Wen-Hwa Chen

IMC growth reaction and its effects on solder joint thermal cycling reliability of 3D chip stacking packaging

Abstract The study aims at assessing the growth reaction of the Ni3Sn4 intermetallic compound (IMC) during bonding process and its dependences on the thermal-cycling reliability of the Cu/Ni/SnAg micro-joints of an advanced 3D chip stacking package under accelerated thermal cycling (ATC) loading. The growth reaction of the IMC during bonding process is also predicted through experiment and classical diffusion theory, and the relation between the IMC thickness and bonding process temperature and time is derived according to the predicted activation energy of the chemical reaction between Sn and Ni by experiment. Moreover, the micro-joint reliability prediction is made using finite element (FE) analysis incorporated with an empirical Coffin–Manson fatigue life prediction model and also ATC experimental test. To facilitate the FE modeling, the temperature-dependent thermoelastic properties of both single crystal and polycrystalline Ni3Sn4 IMC are characterized through molecular dynamics simulation and the Voigt–Reuss bound and Voigt–Reuss–Hill approximation. Results show that monoclinic single crystal Ni3Sn4 reveals a high elastic anisotropy or direction dependence of elasticity. The diffusion reaction of Sn and Ni exhibits that a longer bonding process time and a higher bonding temperature could not only increase the IMC thickness but also vary its surface morphology. In addition, the thermal–mechanical performance of the micro-joints is strongly affected by the geometry and material of IMC layer, where IMC with a thicker thickness, a less Young’s modulus, a smaller CTE and even a more rounded surface morphology can better the reliability.

Strain- and strain-rate-dependent mechanical properties and behaviors of Cu3Sn compound using molecular dynamics simulation

This work aims at investigating the mechanical properties and behaviors of orthorhombic Cu3Sn crystals at room temperature through molecular dynamics (MD) simulation. The focuses are placed on the tensile stress–strain behaviors and properties of the Cu3Sn single crystal and also their dependence on applied strain and strain rate. An attempt to characterize the deformation evolution of the Cu3Sn nanostructure during the stress–strain test is also made. In addition, the elastic properties of bulk polycrystalline Cu3Sn are estimated, as a function of strain rate and applied strain, by using the monocrystal results. The effectiveness of the MD model is demonstrated through comparison with the nanoindentation results and also published theoretical and experimental data. The calculated orthotropic elastic and shear moduli and Poisson’s ratio of Cu3Sn single crystal reveal not only high anisotropy, but also the great effects of applied strain and strain rate only as the strain rate exceeds a threshold value of about 0.072% ps−1. Specifically, raising the strain rate increases the orthotropic elastic properties and also the ultimate tensile and shear strengths of the nanocrystal, whereas increasing the applied strain reduces them.

Effects of Bonding Parameters on the Reliability of Fine-Pitch Cu/Ni/SnAg Micro-Bump Chip-to-Chip Interconnection for Three-Dimensional Chip Stacking

As the demands for portable electronic products increase, through-silicon-via (TSV)-based three-dimensional integrated-circuit (3-D IC) integration is becoming increasingly important. Micro-bump-bonded interconnection is one approach that has great potential to meet this requirement. In this paper, a 30-μm pitch chip-to-chip (C2C) interconnection with Cu/Ni/SnAg micro bumps was assembled using the gap-controllable thermal bonding method. The bonding parameters were evaluated by considering the variation in the contact resistance after bonding. The effects of the bonding time and temperature on the IMC thickness of the fabricated C2C interconnects are also investigated to determine the correlation between its thickness and reliability performance. The reliability of the C2C interconnects with the selected underfill was studied by performing a -55°C- 125°C temperature cycling test (TCT) for 2000 cycles and a 150°C high-temperature storage (HTS) test for 2000 h. The interfaces of the failed samples in the TCT and HTS tests are then inspected by scanning electron microscopy (SEM), which is utilized to obtain cross-sectional images. To validate the experimental results, finite-element (FE) analysis is also conducted to elucidate the interconnect reliability of the C2C interconnection. Results show that consistent bonding quality and stable contact resistance of the fine-pitch C2C interconnection with the micro bumps were achieved by giving the appropriate choice of the bonding parameters, and those bonded joints can thus serve as reliable interconnects for use in 3-D chip stacking.

Drop Impact Reliability Analysis of 3-D Chip-on-Chip Packaging: Numerical Modeling and Experimental Validation

