Estimated reading time: 4 minutes

The Chemical Connection: Through-glass Vias in Glass Substrates

This month’s theme is vias and how best to ensure via quality and reliability. I don’t have much expertise in this process area or much to contribute that most of you don’t already know. However, I’ve recently become peripherally involved in a via technology that may be of more than academic interest to some of us. It entails putting vias in a material not usually associated with PCB manufacturing: through-glass vias (TGVs) in glass substrates.

Using glass as a PCB substrate has been bouncing around for a long time, and interest has grown over the past 10 years. Glass has several advantages over fiberglass/organic resin panels typically used as substrates. Besides having excellent dielectric properties, glass has better dimensional stability over a wider range of environmental conditions, allowing an extremely tight layer-to-layer interconnect overlay. Glass also has greater tolerance for higher temperatures, so glass substrate PCBs can operate reliably over a wider range of temperature differences. There are several other advantages to glass in the design rules for power delivery, but with the higher dimensional stability and heat tolerance, the interconnect density possible with glass increases by an astounding 10 times. One doesn’t need a lot of imagination to realize that this means a lot more capability for our electronic devices in the same space. Prototype devices of double-sided through-hole glass substrate PCBs prove that the technology lives up to the theoretical expectations.

Currently, the major holdup in TVG technology is how to form vias in the glass. Various glass companies have provided glass formulations with the desired dielectric properties in thicknesses varying from 0.3 mm (0.012 inches) to 0.9 mm (0.035 inches). Glass thicknesses less than 0.3 mm are difficult to handle and result in low yields because of breakage. Using conventional drilling techniques is out of the question; the drills cause too much trauma to the glass in the form of cracking and breakage. The only viable alternative at this point is laser drilling, but that has its problems.

It takes a relatively long time to drill a TGV in glass with a laser, even for 0.3 mm glass (and we are talking about drilling millions of vias in a 550 x 510 mm [22 x 20 inch] glass substrate). That’s okay for prototyping to prove the concept, but it’s too slow for production quantities. A faster way is to use laser induction to modify the glass structure through the glass where the via is to go, then etch the weakened glass to form the TGV.

Unfortunately, the best etchant for glass is hydrofluoric acid (HF), which is not fun to work with. With the concentrations needed to etch glass, 5–10% by weight (concentrated HF is 49 wt%), it is not deadly but can cause painful burns. I’ve used a great deal of HF to etch titanium and TV and phone displays in the past 10 years. However, with the proper precautions and training, it is not hard to provide a safe working environment.

Once the laser induction is complete, we etch the vias out by soaking the substrate in a tank of HF at the proper strength for eight to 10 hours. This is still very slow, and even though the unaltered glass is more resistant to the HF, it will still etch to some extent, resulting in undesirable glass thickness variations. Limited testing has shown that spray etching will reduce the via formation time to an hour or less, depending on the thickness of the glass, with less surface thickness variation. I expect to see a lot more testing for etched glass vias in our lab in the near future. Oh joy!

Most glass PCBs are double-sided, but there are also multilayer prototypes. For multi-layers, glass thinner than 0.3 mm is desirable for obvious reasons, but also because as the glass thickness approaches between 0.2 mm and 0.1 mm, it’s possible to use the laser to create the holes for the vias rather than using laser induction and etching to make the vias. The major downside—assuming you can handle the glass without breaking it—is that the laser shot leaves microcracks on its surface. One must remove these microcracks by etching a micron or two from the glass surface with HF after the laser shot. This is the most desirable process because of its speed compared to other via formation methods, provided that thin glass handling problems are overcome. I also expect to soon see more glass thinning tests in the lab. Double oh joy! (You might get the impression from the somewhat sarcastic “Oh joys” that etching tiny holes in glass with hydrofluoric acid is not one of my favorite activities. You would be right.)

If you think you’re having fun with vias in FR-4 substrates now, looking at what’s in store in the future will put things in perspective. The advantages of glass substrates are so substantial that they will arrive probably sooner rather than later.

This column originally appeared in the June 2025 issue of PCB007 Magazine.