PCB Laser Drilling: UV, CO2, and Hybrid Systems Compared
Microvias are what make high-density interconnect possible, and almost all of them are formed by PCB laser drilling rather than by a mechanical bit. A laser removes dielectric without applying mechanical force, which allows holes far too small for a drill to be produced reliably at high speed. The trade-off is that the process only works within a narrow set of material, copper, and thickness conditions, and the machine that produces the hole determines both the cost and the achievable geometry.
Understanding the differences between laser types explains why a fabricator asks for specific stackup parameters, and why a stackup that works at one supplier may not be producible at another.
Why Laser Drilling Replaced Mechanical Drilling for Small Holes
A mechanical drill has a practical minimum diameter, below which the bit breaks and the hole wall is unreliable. For a microvia in the 3 to 6 mil range there is no economically viable mechanical process at all. A laser avoids the mechanical limit entirely, and because the pulse can be positioned and triggered quickly, it also removes the cost penalty that would otherwise apply to very small holes.
The trade-off is depth control. A laser removes material thermally, so the process must stop precisely at the target copper pad. That is achieved by making the dielectric layer thin and uniform, and by using the underlying copper as an optical stop. A stackup with uneven dielectric thickness produces over-drilled and under-drilled vias in the same panel.
The other consequence is a tapered hole profile, wider at the top than at the bottom. The taper affects plating and current capacity, which is why a microvia that must carry current is usually replaced by several microvias in parallel rather than by one larger hole.
UV and CO2 Lasers Behave Differently
The two wavelengths used in production interact with materials in fundamentally different ways. A CO2 laser emits in the infrared, where most polymers absorb strongly and copper reflects almost completely. It removes dielectric very efficiently and can clear glass-reinforced material, but it cannot drill through copper and struggles to produce features below roughly 75 microns.
An ultraviolet laser has a shorter wavelength and removes copper as well as dielectric, but its removal rate for dielectric is lower and its performance on non-uniform reinforced laminate is poorer, because reaching the glass reinforcement requires a higher energy density that can also damage the inner-layer pad. In practice, UV is used where small features or copper removal are needed, and CO2 is used where dielectric removal speed and area throughput matter most.
Hybrid Systems Combine Both
A hybrid laser tool carries a UV head and a CO2 head on the same platform. The UV beam opens the copper and defines the top of the hole, then the CO2 beam removes the exposed dielectric. Because each step uses the wavelength best suited to the material, the combination produces small holes with clean sidewalls at a higher throughput than either source alone could achieve.
The architecture introduces its own constraints. The two beams must be registered to each other with the same accuracy that the microvia requires, and the panel must be processed in fields so that the beam positioning system can move between them. Older machines use a fixed distance between the heads with step-and-repeat positioning, which limits how optimally a given panel can be arranged. Machines with programmable head spacing can adjust the field for the panel layout and therefore use more of the tool’s capacity.
Drilling throughput is governed by two variables: pulse energy, which sets how long one hole takes, and the beam positioning system, which sets how long the tool takes to move between holes. A machine can be limited by either one, which is why quoting a hole count without stating the pattern and the field size is not a meaningful measure of capability.
Materials and Their Drilling Behavior
Polyimide is one of the most laser-friendly substrates, because it ablates cleanly and leaves a residue-free wall that plates well. Nd:YAG and UV sources drill polyimide reliably at dimensions that mechanical processes cannot approach, with the most economical working diameter in the tens of microns and throughput measured in thousands of holes per minute. Direct laser punching can produce holes without an intermediate imaging step, and the resulting wall quality is good enough that no additional desmear treatment is required.
PTFE-based laminates behave differently: they are softer, more prone to smearing, and the optimal geometry differs from a polyimide stack. The practical implication is that the drilling recipe has to be qualified for the specific material system, and a change of laminate in an HDI stack is a process change, not a substitution.
Drilling Ceramic and Rigid Substrates
Ceramic substrates present the opposite problem. Alumina has a high dielectric constant that makes it attractive for microwave circuits, but it is brittle, and any mechanical drilling or routing operation has to be done with minimal applied pressure. Laser drilling removes that mechanical constraint almost entirely.
Short-pulse, low-energy, high-peak-power sources are used for this application, because they remove material by ablation rather than by thermal melting, which avoids the microcracking that mechanical stress would produce. Holes below 100 microns can be produced with acceptable quality, and the technique has been applied successfully to microwave amplifiers operating across the 8 to 18 GHz range, where substrate integrity directly affects loss.
Process Control and Quality Verification
The parameters that must be controlled are pulse energy, pulse width, pulse repetition rate, beam focus and spot size, and the registration between the beam and the target pad. Because the process removes material thermally, small drifts in any of these produce a change in hole diameter or depth that may not be visible until plating reveals it.
Verification is therefore destructive by nature. Cross-sections taken at the start of a production lot show the taper, the depth, and whether the target pad was penetrated cleanly, while electrical test only confirms that a connection exists. Where a via is filled, the fill must also be inspected for voids, since a void inside a stacked via becomes a crack initiation point under thermal cycling. Filling processes and their inspection are covered in electroplating and via filling for HDI.
How Laser Drilling Shapes the Stackup
Because the process depends on thin, uniform dielectric and on copper as a stop layer, the stackup cannot be chosen independently of the drilling capability. Build-up layers must be thin enough for the laser to reach the target, the copper below must be treated so that it reflects or absorbs predictably, and the capture pad must be sized for the registration tolerance of the machine. Related stackup decisions are described in blind and buried via stack selection.
The same constraints explain why laser-drilled structures are handled differently in data preparation. A laser via has a defined shape in the artwork, and the CAM system must convert it into the correct aperture and drill program. The conversion rules are described in HDI board CAM methods.
FAQ
Can a CO2 laser drill through copper? Not in production. It removes dielectric efficiently, but copper reflects the infrared beam almost completely, so the copper layer must be opened first, either by a UV laser or by chemical etching, before the CO2 beam can reach the dielectric.
What determines the smallest microvia a fabricator can produce? The laser wavelength and spot size set the physical limit, but the practical limit is usually the capture pad and the registration accuracy. A hole smaller than the pad can reliably capture is not useful, so the design should be built around the pad geometry the process supports rather than around the smallest hole the machine can make.
Does laser drilling eliminate the need for desmear? It depends on the material and the recipe. Polyimide generally produces a clean wall that plates directly, while epoxy-based laminates can leave resin residue that must be removed before plating. The requirement should be confirmed for the specific laminate rather than assumed from a general rule.
