Search

The conductivity of a nonconductive adhesive (NCA) flip chip interconnect is completely dependent on the direct mechanical contact between the integrated circuit (IC) bump and substrate pad. Cure shrinkage of NCA is critical for the formation of the final contact force in the contacts. However, measurement of the cure shrinkage during cross-linking reaction is fairly difficult. This paper introduces a new, yet simple, approach to measure cure shrinkage of adhesives using a thermo-mechanical analyzer. Isothermal studies of shrinkage change as a function of curing show four distinct regions. First, the thickness of the epoxy decreases due to decreasing viscosity and applied load, followed by a stage where the dimension change is constant as the cross-linking reaction is yet to set in. Once cross-linking begins, the shrinkage reaches a maximum followed by a plateau where the cross-linking reaction has completed. Sharp changes of the slope of cure shrinkage versus degree of cure were observed to coincide with gelation and vitrification. After gelation, a linear relationship between the cure shrinkage and degree of cure was observed to extend until the occurrence of vitrification, which quenches the cross-linking reaction. Applied load in the range of 0.05 N was found to be optimal to minimize measurement errors.

Non-conductive adhesive (NCA) flip chip interconnects are emerging as attractive alternatives to lead or lead-free solder interconnects due to their environmental friendliness, lower processing temperatures, and extendability to fine pitch applications. The electrical connectivity of a NCA interconnect relies solely on the pure mechanical contact between the integrated circuit bump and the substrate pad; the electrical conductivity of the contact depends on the mechanical contact pressure, which in turns depends to a large extend on cure shrinkage characteristic of the NCA. In addition, to reduce the cost and increase the output, NCA is usually ramped up to 200 °C and cured for 1 min during the assembly process. However, fast cure reaction poses a great challenge for the accurate measurement of cure shrinkage. In this paper, to precisely determine the cure shrinkage at high temperature, cure shrinkage was first measured at lower temperatures with slow reaction rate by means of thermomechanical analyzer and then extrapolated to high temperatures. With the increase of cure temperature, the maximum of degree of cure will increase, but the maximum cure shrinkage reduces due to the expansion of materials at higher cure temperature. Furthermore, the slopes of the linear relationship between the cure shrinkage and the degree of cure at different cure temperatures after gelation were found to be similar and independent of the cure temperature. The cure shrinkage from gel-point to complete curing was determined to be 4.275%.

A fundamental understanding of the dynamics of the PCB assembly when subjected to a half-sine acceleration has also been obtained through analyzing the PCB as a spring mass system, a beam, and a plate, respectively. The magnitude of stresses in solder interconnection due to flexing of the PCB is two orders higher than the magnitude of the stresses induced by acceleration and inertia loading the IC package. By ignoring the inertia loading, computational effort to evaluate the interconnection stresses due to PCB flexing can be reduced significantly via a two-step dynamic-static analysis. The dynamic analysis is first performed to evaluate the PCB bending moment adjacent the package, and is followed by a static analysis where the PCB bending moment is applied around the package. Parametric studies performed suggest a fundamental difference in designing for drop impact and designing for temperature cycling. The well-known design rules for temperature cycling—minimizing package length and maximizing interconnection standoff—does not work for drop impact. Instead, drop impact reliability can be enhanced by increasing the interconnection diameter, reducing the modulus of the interconnection materials, reducing the span of the PCB, or using either a very thin or a very thick PCB.

The anisotropic conductive adhesive (ACA) is a promising solder alternative candidate that shows potential for further pitch reduction. Although much work has been published on ACA joint behavior, study on correlation of material properties with reliability performance is still lacking. The main objective in this study was to identify the impact of material properties on reliability, so as to engineer highly reliable microelectronics assemblies. Four representative ACA materials (both film and paste types) with diverse properties were selected. Material properties were characterized as close as possible to "stress test" conditions so as to allow more accurate correlation predictions. Reliability performance was obtained by assembling test chips of 200-/spl mu/m pitch onto BT-substrates, then subjecting them to reliability tests. Correlation analysis was conducted and key material properties that contributed to good reliability performance were identified. Findings indicated that the best properties for high reliability assemblies were: high adhesion strength after subjecting to "stress aging", low coefficient of moisture expansion (CME) and low elastic modulus (E).

