Troubleshooting Glue Joint Failure in Maple
A discussion about various issues that could cause glue-line failure in joints made with freshly machined quartersawn maple. October 2, 2007
Question
I have been unable to achieve anything other than a 100% glue line failure (2 glued pieces being pried apart with no wood failure) in a quartersawn hard maple lamination. The laminations are approximately 3/32" thick, glue surface is the quatersawn face, not the flat sawn edge, and number of laminations vary from 5 to 17. Wood has been processed by knife planer (newly sharpened) and knife planer followed by widebelt sander 80 grit. All with the same failures. Glues have been polyurethane (Gorilla, with both glue surfaces dampened prior to gluing, glue applied to mostly one surface, but experimented with 2 surface gluing, both achieved same failure), Titebond II with same failure, two-part plastic resin (Unibond 800) glue both surfaces with same failure.
All of the failures have a similar appearance - the 2 glue surfaces appear to have been coated with glue, but the glue interface between pieces is failing, or at least that's the appearance. The failed glue surface is smooth, not grainy at all.
Clamping pressure is either by a press with multiple veneer type screw clamps, or pipe clamp at same pressure I've used for years in other woods with no problems. Might the extreme density of the quartersawn face require much less clamping pressure, assuming accurate jointing?
Forum Responses
(Adhesive Forum)
From contributor F:
My first impression was a glue starved joint. But your last sentence sounds reasonable.
From Jeff Pitcher, forum technical advisor:
It sounds like both you and contributor F are on the right track here. If your clamping pressure is too great it will result in a starved joint. The adhesive choice won't make much difference if there's no adhesive in the joint. You also might want to double check your blades, as dense woods burnish easily.
From contributor B:
I don't think it's a glue starved joint, but I do suspect that open time is a likely problem since it happens with all types of glue. And I don't think poly glue is the right choice either. How long do you think the joint is open from application to completion of the clamping for the various multi-layers?
From Professor Gene Wengert, Sawing and Drying Forum technical advisor:
One possible problem at this time of year is that the wood is extremely dry and therefore very adsorptive. Time between spreading and final pressure must be very short. Try gluing up half as many laminations with shorter time between spreading and pressure. Then after these glue, glue the two pieces together into one. There is nothing unusual about q-sawn maple.
From contributor U:
I would and try a different glue. If the joint is stressed (curved), you might try something like Weldwood. It will add a little moisture to your wood, coat both sides. Also the open time is a little more. I would apply a little heat overnight, something like under a tarp with a light bulb.
Gene, I always like the comments from you guys in the Midwest, "this time of year... the wood is extremely dry." Not a problem here. We will get an inch of rain tonight, have 90 mph wind gusts and have 38' seas. We wish it was dry... Come summer, things will dry out, assuming it quits raining.
From Professor Gene Wengert, Sawing and Drying Forum technical advisor:
Sometimes I am too focused on North America, and sometimes too focused on the area east of the Rockies. Thanks for friendly, subtle advice.
From the original questioner:
Thanks for your responses. Regarding too long open time, I get the same failure even when I glue a simple two piece test sample. These are small and are glued and finish clamped within 2 or 3 minutes, so I'm not sure that's the problem.
I'm currently running another batch of test pieces, various glues, all low clamping pressure. As well, I'm testing National Casien's PC2000, which was recommended by a maple butcher block manufacturer.
From Professor Gene Wengert, Sawing and Drying Forum technical advisor:
People do not have trouble gluing maple, so something is not right in the way you are processing the wood.
Are all the surfaces freshly prepared? Are you certain that you are not damaging the wood surface fibers when you use the 80 grit sander? Sometimes it has been shown that such an operation will loosen the surface fibers so that although the glue joint is indeed strong, you have glued the loose fibers on one piece to the loose fibers on the other. When the joint breaks, it looks like 100% glue failure, but if you look under magnification, you will see that there are a great number of fibers. This damage is called "subsurface damage." Try gluing from the planer without sanding.
The best surface for gluing is always a knife-prepared surface (unless it is burnished). The rough sanded surface often does not have high integrity and also has high glue adsorption due to the roughness.
If the saws were over-heating the wood, you would see darkening, so I discount this. Do you have squeeze out? You need to have squeeze out throughout the joint edges, indicating that you used enough glue.
