Polyimides curing

Three different dry etch techniques were investigated isotropic O2 plasma etching in a Tegal 200 reactor, R.I.E. in a parallel-plate in-house modified Tegal AOO reactor and R.I.M. in a Veeco, Model RG-830. The conditions of operation for each system were as follows where time is the time to etch 1.2 p of fully cured polyimide. 

A 3/8 inch diameter aluminum or titanium-tungsten dot pattern WLs fabricated on top of the cured polyimide film to make electrical leakage to substrate measurements for pinhole density estimation. An etch decoration technique was used to visually determine pinhole densities in polyimide films. The polyimide film was cast on substrates comprised of a layer of 200 nm thick alumimmi on blue colored field oxide with a grid pattern for area computation. Replicate holes were etched in the aluminum by a hot phosphoric acid solution. With the polyimide film removed, a good visual contrast was achieved for pinhole density counting. 

On the other hand, bonding of Si to Si was achieved by a low-temperature curing polyimide film [100], or by using an intermediate deposited layer of borophosphosilicate glass and subsequent anodic bonding (300 V, 350°C) [101]. 

Special emphasis will be laid on the very unique information that can be obtained from the study of the incipient metallization stages of a well-defined polymer, i.e. a cured polyimide film. 

Figure 1. Cls XPS core-level spectra of (a) PMDA-ODA starting material, (b) potassium polyamate, (c) polyamic acid and (d) re-cured polyimide. The takeoff angle of electrons was 35° from the sample surface.

If the two different polyamic acid solutions were mixed and then cured, the glass transition temperatures of the cured polyimides are different from those corresponding to separately cured polyimides. These results are interpreted as transamidization which gives random co-polymer(s). C. Feger in Polymeric Materials for Electronic Packaging and Interconnections J.H. Lupinski and R.S. Moore Eds American Chemical Society Washington, DC, 1989 p 114. 

X-ray photoelectron spectroscopy has been used to study the metal polyimide interface formed during room temperature metal deposition. Several mono-layers of Al, Au and Cu were sputter-deposited onto cured polyimide, to a thickness which permitted the observation of both polyimide and metal peaks. Deconvolution of core-level Cls, Nls and Ols polyimide peaks and A12p, Au4f and Cu2p3/2 metal overlayer peaks has demonstrated that chemical reaction occurs at the carbonyl sites for all these metals under the conditions used. In addition, the aromatic nature of the molecular structure at the interface is believed to decrease while the percentage of an isoimide-like component increases. 

In order to monitor the progress of interfacial reactions occurring during the metallization of cured polyimide, x-ray photoemission spectroscopy (XPS or ESCA) was used to reveal electronic core-levels indicative of the environment at the interface and adjacent regions. Evidence of chemical reaction would include the appearance of new peaks with characteristic binding energies (chemical shifts) representative of new or altered chemical states of the element. We can thus ascertain the formation of metal-oxygen chelate complexes (1). 

The interaction of the polyimide chains via carboxylate type bonds is however clearly evident in spectra 7b for the cured polyimide film on polycrystalline silver. In addition to the characteristic vibrational modes of bulk polyimide we find two strong bands at 1600 cm 1 and 1440 cm 1 not present in the bulk spectra in the thick film (7c). These are interpreted as arising from the polyimide-silver carboxylate type interface bonding as discussed above for polyamic acid. The presence of both the asymmetric and symmetric V(C00 stretch could again be due to a preferential monodentate bonding configuration or to the polycrystalline nature of the substrate. 

Cured polyimide resins, (IV), used as coatings for semiconductor devices were prepared by Akiba [4] that had good heat resistance, improved adhesion to substrates, and resistance to thermal deterioration. 

For the cured polyimide, the CF /F ratio was calculated using the integrated peak area results. A value of 0.33 was determined experimentally (vs. the value of 0.33 predicted). Extensions of this approach occurred by calculation of =C=0/N and =C=0/ ratios. Values of 1.08 and 0.32, respectively, were determined experimentally (vs. 2.0 and 0.67 predicted on the basis of structural formula considerations). These results indicated either carbonyl deficiency, an excess of nitrogen or an excess of fluorine in the surface region. 

Table II. The C-ls and K-2p Binding Energy Data for Cured Polyimide (PI5878)KOH-HAc Surfaces...

The surface chemical structure of several thin polyimide films formed by curing of polyamic acid resins was studied using X-ray photoelectron spectroscopy (ESCA or XPS). The surface modifications of one of the polymer systems after exposure to KOH, after exposure to temperature and humidity, after exposure to boiling water, and after exposure to O2 and 02/CF plasmas were also evaluated. The results showed imide bond formation for all cured polyimide systems. It was found that (a) K on the surface of the polyamic acid alters the "normal" imidization process, (b) cured polyimide surfaces are not invarient after T H and boiling water exposures, and (c) extensive modifications of cured polyimide surfaces occur after exposures to plasma environments. Very complex surfaces for these polymer films were illustrated by the C Is, 0 Is, N Is and F Is line characteristics. 

Figure 8- Dielectric Rheological Response of a Curing Polyimide.

Geldermans, P., Goldsmith, C., and Bedetti, F., "Measurement of Stresses Generated during Curing and in Cured Polyimide Films", Ref. 4, pp. 695-711. 

Figure 15.22. CO2 permeability of cured polyimide films containing CH dCH-, )j-NlL montmorillonite vs. volume fraction of filler. [Data from Lan T, Kaviratna D, Pinnavaia T J, Chem. of Mat., 6, No.5, 1994, 573-5. ...

Samples of polyamic acid were obtained commercially (DuPont) as concentrated solutions or were synthesized in this laboratory in N-methylpyrrolidone (NMP), with the polymerization solutions stored under argon until use (7). All dilute solutions were prepared by dilution from the concentrated solution with distilled NMP. Cured polyimide samples were either commercially available films (Kapton), or were cured in this laboratory from either commercial or laboratory synthesized polyamic acids, using a thermal or a combination chemical/thermal cure. Solvents used were all reagent grade, and at times were redistilled before use. 

Measurements on Cured Polyimide. Table II below shows results obtained for the cured polyimide samples dissolved in concentrated sulfuric acid. For the samples cured in this laboratory, where the precursor polyamic acid M is known, the observed molecular weight of the cured polyimide is comparable. This suggests that the final physical properties of the cured polyimide should be determined by the molecular weight of the precursor polyamic acid formed. 

Comparison of Chain Dimensions. Knowledge of the intrinsic viscosities and molecular weights of a series of PMDA/DAPE polyamic acids allows estimation of the unperturbed chain dimensions. Comparison may then be made with the cured polyimide, with results obtained by other workers, and with calculated values. The expressions in the literature using values for [n] obtained in good solvents generally involve extrapolation of [t)]/M toM = 0, where excluded volume effects are presumed to be minimal (12-14). Although this method is not strictly valid, it is useful for comparison purposes when direct measurement of the dimensions is impractical. The data in Table I for the polyamic acid in distilled NMP and in the poorer mixed solvent NMP/dioxane yield (see Figure... 

We have been able to measure both the weight average molecular weight and the intrinsic viscosity for equilibrated polyamic acid precursors in NMP, and for stable solutions of the cured polyimide in concentrated sulfuric acid. These measurements were not complicated by either polyelectrolyte effects in the amide solvent or degradation in the acid, as have been reported by other workers for these systems. The results show that molecular weight achieved in the condensation to the polyamic acid is retained in the final polyimide, at least for moderate molecular weights. The condensation reaction appears to follow the expected kinetics and leads to a most probable distribution at equilibrium. 

---