Durability of metal/polymer interfaces is essential for the long-term reliability of high performance microelectronics packages. Such interfaces undergo stresses during production and in service. In this work, we report on the durability of interfaces formed between reactive metals and polyimides (PI) that have been subjected to stresses simulating both types of environment. The P1 surfaces were treated by Ar RF plasmas prior to metal deposition, and durability was determined by measuring the 90 degree peel strength as a function of environmental exposure. The durability of Cr and Ti/PMDA-ODA interfaces through processing stresses (i.e., large thermal excursions) depends on the PI surface modification and the metal reactivity. For both, we observed interfacial degradation due to oxidation of the metal–the cause is water absorbed by the polyimide. These studies, coupled with water transport measurements, suggest that the physical structure of the interface is the dominant factor. To determine the durability under service environmental stresses (e.g., temperature/humidity), we correlated peel strengths with interfacial chemistry and water uptake. In this case, Ar and O2 plasmas were used. For Ta/BPDA-PDA, the durability depends on the type of plasma treatment. Ar-treated specimens maintain strength through 500 hours T/H stressing whereas those treated by O2 plasma alone fail at 165 hours. The differences here can be explained by the interfacial chemistry–Ta/Ar-etched surfaces form a stable TaG-like structure whereas the Ta/O2-etched surfaces form a metastable, sub-oxide structure that transforms to Ta2O5 during stressing. Ta/PMDA-ODA interfaces fail readily under these conditions due to the increased water uptake of the PI.

We have carried out a comparative study on the thermomechanical properties and interfacial fracture for two polyimides, a rigid-rod like BPDA-PDA and a flexible-chain PMDA-ODA. This study is focused on the correlation between material strength and polymeric chain morphology. A strong correlation was observed, with high chain packing order giving rise to strong material strength. Quantitatively, the mechanical strength of PMDA-ODA thin films along the direction tangential to the film plane is about 1/3 of that of BPDA-PDA, and its thermal expansivity 1.5 to 5 times larger. For fracture, the mechanical strength of the polyimide plays a different role. A softer polyimide such as PMDA-ODA will strengthen the confinement exerted by the metal lines with smaller interfacial strain, thereby decreasing the amount of deformation transmitted through the interface into the metal, and shifting the onset of delamination to a higher strain on the entire structure. However, the same softer polyimide layer also induces a larger amount of plastic deformation, leading to fracture of the film at a smaller strain. For a strong metal/polymer interface, the strength of the interfacial bonds is sufficient to hold the interface intact while the polymer near the interface deforms plastically because it is much softer mechanically than the metal. This continues until the deformation energy accumulated in the polymer is sufficient to delaminate the interface. Under this condition, the locus of failure will be determined by the relative fracture toughness of the polymer compared with the interface, which may very well occur in the polymer very close to the interface.

In this paper we present results on the use of x-ray absorption spectroscopy and in particular extended x-ray absorption fine structure (EXAFS) to study non-destructively buried interfaces between polymers and metals. In particular we study the interaction between isolated metal atoms (Ti or Cr) and the surrounding polymer matrix (polyimide or triazine). The metal atoms have been introduced into the polymer matrix using ion-implantation. We find that Cr forms a π complex with the aromatic structures of polyimide and triazine in agreement with model compound and XPS data. Ti on the other hand seems to form Ti-O-C bonds with both polymers.

Polyimide-on-metal interfaces relevant to electronics packaging applications have been studied by Fourier transform infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS). Two polyimides with different backbone structures were considered: (i) biphenyltetracarboxylic dianhydride (BPDA) - p-phenylene diamine (PDA) and (ii) 4,4′-(hexafluroisopropylidene)-bis(phthalic anhydride) (6FDA) - 4,4′-bis(4-aminophenoxy) biphenyl (APBP). Thin polymer films (10–100 nm) were cured on Al, Cr and Cu surfaces at different temperatures to allow probing of chemical interactions by surface spectroscopies. Anhydride intermediates were detected on all three substrates for both polyimides. Carboxylate intermediates were detected in BPDA-PDA films on Cr and Cu and resulted in increased adhesion at these interfaces. XPS data revealed a nitrogen deficiency at the BPDAPDA/Cr interface. A mechanism explaining the formation of this intermediate and the deficiency of nitrogen is proposed. Hydrolytic stability of the interface was investigated using ex situ total internal reflection FTIR spectroscopy and XPS. The results show that the fluorinated polyimide, 6FDA-APBP, is more stable than BPDA-PDA when treated at 85 °C and 85% relative humidity for 100 hours.

