Search

In this study, tensile tests of 63Sn/37Pb solder were carried out at various strain rates from [10.sup.-5] [s.sup.-1] to [10.sup.-1] [s.sup.-1] over a wide temperature range from -40 [degrees] C to 125 [degrees] C to study the effect of strain rate and testing temperature on the mechanical properties in a systematic manner. based on these experimental data, a set of empirical formulae was derived by a statistical method to describe the effect of temperature and strain rate in a quantitative manner and explain the variation in the mechanical properties published in other reports. It is concluded that the empirical formulae can be used to characterize the mechanical properties of 63Sn/37Pb over a wide range of temperatures and strain rates.

In this study, both two-dimensional and three-dimensional finite element analyses were used to study the stress distribution in and deflection of the flip chip assembly under thermal loading. It is found that the three-dimensional results compared favorably with experimental measurements, while the two-dimensional results consistently overestimate both stresses and deflection. Among the two-dimensional models, the plane stress assumption seems to yield results closer to the full three-dimensional predictions. Furthermore, three-dimensional models were used to investigate the effect of printed wiring board size on the overall deflection of the flip-chip assembly. This size effect of the printed wiring board has significant implications on the design of multi-chip modules. The results indicate that a square array placement pattern is preferable to a staggered array for multiple chip modules in order to reduce mechanical interaction between chips. For square arrays, such mechanical interaction between chips can be neglected when the minimum distance between adjacent chips is more than 2 times the chip size.

In the electronic industry, the dominant failure mode for solder joints is assumed to be thermal cycling. When semiconductor devices are used in vibrating environment, such as automotive and military applications, dynamic stresses contribute to the failure mechanism of the solder joint, and can become the dominant failure mode. In this paper, a damage mechanics based unified constitutive model for Pb40/Sn60 solder joints has been developed to accurately predict the thermomechanical behavior of solder joints under concurrent thermal and dynamic loading. It is shown that simultaneous application of thermal and dynamic loads significantly shorten the fatigue life. Hence, damage induced in the solder joint by the vibrations have to be included, in fatigue life predictions to correctly predict the reliability of solder joints. The common practice of relating only thermal cycling induced inelastic strain to fatigue life can be inadequate to predict solder joint reliability. A series of parametric studies were conducted to show that contrary to popular opinion all dynamic loading induced strains are not elastic. Hence, vibrations can significantly affect the fatigue life and reliability of solder joints in spite of their small mass.

Environmental Scanning Electron Microscopy (E-SEM) is a technique which provides the capability to investigate surface morphologies over a variety of controllable environmental conditions. This paper discusses the principal of the E-SEM and presents three experimental investigations of failure mechanisms in electronic packages using E-SEM techniques. The first is a study of thermally induced failures of MMIC devices. The second investigation focuses on mechano-sorptive properties of multilayer thin-film polyimides. The third investigation concentrates on relative humidity and thermal cycling effects on interfacial bonding characteristics of resin/fiber interfaces in printed wiring boards, and microcracks surrounding plated-through-holes.

Several workers have demonstrated the presence of singular stress fields near the edge of a bimaterial interface subject to thermal gradients. Many of the analyses in the literature are limited to linear elastic materials and are useful as first estimates of stresses in such assemblies. However, a time-dependent stress analysis is necessary when viscoelastic materials such as polymer films are bonded to elastic substrates. This paper shows the relevance of viscoelastic modeling in demonstrating the effects of combined temperature and moisture loading on bimaterial interfaces and viscoelastic films sandwiched between elastic substrates. A 2-D finite element method for linear hygrothermoviscoelasticity based on an incremental creep strain rate form is developed. The correspondence principle of linear viscoelasticity is invoked and comparisons between FEA and analytical solutions are demonstrated. Moisture is modeled using Fick's law and the thermorheologically simple material (TSM) postulate is extended to incorporate a moisture shift function in analogy with temperature. The analysis of a polyimide film sandwiched between elastic substrates subject to a thermal cycle with diffusion from the free edge shows stress reversals for the peel stresses that may explain failure modes not anticipated by an elastic analysis.

Thermo-mechanical behavior of various levels of electronic packaging products is studied by moire and microscopic moire interferometry. The global deformations of packages with complex geometries and the local deformations of solder interconnections are determined by displacement measurements of high sensitivity and high spatial resolution. Several packaging studies are reviewed. They include analyses of thin small outline package, leadless chip carrier package, surface mount array package, chip/organic carrier package, deformation near a plated through hole, and determination of an effective CTE. In-situ and quantitative nature of the methods leads to more accurate and realistic understanding of the macro and micro mechanical behavior of packaging assemblies and interconnections, which in turn, facilitates design evaluation and optimization at an early stage of product development.

