Glue Line “Creep” Study
Analysis of Factors Contributing to Glue Line Transmission “glue-creep” in Wood Joinery
Glue line telegraphing (glue-creep) occurs on exposed wood joints as a thin, slightly raised bump along adjacent glued surfaces. This study was designed to evaluate what factors may contribute to this taking place for wood to wood applications. I tested a number of scenarios across a variety of six wood species, seven adhesive types and four finishes. Over a six week period of preparation, glue-up, finishing and acclimation within a stable environment, no visible glue-line creep appeared. Future seasonal changes and wood movement may change the initial results of this test. Subsequent test samples may be created to access other factors relating to wood grain orientation, shorter cure-to-finish time and application of water-based finishes may also produce visible glue-line creep.
A great deal of speculation exists around glue line telegraphing in wood, also known as “creep”. Creep occurs once a glue has dried along a wood joint where the glue line is exposed. The “line” can often be visible, but is more likely something that can be felt by hand along the surface where two wood boards have been joined together such as the seam of a table top or door panel, or other exposed joinery elements like dovetails. For instances where the glue itself is hidden within a joint like a mortise and tenon, it is likely that some of the same forces at work for exposed joints are at work, but because it is an internal joint, it can not be visibly observed or felt with the hand.
Woodworkers, contractors and consumers in general have identified glue creep as an occurrence that takes place across a wide range of situations. While adhesive types are typically the focus of why it may or may not occur, adhesive types do not show to a consistent influence in preventing glue creep from taking place. In fact, at the time of this article, Google search results of “Glue Creep” from page one provide a list of results that definitively attribute glue creep to no less than 5 different manufactured adhesives, along with a series of environmental, mechanical and chemical mechanisms that make glue creep occur.
Suggested Mechanisms Contributing to Glue Creep found in Variety of Publications included:
- Moisture content of wood
- Temperature at glue-up
- Humiditiy at glue-up
- Cure time
- Seasonal wood movement
- Assembly location/Final location
- Clamping force applied
- Porosity of wood species
- Method of adhesive application
- Clamp time
- Type of wood species (oily)
- Grain orientation of adjacent pieces
- Surface Prep (plane, scrape, sand)
- Type of adhesive (PVA, Polyurethane, CA, Epoxy, Urea, etc)
- Finish type
- Combinations of adhesive and finish
- Brand (adhesive formulation)
Because of the wide array of theories along with a lack of any scientific methodologies, I set out to consistently test a limited number of influencing factors that represented a more practical “real world” set of scenarios. By testing adhesive types, wood species, finish type and environmental changes over time (seasonal movement), I’ll be testing the four most common and controllable factors for most applications.
The duration of this test from start to the date of this publication has been 6 weeks. The test utilized 7 different adhesive products, 6 species of domestic/tropical hardwoods along with 4 types of finish (including one unfinished control). Each wood species tested was from a single board, and had consistent quarter/rift sawn grain orientation for every piece. Consistent, uniform clamping pressure was applied by hand, and evey sample was given a full 24 hours of cure time before clean up, sanding and finishing.
Currently, none of the six wood samples are showing any sign of creep. The pieces have remained in a below ground basement workshop, which maintains a fairly consistent temperature of 68 degrees year round, and a relative humidity of ~35-45%. If glue creep is to occur moving forward, I speculate that temperature and humidity level changes of the seasons or of location are likely to influence subsequent glue line creep.
Furthermore, reducing the glue-cure time prior to sanding and finishing may also impact glue line creep. I suspect unclamping after 2-3 hours and moving immediately to sanding and finishing may produce a different result as well.
Materials and Methods
All wood samples mositure content was measured below 12% at the time of milling. Material consisted of:
- 4 domestic hardwood species: Poplar, Walnut, Cherry and Ash
- 2 tropical hardwood species: African Mahogany and Purple Heart
Each species was cut from a single board and milled to 3/4” thick and 14 “ long. Then, eight equal rip cuts were made from each board at 7/8” thick, allowing for 7 individual, adjacent glue lines per species. Every surface that was to receive glue was also given a few light passes with a hand plane to ensure an even and flat surface across all pieces. Every sample sat for a minimum of 24 hours prior to glue-up.
