Finish Failure at Door Frame Joints

What caused this defect: Wood moisture, glue characteristics, finish issues, or all three? August 17, 2009

Can anyone tell me what causes the finish to crack and start to chip (but only where the rail meets the stile)? Is this an issue with the finish or the joinery?

The wood is maple and the rails/stiles are joined with glue only. My finish schedule is sprayed on dye stain, pigmented wiping stain, toner, Mohawk vinyl sealer, and 2 coats of Mohawk pre-cat lacquer.

Click here for higher quality, full size image

Forum Responses
(Finishing Forum)
From contributor S:
It shouldn't have anything to do with your coating. Picture is not real clear but it looks like possibly wood movement at the join. What type of glue was used? Are you certain there wasn't any glue left on the face before spraying? Are all the doors exhibiting this?

From contributor J:
I have had this happen for two reasons - one, moisture ran down the door panels and settled in the cracks and corners, and two, excessive wood movement. But most of the time this happens because of moisture.

From Professor Gene Wengert, Sawing and Drying Forum technical advisor:

I also vote for a moisture change that caused wood movement.

From contributor R:
It's hard to tell from a picture but it looks like there is a quirk at that joint. Did the edges get eased? If so, this could be finish that has puddled in that void and exceeded the max mil thickness, also called mud cracking. Otherwise I'll go with moisture.

From the original questioner:
Thanks for the responses.
1) I'm using Titebond II glue.
2) I don't think I left any glue on the surface. After I clamp the wood, I use water and paper towels to wipe the squeeze-out off. I also sand quite a bit after gluing.
3) Not all doors do it. It may not have anything to do with it, but the doors that do it the worst are small, square doors that are rarely utilized. It occurs less on the more traditional doors where the stiles are much longer than the rails (and also receive more use).
4) The RH should be pretty standard in that it is in a house that has always been inhabited. I have never seen moisture pooling on the doors.
5) The edges are not eased - it is a smooth joint.

If I am getting movement, would nailing in addition to gluing help? I've added another picture, because often it seems to start at both the top and the bottom of the joint and work towards the middle.

Click here for higher quality, full size image

From contributor M:
It is definitely due to cross-grain swelling and shrinking from humidity changes. Your last photo proves it. The center of the rail is staying put while the two edges are swelling and shrinking outward from the center. When a cross-grain piece (rail) is glued to a with-the-grain piece (stile), of course the movement will break the glue bond first where the movement is the greatest - at the edges of the rail.

From contributor E:
I'll add one more thing - you may want to avoid cleaning the joint with water and paper towels. This could allow water to weaken the glue near the surface of the joint. I always wait until the glue has settled and becomes gummy, then just shave it with a chisel. It's quicker, cleaner, and won't adversely affect a joint.

From Professor Gene Wengert, Sawing and Drying Forum technical advisor:
Avoid using the water washing. The last photo is a classic example of moisture drying out. To avoid the moisture problem, it would be important to ascertain the MC accurately before assembly and make sure that the RH or EMC of the air in the manufacturing plane agree within 2% MC with that MC.

Perhaps you already know that the top and bottom edges of a rail have more exposed wood, so they dry out faster (if the MC is a bit too high and the plant's air is a bit drier) than the center of the piece. With the drying comes localized shrinkage, which can open a joint. The forces can be so large that they can break a glue bond. Also, the piece that the rail is joined to has the grain running at 90 degrees to the rail, and so that piece does not change length with changes in MC. So, one piece is trying to move and the other is not. Stress builds up and the joint opens.

The key is to make sure that the MC is correct just before assembling, because without MC changes, there will be no movement. Any MC measurements after the problem shows up will not tell us if the wood was wet prior to gluing, so measure the MC before gluing. Use a digital moisture meter; in this case, a pin type meter might be best. Use a name brand meter, costing about $200. Remember that any readings under 6.5% MC are not accurate. Use a $30 RH meter from Radio Shack to monitor the plant's RH. We do not need a more accurate RH meter.

