Diffusion Bonding: The Process (part 1)
One of the many vacuum processes to take note of today is that of diffusion bonding (aka diffusion welding or thermo-compression bonding). This technology, although highly specialized, continues to gain in popularity with design engineers and is finding more applications throughout manufacturing. Let’s learn more.
There are many types of diffusion bonds requiring a variety of thermal profiles (Fig. 1). These can be categorized as follows.
Solid-State Diffusion Bonding
A solid-state bond can be defined as a method in which two or more mating parts are joined together without the use of an interface layer (i.e., without the application of a material between the parts applied by plating, sputtering, ion implanting or brazing or in the form of a foil). The resulting interface must be joined at or below the melting point of either parent material or any resulting eutectic that may form (i.e., Tbond < Tmelting point).
Typically, one sees a 50% or greater increase in grain size. The total part strain must be in the order of 2-4% to achieve a high-quality hermetic seal. The strain requirement means that considerable load (i.e., unit-normal force) must be applied to the bonding surfaces in order to achieve proper results.
Activated Diffusion Bonding
In the activated diffusion-bonding method, the surfaces to be bonded are coated with a solid material, typically having a smaller atomic diameter and higher vapor pressure than the parent material. The resulting bond is enhanced by the presence and diffusion/mass transport rate of this secondary material. The process is run at or below the melting point of either parent material or any resulting eutectic that may form (T bond < T second material solubility).
Activated diffusion bonding is routinely performed on materials that suffer strength disadvantages due to grain growth or structures that may not be heavily strained. Parts having poor surface finish, materials that are degraded by grain growth (e.g., nickel- and cobalt-based superalloys) or bonds requiring ultra-low hermetic-seal values are typically bonded using this method. This is particularly true when the structure contains very tiny embedded passages that could become plugged or blocked by a bonding method relying on liquid generated during the bonding operation.
Liquid Interface Diffusion (LID) Bonding
In this bonding method, one purposefully places a secondary material at the bond interface with the intention that it will melt and produce a liquid during the bonding operation (similar to conventional brazing), but it should be emphasized that there are significant differences between these two processes.
Specifically, the starting interface in the LID bond is typically very thin – on the order of 3.81 microns (0.00015 inches) thick. By comparison, a typical braze joint is usually at least 38.1 microns (0.0015 inches) thick. Because the starting thickness is so thin, it is also normal that the second (or “B”) material is nearly completely diffused out into the lattice of the parent material during LID bonding. The liquid flows outward into the surface asperities, helping to minimize the voids that may otherwise be caused from parts with poor surface finish or minor defects (such as burrs).
In a typical braze joint (Fig. 2), the microstructure typically shows a defined layer of the braze-alloy chemistry and has fillets at the outer interfaces at each layer. The resulting strength of the diffusion-brazed joint and the LID bond joint can often vary by substantial amounts. This is due to the relative amount of actual parent-to-parent metal surface being bonded. In addition, the added LID bonding layer may be selected to promote the formation of a eutectic with the parent metal so that the bonding temperature can be lowered to just above the eutectic of the resulting system (T bond > T eutectic or T liquidus of the second material).
The resulting diffusion rate occurs at three to five orders of magnitude faster than solid-state diffusion bonding. Therefore, LID bonding may be done more quickly and at lower strain energy than solid-state or activated diffusion-bonding methods. LID bonding is used for large parts with large internal-passage geometries – over 3.175 mm (0.125 inches). LID bonding is also used in instances where the risk of plugging of the passages is a concern or where part size is such that there may not be enough mechanical force available from the differential-expansion tooling or hydraulic system/rams to achieve a good bond.
An LID bond joint can be quite thin and produces excellent diffusion interfaces with the parent material. This type of bond can be used directly adjacent to passages that are quite small, such as 0.16 mm x 0.16 mm (0.0625 inch x 0.0625 inch), with very little risk of plugging. The actual amount of liquid that is generated can be controlled using slow ramp rates –18˚C/hour (30˚F/hour) for example. Even at pressures up to 2,760 kPa (400 psi) unit-normal stress and very little parent-metal strain (less than 0.1%) occurs, and only a small percentage of the alloy is displaced.
Transient Liquid Phase (TLP or TLID) Diffusion Bonding
TLP bonding involves surfaces that may be similar in quality and chemistry to the thin-layer LID bonding method, but the process prevents liquid from actually forming to any significant amount (i.e., retards the actual melting of the braze alloy by controlling the in-situ eutectic formed). As a result, the risk of plugging a small passage can be substantially reduced while otherwise still allowing the successful bonding of parts that may have a poor surface finish.
Slow ramp/heating rates can be continued above the solidus temperature of the second material all the way up to temperatures well above the liquidus of the interface system provided that the ultimate temperature never exceeds the incipient melting point of the parent material.
Vacuum diffusion bonding has made great strides in recent years and is a technology that design engineers and heat treaters need to know more about. Past limitations such as dirty materials, uneven contact surfaces and poor equipment design (yielding uneven pressure) have been overcome, and the result is a robust process worthy of consideration.
A look at the equipment being used to perform these processes.
- Herring, Daniel H., Vacuum Heat Treatment, Volume II, BNP Media, 2016
- Hubele, Norm, Refrac Systems (www.refrac.com), technical and editorial contributions and private correspondence
Vacuum Heat Treatment Volume II
Vacuum Heat Treatment Volume II by Dan Herring covers new topics and expands on those already talked about previously, making this two-volume set the comprehensive resource on the subject of vacuum heat treatment. The book will provide readers with a wide range of useful information, both from a practical and technical standpoint on the subject and give them the opportunity to reach out to sponsors to learn even more. Readers will be able to make better and more informed decisions about their equipment, process and service needs. What makes this book unique to the heat-treatment industry is that it is written specifically for the heat treater, engineer and metallurgist by one of their own.
Order online at: www.industrialheating.com/products