What residual stress is
Residual stress is stress that stays inside a material when nothing is pushing on it from outside. A film can be pulled or pushed by its own structure after it has been made. If the film wants to be smaller than the substrate allows, it is stretched, which is called tensile stress. If it wants to be larger, it is squeezed, which is called compressive stress. Both are normal and neither is automatically bad. Trouble arises when the stress is large relative to what the film or interface can tolerate.
A simple picture helps. Imagine a thin sheet glued to a thicker plate. If the sheet shrinks as the glue sets, it tugs on the plate and the whole piece may bow. The sheet is in tension, and the plate bends to balance it. A coated part behaves the same way, usually on a much smaller scale.
Where the stress comes from
Two main sources are usually considered, along with a few smaller ones.
Thermal stress. Many coatings are made at high temperature and then cooled. If the film and substrate shrink by different amounts, one is forced to stretch or compress to stay attached. The size of the effect depends on the difference in thermal expansion, the temperature drop and the stiffness of the materials. A diamond film on a metal substrate is a classic case: diamond expands very little when heated, while most metals expand much more, so on cooling the substrate tends to shrink more and squeeze the film into compression. Cemented carbide is closer to diamond in expansion than steel is, which is one reason it is a popular substrate for coated cutting tools.
Intrinsic (growth) stress. This arises during deposition itself. As grains nucleate and merge, the gaps between them close, pulling the film into tension. Energetic particle bombardment, common in sputtering and arc processes, can pack atoms more tightly and push films into compression. Trapped gas, such as hydrogen in diamond films, and defects can also contribute. Intrinsic stress depends on process conditions such as pressure, temperature and deposition rate, which is why two films of the same material from different reactors may carry quite different stress.
Other contributions. Phase changes in the substrate, chemical reactions at the interface, and later wear or heating in service can all add or release stress.
What stress does
Stress alone is not damage, but it stores elastic energy. When there is enough, the film can release that energy by changing shape or breaking.
- Cracking. Films in tension tend to form networks of cracks, often in a pattern like dried mud. The cracks run through the film thickness.
- Buckling and delamination. Films in compression can lift off the substrate and form blisters or ripples, sometimes spreading into larger flakes. This is a common reason coatings peel from edges or corners.
- Curling or bending. A thin, free film or thin substrate can bow. In electronics or optics, this distortion may be a problem in itself.
- Reduced lifetime. Even without visible failure, stress adds to the load applied in service, which can shorten life under wear or impact.
Moderate compressive stress is often considered helpful. A compressed surface resists the opening of cracks, which is why compressive hard coatings are widely used. Very high compression, however, can make the film peel. Tension is generally the more dangerous case for brittle materials.
Why thickness changes things
Stored elastic energy rises as a film gets thicker. A thin film may carry high stress without cracking, while a thicker film with the same stress can release enough energy to crack or peel. That is one reason there is often a practical maximum thickness for a given coating system. Coating thickness is therefore a design trade-off, and thicker is not simply better.
Measuring stress
Several techniques are common.
- Substrate curvature. A thin substrate is measured before and after coating. The change in bow, interpreted with elastic constants, gives an average film stress. This is a classic and widely used method.
- X-ray diffraction. Stress changes the spacing between atomic planes slightly. By measuring those shifts at different tilt angles, the strain, and thus the stress, in crystalline films can be estimated.
- Raman spectroscopy. In diamond, the position of the main Raman peak shifts with stress, which gives a convenient, non-contact estimate for diamond films. Raman spectroscopy is also used to check diamond quality, so one measurement can serve both purposes.
Different methods can disagree somewhat because they probe different depths and make different assumptions, so values should be treated as estimates.
Reducing harmful stress
Producers have several options, usually combined.
- Match expansion. Choose a substrate with expansion close to the film, or add interlayers that bridge the difference.
- Adjust process settings. Temperature, pressure, bias voltage and growth rate all change intrinsic stress.
- Control cooling. Slower cooling can reduce temperature gradients, though it cannot remove the expansion mismatch.
- Use multilayers or gradients. Alternating or graded layers can spread out stress and stop cracks from running straight through.
- Anneal. Heat treatment can relax some stress in certain systems.
- Improve adhesion. A well-bonded film tolerates more stress before it lets go, tying this subject closely to coating adhesion.
Reading failures
The pattern of failure offers clues. Regular crack networks suggest tensile stress. Blisters and edge lifting suggest compression with weak adhesion. In failure analysis, these clues are combined with measurements to decide whether to change the process, the substrate preparation or the coating design.