From contributor D:
When you wet the wood surfaces, let them dry before you apply glue. That is, let the dark of the fresh water disappear for a few minutes before you spread glue. Simple, but often overlooked. When urethane glue first came out, I did tests where I wet both pieces of wood and immediately glued and clamped - they all failed, and I didn't use urethane for a year or two until I tried a different method.
If I were in your shoes, I'd go to different glues first, with my faith in urethane shaken.
Do us a favor and keep us posted. With this many plausible suggestions, it will be interesting to hear how it shakes out.
From the original questioner:
Here are the results of my last test. Same quartersawn maple used on all samples.
1- Fresh from planer (new knives) - National Casien PC2002 - applied 1 side - clamped within 2 minutes - light clamp pressure - 24 hour cure 70 degree cure room. Result: same 100% (or apparent 100%) failure.
2 - Planed and sanded at 80 grit - polyurethane (Gorilla) - glue surfaces wiped down with acetone and left 15 minutes to evaporate - surfaces dampened - glue applied top one side while surface still has the dampened appearance - clamped within 3 minutes - light clamp pressure - 70 degree cure room for 24 hours. Result: same failure as all the rest.
3 - Planed and lightly hand sanded with 60 grit-cross grain - Titebond II heavily applied to 1 side - light clamp pressure - same cure temp/time as all above. Result: glue joint appears to have held, wood failure leaving a tearout pattern similar to tearout pattern that sometimes happens in planing quartersawn maple.
I'm curious about the failures, so I'm going to continue testing the above failed joints by reapplying glue to the failed glue line, wondering if that glue line would serve as a glue size.
Gene: the thought about contamination makes sense, but I need some empirical way to figure out what/where it's coming from (potentially) and how to deal with it. I will try wiping down knives/bed/rollers of planer with acetone. The infeed roller is rubber (or rather I think some type of polyurethane, yellow color) which I set this planer up with when I rebuilt it (Delta rc22 circa 1960's 2-1/2" planer head, 3 knife). Planed surface, even with fresh knives, has a polished appearance on hardwoods in this planer.
For the record, I took some South American mahogany, which is what I usually glue up here, straight from the planer (same slightly polished appearance), straight to any of the glues, and got excellent bonding. Mahogany has all that nice open grain.
From Professor Gene Wengert, Sawing and Drying Forum technical advisor:
So, I am tempted to conclude that your knives are burnishing the wood and that your 80 grit sanding is creating loose fibers. If this is true, then light sanding by hand after planing will eliminate the burnishing and not create loose fibers or sub-surface damage. Burnishing is a problem with maple and other dense woods, but seldom with softer wood like mahogany. Can you try faster feeding in the planer? Make sure you do not take a very light cut, as this burnishes the fibers rather than cutting them. Do you know your clearance angle? Do you know if the knives have been jointed?
From the original questioner:
Gene, with the machinery I have, taking the stock down to 3/32 has to be done in light passes or the stock will be completely chewed up. So, that is definitely an issue. One interesting point on the burnishing is that even after going through the 80 grit version of the sanding (wide belt, not hand) the sanded wood still has a bit of a shine to it, though not as pronounced as with the planer, but a surprising amount nonetheless.
This planer is not equipped with jointing fixtures, so the knives have not been jointed. I'm not sure on the clearance angle, but there is only a primary bevel, no secondary bevel, and no jointing, so I assume I'm not pounding the surface with negative clearance. However, the planer has only one feed speed and it's quite slow at 12ft/min.
The more I think about it, the more I think burnishing is the problem, so I will try an experiment. I'll run a sample of the same stock at a faster feed rate by hand through the jointer and glue it up.
I've been talking to a local shop with a gang oscillating, I guess you'd call it a jig saw. They produce 1/8" sawn veneer faces for door stiles, and he calls the sawn finish a surface ready for gluing. My question is, does this sawn face present the same ripped grain problem presented by the wide belt sanding? Is it actually a decent glue surface? If it is a decent glue surface, is glue applied more heavily than a planed surface?
From contributor W:
I think we are getting somewhere through our collective vast experience. Burnishing is a frequent problem, with sanding belts as well as knives. In my stroke sander I notice burnishing when the belts start to get loaded, long before they are dull or the loading is very evident visually. Other variables are feed speed and belt speed. I check for burnishing by looking along the surface at a low angle toward a far away light source, with the work piece in a dark corner. 3/32 is I think a difficult thickness in any planer running hard maple, so the need to take light cuts may produce burnishing. Look to a sanded surface with a fresh belt at a slow or hand rate.