Optical adhesives combine the traditional function of structural attachment with a more advanced function of providing an optical path between optical interconnects. This article aims to characterize refractive index and birefringence of such adhesives under environmental exposure to different temperature conditions. Optical time domain reflectometery (OTDR) and prism coupling methods were employed to measure optical properties of an optical adhesive. Thermo-optic coefficient (dn/dT) of the adhesive was observed to decrease noticeably from −2 × 10−4°C−1 to −4 × 10−4°C−1 around the glass transition temperature (Tg ∼ 78°C). It is observed that refractive indices for both TE and TM modes increase with increasing annealing temperature, but the birefringence (TE − TM) is decreasing. This suggests that the material has become more isotropic due to the annealing. The environmental changes in optical properties of the adhesive are discussed in the light of Lorentz–Lorenz equations. © 2005 Wiley Periodicals, Inc. J Appl Polym Sci 98: 950–956, 2005

Driven by miniaturization, cost and environmental friendliness, non-conductive adhesive (NCA) is emerging as one of the potential alternatives for solder-less flip-chip interconnections. It offers inherent characteristics such as finer pitch interconnect, green process, good electrical performance, lower cost and low processing temperature compatible with organic substrates. A reliability evaluation of NCA for flip chip application was carried out to investigate the degradation of the NCA joint in environmental tests. Finite element analysis was also performed to understand the observed phenomenon. In order to obtain accurate material properties for precise finite element simulation, NCA pastes are characterized, as close as possible to reliability test condition, in terms of their thermo-mechanical, moisture diffusion, and viscoelastic properties. The results showed that failures occurring in NCA joints are caused by moisture induced hygroscopic swelling and stress relaxation. All these factors affect the degradation of the compressive force that maintains the mechanical contact in NCA flip chip. The common failure modes observed were interfacial delamination, bump/pad opening as well as cracking.

Adhesives interconnection systems are now commonly used in flip chip package for bonding metallic bump to substrate metallization. However, one major limitation of these adhesives is the problem of degradation in the mechanical contact between bump and substrate metallization which causes drastic increase in contact resistance. This problem is caused by the stress relaxation of the adhesive over a period of time. In addition, because these devices are subjected to variations in electrical and thermal loads, they experience corresponding cyclic mechanical straining which results in cyclic relaxation that has distinguished characteristic apart from static relaxation. The study showed that cyclic relaxation has dependency on hold time, no-load time and ramp rate. PSpice modeling has been used to understand the cyclic relaxation phenomenon in the adhesives. The accuracy of PSpice modeling used to mimic the linear viscoelasticity behavior of the adhesive has been found to be within 5% error compared to the experimental results.

The dynamic response of the printed circuit board (PCB) in a standard board-level drop impact test has been modeled as a spring-mass system, a beam, and a plate. Analytical solutions for the time-response and amplification of the deflection, bending moment, and acceleration at any point on the PCB have been developed and validated with finite element analysis. The analyses have shown that (i) the response of the PCB was dominated by the fundamental mode and (ii) the response of the PCB depends heavily on the ratio between the frequency of the PCB and the input acceleration pulse. The bending moment on the PCB has been shown beyond doubt to be responsible for the interconnection stress; the maximum moment on the PCB can be most effectively reduced through reducing the PCB thickness. The rapid receding of the higher modes in the moment response suggests that it can be adequately and effectively modeled using the standard implicit time-integration code.

According to the latest ITRS roadmap, the pitch of area array packages is expected to decrease to 100 /spl mu/m by 2009. Simultaneously, the electrical performance of these interconnections needs to be improved to support data rates in excess of 10 Gbps, while guaranteeing thermomechanical reliability and lowering the cost. These requirements are challenging, thus, needing innovative interconnection designs and technologies. This paper describes the development of three interconnection schemes for wafer-level packages (WLPs) at 100-/spl mu/m pitch, involving rigid, compliant, and semicompliant interconnection technologies, extending the state of the art in each. Extensive electrical and mechanical modeling was carried out to optimize the geometry of the interconnections with respect to electrical performance and thermomechanical reliability. It was found that the requirements of electrical performance often conflict with those of thermomechanical reliability and the final "optimum" design is a tradeoff between the two. For the three interconnection schemes proposed, it was found that the electrical requirements can be met fairly well but acceptable mechanical reliability may require organic boards with a coefficient of thermal expansion of 10 ppm/K or lower.

This paper presents a nonlinear finite element analysis on board level solder joint reliability enhancement of a double-bump wafer level chip scale package (CSP). A viscoplastic constitutive relation is adopted for the solders to account for its time and temperature dependence in thermal cycling. The fatigue life of solder joint is estimated by the modified Coffin-Manson equation, which has been verified by experimental results using one of the double-bump wafer level CSP packages as the test vehicle. A series of parametric studies were performed by changing the Sn/Ag inner bump size (UBM pad size and standoff height), the eutectic Sn/Pb external solder joint size (pad size and standoff height), pitch, die thickness, and the encapsulant thickness. The results obtained from the modeling are useful to form design guidelines for board level reliability enhancement of the wafer level CSP packages.

