Diffusion and self-adhesion of the polyimide PMDA-ODA H. R. Brown, A. C. M. Yang, T. P. Russell and W. Volksen IBM Almaden Research Center, 650 Harry Road, San Jose, California 95120-6099, USA and E. J. Kramer Come//University, Department of Materials Science and Engineering, Ithaca, New York 14853, USA (Received 21 October 1987; revised 18 March 1988: accepted 18 March 1988) The relationship between the interdiffusion of two layers of the polyimide PMDA-ODA and the adhesion between the two layers has been examined. The polyimide layers were made by successively depositing a polyamic acid layer from solution and then curing the layers at elevated temperature to the polyimide. Diffusion occurred during the curing process of the second layer and was controlled by the cure schedule. The interdiffusion was measured using forward recoil spectrometry (FRES) and the adhesion measured by peel tests. Good correlation was found between the interdiffusion distance and the adhesion. It was found that a large diffusion distance, at least 200 nm, was required to obtain a bond whose strength was equal to that of bulk material. (Keywords: diffusion; self-adhesion; polyimide) INTRODUCTION The adhesion of polymers to themselves (self-adhesion or autohesion) is a subject of technical importance that has inspired considerable scientific interest. There seems to be general acceptance of the primary role of diffusion across the interface to allow chains to entangle as the main mechanism of adhesion in most situations ~-3. These situations include the adhesion (tack) that is observed when two surfaces of an uncrosslinked elastomer are pressed together and the welding or crack healing that occurs when a rigid polymer is heated above its glass transition or melting temperature. Several theories of tack or healing have been proposed recently based on the reptation model of polymer diffusion *-1°. The main differences between these models are in the assumptions that have been employed on the relationships between failure stress or energy and the number of chain segments or total length of the polymer chains crossing the interface. Although differing in detail these theories all assume that the time necessary for one tube renewal, which corresponds to a diffusion distance of a radius of gyration of the chain, is sufficient to return the strength of the interface to that of the bulk material. The experimental situation is not entirely clear, even in non-crystallizing elastomers where the failure processes are not as complex as those found in glassy or semi- crystalline polymers. From studies on mixtures of uncrosslinked polyisobutylene with a crosslinked butyl rubber, Ellul and Gent ~1 delineated the relative importance of adsorption and diffusion in that system. Bothe and Rehage ~2 and Wool ~3 have summarized the evidence for the diffusion model of tack but in neither case were accurate measurements of diffusion available so that the theories, and particularly the assumptions of diffusion distance required for complete healing, could not be evaluated quantitatively. Recently, however, Roland and B6hm x4 measured the diffusion across an interface of polybutadiene using neutron scattering and found that the autohesion of the polybutadiene continued to increase over times two orders of magnitude greater than that required for diffusion distances of one molecular radius of gyration. They speculated that a minor quantity of branched chains that diffuse very slowly might be responsible for the interfacial strength. Welding or crack healing experiments on glassy polymers have been published by Kausch and co- workers 15-17 and Wool and O'Connor 9. Both groups found that the strength (as measured by critical strain energy release rate) of an interface increases as the square root of healing time, a result that is consistent with the diffusion theory. Again the diffusion was not measured directly under the same conditions as the crack healing and, therefore, the assumptions on the amount of interdiffusion required to restore full strength were not tested. In addition crack tip studies using optical interference techniques 16 show that the failure mode of partially healed materials is complicated. This complexity probably occurs because there is no driving force for a crazing type of failure to follow precisely along the weakest line as the onset of strain softening ahead of the craze tip may not be greatly affected by reduced entanglement. Experiments by Robertson la using thin films of low molecular weight polystyrene to bond two acrylic sheets where the locus of failure did not necessarily keep to the polystyrene re-emphasize this point. The self-adhesion of the aromatic polyimide made from pyromellitic dianhydride and oxydianiline (PMDA- ODA) is technically important. It is not, however, a 0032-3861/88/1018074)5503.00 © 1988 Butterworth & Co. (Publishers)Ltd. POLYMER, 1988, Vol 29, October 1807