In this work, we present a systematic approach to the problems in polymer metallization. There are three main components to our study which include: assessment of the limits of metal polymer interaction, consideration of metal-metal interaction or aggregration at the interface limit, and modification of the copper surfaces to increase reactivity. All of these phenomena are assessed through application of semi-empirical electronic structure calculations with the ultimate objective being to effect a strategy for better metal deposition with the primary metal being copper. Polymer models include polyimide, epoxy resins, polyolefin ketones, and nitrile functionalized polymers. In the first component study, orbital interactions which affect adhesion are compared for a variety of polymer models and transition metals. This initial study is then extended to consider the potential for metallization by consideration of competition of metal-polymer interactions and metal-metal interactions as a metal lattice “grows” at the metal/polymer interface. Finally modification of the metal/polymer interface is assessed by considering the effect of oxidation or codeposition of dissimilar metals to augment the electronic properties and thus affinity of copper for the polymer. The general trends produced in this work are consistent with experimental measurements of adhesion and have been validated by spectral measurements. This work clearly shows the molecular level limitations of polymer-metal interactions for a series of important systems.

This work examines the adhesion of coatings derived from divinylsiloxane bisbenzocyclobutene, mixed stereo and positional isomers of 1,3-bis(2-bicyclo[4.2.0]octa-1,3,5-trien-3-ylethenyl)-1,1,3,3-tetramethyl disiloxane (CAS 117732–87–3), on oxidized silicon substrates treated with silane coupling agents.This material, commercially available as Cyclotene™ 3022, can be used in the construction of high performance electronic circuits, such as multichip modules. Silane coupling agents examined in this study were 3-aminopropyltriethoxysilane (CAS 01760-24-3)(APTES)(, vinyltriethoxysilane (CAS 00078–08–0)(VTES), and 3-methacryloxypropyl trimethoxysilane (CAS 02530–85–0) (MOP-TMS).

Measurement of the interfacial adhesion was performed using microindentation. Bond strengths obtained by this method exceed 200 MPa for the most effective coupling agents. However, these high bond strengths were not found to correlate with acceptable adhesive performance in all cases. In addition to the choice and preparation of the coupling agent, process related chemical exposure has been found to be a key element in the observed adhesive performance. The effect of the cure schedule for the thermoset coating has also been found to be a controlling factor. A short cycle test vehicle was developed consisting of a single 20 gIm polymer layer etched with anisotropic sidewalls. This test vehicle was used to evaluate the efficacy of the coupling agents during process exposures and subsequent thermal shock testing. A solution of MOP-TMS pre-hydrolyzed in methanol was found to produce the most reliable interface with high bond strength.

Glass fiber reinforced epoxy laminates have been widely applied in the electronic packaging industry as multilayer printed wiring boards (PWBs). Previous investigations indicate that, the electric insulation between printed through hole (PTH) and conductive lines can be lost if the boards are subjected to certain environmental conditions [1,2,3]. The insulation breakdown is caused by copper filament formation along the glass fibers once the filament bridges any two nearby circuittraces, such as a PTH and a power plane, or between two PTHs. The conductive filament formation is a two step process: path formation and electrochemical reaction. The path formation is caused by degradation of the glass-epoxy interface, which provides the copper ion migration path for the second step. Humidity and temperature are the key factors leading to interfacial degradation. Usually, humidity is of more concern because water absorption helps the path formation. Among board materials, epoxy resin, as the matrix material, is more susceptive to moisture absorption than glass fiber or copper [3,4]. Operating temperature that changes the degree of moisture absorption [5,6], and thermal cycling that enhances water absorption, and stress induced from thermal-mismatches [7], are also considered as the important factors to the interface degradation.