Moisture induced swelling and sorption characteristics of four types of epoxy molding compounds used in the packaging of semiconductor devices were experimentally investigated. The hygroscopic strain with respect to moisture content was found to be linear except at the initial stages of desorption where anomalous trends were observed. The swelling coefficient values obtained from the "stable" swelling regions were found to range from 0.3 to 0.6 (%L/%M) at 85/spl deg/C for the four types of molding compounds tested. It was also found that the swelling coefficient increased with temperature for all four types of molding compounds tested. The significance impact of hygroscopic mismatch strain was investigated by comparing the strains induced due to mismatch of the coefficient of hygroscopic swelling and the coefficient of thermal expansion for each type of molding compound when attached to a copper lead frame. The hygroscopic and thermal mismatch strains were compared using the swelling coefficient and CTE values for each type of molding compound and adjacent material (i.e., copper lead frame). Hygroscopic mismatch strains were found to be highly significant relative to the thermal mismatch strains, and they should be accounted for in the reliability modeling of packages subjected to accelerated testing. The effect of hygroscopic mismatch strains is often ignored in the reliability tests and models. In this study it is found that the hygroscopic strains can be comparable to, if not higher than, thermal mismatch strains.

This paper describes a comprehensive treatment of moisture induced failure in integrated circuit (IC) packaging with emphasis on recent advances. This includes advanced technique for modeling moisture diffusion under dynamic boundary conditions such as experienced by packages during solder reflow, autoclave, and temperature-humidity cycling; advanced characterization technique for moisture sorption and diffusion properties of packaging materials including effect of edge diffusion on transverse diffusivity, anisotropic diffusivity in organic laminates, impact of non-Fickian sorption; advanced techniques for modeling vapor pressure during solder reflow; advanced techniques for modeling dynamic delamination propagation during solder reflow; interfacial fracture strength as a function of temperature and moisture; as well as plastic analysis of popcorn cracking.

A reliable technique for characterizing the hygroscopic swelling of materials has been developed and used to characterize a number of packaging materials. Using these data, hygroscopic stress modeling were performed. The hygroscopic stress induced through moisture conditioning was found to be significant compared to the thermal stress during solder reflow. Hygroscopic stress in over-molded wire bond PBGA and molded Flip Chip PBGA was found to be 1.3 times to 1.5 times that of thermal stress. Hygroscopic swelling of the underfill in FCPBGA was found to be the main failure driver during autoclave test. Autoclave performance of FCPBGA package assembled with different underfills and chips were analyzed. Excellent correlation was found between autoclave performance and the hygroscopic swelling characteristics of the underfills.

An analytical proof has been developed which demonstrates the equality of radial and edge plastic work rate in an arbitrarily shaped plate that is collapsing in pseudo-conical mode. This new-found equality relation eliminates the need to compute the radial plastic work rate, substantially reducing the effort required to compute the upper-bound plastic collapse load using the yield-line theory.

New quadrilateral finite elements capable of continuous modelling of crack propagation are presented. The crack extension is modelled by moving the mid-node on the side of the element that coincides with the crack plane in the direction, and with the velocity of the crack propagation. Both Serendipity and Lagrangian elements are generated and prescribed with either parabolic or elliptic displacement fields, the latter capable of generating singularities. Remarkably accurate stress intensity factors for a large range of crack length/specimen width ratios are obtained using the Lagrangian elements with elliptic displacement fields. These elements have been found to be very successful in the modelling of dynamic crack propagation.

Moving node finite elements are useful in determining the stress intensity factors using strain energy concepts; in this paper the material kinetic energy is included so that the effects of the material inertia during dynamic crack propagation can be assessed. This is used to determine crack propagation speeds and these compare favourably with those quoted in the literature.

In this paper, a series of PCB strain-to-fatigue life characteristics were generated at a constant bending frequency of 100Hz for different material combinations: eutectic SnPb solder, SAC101 solder, and In/Ni-doped SAC101 solder, assembled on PCB with OSP and ENIG finish respectively. As expected, SAC101 solder exhibits similar performance as eutectic SnPb solder, and solder joints on OSP finish show better performance than solder joints on ENIG. The effect of bending frequency in the range of 30Hz to 150Hz was investigated and the fatigue life was found to decrease with increasing bending frequency. Failure analysis has shown that failures tend to become more brittle with increasing cyclic bending frequency. The PCB strain-to-fatigue life characteristics were converted to peeling stress in the solder joint -to-fatigue life (S-N) using 3D finite element modeling supported by strain rate dependent stress-strain properties of the solder alloy. The S-N characteristics so obtained were found to be independent of the bending frequency.

Due to rapid and continual changes in the components and designs of portable electronic devices, it is essential to have the ability to perform quick, cost-effective and accurate assessments of the drop impact performance of component interconnections for portable applications. In order to understand the phenomena of solder joint fatigue under impact-induced PCB bending, a tester was developed specifically for applying high bending strains at high strain rates to a PCB assembly. Fatigue characteristics of solder joints were determined for various PCB strains, allowing the construction of fatigue E-N (strain vs. cycles-to-failure) curves. Fatigue data was also generated for several bending frequencies in a study of the effect of strain rate on drop impact performance. Higher bending frequencies were found to reduce fatigue life. A comprehensive test method would also have to be capable of applying variable loading levels in order to replicate the complex loading spectrums seen in actual product drops. A variable amplitude study is presented, where crack propagation tracking shows the effect of changes in load levels on the crack growth rate. Finally, a study on the effect of low temperatures on drop impact reliability is presented, where fatigue characteristics were obtained using the bend tester fitted with a cold temperature chamber.

