dominant loss mechanism

This is why:

• RF PCB
• high-speed backplanes
• 112G interconnects

are extremely sensitive to copper surface condition.

Subtractive processes cannot easily control this independently from adhesion requirements.

6. mSAP: From "Removing Copper" to "Engineering Conductors"

mSAP changes the paradigm.

Instead of starting with thick copper and removing it, mSAP:

1. starts with ultra-thin seed copper
2. defines geometry with high-resolution resist
3. electroplates copper only where needed
4. removes seed layer

This creates:

• near-vertical sidewalls
• precise trace width
• smooth copper surfaces

The key shift is: geometry is defined by lithography and plating—not by etching chemistry

This gives:

• deterministic control
• reduced variation
• repeatable structure

7. Ultra-Fine Line Scaling: Where Subtractive Completely Breaks Down

As design moves toward:

• 30 μm
• 20 μm
• 15 μm
• even <10 μm

subtractive etching faces:

• excessive undercut
• poor yield
• unstable process window

mSAP, however, is inherently suited for fine geometry because:

• it builds features rather than carving them out
• it avoids lateral material loss
• it maintains uniform cross-section

So the transition is not gradual. 

It is a hard boundary

Below a certain scale: subtractive etching is no longer viable for consistent production

8. Impedance Control at 112G and Beyond: Why Geometry Precision Is Everything

At ultra-high speeds:

• impedance tolerance windows are extremely tight
• channel loss budgets are minimal
• timing margins are narrow

In this regime:

• ±2–3 μm variation can matter
• surface roughness directly affects eye diagram
• geometry consistency across channels is critical

mSAP provides:

• consistent conductor width
• smooth sidewalls
• better predictability

This enables:

• tighter impedance control
• improved insertion loss
• better channel matching

Without this level of control, high-speed systems become unstable or require overdesign

9. Yield, Repeatability, and Process Window: The Manufacturing Reality

Beyond physics, there is a manufacturing reality.

Subtractive etching:

• sensitive to chemistry variation
• sensitive to panel loading
• narrow process window at fine geometries

Result:

• yield instability
• panel-to-panel variation
• difficulty scaling to volume

mSAP:

• relies more on controlled deposition
• has tighter lithography-defined geometry
• offers better repeatability

For production: this means higher yield and lower variability

10. The Strategic Conclusion: mSAP Is Not Optional for High-End PCBs

Subtractive etching still has a role:

• standard multilayer boards
• moderate-speed applications
• larger geometries

But for:

• HDI PCB ultra-fine routing
• High-Speed PCB above 56G / 112G
• mmWave RF
• advanced packaging interfaces

mSAP is no longer optional

It becomes: a foundational technology for enabling next-generation electronics

Technical Summary

The transition from subtractive etching to mSAP is driven by fundamental physics and electrical performance requirements.

The key conclusions are:

• Subtractive etching is limited by isotropic chemistry and unavoidable undercut
• Geometry errors directly translate into electrical performance degradation
• Copper roughness becomes a dominant loss factor at high frequency
• mSAP enables precise, vertical, and smooth conductor formation
• Ultra-fine geometries below ~30 μm require additive processes
• High-speed and RF systems demand geometry consistency that subtractive cannot provide
• Manufacturing yield and repeatability favor mSAP at advanced nodes

mSAP is not the future—it is the necessary foundation for high-end PCB performance in the present generation of electronics.