From contributor N:
Perhaps an obvious comment - is it warm enough in your workshop? We glued up some doors with polyurethane last week and all the joints failed, with hindsight because it was only 45 degrees F in the workshop. PVA joints at the same time chalked and failed too.
From the original questioner:
No, the temp is controlled in the shop, plus we have a conditioning/drying room which we set for 70-75 degrees for glue cure. The problem is burnishing. Further, the density of this wood combined with the burnishing made clamp pressure more sensitive than it should have been; the glue could not be absorbed by the wood surface and the only place it had to go was out of the joint.
Removing the burnishing is time consuming and only partially effective. The problem I'm trying to resolve for the future is how to produce this 3/32" sawn quartersawn veneer without burnishing. All the machinery I have (including local subs) will create either a burnished or compromised (wide belt sanding) glue surface. The only option I can think of is the oscillating gang band/jig saw I mentioned above, assuming this would give a suitable glue surface. Anybody know anything about quality of cut these things give relative to appropriate glue joint surfaces?
From contributor Y:
Holy cow, gluing maple shouldn't be this hard! Now I'm worrying about maple glue joints on work I've done in the past. We've often glued sanded maple together, and haven't had any of the issues described here.
One thought to add to the mix: It could be not enough clamping pressure. Reviewing 'Understanding Wood,' it says that the denser the wood, the greater the pressure required for an adequate joint, recommending about 200 psi for maple. I quote, '...gluing up one square foot of maple requires pressure of 28,800 pounds. Over 14 tons! This would require, for an optimal glue line, 15 or 20 C clamps, or about 50 quick-set clamps. Conversely, the most powerful C-clamps can press only 10 to 11 square inches of glueline in maple.'
From Professor Gene Wengert, Sawing and Drying Forum technical advisor:
If a wood surface is slightly burnished, it can be restored by simple light hand sanding with 120 grit (I do not like the coarser grits).
There is a pretty good test to see if the surface is active for gluing. Put a water drop on the surface and it should disperse within a minute or two. If the drop is still standing like water drops on a newly waxed hood of a car, then you will have gluing problems.
From the original questioner:
In reading through the entire section of "Understanding wood" that was referenced, Hoadley says that "if mating surfaces were perfect in terms of machining and spread, pressure wouldn't be necessary. The rubbed joint attests to this. But unevenness of spread and irregularity of surfaces usually require considerable force to press properly."
I believe the clamp pressure on dense wood that you reference relates to the force needed to bring dense unyielding lumber into intimate contact. Force is not necessarily required to force the glue into the cell structure. At least that's my reading of the passage you mentioned.
In this case, and especially on the failed sample pieces which were simply 2 very small pieces of 3/32 veneer, the appropriate amount of force required would be the amount required to bring these two very flexible pieces into intimate contact, i.e. not a heck of a lot.
Subsequent tests have all pointed to burnishing as the problem. The most successful burnishing removal system so far has been a well tuned, freshly sharpened plane. I have not tried Gene's 120 grit hand sanding yet and will experiment with it so I have as thorough an understanding of this business as possible for the next time.
From Professor Gene Wengert, Sawing and Drying Forum technical advisor:
You are correct in that the glue is not forced into the wood with pressure. Instead pressure forces the two pieces close enough together, squeezes glues into the nooks and crannies, and squeezes out any excess glue.
From contributor I:
I have been following this discussion and have hesitated to contribute because my experience is with thin pieces for musical instruments. I glue up highly figured hard maple for musical instruments and I've always used hide glue from a glue pot. I thin it down a bit for hard maple, thin compared to what I use for mahogany, soft maple or walnut, and size both pieces with hot glue, then put glue on both and clamp them together. I leave them clamped for 24 hours. I've got instruments out there that have been in use for 15 years without a glue failure.
From contributor K:
Gorilla glue is not all it is cracked up to be. PVA/EVA glue up on maple should be simple, but when we are having issues with a species, we use a PUR hot melt and have so far not been let down. PURs are expensive, requiring a special hot melt gun dispenser, and are very thick, making fine glue lines difficult. They are not well suited for glue ups requiring over 30-60 sec of open time, and are hard to use on edge glue ups over 1" thick, but they will stick to virtually anything, with minimal pressure.
