Electronic package systems for information processing include the functions of electrical interconnection, cooling and physical support for the sets of IC chips plus other components in electronic systems. There are several distinct packaging technology worlds, ranging from that of the large high performance mainframe systems to that of the supermini and to that of the small workstation or home computer. In all cases, the package system is a major cost factor; for high performance systems, it is also a major performance factor.

An increased understanding and utilization of suhstrate properties will he required to improve future performance and reliability of microelectronic devices. Previously, only the passive substrate/package properties were required for micro devices. Newer designs are utilizing both the active and passive properties of the ceramic materials. These new demands are being driven by end user circuit needs, and will generate major changes in the processing and fabrication of oxide materials.

The status of oxide raw material availability, and the characterization of these materials is reviewed. The processing, and fabrication of substrate materials and the interactions with end use properties are discussed. The future expectations and demand on the materials and processes are forecasted. Substrate properties and their impact on final circuit performance are emphasized.

A brief description of a set of processes and materials to make printed circuit panels will be given. Then we will discuss the multi-disciplines required for forming a scientific base and provide examples where the fundamentalist and often times specialist is providing the scientific data needed by the packaging engineer. Control techniques, characterization or otherwise basic information helpful in decision making is provided to the development and production groups. Specifically, the state of resin cure at critical steps in the process is used to demonstrate the inter-relation between cure, flow, and mechanical properties in affecting printed circuit board integrity. Finally, the complex role of moisture solubility and diffusion from the process environment in perturbing the chemical and mechanical state is discussed.

Multilayer ceramics (MLC) technology is the basis for several major products today, which include capacitors and packages for integrated circuits. This technology was originally developed to produce miniature capacitors for the micromodule program. When these goals were achieved, new applications in packaging were found. The MLC technology is now the building block in IBM's medium and large computers. The current state of this technology will be described. It's advancement will require improved understanding of the sciences involved in each of the process steps. Several of these technical areas will be discus-ed, particularly the relationships among the electrical properties of the materials and the performance of the package. As the dimensions become smaller, more concern will have to be shown for the defect levels. This implies improved control of material preparation and processing.

This presentation will review the development of hybrid circuit technology, a technology which is now widely accepted for all types of electronic products. The industry has matured and is predicted to show steady growth in the future.

Thin ceramic structures can be efficiently formed by the tape casting process, using a doctor blade for thickness control. A concentrated suspension of the ceramic precursor powder is made in a solvent system, with a polymeric binder, a dispersing agent, and other additives. For high performance ceramic applications, a non-aqueous solvent is generally preferred. The physical chemistry background leading to this choice is discussed, particularly in regard to the competing adsorptions of solvent, dispersant, and binder. Glyceryl trioleate is shown to be a useful dispersant for such systems, allowing dense compacts to be achieved, even with submicron particle sizes. Thus a refractory powder can be fired to nearly theoretical density at practical temperatures.

For the past two years we have been investigating the tape casting of BaTiO3 Specifically we have been interested in developing a useful tape casting formulation and procedure and in studying the effects of powder particle size distribution and dispersion processes on the forming of BaTiO3 tapes.

The formulation we have developed is non-aqueous. An MEKethanol mixture is the solvent and an acrylic binder is used. Two dispersants are being used, a phosphate ester and Menhaden fish oil. Ultrasonic dispersion was found to be effective in breaking up weak powder agglomerates. We have found it important to add the dispersant before adding other organic components to obtain the best dispersion and strongest tapes. The drying method is also important to tape strength. The strongest tapes resulted when the tape was removed from the glass plate soon after casting. We have also demonstrated that for forming purposes a wide particle size distribution is preferred.

In the last few years we have succeeded in developing three separate groups of near-zero-expansion ceramics. These are based on: (1) Modification by crystalline solution of GeO2 for SiO2 in the well-known cordierite phase. (2) Discovery of a new structural family known as [NZP] or [CTP], many members of which have near-zero values of a and very extensive crystalline solution possibilities permitting continuous control of α. (3) Di- (or multi) pfiasic composites of phases, one with negative and one with positive a to give a net-zero-α ceramic material.

The raw materials and clean-room processing techniques required to produce a low alpha-particle emitting insulating material have been investigated with the goal of producing a dense fired alumina ceramic with a reproducible, low alpha-particle flux (≤ 0.01 α/cm2hr). Various sources of A12 O3, MgO, and SiO2, the primary constituents of the alumina body, were tested for the alpha-particle emitting elements, U and Th, and for total alpha-particle flux. Many materials contain U concentrations with corresponding alpha-particle fluxes of 0.01α/cm2hr. Die pressing, filter casting and isopressing have been used under class 100 clean-room conditions to process powder compacts and the benefits and problems of each were evaluated. Analysis of the material before, at various stages during processing, and after firing has shown that dense fired Al2O3 ceramics can be produced with no detectable increase above that of the source powder in the U and Th contents, or the alpha-particle flux.

A newly developed amorphous, corrosion resistant Ta2O5 coating in saturated H2/H2O/Cl2 environment was studied by scanning and transmission electron microscopy and surface spectroscopy. Although Ta2O5 films formed on polished Al single crystal surfaces by sputter deposition of Ta and subsequent anodization effectively prevent the Al from corrosion by water saturated with chlorine, Al alloy surfaces are not protected, since the anodic film formed over grain boundaries, processing lines, and emergent Precipitates is poorly adherent, thus providing loci for corrosion. This problem was eliminated by casting a 400Å layer of tantalum oxyhydroxide polymer from ethanol solution onto the Al alloy substrate and curing to a Ta2O5 layer that effectively resisted attack by wet chlorine.

