Cold or contact welding was first recognized as a general materials phenomenon in the 1940s. It was then discovered that two clean, flat surfaces of similar metal would strongly adhere if brought into contact under vacuum. It is now known that the force of adhesion following first contact call be augmented by pressing the metals tightly together, increasing the duration of contact, raising the temperature of the workpieces, or any combination of the above. Research has shown that even for very smooth metals, only the high points of each surface, called "asperites," touch the opposing piece. Perhaps as little as a few thousandths of a percent of the total surface is involved. However, these small areas of taction develop powerful molecular connections - electron microscope investigations of contact points reveal that an actual welding of the two surfaces takes place after which it is impossible to discern the former asperitic interface. If the original surfaces are sufficiently smooth the metallic forces between them eventually draw the two pieces completely together and eliminate even the macroscopic interface.
Exposure to oxygen or certain other reactive compounds produces surface layers which reduce or completely eliminate the cold welding effect. This is especially true if, say, a metal oxide has mechanical properties similar to those of the parent element (or softer), in which case surface deformations do not crack the oxide film.
Powders in powder metallurgy use cold welding to best advantage because they present large surface areas over which vacuum contact can occur. For instance, a 1 cm cube of metal comminuted into 240-100 mesh-sieved particles (60-149 μm) yields approximately 1.25×106 grains having a total surface area of 320 cm2. This powder, reassembled as a cube, would be about twice as big as before since half the volume consists of voids.
If a strong final product is desired, it is important to obtain minimum porosity (that is, high starting density) in the initial powder-formed mass. Minimum porosity results in less dimensional change upon compression of the workpiece as well as lower pressures, decreased temperatures, and less time to prepare a given part. Careful vibratory settling reduces porosity in monodiameter powders to less than 40%. A decrease in average grain size does not decrease porosity, although large increases in net grain area will enhance the contact welding effect and markedly improve the "green strength" of relatively uncompressed powder. In space applications cold welding in the forming stage may be adequate to produce usable hard parts, and molds may not even be required to hold the components for subsequent operations such as sintering.
Hard monodiameter spheres packed like cannonballs into body-centered arrays give a porosity of about 25%, significantly lower than the ultimate minimum of 35% for vibrated collections of monodiameter spheres. (The use of irregularly shaped particles produces even more porous powders.) Porosity further may be reduced by using a selected range of grain sizes, typically 3-6 carefully chosen gauges in most terrestrial applications. Theoretically, this should permit less than 4% porosity in the starting powder, but with binary or tertiary mixtures 15-20% is more the rule. Powders comprised of particles having a wide range of sizes, in theory can approach 0% porosity as the finest grains are introduced. But powder mixtures do not naturally pack to the closest configuration even if free movement is induced by vibration or shaking. Gravitational differential settling of the mixture tends to segregate grains in the compress, and some degree of cold welding occurs immediately upon formation of the powder compress which generates internal frictions that strongly impede further compaction. Considerable theoretical and practical analyses already exist to assist in understanding the packing of powders.
Moderate forces applied to a powder mass immediately cause grain rearrangements and superior packing. Specifically, pressures of 105 Pa (N/m2) decrease porosity by 1-4%; increasing the force to 107 Pa gains only an additional 1-2%. However, at still higher pressures or if heat is applied the distinct physical effects of particle deformation and mass flow become significant. Considerably greater force is required mechanically to close all remaining voids by plastic flow of the compressed metal.