This paper presents a study of the resistance of solder joints to failure when subjected to strain rates that simulate the conditions of drop impact on a portable electronic product. Two test methods are used in this study: board-level drop/shock test (BLDT) and component-level ball impact test (BIT). The performance of 12 material combinations consisting of 6 solder alloys and 2 pad finishes were investigated using these two test methods, and analysis of correlations between the methods was performed. Quantitative correlation and sensitivity coefficients for the failure modes and the measured characteristics, namely, number of drops to failure for BLDT and peak load, total fracture energy, and energy to peak load for BIT, were evaluated. Analysis of the test results indicates that there is a lack of universal correlation between BLDT and BIT. Nevertheless, BIT can still serve as a test methodology for quality assurance in view of the strong correlation between the measured BIT characteristics and the failure mode. The total fracture energy parameter is preferred over the peak load and energy to peak load due to its higher sensitivity and reduced susceptibility to measurement error.

This paper describes the use of a fracture mechanics-based modeling approach to explain two interesting experimental observations in solder joints subjected to high speed cyclic bending, namely 1) the evolution of the crack front from a positive to a negative curvature; and 2) the initial acceleration and the subsequent retardation of the crack growth rate. Two- dimensional crack fronts in the solder joint were modeled, and J-integral along the crack front was computed. The modeling results have provided physical insights into the experimentally observed crack growth phenomena.

The higher stiffness of Pb-free SAC solders makes Pb-free assemblies more sensitive to drop impact. In order to be able to optimise the drop test performance, it is necessary to have better insight into the crack propagation in the Pb-free solder joints. This study combines on-line resistance measurements of a solder joint during drop testing and high speed bend testing, failure analysis of the assembly with dye-and-pry method and with cross-sections , and electrical FE simulation. The result is a fingerprint of the crack propagation during consecutive loading cycles. The carrier in the study is BGA's, a critical component family for drop impact. Combinations of solder alloys and pad finishes, SnPb on OSP, SAC305 on ENIG, SAC101 on OSP and SAC101(d) on ENIG are studied regarding the failure mode and crack propagation. This study demonstrates that for the large majority of Pb-free solder joints, there is a negligible initiation period; cracks start forming at the first PCB bending cycle. The presence of large cracks, especially at both sides can increase the compliance of the joint and slow down crack growth. Even if large cracks are present, the resistance increase is less than 1 mOmega per interconnect, which is far from the 100Omega that is often taken as a failure criterion. Brittle joints as found with SAC305 on ENIG have erratic propagation rates while ductile joints are much more predictable. Therefore, the way to optimise the drop test performance of a Pb-free BGA assembly, is to prolong the crack propagation within the ductile solder material.

The Highly Accelerated Robustness Test, HART, is an approach that allows the assessment or the robustness of adhesive interconnects at a throughput rate that is 10 times faster than the conventional accelerated humidity test, without going to higher stress levels. The test is based on the physics of a conductive adhesive interconnect. In stead of using the contact resistance drift over time in a static temperature- humidity environment, HART uses the effect of a changing temperature or humidity environment on the contact resistance as a characteristic of the robustness of the interconnect. HART using moisture as stressor is by far the most powerful approach of the two. Both the response and hysteresis in a cyclic humidity environment are proposed as new criteria for the interconnect robustness. Flip chip on flex interconnections with conductive and non-conductive adhesives are used as a carrier. Different contact material combinations and different adhesives are used in this study.

This manuscript presents some of the research works performed by a consortium of companies for the studies of the drop-impact reliability of electronic components. The extensive works covers test methodologies and failure model. Under the test methodologies, the board level high speed cyclic bend test and the component level ball impact shear test were correlated with the board level drop/shock test using 23 groups of diverse solder joints. Under the failure model, the stress-strain characteristics of 4 solder alloys were generated for the medium strain rate regime (up to 300 s-1); the S-N characteristics of 6 solder joints between the PCB fibre strains of 1x10-3 to 3x10-3 at bending frequency of 100 Hz were generated; the S-N characteristics of the solder joints as a function of bending frequency (from 30 Hz to 150 Hz) and test temperatures (-10degC and 25degC) were also investigated. The characteristics of crack propagation were correlated with the observed crack path in the solder joints. In addition, a robust solder joint design was proposed and validated.