In an earlier work, thermal stresses in soldered pins were calculated under the assumption of an elastically deforming solder barrel. In the present analysis, the nonlinear, temperature dependent stress-strain curve of solder, without creep, is taken into account, yielding an elasto-plastic foundation modulus function. For the thermal mismatch loading, the total temperature range [Delta]T (e.g., 0-100 [degrees] C) is divided into a finite number n of temperature stages, and the final responses are superposed from the n contributions each of which done using an isothermal stress calculation. Three solders (Sn-Pb solders 63/37, 70/30, and 10/90), were investigated; the solder barrel thickness and mismatch temperature range were also variable parameters. Comparisons with elastic solder foundation calculation results show that thermal and inelastic solder stress/strain dependence should be included especially in the higher (over 75 [degrees] C) temperature range.

Failures in electronic packaging under thermal fatigue often result from cracking in solder joints due to creep/fatigue crack growth. A nonlinear, time-dependent finite element analysis was performed to study the effect of critical design parameters on thermal reliability of leadless chip capacitor or resistor solder joints. The shear strain range based on thermal hysteresis response was used to study the sensitivity of various parameters, such as solder stand-off height, fillet geometry, Cu-pad length, and component length and thickness. The results were used as guidelines for designing reliable solder joints. In addition, an analytical model for the solder joint assembly was derived. It can be used as an engineering approach for rapid assessment of large numbers of design parameters. The accuracy and effectiveness of the analytical model were evaluated by comparing with finite element results.

This study correlates previously published fatigue life data with the hysteresis energy, and compares this to the previous correlation's with the applied plastic strain. It was found that, while the hysteresis energy could be used to describe the fatigue life, corrections must be made to account for the temperature and strain rate dependence of the flow stress. Because of these factors, it is believed that the plastic strain, and not the hysteresis energy, is the true governing factor in determining the fatigue life. Part I (described here) covers only the influence of temperature and cycle frequency for symmetric cycling. Part II, to be published, will deal with hold time and asymmetric cycling.

The intense, localized stress field produced by a temperature load in a multilayered structure may be significantly affected by the local geometry of the free edge. We examine here the stepped edge problem associated with bonding an elastic layer (silicon chip) to a single or multilayer substrate with a slightly larger length. Stress functions are introduced in various rectangular regions and the continuity of tractions are enforced across all inter-region boundaries. Furthermore, continuity of displacements is enforced across the junction of the two segments of the base laminate. The analysis results indicate that even a minute protrusion of the edge of the base laminate relative to the attached chip may cause significant changes in the peeling and shearing stresses in the end region of the interface.

The ability of solder joint life-prediction algorithms to predict the failure of solder joints due to temperature-cycling induced creep-fatigue has been investigated using representative leadless chip carriers (LCCs) as the test vehicle. Four different algorithms are assessed: the classic Coffin-Manson algorithm, a modified Coffin-Manson algorithm with dependency on peak stress, and two strain-energy based algorithms. JPL's special purpose nonlinear finite element computer program was used to dynamically simulate the solder joint response to the standard NASA temperature cycling environment, which ranges from -55 [degrees] C to +100 [degrees] C with a 4-hour period. The computed stress-strain history provided the inputs needed by each of the failure algorithms. To test the accuracy of the analytical predictions, three different sizes of LCCs (68 pins, 28 pins, and 20 pins) were subjected to an experimental test program using the same 4-hour temperature cycle as used in the analytical predictions. The three different sized ceramic packages, each with a 50-mil pitch, provided a range of cyclic strain ranges and solder fillet geometries so as to test the algorithms against realistic electronic packaging variables. The study highlights limitations in the historical Coffin-Manson relationship, and points up possible improvements associated with incorporating a stress modifier into the Coffin-Manson equation. This modification is also somewhat simpler and more accurate than the energy-density based algorithms, which also performed quite well.