Epoxy/glass interfacial debonding can be introduced near PTHs by thermal cycling [8]. The interfacial debonding between glass fibers and the epoxy matrix near PTHs accelerates the electromigration process. In order to study the interfacial debonding mechanisms, thermal cycling tests of PWBs were conducted and the dynamic process of interfacial debonding was characterized using environmental scanning electron microscope (ESEM). This paper presents the experimental results with discussion of the possible interfacial debonding mechanisms.

Adhesion of Cu and Cr to polyimide was studied by 90° peel test, AES, and X-ray diffraction. The two necessary conditions of the high adhesion strength were the presence of Cr interlayer and the preactivation of the polymer surface by RF plasma treatment. The Cr is thought to react with activated polyimide to form carbide-like bonds which enhance the adhesion strength drastically. Correlations between the peel strength and C260/Cr527 Auger PHR from the peeled metal strip were good. Qualitatively, the peel strength is related to the amount of plastic deformation done on the metal strip during peeling.

Mechanical modelling of microelectronic packages for design optimization requires the quantitative knowledge of a large number of physical parameters. Because of the composite nature of many components, and because relevant materials properties in the microscale often deviate from the bulk values, there is therefore a need for a variety of micromechanical testing techniques. These techniques include tension, compression, shear, bending, and fatigue testing of samples with sizes in the 10 micron range. A novel technique has been developed for the testing the z-direction properties of thin films with displacements in the sub-micron range. Mechanical properties of thin films and surface regions may advantageously be probed by a continuous microindentation tester. All elastic properties of both isotropic and anisotropic packaging materials and components may be evaluated directly by the above methods, and the inelastic deformation behavior determined over many orders of magnitude of strain rate by load relaxation measurements. The above techniques and examples of their application will be discussed.

Flip chip interconnect systems are becoming increasingly popular in the electronics industry due to their low profile and high densities. During temperature cycles, the differential expansions between various assembly members of a flip chip interconnect system produce mechanical stresses that are the driving force for failures. Such failures can be a significant reliability concern.

Accelerated test methodologies for flip chip interconnect systems assess the reliability of existing interconnects and identify potential reliability concerns in future interconnect designs. Traditionally, such methodologies have relied on test methods such as temperature cycling to determine the mechanical integrity of the flip chip interconnect. However, application of such methods to new interconnect materials, geometry and processes requires characterization of basicmechanical behavior of the interconnect system.

In this work, the feasibility of using a micromechanical test methodology for flip chip applications is examined. 90PbSn solders is used as an example. The data obtained from such flip chip solder interconnects is validated using a damage integral methodology. The measured data is shown to adequately describe published thermal stress profiles and thermal fatigue life data measured using solder joints of the same composition. Finally, some considerations in the application of micromechanical measurements to determination of acceleration factors and development of accelerated test methods are discussed.

This benchmark study examined the effect of four different backmetal/ leadframe schemes on the time to failure from autoclave testing for surface mount, small signal transistors in the SC59 package configuration. The four schemes examined included die with Au backmetal bonded to a Cu plated Alloy 42 leadframe, Au backmetal bonded to a Ag striped, Ni plated, Cu plated alloy 42 leadframe, Au backmetal with a thin, permeable Ni layer bonded to a Ag striped, Ni plated Cu leadframe and V/Ni/AuGe/Au backmetal bonded to a Cu leadframe. Devices with AuSi eutectic bonded directly to the Cu plated Alloy 42 leadframe separated between the Si die and the AuSi eutectic after 240 hours of autoclave testing which produced VCESAT failures. Identical devices bonded to Ag striped, Ni plated, Cu plated Alloy 42 leadframes had the same failure mechanism after 624 hours of autoclave testing. Adding a permeable Ni layer to the AuSi eutectic structure and bonding to a Ag striped, Ni plated Cu leadframe increased the lifetime to 744 hours but still produced the same failure mechanism. Devices using a V/Ni/AuGeJAu bonded to a Cu leadframe remained electrically good and physically intact after 864 hours of autoclave testing. These results indicate the use of leadframe plating and/or backmetal barrier metals is needed in order to produce devices that can withstand extended reliability testing.