To provide a comprehensive understanding of the field effect in silicon nanowire (SiNW) sensors, we take a systematic approach to fine tune the distance of a charge layer by controlling the hybridization sites of DNA to the SiNW preimmobilized with peptide nucleic acid (PNA) capture probes. Six target DNAs of the same length, but differentiated successively by three bases in the complementary segment, are hybridized to the PNA. Fluorescent images show that the hybridization occurs exclusively on the SiNW surface between the target DNAs and the PNA. However, the field-effect response of the SiNW sensor decreases as the DNA (charge layer) moves away from the SiNW surface. Theoretical analysis shows that the field effect of the SiNW sensor relies primarily on the location of the charge layer. A maximum of 102% change in resistance is estimated based on the shortest distance of the DNA charge layer (4.7 A) to the SiNW surface.

Closed-form solutions for the dynamic responses of the printed circuit board (PCB) subjected to half-sine support excitation and with damping have been derived for the PCB that is modeled as an Euler beam. The analytical solutions have been used to investigate the effects of frequency ratio, damping ratio and distortion of the excitation pulse. Responses of the PCB were found to decrease non-linearly with increasing damping ratio. They are also dominated by its fundamental mode while the higher modes tend to decay more rapidly.

Medium strain rate (0.1/s to 300/s) constitutive properties are needed for accurate modeling of drop impact conditions. This study presents an experimental procedure for obtaining the material properties of solder alloys at these medium strain rates. The effect of grain size on the medium strain rate behavior of solder alloys is also studied, using SnPb alloy as an example. Grain size variation is achieved through different aging conditions. For the lower strain rates tested, yield stress generally increases with aging. Interestingly, for the higher strain rates, the yield stress decreases with less severe aging but increases with more severe aging. In addition, aging causes a decrease in the strain rate sensitivity of the solder.

A high-speed shear tester has been used to evaluate the interface strength of Pb-free solder balls with varying Ag and Cu content to a BGA laminate interposer with different finishes. The shear rate used is 0.45 m/s. Components with SAC405 and SnPb solder balls have been used as a reference. Results are compared with SAC305, SAC305 with NiGe addition, Castin 258 (2.5Ag-0.8Cu-0.5Sb), SAC105 (1.0Ag-0.5Ag), SAC101 (1.0Ag-0.1Cu-0.02Ni-xIn), SACX (0.3Ag-0.7Cu-0.1Bi), LF35 (1.2Ag-0.5Cu) and Sn3.5Ag. Electroplated NiAu finishes on the BGA laminate interposer studied come from different suppliers. Three failure modes are observed in the high speed shear test: brittle fracture in the intermetallic layer at the ball/interposer interface, ductile fracture in the solder bump and peel off of the solder pad from the interposer. Analysis of the fracture interface shows that almost all Pb-free solder alloys fail either in the intermetallic layer or by pad peel off. Only SAC101 and Sn3.5Ag show ductile fracture in the bump or pad peel off, but no fracture in the intermetallic layer. This is comparable with results for eutectic SnPb under these test conditions. A JEDEC board level drop test on same packaging with SAC101, SAC305 and SnPbAg shows that the number of drops to failure for packages with SAC 101 is also higher than those with SAC 405 and even slightly higher than with eutectic SnPb solder balls. Failure analysis is done to understand the difference in shock resistance with different solder bumps. From the observations in both of the tests, it is evidenced that the high speed shear test, when combined with proper analysis of the failure mode, is a simple but very powerful tool to evaluate the resistance of Pb-free BGA solder joints to brittle fracture under shock impact.

In portable electronics or during the transport of other electronics, shock impact like drop is not easy to be avoided. Therefore, the ability of electronic packages and assemblies to resist the sudden shock impact is becoming a growing concern. Recently, some outer balls of lead free BGA packages even ' tended to come off almost spontaneously during handling/transportation and the failures have typically a brittle nature. To overcome this, some efforts were done to improve the solder interconnects robustness to resist the shock. In this paper, a high-speed shear test is carried out to evaluate the robustness of lead-free solder interconnects of BGA packages under a shock impact. Five Pb-free solder compositions of SAC405, SAC305 + NiGe, CASTIN, and SACX are evaluated with eutectic SnPb as a reference. All BGA packages in the tests are with NiAu finished substrates. It is found that interconnects with reduced silver and copper content (SAC 101) outperform interconnects with other Pb-free solder alloys in the high-speed shear test. From the alloys studied, only BGA packages with SAC101 don't show any brittle failure in the IMC interface similar to packages assembled with eutectic SnPb. A JEDEC based board level drop test is also carried to see the shock impact resistance of SAC101 with SAC305 and SnPbAg as references. In the drop test, the number of drops to failure for packages with SAC101 is also higher than those with SAC 305 and it is similar to the number with eutectic SnPb solder balls. It also shows that packages with Cu-OSP finished substrates outperform those with NiAu finished substrates.