The plastic deformation kinetics of 95.5Sn4CuO.5Ag solder joints were determined in monotonic loading shear over the temperature range of 25 [degrees]-150 [degrees] C using three types of tests: (a) constant shear rate, (b) constant shear stress (creep), and (c) differential tests (changes in shear rate or temperature during an otherwise isothermal constant shear rate test). The deformation kinetics were evaluated in terms of the Dorn high temperature plastic deformation equation [[first derivative of [Gamma] with respect to time].sub.p] = A [Mu]b/kT D[(b/d).sup.p] [([Tau]/[Mu]).sup.n] where [[first derivative of [Gamma] with respect to time].sub.p] is the shear rate, [Mu] the shear modulus, b the Burgers vector, D the appropriate diffusion coefficient, d the grain size and [Tau] the shear stress. A, p, and n are constants whose values depend on the rate controlling mechanism. It was found that n increased with stress from [approximately]4 at 2 MPa to [approximately]20 at 25 MPa, relatively independent of temperature. The activation [Delta]H was determined to be 21.1 [+ or -] 2 kcal/mole. The constant A, however, decreased with temperature from a value of [approximately][10.sup.18] at 25 [degrees] C to [approximately][10.sup.10] at 150 [degrees] C. The values of n and [Delta]H suggest that dislocation glide and climb is the rate controlling mechanism and hence that p [approximately equal to] 0. It is speculated that the large decrease in A with temperature may be the result of an effect on the microstructure.

An experimental method is presented to study laser weld induced thermal strain using digital image analysis enhanced moire interferometry. A phenomenon that occurs in the assembly of optical components is that the final optimum coupled power will randomly change upon completion of the laser weld process. The change in power is due to residual thermal stresses being generated in the welded components. For single mode devices, relative motions of the components in the order of 1 [[micro]meter] could result in a 1 dB degradation of coupled power. The behavior of thermal strain is unpredictable since the relative orientation of the optical components at the optimum alignment is random. The goal of this investigation was to perform a baseline study of parameters affecting laser weld thermal strain. The first phase of the work was to study thermal strain induced by a single weld on a flat Kovar plate. The results show that thermal strain is independent of material inhomogeneity. However, this investigation did reveal asymmetry of the power distribution in the weld laser with a principal axes offset +30 deg from horizontal. The second phase of the experiment was to characterize thermal strain resulting from welding on an interface of two Kovar plates. The results indicate that thermal strain at the center of two welds is not affected by welds that are greater than 1 mm apart. Also, thermal strain levels at locations adjacent to the weld are not significantly affected by weld separation distance. This study successfully demonstrated that digital image analysis enhanced moire interferometry can be used in the study of laser weld induced thermal strain. Digital image processing, fractional fringe analysis, and high frequency specimen gratings increase sensitivity levels to enable the technique to be used to characterize submicron thermal distortions.

The local coefficient of thermal expansion (CTE) mismatch between compliant surface mount component leads and the solder that is used to attach the components to a printed wiring board can dramatically influence the thermal fatigue life of the solder joint. To quantify the contribution of the local CTE mismatch to the overall thermal fatigue damage of the solder joint, a finite element thermal fatigue simulation using an energy partitioning technique is used to compare four different lead end shapes. The four lead configurations considered are J-lead, and three gullwings leads; one with the foot parallel to the board surface, one with the foot sloped slightly downward towards the board, and one with the foot sloped slightly upward. The dimensions of the leads are purposely chosen so that the in-plane compliance is equal for the different lead shapes, only the shape of the lead end varied. Comparisons are first made with equal solder join heights and then the solder height of the gull-wing lead is varied between 0.05 to 0.23 mm (2 to 9 mils). The influence of the solder wetting angle is also investigated.

An elastic analysis is developed to provide estimates of the stresses in area array solder joints during thermal excursions. The analysis provides a rapid quantitative evaluation of the important stresses during elastic deformation, and may serve as a basis for the qualitative understanding of the loading of interconnects during thermal and power cycles.

Interlaminar stresses near the free edge of a multi-layered structure under thermal and mechanical loads are significantly affected by nonlinear and inelastic material properties. Most previous studies of the subject ignored such effects and obtained singular or extremely severe and localized stress fields in boundary regions based strictly on the assumption of linearly elastic stress-strain relation. In the present paper, a variational method, using approximate stress functions and the principle of complementary energy, is developed to study the thermal stress in a three-layer beam including a thin, compliant, non-linearly elastic middle layer. It is found that the elastic softening behavior of the thin layer results in dispersion of the interlaminar stresses and widening of the boundary region. Hence the use of toughened, compliant bonding layers may produce a beneficial effect by alleviating local concentration of interlaminar stresses.