The flexibility and cost advantages of the wire bumping process are making it increasingly attractive as an alternative bumping method for low I/O flip chip technologies. In this study, electrical measurements from DC to 8.1 GHz and die shear experiments are used to assess the thermo-mechanical integrity of Au/2%Pd wire bumped GaAs/Al2O3 flip chip modules. Finite Element Modeling results for in-process and in-service temperature cycling indicated that the maximum stress of the flip chip module is well below the failure stresses for GaAs, Al2O3 and the Au interconnect. These results are validated by data from thermal cycling experiments. Average shear strength after thermal cycle, thermal shock, and thermal aging (MIL-STD-883C, Condition C) was 64 grams/bump. Electrical characterization (from 45 MHz to 8.1 GHz) using a coplanar waveguide flip chip test structure showed no thermal stress-induced microwave energy losses.

Nanoindentation has been used to characterize polycrystalline silicon serpentine structures which will become the “framework” for piezoelectric meander line microactuators. These structures are fabricated using silicon wafer surface micromachining processing techniques, and they are suspended about 1 micron above the wafer surface. Normally, nanoindentation equipment is used for materials scratch-testing; however, this equipment was modified so that a probe tip could make contact to the sides of the meander lines, thus enabling known forces to be applied parallel to the wafer surface. For one type of device, the nanoindentaion technique yielded a spring constant of 400 N/m, while simple beam theory gave 1300 N/m and the finite element analysis software ANSYS gave 1200 N/m. This measurement technique holds the promise of being able to apply a known force and then countering that force with piezoelectric actuation to determine the forces that future piezoelectric microactuators will be able to provide.

To create a model for the release process of solder paste during stencil printing for surface mount applications it is necessary to determine the shear stress developed at the interface between the solder paste and stencil sidewall. An experiment has been developed to determine the value of the shear stress for solder paste samples. For the purpose of this experiment a Micro-mechanical tester has been adapted and programmed to provide both a shear thinning cycle and a pull-off cycle that simulate aperture fill and stencil lift-off. The shear stress developed at the solder/stencil-sidewall interface is estimated from the data obtained during the pull-off portion of the Micro-mechanical test procedure. The micro-mechanical tester is fitted with a set of parallel plates that can be adjusted for plate separation and surface roughness.

The experiment consists of two parts: (1) the shear thinning cycle and (2) the horizontal pull-off. After application of the solder paste and adjustment of plate separation, a back and forth movement of the upper plate provides shear thinning of the paste. This step is necessary to simulate the shear thinning that occurs from the application of squeege pressure during aperture fill. The horizontal pull-off then simulates the lift-off step of the stencil printing procedure. During the horizontal pull-off data is. taken which allows calculation of the force developed as the upper plate is pulled away from the lower. Results from this experiment show the values of shear stress that develop during pull-off with a variation of surface treatments and plate separations.

This paper will present the experimental set-up, a description of the relationship between this experiment and the actual stencil lift-off process, and shear stress data that has been acquired for a variety of solder pastes and plate separations.

The High Density Electronics Center (HiDEC) at the University of Arkansas is developing advanced electronic packaging methods including multi-chip modules which utilize high temperature superconductor interconnects. A wide variety of materials have been proposed and investigated for these applications. Good knowledge of the mechanical properties of the candidate thin film materials is important for reliable device design.

Two new techniques have been developed to measure strain at fracture of thin film materials. The first is an acoustic technique in which an accelerometer is attached to the film. The film is then gradually strained. When the film fractures, an acoustic wave propagates through the film thereby creating a transient signal from the accelerometer. This signal is used to trigger an oscilloscope. The second technique is similar except that it utilizes a laser probe to detect the acoustic wave. In this photo-acoustic technique, the laser is reflected from the film onto a bi-sectional photodetector while the film is gradually strained. Upon film fracture, the acoustic wave modulates the laser beam and a transient voltage pulse is produced by the photodetector. Both techniques are capable of detecting extremely small cracks.

We describe an experimental method capable of evaluating the in- and out-of-plane components of the anisotropic thermal diffusivity in supported and unsupported thin polymer films. The technique is used to quantify the in-plane thermal diffusivity in films of PMDA/ODA (Dupont's P12545), BTDA/ODA/MPD (Dupont's P12555), BPDA/PPD (Dupont's P12611), BCB (Dow's Cyclotene 3022) and HFDA-APBP (Amoco's UD4212) with thicknesses in the 1–10 micron range.