I guess the answer to that also depends on whether you’re intent on using actual plastic glue, such as Cyanoacrylate, or plastic cement. In the case of crazy glue, you’re adding material to create the bond so a channel might help. In the case of cement, you’re using a solvent-based material that will literally dissolve the two materials and weld them together. So in that case, you want to do the opposite and have not only no channel but print on a smooth plate where the bonding surfaces will mate.
If you’re using glue, then I suppose the amount of channel needed would be a trial and error. 0.1mm sounds as good as any place to start, but then again, will the resolution of the filament produce a pure channel or just a slight indent?
Personally, I have found that glue and even epoxies perform very poorly with plastics. Cement, on the other hand, produces a true chemical weld. I have conducted experiments with both. When you bond two items together, the cemented bond is actually stronger than the surrounding material being bonded. I’ve tested it, and when you try to pry two cemented materials apart, the surrounding plastic fails before the bond does, just like a metal weld would.
The material I found to be the best for PLA, PETG and PC is the following:
However, if you ever try to adhere dissimilar plastics or even attempt to bond TPU to plastic, through trial and error, I found this glue.
It works because it is designed to bond dissimilar plastics together. Although it does not claim to support PETG, I have successfully used it on Silicone/PETG and TPU/PETG combinations. One such project was to create a silicone wiper similar to the one used in the A1 for my P1P. It actually bonded so well that the bond was stronger than both the silicone and the PETG that supported it. The silicone tore first, leaving behind a patch of bonded material. This proves that the glue was stronger than the product it was bonding.