After finishing, there will still be moisture changes (no finish is an excellent moisture vapor sealer), but they may happen more slowly with an excellent finish. You didn't ask, but I think that a hot melt PUR adhesive should be evaluated for this application. Make sure you consult with the PUR manufacturer to find out the proper pressure, curing time, open time and so on.

From contributor G:
The first photo to me looks like glue starvation on the edge of the panel groove. Hard to get glue here without having the panel stick, even harder to remove the squeeze out. Also make sure you are not sanding too soon after assembly. If the joint is sanded flush when still wet, it will leave a raised joint later and possibly lead to what is shown in the photos.

From contributor A:
Titebond 2 can cause a lot of problems. First it cold flows/creeps. Second, I just took a look at a cherry clock case I whacked together about 10 years ago. I glued it with Titebond 2 and then gave it a couple of coats of Minwax poly. Within a short amount of time the glue crept out of the joint and turned reddish. It's a face grain/edge grain joint like you would have on a face frame cabinet. I only use Titebond 1 for interior. Titebond 3 for exterior. Titebond 2 is obsolete.

From Professor Gene Wengert, Sawing and Drying Forum technical advisor:

Adhesives creep under load, and not just sitting around. So doors, tables, cabinets, and similar will not experience adhesive creep, as they are not under high loads continually.

All the Titebonds will behave similarly with regard to creep, but unless you have a joint under a high load, you will not see any creep. What some people call creep is not creep but is the result of size change when the MC changes. Moisture size changes are very large compared to creep size changes. Creep occurs over a few years, not within a week or two or even several months.

Note that TB I is not as resistant to wetting as TB II or III. But again, a cabinet door is not going to get wet long enough and continually to merit using TB II or III. All the TB adhesives will make a joint stronger than the wood will almost all adhesives. So why spend more money if you cannot take advantage of the extra or special properties?
The above discussion appeared earlier in the Adhesives Forum and was reviewed and supported by Jeff Pitcher, the forum technical advisor.

As a test, I put three large drops of TB I, II and III on a piece of wood and then after it dried have hung it so that gravity will make it creep. So far, nothing. I am not sure where the creep information for an adhesive in furniture, cabinet doors and the like comes from. Certainly, a joint can creep in a laminated beam under stress (but it takes time) and it must be a carryover from this concept. Wood also creeps, as many of us have seen when looking at old flooring beams in a house or barn. Also, a heated joint with some PVA adhesives can creep under stress as the adhesive can soften. Similarly, moisture softens some PVAs. Nevertheless, moisture movement is far greater than creep movement.

One approach that has been tried to avoid this problem, especially with a wider bottom rail, is to pin the joint with a fastener at top and bottom. Unfortunately, when the rail loses moisture and shrinks in width, the fasteners hold real well and the wood will crack between the two fasteners. I have also seen a design (mortise and tenon) where the amount of wood (thickness) of the stile on top of the tongue from the rail is very small. This allows the rail and stile joint pieces at the face to move together more easily. In any case, the joint needs to be prepared within minutes before the pieces are glued to avoid any size changes due to unequal MCs and EMCs. The moisture of both must be the same and should match the shop's EMC and the customer's EMC, so any moisture changes are minimal.

From contributor J:
Titebond 2 has a certain amount of flexibility to it. I would dare to guess that it is far more flexible than the coating that was put over it. That said, Titebond one is more brittle (less flexible) and would to my thinking allow less flex (or expansion/contraction) at the joints. I don't think this is a matter of science experiments, as people have been successfully making cabinet doors long before this thread. A less flexible adhesive should help to some degree and while this case may not fit all of the qualities of cold creep, I believe TB2 is allowing too much movement directly at the joint.