A wafer level packaging technique has been developed with an inherent advantage of excellent solder joint co-planarity essential for wafer level test and burn-in. A suitable weak metallization scheme has also been established for the detachment process. During the fabrication process, the compliancy of the solder joint is enhanced by stretching the solder joint to achieve high aspect ratio and small shape factor. Thermal cycling reliability of these hourglass-shaped, stretch solder joints has been found to be considerably better than that of the conventional spherical-shaped solder bumps.

Good correlation has been established between high speed shearing of solder joint at component level and board level drop tests, endorsing high speed shearing as a viable quality assurance test for manufacturing and incoming inspection. The high speed shear characteristics of solder joints under different test conditions (shear speed, shear angle, and temperature) and aging conditions (multiple reflow, temperature humidity, and salt spray) have been evaluated. Preliminary S-N characteristic for SnPb/spl I.bar/OSP and SnAg/spl I.bar/OSP solder joints have been generated using high speed cyclic bends test. These could be devolved into a life prediction model for board level solder joints in product drop impact.

This study performs experimental tracking of crack propagation (stage II fatigue) in a single solder interconnection during drop impact. A high resolution, highspeed four-point resistance measurement system is used for tracking crack progression. Results indicate that most of the drop impact low-cycle fatigue life of the solder joint is spent in the crack initiation stage. Cross-sections and fractographs show that various mixes of intermetallic and bulk failure are possible within a single joint. Several failure mechanisms are suggested. Comparisons of failure modes are also made between drop impact, quasi-static loading and high-cycle fatigue. A thorough understanding of failure mechanisms is important for the development of failure criteria and life prediction models for drop impact.

The objective of this study is to obtain experimental failure models governing solder joint failure during drop impact testing of board assemblies. A high-speed bend test was developed to perform displacement-controlled bend test of board assemblies at the high flexing frequencies of drop impact. These test frequencies and amplitudes are not achievable by conventional universal testers. Experimental data was obtained for various PCB strain amplitudes, flexural frequencies solder alloys and pad finishes. Results from the high-speed bend tests are used to construct constant amplitude power law fatigue curves. Solder joint reliability found to be dependent on the test frequency, and therefore strain rate. The experimental failure data from these high-speed bend tests are a required basis for a drop impact failure criterion which can take into account frequency and amplitude effects and which is general enough to be applied to product level testing.

A cost-effective wafer level packaging technique termed "stretch and break", based on stretching and detachment of solder interconnections, has been established. Excellent co-planarity, essential for wafer level test and burn-in, is inherent in the process. The technique allows the freedom to use solder materials of up to 400/spl deg/C melting temperature for forming the interconnection. The shape of the interconnection can also be easily manipulated for optimum performance. The mechanical and thermal cycling reliabilities of the stretched solder interconnection have been found to be significantly better than those of conventional solder joints.

This paper provides a comprehensive summary of research activities focusing on the reliability of board level interconnections in drop impact, covering experimental work together with analytical and numerical modeling studies. These activities involve investigating (i) the dynamics of the PCB in product level and board level drop impact; (ii) the failure drivers: inertia force, differential flexing, and knocking; (iii) the relations among the PCB flexural modes, PCB strain, and interconnection stress/strain; (iv) high-strain-rate material characteristics; and (v) failure models

This paper presents a comprehensive methodology to model the static temperature-humidity (TH) ageing test (85°C/85%RH over 1000 hours) of flip chip on flex interconnections with non-conductive adhesives (NCAs). Nonconductive adhesives, being a special form of conductive adhesives, are chosen, as they allow bringing the pitch further down. The methodology combines experimental techniques for material characterization, finite element modeling (FEM) and model validation. A non-conductive adhesive (NCA) has been characterized using several techniques. The thermomechanical properties and the moisture properties were obtained for the NCA. A temperature dependent viscoelastic constitutive model was also obtained for the NCA. The viscoelastic model was defined by the Prony series expansion. The shift factor was approximated by the Williams-Landel-Ferry (WLF) equation. Finite element modeling has been performed to analyze the flip chip interconnects on flex with the NCA under process condition and reliability ageing conditions. The viscoelastic constitutive relation has been used to model the NCA in ageing modeling. An integrated process-ageing modeling methodology has been developed to combine the thermo-mechanical stress and hygro-mechanical stress, followed by stress relaxation analysis. To verify the finite element models, the static TH ageing test (85°C/85%RH) were also performed. The contact resistance was monitored with high measuring resolution during the accelerated test. The simulation results are good agreement with the experimental results. The approach developed in this paper can be used to provide guidelines with respect to adhesive material properties, assembly process parameters and good reliability performances.Copyright © 2005 by ASME

This paper presents package warpage and interconnect fatigue life estimates of stretched-solder-column interconnections in a 20mm/spl times/20mm package with 100 /spl mu/m pitch. The modelling was carried out in two stages using the macro-micro modelling approach. The effect of interconnection height, as well as chip thickness and substrate CTE (coefficient of thermal expansion) on fatigue life is discussed. The results suggest that the substrate CTE will continue to be the key factor in determining the thermomechanical reliability of a large package with small interconnections.