Leadless chip resistors (LCR) made by two different manufacturers and surface mounted on glass/epoxy printed circuit board (PCB) were subjected to thermal cycling between -55 [degrees] C to 125 [degrees] C in order to induce thermal fatigue failure/damage. The test units were subjected to a maximum of 250 thermal cycles. Solder joints in both types of LCRs were examined in scanning electron microscope and a relative comparison of the extent of fatigue damage is presented. The failure mechanism is associated with cracking in the eutectic composition Sn/Pb solder initiated at the stress concentration sites. A nonlinear, time-dependent finite element modeling analysis has been performed to determine critical stress concentration sites in the solder joint. Key parameters leading to the initiation of solder damage are identified, and recommendations are made to improve the design in terms of solder configuration such as the radius of corner of the alumina substrate and the standoff height.

Long-term reliability of electronic packaging has become a greater challenge as a result of ever increasing power requirements and the decreasing size of electronic packages. In this study, the effects of three variables on plated-through hole (PTH) design have been investigated on the thermal cycling fatigue lives in four-layered printed wiring boards (PWB's). These three variables were evaluated at two levels each: (a) hole size (0.030 and 0.040 in.), (b) internal pad (presence or absence), and (c) epoxy-plugged holes (plugged or unplugged). The electrical resistance was measured on 40 test boards with [2.sup.3] design of 8 daisy-chain PTH nets each. Full factorial analysis and analysis of variance indicate that all three factors had significant influence on PTH fatigue life, but no two-factor or three-factor interactions were found. Metallurgical analysis reveals that the failure mechanism is barrel cracking near the internal pad. This mechanism has been illustrated by a finite element analysis in this study and correlated by a SEM stereoimaging analysis in the literature. The increase of electrical resistance with thermal cycles correlates well with an analytical barrel crack model. The crack length in each net at specific cycles is calculated, but fails to match predictions from a fracture mechanics model.

A new method is presented to compute the mechanical compliance and thermal stress in surface mount leads of arbitrary shape. It involves expressing the profile polynomially by a computer-generated spline curve, and numerically integrating the generalized Castigliano formulas. Comparisons to data in the literature, generated by standard analytical and experimental techniques, show excellent agreement. An example is presented to show typical results obtainable by using the model.

The paper considers the distribution of temperature and thermal stresses in a bilayer slab which consists of a substrate and a thin coating subjected to a local short heat pulse. The heat source is distributed uniformly over a circular spot on a surface of the coating. The temperature and thermal stresses distributions in a slab are found analytically in a closed form using Laplace and Hankel integration transforms. The analytic relation is obtained for the evaluation of components of stress tensor in a coating as a function of a Fourier number and radius of heat source.

The out-of-plane displacement field around two plated holes with different pad diameters in an FR-4 printed wiring board was measured for a single 30 degrees C-210 degrees C-30 degrees C temperature cycle using electro-optic holographic interferometry. At the end of the temperature cycle, the outside edge of the pad was raised above the level of the laminate and the inside edge was depressed below the level of the laminate. This indicates that the barrel is plastically deformed in compression to a total strain of approximately 0.58-0.66 percent which is well above typical yield strains of 0.2 percent. The smaller diameter pad was inclined more than the large diameter pad, but the residual compressive strain in the barrel was roughly the same. Both the residual compressive strain and the inward inclination of the pad are in conflict with the predictions of most finite element models of plated hole deformation. However, there were cracks at the pad-barrel interface which are not included in finite element models. The residual compressive deformation of the barrel is attributed to inelastic deformation of the FR-4 matrix at the high end of the thermal cycle. The stress in the barrel was estimated using an approximate elastic analysis of pad deflections. The estimated stress for different hole diameters for the same pad diameter was roughly proportional to the ratio of their barrel plating cross-sectional areas for a 30-150 degrees C temperature change. The elastic analysis is shown to predict (unrealistic) tensile barrel stresses at the end of the full temperature cycle.

An approximate thermal contact conductance correlation which does not depend upon the surface asperity slope was developed. Published surface texture data for 65 specimens were used to establish a relationship between the average roughness and the RMS asperity slope, which was then used to develop a new approximate thermal contact conductance correlation. The investigation was conducted for a range of surface roughness typical of contacting surfaces. Comparison to limited test data and to 2080 simulated contact joints, indicates the new approximate thermal contact conductance correlation has an expected RMS error of approximately 23 percent.

During normal power cycling of the electronic equipment, the differing coefficients of thermal expansion result in differential elongations. Because each level of packaging is subject to mounting constraints, the differential strains result in bending and shear stresses. Repeated duty cycling can cause fatigue at joints, at interfaces between different materials, at interconnection locations, or cause delamination of composite materials. Accelerated Thermal Cycling (ATC) is done to simulate the fatigue failures that may arise because of this power cycling. The current practice is to determine ATC stresses by assuming that the temperatures of various layers are equal and constant. In this study, we have relaxed the isothermal assumption and we provide results for thermal stresses and strains in a first level package. This is accomplished by accurately determining the transient temperature fields in various layers of the package. Temperature variations for different heat transfer coefficients have also been calculated. The results indicate that realistic estimates of thermal stresses and strains are only possible with models that allow for temperature variation in the body of the package. High equivalent stress values are obtained at the chip-heat sink interface and in the bumps connecting the leads to the chip.