From contributor A:
Gene, I am using terms that most woodworkers typically use to describe glue behavior. I am quite familiar with structural/engineering terms and definitions like creep. Creep is basically how much a material stretches when under sustained loads up around its yield strength. For instance, Kevlar fibre does not creep. Spectra (UHMW Polyethelene) fibre has an amazing amount of creep. Stretching Saran wrap in your hands is a great example of creep.

What we see in practice is the wood parts are actually compressing the dried glue out of the joint, causing an obvious raised area. Allowing it to dry for long periods of time (2 weeks) then sanding it flat will often result in another raised glue line in the future. I have seen this a thousand times in my career.

White glue is the worst (nice and gummy)
Titebond 3 behaves much like the white glue
Titebond 2 will get squeezed out of joints
Titebond 1 will not.

I will take a photo of my cherry lamp if need be. If the glue joint was creeping then there would be a depression rather than raising.

From Professor Gene Wengert, Sawing and Drying Forum technical advisor:
First, let's clarify that what is referred to as "white glue." When the second generation of PVA adhesives came on the market, they colored them. Not to lose a market - people making the original white glue also colored theirs. So, glue color is not an indicator of what type of PVA we are dealing with.

Second, there are at least two reasons why a glue joint is raised. One is that the glue joint itself is rather rigid, so when the surrounding wood dries and shrinks a bit, the area right at the joint does not (as the glue makes it rigid and unyielding). Second, when wood expands, especially the surface while the core is not changing, the surface will create a raised bump at the joint area. We see this often with a t&g floor. Third, if heat is involved, it is possible for the adhesive to flow and any excess to be squeezed out of a joint under stress.

Of the three TB adhesives, TB II and III are more stable with heat and moisture. TB I will be the one that is softer than the other two, but all three are very rigid.

A good glue joint has an original thickness of 0.002 to 0.006 before the moisture diffuses out. After curing the joint is even thinner and the amount of adhesive that is not chemically tied up with the wood molecules is extremely small. The amount of adhesive available to squeeze out is minute.

Again, the wood is going to move much more than the adhesive, even if there were forces trying to make the adhesive creep. It is the wood that is moving and that "creeps." Contributor A, I have no doubt that you have observed a glue joint issue 1000s of times. I do believe we are talking about the same final results. We seem to be using different terminology for the same event. I do believe that my explanation is a more accurate description of why this is happening.

From contributor A:
Gene, I concur wholeheartedly.

From contributor N:
I have had the same problem with a particular tinted pre-cat lacquer. Have you tried a different finish? Maybe one with a little more flexibility?

From contributor L:
I believe the whitish marks are a breakdown of the pre-cat lacquer due to moisture. The moisture can be either acting upon the inside surface of the lacquer or the outside surface, or a combination of the two, especially if it is in a kitchen where the atmosphere can get steamy, surfaces can get excessive cleaning, and the lacquer is of insufficient thickness to resist passage of moisture through the slightly porous pre-cat lacquer. A much safer approach is to use much more impervious acid catalyst lacquer, especially in a kitchen or bathroom environment. Pre-cat is neither impervious enough on the one extreme, nor breathable enough (like an oil finish) on the other, so only suitable for well-dried timber in dry environments.

From contributor B:
This looks like moisture damage. The reason I say this is due to the white haze in the film, the trademark of water under your film. My first two guesses for this are:

1: Water dripped down into the floating panel and got into your joint.
2: The door got stained, and stain fell heavy into your joint. Not a big deal if you let it dry overnight, but if the door was sealed before that heavy glob of stain in that joint dried, then you're gonna get this problem.

If the stain caused it, then chances are this flaw won't show for a day or two after it was sealed. If a few days pass after sealing, and the join doesn't haze like that, then it isn't excess stain moisture in the crack, and chances are the homeowner tried to clean the cabinets with an overly wet sponge, and water soaked down into the center panel crack causing it, or if the cabinet is near the sink, it could be that water from the sink dripped into the door and caused it.