Accurate material models are necessary to obtain meaningful results in computational simulation. The results of materials testing and the process of developing a material model for solder under impact loading are presented in this paper. The material model is to be used in the simulation of electronic packages under drop impacts. As such, it is necessary to incorporate the effects of strain rate sensitivity in the model. Dynamic material properties of Sn63/Pb37 solder were obtained by testing the solder using the split Hopkinson pressure bars (SHPB). The accuracy of the material models was evaluated by subjecting single solder balls to impulsive loads on the SHPB and comparing the test results with FEM. Examination of the fractured surfaces showed that there is a transition from ductile to brittle fracture as strain rate is increased. SHPB tests on single solder balls also showed that the stiffness of the solder balls is strongly dependent on rates of deformation.

Many failures in plastic IC packages involve delamination. Fracture mechanics has been applied to study such failures. However, in order to be able to predict delaminations, the adhesion strength or more specifically the fracture toughness of the particular interface needs to be measured as a function of temperature, moisture content and mode mixity. Several methods have been described in the literature but most of them are too complex for wide implementation by industry. In this paper, a new and relatively easy method for measuring the fracture toughness of a polymer-metal interface is presented. It is a modifcation of the widely-used button shear test. The same equipment can be employed although the specimens needs to be modified. It is shown that a range of mode mixity can he obtained by varying the shearing height.

As microsystems continue to move towards higher speed and microminiaturization, the demand for interconnection density both on the IC and the package increases tremendously. With the shift towards nano ICs by 2004 with

This paper presents a comprehensive methodology to model the assembly process of flip chip on flex interconnections with non-conductive adhesive (NCA). The methodology combines experimental techniques for material characterization, finite element modeling and model validation. The non-conductive adhesive material has been characterized using several techniques. A unique experimental technique has been developed to measure the cure force. A 2-D axisymmetric finite element model is used for analysis of flip chip on flex package with nonconductive adhesive, which takes into account assembly force, cure shrinkage, adhesive modulus buildup, removal of assembly force and cooling down to room temperature. The relationship between the contact pressure obtained from finite element simulation and the measured bump contact resistance has been established through the development of a dedicated experimental setup, which uses a microforce-tester combined with a digital multimeter and nano-voltmeter. The process modeling has been validated by comparing the predicted contact resistance value and the measured contact resistance value after assembly process. The approach developed in this paper can be used to provide guidelines with respect to adhesive material properties, assembly process parameters and good reliability performances.

Solder alternative technologies for flip-chip interconnections are fast emerging due to the drive for environmental friendly processes. The anisotropic conductive adhesive (ACA) is one promising solder alternative candidate that shows potential for further pitch reduction. The main concern with ACA interconnection is their long-term reliability performance especially in humid environments. Although much work has been published on ACA joint behaviour, studies on correlation of material properties with reliability performance is still lacking. The main objective in this study was to identify the impact of material properties on reliability, so as to engineer highly reliable microelectronics assemblies. Four representative ACA materials (both film and paste types) with diverse properties were selected and reliability tests were carried out. Correlation analysis identified key material properties that contribute to good reliability performance. Findings indicate that the best properties for high reliability assemblies are: high adhesion strength after subjecting to reliability test conditions, low coefficient of moisture expansion and low storage modulus.

Quasi-static bend tests and drop tests of surface mounted MCMs on printed circuit boards (PCBs) are conducted to compare the failure patterns from the two types of tests. The boards are 120/spl times/90 mm in dimension. The modules are 40/spl times/40 mm with a brass heat spreader of the same dimension on top. Five-point bend tests are conducted with different ramp-rates from 0.1 to 10.0 mm/min to determine the effect of different PCB strain rates on the reliability of the solder interconnections. Drop tests are conducted for the same batch of PCBs at 1.0 m and 1.5 m drop heights and PCB strains from drop tests are compared with those taken from the quasi-static tests. A failure analysis is also conducted to determine the possible mechanisms of failure. Findings show that solder joint failure in five-point bend tests is dependent on the ramp-rate. The tests also show that mass inertia of the heat-spreader may cause solder joints to fail at lower PCB strain levels than expected especially for packages with heavier heat-spreaders which are needed for higher heat dissipation.

There is a need for more detailed studies on the drop impact reliability of electronic packages, in view of the fact that solder interconnections are becoming smaller, and portable electronic products such as mobile phones can be easily dropped during usage. This paper investigates the mechanical response of printed circuit boards (PCBs) inside portable consumer electronic products, when such products are subjected to drop impact. The response of the PCB is studied because it is one possible factor causing the failure of the interconnections with IC packages. The board response is measured via accelerations and strains at specified locations on the board. The drop test experimental results are consistent with drop orientation and physical structure of the products.