Seven different glues were applied according to the manufacturers instructions. The brands included were sequentially applied in the following order for each wood sample:
- Gorilla Glue (Polyurethane)
- Gorilla Wood Glue (Type II PVA)
- Titebond I (PVA)
- Titebond II (PVA)
- Titebond III (PVA)
- Titebond Liquid Hide Glue
- Elmer’s Wood Max (PVA)
Each board was glued up and clamped using three equally placed bar clamps. Every board was given a full 24 hours of cure time before unclamping and 48 hours before scraping and sanding the surface with a random orbiting sander with 150 and 220 grit. Another 24 hours passed following sanding before the first sample of finish was applied. Next, each board was broken out into four total surfaces for finish application, two halves for each side. Commercially available finishes tested included:
- Minwax Wipe on Poly (satin finish): two coats applied with a clean rag, with light sanding between coats.
- Watco Danish Oil: two coats applied with a clean rag, wet sanded to fill in the pores, with excess removed per manufacturing instructions.
- Zinsser Bulls-Eye Shellac: two coats applied (spray finish), with light sanding between coats.
- Unfinished surface: serving as a finish control group
Across all six wood samples, none of the seven glue varieties produced noticeable glue creep. Although some of the lighter and more smooth grained woods produced a visible glue line (particulalry cherry and poplar), neither was consistent from one glue type to the next, with exception to the darker-tinted Titebond III.
Several factors may ultimately contribute to the current samples of this test to eventually producing glue creep:
- Fluctuating changes in temperature or humidity due to seasonal wood movement. Wood is an organic substance, and as a result, it is susceptable to move because of that.
- Subsequent interaction between the finish and glued surface, pushing the glue line to “creep” above the surface of the wood.
- Further changes in the chemical composition of the glue over time, impacting its viscosity, maleability and bonding properties.
Additionally, other factors relating to the preparation of material during the fabrication process may also drive glue creep. These factors would need to be tested in subsequent scenarios on new samples of wood. This could include, but is not limited to:
- Duration of clamping and curing time of the glue: Shorter periods of clamping and cure, prior to sanding and application of finish may inadvertantly impact the cure and stabilization of the adhesive, and as a result force the glue to expand out of the joint following the finish process.
- Overall orientation of the individual boards grain direction: This test included adjacent pieces glued up from the same board. None of the samples contained end grain orientation that was perpindicular to the next. An extreme example of this would be to have one board in a quarter sawn orientation (grain direction perpindicular to the board’s face) with the adjacent glued up board in a flat sawn orientaion (grain direction parallel to the board’s face). This would represent two extremes of tangential movement along the wood’s cellular structure. It is important to note here, that it would more likely be the wood’s contraction/expansion, versus the glue iteself moving – thus causing the glue line to “creep”.
- Board joinery: The overall quality and squareness (90 degree) of edges on adjacent surfaces. If the one edge of the board is out of square, a larger portion of glue may be exposed to the leading edge, causing the line to be “pushed” out over time.
- Clamping pressure: too little pressure may prevent the adhesive from producing an optimal joint between the two surfaces. Conversely, too much clamping pressure may ultimately “starve” the glue joint and push all of the glue out. It’s more likely that this would produce a sub-optimal bond, versus glue-creep.
- Use of a water-based finish: while this may potentially impact the water based PVA products, it’s unlikely that this would impact the Polyurethane or Hide glue products.
- Wood moisture content: if the wood’s moisture content is not in equilibrium with the surrounding environment, further drying and shrinkage would potentially produce a similar effect of the glue line to creep out, when in reality, the wood iteself is moving.
Moving forward, the current samples should be monitored on a monthly basis for a period of 8-12 months to expose them to a full cycle of seasonal changes. Subsequent tests could include testing a greater variation of grain orientation in the wood pieces that are glued together as well as sanding and finishing the samples before allowing a full 24 hours of cure time.
I’ll post an update on my progress testing a few other factors in the coming months, as well as look at the overall effect of seasonal changes. I’m open to discussion from other folks on their experiences.
* Author’s note: I was hired by Gorilla Glue to conduct this study as part of an ongoing effort to address customer service questions, however, that did not impact my methodologies or analysis of the results from this first round of testing.