You are here
Failure of Bonded Glass
I work for an optics company and many of its products contain lenses that are bonded together with a very thin (~0.010mm) UV curing adhesive (Norland Adhesives NOA61 is the most common). Up until recently we have never had any issues with the strength of the bond. However we have recently had a particular design fail during a temperature cycle so I am tasked with looking at why, and how to keep it from happening in the future. The temperature range isn't too scary, -40 to +85C.
The bonded surfaces are often curved so the normal bonded window calculation isn't too close these cases. I've been doing a bunch of searches but I can't find anything pertaining to this sort of condition (i.e. very thin bond between two thick pieces of curved glass).
I've tried correlating the curvature, CTE, and other parameters in an attempt to get an idea of where this design is drastically different but so far nothing jumps out. I've modelled the situation in Abaqus but the results are not near the published strength and the numbers seem reasonable. So I am assumingt that I am just looking at the problem incorrectly.
Has anyone here done an analysis for situations like this, or has seen some papers that they can point me to?
Any help is appreciated.
Dan
is the bond strength
is the bond strength temperature-dependent? Is it possible the temperature change is affecting the fracture toughness of the adhesive?
what do the failures look like? Does separation consistently occur along the convex or concave interface surface, inside the glass material, or inside the adhesive?
Hi wingnut1, I am
Hi wingnut1,
I am sure the bond strength is temperature dependent but that data is not available for any of the adhesives that we've used and to be honest I am not sure how to really test it accurately enough. But we have made many, many, bonded lenses that have gone through the same temperature cycles and have not failed. So even though the bond is probably weaker it has proven to work in th past (the vast majority of times).
The failure is consistently at the edge of the bond and generally forms a ring around the entire cicumference, though some have gone towards the center they are the exception it seems. This matches my FEA model results but the values don't seem too high. Unless there is a stress concentration happening at the edge that is ~10.
Unfortunately manufacturing department has been able to get the lenses apart in a way that we can tell which surface (either lens surface or even the bond layer itself) is the cuprit.
if you have made many
if you have made many successful bonded lenses, what is different about the one that is currently failing? geometry? materials?
The materials are
The materials are different, but not ones that we haven't used before in similar cases. This only major difference is the radius of the surfaces on this lens are shorter than our 'typical' design, but not our shortest either. That was why I was originally trying to work out a correlation between the radius and stress. But I've looked at other designs that are pretty close to this that didn't fail and, by any comparison I've thought of, they look like they should fail too...but they don't. That is why I started searching for papers looking at similar geometeries, but I haven't found any that are really even close.
I am starting to think it is really the tolerance on the radii that is playing a bigger role that I originally thought. We have some measurements before and after bonding for this lens but so far we don't have that data for other lenses. Since the bond layer is so small the error in the radii might be causing more stress in the bond, or the glass at the bond.
bond failure
Hi Dan, can you please share with me your final findings? I have almost the same problem you had.
Thank you,
-BO-