Monitoring the cure of a composite matrix resin with microdielectrometry

This paper presents a study of the cure of glass and graphite fiber epoxy composites using dielectric monitoring techniques. Initial results reported here deal with the neat resin and the relationship between its conductivity and corresponding changes in glass transition temperature during cure.     

Monitoring of laminate cure with microdielectrometry

We report the use of microdielectrometry for monitoring the cure of the matrix resins in epoxy-glass and epoxy-graphite composites. The multi-ply laminates are cured in a press using ramped temperatures. The microdielectrometer sensor is embedded in a cavity made by cutting a hole in each of the inner plies. For a brominated-epoxy glass-reinforced prepreg ramped to final cure temperatures as high as 200°C and at pressures up to 60 psi, a reproducible pattern of cure events is observed beginning with the flow of resin onto the sensor electrodes and ending with the characteristic stabilizing of the loss factor late in cure. The apparent dielectric properties (permittivity and loss factor) show good reproducibility for identically prepared samples, but their characteristics differ from those observed in neat resin. A ten-fold increase in resin conductivity is noted when cures of glass-reinforced prepreg are compared to cures of neat resin flaked off the prepreg. Special techniques are required in sample preparation when monitoring the cure of prepregs containing conductive fibers such as graphite.     

Computer-aided generation of nonlinear reduced-order dynamicmacromodels. II. Stress-stiffened case

For pt. I see ibid., vol. 9, no. 2, p. 262-9 (2000). Reduced-order dynamic macromodels to describe the behavior of microelectromechanical system structures with stress stiffening are presented in this paper. The approach is based on potential and kinetic energy representations of selected fundamental modes of motion, modified to take account of stress stiffening. Energy data are calculated by several finite-element runs, fitted to polynomial functions, and used to develop the equations of motion according to Lagrangian mechanics. The accuracy and restrictions of these macromodels are shown     

Sodium transport in polyimide-SiO2 systems

The use of polyimide (PI) in integrated circuits either as a passivant or as an interlevel dielectric requires an understanding of the potential for transfer of contaminant ions, always present to some degree in device-grade PI, from the PI to other device constituents. Sodium, in particular, if transferred from PI to SiO₂ where it is relatively mobile at typical device operating temperatures, could affect device thresholds and create a reliability problem. In this paper, the quasistatic Q-V technique of Brown using metal-PI-SiO₂-Si structures has been applied to the quantitative study of the transfer of Na between PI and SiO₂. Removal of the PI after cure or after application of bias-temperature stress has been used to demonstrate that (1) quantitative, reversible ion transfer between PI and SiO₂ does occur at temperatures above 200°C in the presence of an applied bias, but (2) in the absence of an applied bias, Na ions are not transferred from PI to SiO₂ during PI cure at 350°C. It is also shown that the kinetics of Na transfer from PI to SiO₂ in the presence of a bias are dominated by an activated diffusion process, with an activation energy on the order of 2.0eV and a prefactor on the order of 2x10⁹ cm²/s.     

Origin and prevention of high contact resistance in multilevel metal-polyimide structures

When polyimide is used as the insulating dielectric in multilevel-metal structures, a high contact resistance can result within the interconnecting vias. This paper examines the particular case of oxygen plasma patterning of the polyimide using a photoresist mask. Auger analysis in combination with compositional depth profiling was employed on a series of samples to measure surface composition of etched vias in polyimide. Results show two effects which, together, can account for high contact resistance: first, there is a thicker than normal aluminum oxide layer on the first level metal surface (due to exposure to the oxygen plasma); second, there is a thin, etch-resistant carbonaceous film (due to redeposition of organic material during plasma etching) that prevents oxide thinning through chemical means. It was found that by lowering the plasma pressure to 50 mTorr near the end of the etch, the organic film can be removed. In the absence of the carbonaceous layer, the oxide can then be chemically thinned to produce clean aluminum surfaces within the vias.     

Optical waveguiding as a method for characterizing the effect of extended cure and moisture on polyimide films

Integrated optics techniques (i.e., guiding light through a thin film) have been used to study the effects of extended cure and moisture exposure on polyimide films as manifested in changes in the TE (in-plane) and TM (out-of-plane) refractive indices and in their difference (birefringence). Mechanical testing of these films has been done using the load-deflection technique. Du Pont polyimide Pyralin 2555 and 2556 (benzophenone tetracarboxylic dianhydride-oxydianiline/metaphenylene diamine), cured repeatedly at 400°C and 375°C respectively, show an increase in refractive index as a function of cure (possibly due to densification), as well as an increase in birefringence (due to increased residual stress). Upon exposure to moisture, the TM and TE refractive indices both increase (absorption of the polar water molecule), but the TE index does not increase as much. Mechanical testing has shown that the residual stress in the polyimide film increases with extended cure and decreases with moisture exposure due to moisture-induced swelling of the film. A model is proposed which relates the optical results to the mechanical measurements. Based on our data, a preliminary stress-optic coefficient is reported.     

Dielectric analysis of the cure of thermosetting epoxy/amine systems

A low molecular weight epoxy resin is cured isothermally with an aromatic amine hardener, and the dielectric properties are measured as a function of the frequency, reaction time, and cure temperature. At specific stages in the cure, small samples from the reacting mixture are quenched and subsequently analyzed for the glass transition temperature and epoxy group conversion by differential scanning calorimetry. In this manner, the change In dielectric properties can be directly correlated with the network structure. The ionic conductivity is modeled as a function of the cure temperature and the cure-dependent glass transition temperature using a Williams-Landel-Ferry (WLF) relation. Combining this WLF relation with the DiBenedetto equation, a comprehensive model relating conductivity with the extent of reaction and cure temperature has been developed.     

Moisture diffusion in poltimide films in integrated circuits

This paper reports the use of planar aluminum-PI-aluminum capacitors, in which the variation of low frequency dielectric permittivity is shown to be proportional to absorbed moisture, to carry out in-situ studies of moisture uptake and diffusion in PMDA-ODA device-grade polyimide films. It is found that the equilibrium amount of moisture uptake depends on ambient relative humidity in the temperature range 20-80‡C, and that the diffusion kinetics obey a Fickian model with a diffusion coefficient in the range 3-5 x 10^-9 cm² /sec at room temperature. The diffusion coefficient is temperature dependent, with an activation energy of 0.32-0.34 eV, and is weakly dependent on absorbed moisture at fixed temperature. An asymmetry is observed between absorption and desorption kinetics which correlates with the moisture-dependent diffusion coefficient.     

Application of the Island Blister Test for Thin Film Adhesion Measurement

The island blister test has recently been proposed as an adhesion test which allows the peel of thin, well-adhered films without exceeding the tensile strength of the film. The island blister test site is a modification of the standard blister test site, consisting of a suspended membrane of film with an “island” of substrate at the film center. The membrane support and island are secured to a rigid plate and the film is pressurized, peeling the film inward off the island. A model for this inward or “annular” peel indicates that even for systems of good adhesion, peel can be initiated at low enough pressures to prevent film failure by making the center island sufficiently small relative to the size of the film.

We have fabricated island blister test sites using micromachining techniques and have used them to measure the debond energy of polymer films on various substrates. The peel data obtained from these island sites match well to the behavior predicted by a simple fracture mechanics analysis. This paper reports the fabrication of the island test sites, the experimental verification of the test, and the results of application of the test to polyimide films on metallic and polymeric substrates.