Amorphous, anodic Ta2O5 films of 400 A thickness deposited on deformable polymer substrates deformed by microshear banding even at 5–10% tensile strain. Electron beam evaporated Ta2O5 (EBE) were considerably more brittle in tension on the same substrates, fracturing at strains less than 1%. Even though both films were amorphous to electron diffraction the EBE vibrational spectrum in the Ta2O5 region (900 cm-1 - 400 cm-1) was more reminiscent of thermally oxidized material that appears to have greater microstructural order. A significant dynamic mechanical peak was observed at 60–90°C for both Al and Ta foils coated with either anodic or EBE Ta2O5 films. This process was assigned to relaxation of interfacial stresses induced by atomic motion in the oxide in this temperature range. In the face of the rather similar dynamic mechanical spectra we suspect that the presence of a greater Packing heterogeneity might explain the greater brittleness of the EBE films.

Measurements are reported of the contact angles of benzene-derived carbon fibers with de-ionized water, aminopropyltriethoxy silane and du Pont 5878 polyimide solutions. Both pristine fibers and fibers implanted with 5×1012 cm-2 and 1×1015 cm-2 fluences of 31p ions were investigated. Such implantation was observed to decrease the contact angle in all three solutions. The ion implantation had no significant effect on tensile strength, but did result in increased structural disorder (as reflected in the width of the main wide angle X-ray diffraction peak) and increased surface roughness. The increased disorder and roughness are believed to be principally responsible for the observed change in wetting angle.

Grain boundary cavity growth in solder joints during thermal fatigue is analyzed. The stress cycle profile is estimated based on a geometrically simplified model of a ceramic chip carrier - printed circuit board assembly and a state variable equation for plastic flow in the solder.

With proper sample preparation, high temperature x-ray diffraction can be used to study in-situ the reactions occurring at a solder/metal interface. We have applied this technique to an investigation of the reaction kinetics between copper and nickel metal and 60/40 Sn/Pb solder. The copper and nickel are shown to follow similar pathways, each having a complex reaction profile that involves an initial “hold” of little reactivity, followed by a two-step diffusion controlled reaction. Activation energies were obtained from Arrhenius-type plots, and result in values of 6.8 kcal/mole for the nickel/solder reaction, and 13 kcal/mole for the copper/solder reaction. These results are compared with those obtained by previous investigators, and discussed in terms of the growth of the different intermetallic phases.

The growth of intermetallic compounds and the strength of Cu/ Sn and Cu/ 60Sn40Pb butt joints were studied as a function of isothermal aging. The effects of single-crystal (100), (110), and (111) oriented copper on the growth rates of Cu3Sn and Cu6Sn5 intermetallic compounds are characterized and the influence of elevated temperature aging on the tensile strength of butt joints analyzed. Substrate orientation appears to influence the growth rate. Metallographic measurements showed that the intermetallic compounds grew at a rate proportional to the square root of time. Tensile tests of aged butt joints revealed a more complex time dependence.

Interdiffusion and intermetallic formation in Ni-Sn interfacial zones are examined by X-ray diffraction in samples prepared by electroplating of Sn at room temperature. For the case of plating directly onto electropolished nickel, only very sluggish formation of Ni3Sn4 was observed at 190 C. In contrast, when the nickel surface is chemicaly or chemically-abrasively activated prior to plating, a thin layer of Ni3Sn forms at the initial interface at room temperature, and subsequent annealing at 100 and 190 C produces all intermetallics predicted by the equilibrium phase diagram including Ni3Sn, indicating that the absence of Ni3Sn usually observed arises from its failure to nucleate.

Titanium copper compound formations have been studied via Seeman Bohlin x-ray diffractometry and Rutherford backscattering spectrometry. The heat treatment temperature range was 350°C to 475°C and th heat treatment times ranged from 0 to 240 minutes. Tetragonal gamma TiCu formed first at ˜350°C and was sequentially followed by orthorhombic TiCu3 at ˜400°C. From Rutherford backscattering spectrometry analysis it was concluded that the second TiCu3 compound deviated from ideal stoichiometry to a nonstoichiometric Ti0.88Cu3.12 composition

Stress concentrations that develop in metal/ceramic bonded systems have been evaluated and shown to encourage crack propagation at, or near, the edge of bonded interfaces. Experimental indentation observations on Nb/Al2O3 confirm the existence of the predicted stress concentrations. In this system, failure was invariably observed to initiate in the ceramic, such that quasi-static cracks located at the interface exhibited crack blunting. However, substantial dynamic reductions in the crack growth resistance of the interface are inferred from fracture surface observations. Implications for the optimal strengths of ceramic/metal bonded systems are presented.

Interfaces are important in ceramic and other composite materials because they can be used to control cracks, the major source of weakness and unreliability in brittle solids. Brittle materials fracture catastrophically, but may be converted into tough composites by the injection of interfaces which retard or deflect the cracks. This paper examines the behaviour of cracks at interfaces and demonstrates several mechanisms for crack control in brittle systems. Crack stopping, dislocation formation, and crack deflection at interfaces have been illustrated by experiments on rubber models and analysed by the energy balance theory of fracture.

Thermal expansion mismatch causes serious problems in operating device reliability because of the various stresses imposed on the joint materials undergoing temperature changes. Silicon has a low expansion coefficient so strain buffers are often employed between the silicon and the outer metal covers with relatively high expansion coefficients. These buffers can be metals like molybdenum, ceramics like alumina, or a brush structure of copper, with soft solders at the joints. However, solders themselves have finite fatigue lives which can be estimated from material and dimensional data.