In this paper, we report the experimental results for calibration of a 2D piezoresistive stress sensor using a four-point bending fixture. The sensor on the [100] test chips was able to accurately measure plane stress components in a temperature compensated manner. The resistance of stress sensors was found to vary linearly with the applied stress. The piezoresistive coefficient was calculated and found to coincide with the reported values for silicon. A further study of the thermally induced stresses was also carried out to determine the resistance change that varied linearly with temperature.

Moisture diffusion characterization at high temperatures in polymeric packaging materials is important since moisture adsorption of polymeric packaging materials plays a determining role in "popcom cracking" of IC packages during the solder reflow process. Karl Fischer Titration is proposed as an effective method for characterizing moisture desorption of polymeric packaging materials in this paper. Samples of molding compound, underfill and die attach materials were preconditioned at 85C/%85RH till saturation. The rate of moisture diffusion was then measured at 5 different temperatures using both Karl Fischer Titration and Thermo-gravimetric Analysis (TGA) methods and compared. The differences at high temperatures was attributed to the outgassing of polymeric materials at high temperatures. This was confirmed using Gas Chromatography/Mass Spectrometry (GC/MS). This questioned the use of TGA or any other conventional gravimetric methods for measuring moisture desorption properties at high temperatures. The differences at lower temperatures was attributed to the low sensitivity of Karl Fischer Titration at relatively lower temperatures.

Industry demand for high power multi-chip packages is increasing to realize multifunctional compact systems. In line with industry requirements a multi-chip package suitable for high performance devices is developed. The package has 1296 I/Os. It is capable of handling 2 GHz signal speed and dissipating 75 W power with an external thermal solution. A concurrent design approach is adopted for the package design. Electrical, thermal, structural issues and assembly, and materials limitations are considered in developing this package. Package level thermal design challenges involve optimization of package structure and selection of stable thermal interface material. As a part of this project, a simulation model of the package along with the external thermal solution is developed and validated by measurements. The optimized heat spreader has been designed by performing a parametric study with the validated model. The type of thermal interface material (TIM) suitable to the package has also been identified by measurements. Desired thermal performance of the package has been achieved through design optimization of the package and selection of suitable TIM. Thermal modeling and measurement methodologies, validation and parametric study results are presented in the paper. Analysis of TIMs and measured thermal performance results of packages assembled with different type of TIMs are also discussed in the paper.

The rate of moisture diffusion in the IC packaging materials was found to be dependent not only on the concentration gradient as described by Fick's Law, but also on the moisture concentration itself. This non-Fickian behaviour poses a challenge for the characterisation of moisture diffusivity of packaging materials which is conventionally based on Fick's diffusion law. This paper presents an integrated non-linear finite element diffusion modeling and optimisation procedure that is capable of characterising moisture diffusivity as a function of moisture concentration from a single moisture sorption experiment.

Popcorn package cracking is a major package reliability issue for surface mount IC packages. Although many works have been done in the quest for a solution to this problem, the success is only limited. This paper reports on the Type II cracking mechanism in a TQFP-208 package and discusses a new leadframe design which can pass the stringent JEDEC J-std-020 Level 1 moisture sensitivity test. During soldering reflow exposure, delamination was found to initiate at die pad shoulder area and propagated into the die attach/die pad interface. In the worst case, the delamination extended to mold compound cracking at the tip of the die pad. Finite Element Analysis (FEA) suggests that the interfacial stresses along die attach/die pad interface is highest at the die attach fillet and the mold compound next to it. A die attach with a higher modulus provides stronger coupling between the silicon die and copper die pad, reduces relative displacement and hence preventing initiation of mold compound/die pad interfacial delamination at die pad shoulder area. However, this is only true if the die attach is also a strong adhesive to the die pad. To overcome this constraint, a new leadframe design is made with die pad which eliminates the die pad shoulder or die attach fillet. One of such design is a "small window pad" which minimises die pad area yet provides good mechanical stability for mold encapsulation and handling during assembly process prior to molding. The new die pad design is found to be applicable to both TQFP as well as QFP packages and passes the JEDEC J-std-020 Level 1 moisture sensitivity test.

This paper presents a new methodology for the mechanical characterisation of solders used in integrated circuit (IC) packages. The methodology is formulated on the basis of package-level solder joints (BGA packages) to properly account for the effects of microstructure and intermetallic compound distribution. The methodology is implemented on the lead-free Sn-3.5Ag solder. In the indentation tests, the lead-free solder is found to be sensitive to the applied strain rates as well as the thermal cycling conditions. Finite element analyses, together with the experimental results are used to extract the elastic modulus and the elastic-plastic constitutive relations. The methodology has shown that the extracted properties are in good agreement with other publications.Copyright © 2003 by ASME