We have noticed that prepping a glue surface on a wide belt sander tends to create a convex gluing surface. I assume that this is due to slight buildup on the belt in the center of the prepped surface where the majority of the stock is being sanded versus less build up on the edges of the surface, allowing slightly better contact with the belt and more material to be removed. This has never given us a good clean joint. It's like trying to glue two round surfaces together.
Aside from shelf life, temperature being too low (does your glue chalk?), moisture between 7-11% on most PVAs, and burnishing I wouldn't know. If you continue to have problems I would get up with Gene Wengert down the road and have him tour your plant - he knows what he is talking about, and his structural criticism can save you thousands in callbacks down the road.
From contributor W:
Now we're touching on something I've always wondered about. Does the convex surface result when the belt is fresh? And is it true on small sections, say the face frame stiles run on edge, with the 7/8 inch dimension being sanded? Do you run frame members, sanding through the joint, with nice tight glue joints resulting?
From the original questioner:
Something you might find interesting regarding this subject: these 3/32 hard maple laminations are part of a piano bridge, a complicated affair that has to withstand considerable cracking forces from many small pins driven into it. In talking with other piano technicians who do this type of work, when I mentioned the problem I was having, to a man they knew exactly what I was talking about. They all still deal with these laminations by means of some kind of "toothing" cutter, i.e. modified, grooved chisels, etc. And interestingly, my understanding is that Steinway still prepares all their laminations similarly, except with a dedicated rolling, knuling toothing cutter. I'm not sure if this is a remnant of 19th century practice that they are sticking with because it has worked, or because they have empirically proved it, but there it is.
Hoadley seems to feel in "Understanding Wood" that toothed planing such as described above does not improve adhesive bond quality.
From contributor M:
We do several types of laminated glue ups. The species are oak, maple, santos and jatoba. All glued with a PVA. This includes a radiused bar rail. Widebelt to 120, then glue. Width of rail is 3.5". Two legs of 12' plus the radius. No problems. If not beaten to death by now, burnishing sounds like the culprit, unless you tighten the clamps to the point of no more.
From contributor K
Never tried on a 7/8" surface, but yes, the belts are fresh, either 80 or 100 grit. We edge glue 5/4" stock primarily. All 5/4" glue up surfaces come straight off the straight line or gang rip saw. We experienced the problem on face glue ups on newel and baluster blanks. Surface being prepped was 2" - 6.5". How we were getting the problem is only my personal theory, but since we changed to prepping wide stock on a finish planer, the problem went away.
From Professor Gene Wengert, Sawing and Drying Forum technical advisor:
Wouldn't the tooth cutter increase surface area and thereby increase joint strength? Perhaps Hoadley is thinking that a good joint is already stronger than the wood, so a better joint is of no benefit.
From contributor Y:
I think Hoadley was talking about what he called "toothed planing," where the grooves serve to roughen the surface and maybe help glue spread. Since the grooves wouldn't correspond to raised ridges on the mating surface, there is no increase in mating surface area, but in fact probably the opposite.
From the original questioner:
Perhaps in a piano's multiple laminations which must survive generations, the toothing is not so much to increase the adhesive bond, as to be redundant. Perfect or near perfect joints in all of the many laminations are pretty hard to achieve with 100% success rate, as I have recently experienced. The cost of fixing a delaminated piece, especially in the soundboard system or case, would be prohibitive, not to mention really piss off someone who has paid $30,000 + for a piano. It would be interesting to try and prove whether this practice is useful in this particular case, or still a waste of effort.
The comments below were added after this Forum discussion was archived as a Knowledge Base article (add your comment).
Comment from contributor T:
A fast way to check the gluing surface is to place a drop of water onto the gluing surface. If the water is absorbed into the wood, the adhesive will penetrate. If the water just lays on the surface, the gluing surface could be burnished.
Comment from contributor A:
Actually, 12 ft per minute is zooming right along:
12 foot = 144 inches, and 144 inches in 1 minute = 144 inches in 60 seconds, which is 2.4 inches per second. If the machine is direct drive and the motor is running at 1725 rpm then those 2.4" per second represent just 28.75 revolutions. If there are two knives, that's just 57.5 cuts per inch. Rough planing is usually done at about 60 cuts per inch and finish planing at 90 cpi. So at just 57 cpi your planer is zipping right along and burnishing is probably not going to happen.
