Estimated reading time: 5 minutes

The Chemical Connection: Managing Cumulative Process Variations for Fun and Profit

Tolerances have become increasingly tight as PCBs have grown more complex and reduced in size, weight, and power consumption. Even if we maintain a traditional ±10% variation in etched line width, the tolerance in a 50 µm (0.002" or 2 mils) wide etched line is only 5 µm (0.0002” or 0.2 mils). That’s not very much and it gets even worse as the line and space requirements drop to 25 µm lines and spaces.

Lately, I’ve seen some tolerances as low as ±5% but so far, that’s more of a wish than what is acceptable. Unfortunately, it won’t be long before that wish becomes a demand. So, what is the best way to manage the cumulative tolerances? Each process step contributes to the total variation in the final product and the variation from Step 2 in the process is added to variation from Step 1 and so on down the line to the final step. To the best of my knowledge, there has never been a process step that reduced overall variation. The variations introduced at each process step directly affect the tolerances of the final product. Reducing these variations at each step will lead to less variation in the final product and the ability to hold tighter tolerances. So, the key to managing cumulative tolerances is controlling the total accumulated variation at each process step.

Given the number of processes (drilling, etching, plated through-hole, etc.) required to produce a finished PCB and the numbers and types of process variations involved, I find it wondrous indeed that anyone manages to get any product out the door at all. I will limit this column’s discussion to the etch process, which, of course, is my area of expertise (and comfort zone). I will also limit it to the basic PCB etch process of surface prep, lamination, exposure, developing, and etching, and give some hard-earned insight on things to look at to reduce the total process variation. (Note: Semi-additive process panels have their own process variations that are better referred to the plating engineers).

We will consider the final etched and stripped board as the finished product for this process. It's also the first place where a detailed inspection can easily determine whether the product meets specs. As a result, there is a tendency to focus efforts on reducing process variations in the etch process. This is not unjustified as the etcher is the major source of process variations for the etch process, but keep in mind it is not the only source of variation. Let’s take a quick look at some of the variations that can occur at non-etching process steps.

Step 1: Foil Thickness

Foil suppliers typically promise a maximum variation of ±10% in foil thickness. Using various non-destructive thickness testing methods (Beta backscatter, XRF, and eddy current) over the years, I have found that the actual variations are closer to 3–5% so the suppliers are clearly giving themselves some wiggle room. Normally, this doesn’t make much difference in the final product, but as line and space requirements drop to 25 µm or less, this may need to be considered. There isn't much you can do about incoming foil thickness, but you should be aware of it.

Step 2: Surface Prep for Lamination

Typically, this consists of spraying an alkaline solution of some sort to remove any oils and/or organics that could interfere with resist adhesion, followed by a micro-etch to roughen the copper surface and provide a slightly acidic surface for improved dry film adhesion. Although the micro-etch only removes the top few microns of the copper surface, it can contribute to the copper thickness variation. Again, this usually has little effect on the final etched line width but may be an important contributor as lines and spaces fall into the 25 µm level and below. I have been using reverse treat (RT) foil (laminated with the rough side up). It’s more costly, but it eliminates the need for a micro-etch to roughen the surface of the copper and any variation of the foil thickness that this might entail.

The dry film suppliers give a range of laminator roll temperatures, drive speeds, air pressures, etc., with recommended values for each for best lamination results. I generally follow the recommended settings for copper foils with good results. I recommend checking the exit temperature of the panels frequently to determine immediately whether your settings are okay; infrared heat guns are relatively inexpensive. Dry film suppliers also give a ±10% variation on film thickness. I suspect there is a fair amount of wiggle room here also.

Exposure units are best left to the unit supplier, but if you’re using mercury vapor lamp exposure units (probably still most of us), getting a radiometer to check energy levels across the exposure tray is a good investment. Uneven exposure across the panel does not help final tolerances. LDI and LED units are another thing entirely. We have just gotten the use of a high-tech LED exposure unit for a government project (super-duper, ultra-galactic top secret, so don’t ask). Talk about a steep learning curve! I might have more on that later.

Step 3: Developing

This is the simplest, most foolproof step in the process. If the exposure has been done properly, then the working window is extremely wide. Almost anything that doesn’t leave resist in the traces or lift the resist off the panel is okay. For good results, you just need a simple pH monitor. Money spent on planned developer improvements might be better spent on improving the exposure process.

Step 4: The Etch Process

We get to the etch process just as we are running out of column space, but that would take a book to write about, and I have no plans for a book at the moment. I find doing a monthly 1,200-word max column quite enough of a challenge. However, here’s a tip, gained from long experience, that will save you some time: Simply changing spray nozzles in the etcher or changing etch chemistry will not improve your etch tolerances. You may get a faster etch rate or better etch factors, but the etched line width tolerances will not be improved. There is no process step I know of that improves total variation. As the computer boys say, “garbage in, garbage out.”

This column originally appeared in the March 2026 issue of I-Connect007 Magazine.