Solder is being extensively used in electronic packages for both electrical and mechanical connection. Solder joints are subjected to severe operating conditions and hence their reliability is very critical for the packages. Simulation is very effective for understanding, predicting and design improvement of electronic packages where solder is the prime joiner, however all the material response complexities of solder over the temperature regime it is subjected to should be modelled. Solder in electronic components very often operates at 0.6 to 0.8 times it's melting point. In this regime the time dependent material response (creep and stress relaxation) is very significant and can no longer be ignored. Also the plastic and creep responses of solder alloys have a very strong dependence on temperature. A nonlinear finite element based methodology has been developed for simulation of solder alloy over its full regime of material behavior, also accounting for the strong temperature dependence. This includes elastic, time independent plastic, and time dependent viscoplastic response. The methodology has been used for predicting the response of a flip chip with 95Pb5 percent Sn peripheral bumps subjected to thermal cycling. Correlation is observed between the location of failure in the bump and the maximum inelastic strain. The importance of doing a multicycle simulation is demonstrated and a direction is indicated for the bump size modification for flip chip bump design improvements for greater reliability.

The isothermal fatigue of the interconnections between a leaded chip carrier (LCC) and a printed wiring board (PWB) is studied. An LCC connected to a PWB by tin-based solder joints was subjected to simulated thermal cycling displacements at varying temperature levels. The electrical resistance of the solder specimens were also measured and recorded. It was found that high cycling frequency improved the fatigue life and reduced the hysteresis load of the specimen.

The characteristics of a high pin count plastic-quad-flat-pack (PQFP) finite element model is discussed. The PQFP three-dimensional model was developed to provide a systematic integrated circuit package analysis technique. It is capable of analyzing different thermomechanical stresses and thermal characterization experienced by the package under various manufacturing and testing conditions, such as thermal cycling and forced air cooling. These features make PQFP a very useful tool in electronic packaging systems.

A method for determining the interlaminar thermal stresses in laminated multi-layered beams under varying temperature levels is discussed. Thermal expansion in every beam layer is calculated using high-order polynomial expansions of the stress functions in conjunction with a computer program. This technique can be very useful in analyzing the thermal expansion behavior of multi-layered beams with the least amount of effort.

The influence of the bonding layer thickness of multi-layered electronic components on thermal stresses is studied. Since solder and adhesive materials are commonly used in the fabrication of electronic components, their effects on the components' performance were investigated. The alternating technique was applied on a three-layer composite strip to determine the material's thermal stress distribution with regard to its dimension. It was found that solder joint dimension has significant effects on the edge stress value.

A technique for measuring the mechanical behavior of high lead lead-tin solder joints under varying temperature levels was evaluated. Solder joint specimen were fabricated using different kinds of alpha metals and their initial characteristics were analyzed by shear and strain equations. The specimen were then subjected to various temperature profiles using a laboratory thermal chamber and their behavioral change was studied. Results indicate that the technique can be useful in investigating the thermomechanical characteristics of different kinds of alloys and related materials.

The isothermal and thermomechanical fatigue characteristics of 63Sn/37Pb solder are studied. Since solder joints undergo thermomechanical stresses during actual operation, it is important to determine the effects of such stresses. Solder specimen were subjected to isothermal and thermomechanical tests under varying temperature levels and their fatigue lives were calculated through numerical approaches. Results reveal that grain boundary is the main cause of 63Sn/37Pb solder failures.

Reliability of epoxy molding compounds used in plastic packages of integrated circuit (IC) devices depends to a great extent on the level of thermal stresses. These are due primarily to the thermal expansion (contraction) mismatch of the epoxy and the silicon materials. In this analysis we assess the effect of silica fillers on the level of thermal stresses. We conclude that thermal stresses in the compound can indeed be reduced by the application of appropriate fillers. We found that the filler volume concentration does not have to be larger than 30 percent to keep the thermal stresses at a sufficiently low level. This number is close to the filler volume concentration of 30-40 percent in commercially available molding compounds. The obtained results and recommendations can be helpful in the analysis of stresses in, and physical design of, plastic packages. [DOI: 10.1115/1.1349422]