This paper aims at investigating the drop impact solder interconnect reliability of an advanced ultra-fine-pitch 3-D integrated circuit chip stacking packaging in accordance with JEDEC board-level test specification through finite-element (FE) simulation and experimental testing. To characterize the transient dynamic responses of the package, ANSYS/LS-DYNA incorporated with the Input-G method is applied. To well simulate the mechanical behaviors of the solder interconnects, a strain-rate-dependent elastoplastic Johnson-Cook constitutive model for the Sn3.5Ag solder is applied. In addition, an inverse calculation is carried out to identify the overall structural damping of the dynamic system. Furthermore, a JEDEC-compliant drop tester is used to conduct the drop test, where failure analysis is performed using an optical microscope. Moreover, a simplified Darveaux fatigue life prediction model is constructed based on the calculated strain energy densities at different JEDEC test conditions together with the corresponding drop test data. To demonstrate the validity of the developed fatigue life prediction model, a confirmatory experiment is performed. Finally, parametric FE study incorporated with experimental design is performed to seek a design guideline for enhanced solder interconnect reliability under drop impact. Both the experimental and simulation data reveal that underfill can greatly enhance the drop impact reliability of the solder interconnects. In addition, the cornered interconnects and even package in a one-component configuration would fail earlier than the central ones, with a cohesive fracture in the Sn3.5Ag solder rather than the intermetallic compound (IMC) and its interface. Besides, an increasing IMC thickness reduces the drop impact solder interconnect reliability while enhancing the thermal cycling one.

Interconnect Reliability Characterization of a High-Density 3-D Chip-on-Chip Interconnect Technology

This paper investigates the solder interconnect reliability of a high-density 3-D chip-on-chip technology under an accelerated thermal cycling (ATC) test condition through finite element (FE) modeling and experimental validation. The fabrication of the 3-D chip-on-chip technology is accomplished with a two-step gap control bonding process to minimize the solder squeezing phenomenon. The alternative goal of this paper is placed on the influences of underfill on the interconnect failure mechanism and reliability. With the calculated plastic strain, the thermal fatigue life of the most critical solder interconnect can be estimated through an empirical Coffin-Manson fatigue life prediction model. The effectiveness of the proposed FE modeling is demonstrated through ATC tests. Finally, to identify the parameters most affecting the lead-free solder interconnect reliability, both parametric FE analysis and a simulation-based experimental design scheme based on a response surface methodology are carried out with the validated FE model. Both the numerical and experimental results that underfill can not only change the interconnect failure mechanism from an interfacial crack between the Al pad and the copper (Cu) layer of the under bump metallurgy to a cohesive solder failure, but also greatly improve the solder interconnect thermal fatigue life by as much as 2.5 times. Furthermore, the experimental design demonstrates that both the Youngs modulus of intermetallic compound and thermal expansion coefficient of underfill are identified as the parameters most affecting the solder interconnect reliability of the 3-D chip-on-chip technology.

Hygro-thermo-mechanical Behavior of Adhesive-BasedFlexible Chip-on-Flex Packaging

Although adhesive-based chip-on-flex (COF) packaging technologies have many advantageous features, such as flexibility and compatibility with standard semiconductor and microelectronics packaging processes, the low hygro-thermal resistance leads to reliability concerns. Thus, finite element (FE) modeling and experimental testing have been used to investigate the effects of temperature and humidity conditions on the hygro-thermo-mechanical behavior of a thin flexible anisotropic conductive adhesive (ACA)-based COF packaging technology. The investigation starts from process modeling of the thermo-mechanical behavior of the technology during the ACA bonding process. The validity of the process modeling is demonstrated by temperature and warpage experiments. Furthermore, three-dimensional (3-D) transient moisture diffusion FE analysis through a thermal–moisture analogy based on the “wetness” technique is performed to evaluate the moisture distribution, in which the moisture properties of the polyimide (PI) substrate are obtained through a moisture absorption experiment. Then, the effect of the moisture properties of the ACA adhesive and PI substrate on the moisture diffusion behavior is examined. Finally, following process modeling, 3-D hygro-thermo-mechanical FE analysis under a constant temperature and humidity condition is undertaken to assess the influence of hygro-thermal aging and stress relaxation of the ACA adhesive on the long-term contact performance of the interconnects.

Crystal size and direction dependence of the elastic properties of Cu3Sn through molecular dynamics simulation and nanoindentation testing

Abstract The study aims at exploring the elastic properties of orthorhombic Cu 3 Sn crystals through a proposed molecular dynamics (MD) simulation model based on the modified embedded atom method (MEAM) and nanoindentation testing. The focuses of the study are placed on their dependence on the crystal size and direction. The electronic nature of single crystal Cu 3 Sn is also examined by using first-principles calculations based on density function theory (DFT). According to continuum mechanics, the elastic stiffness coefficients of the single crystal Cu 3 Sn are derived from the calculated energy, and used in the generalized Hook’s law in compliance form to compute the associated elastic constants. The simulated elastic properties are compared with the results of the published first-principles calculations. For comparison with the present nanoindentation finding and the other published experimental data, the effective elastic properties of the polycrystalline Cu 3 Sn together with their size dependence are also derived using the Voigt–Reuss bounds and Voigt–Reuss–Hill average based on the calculated single crystal data. The simulation results show that the orthorhombic Cu 3 Sn crystals exhibit a high elastic anisotropy, which has been also confirmed by the electronic structure analysis, and also a strong size and direction dependence of elasticity. In addition, the calculated effective elastic properties of the polycrystalline Cu 3 Sn agree well with the present nanoindentation results and the published theoretical/experimental data.

Investigation of Electrical Contact Mechanism for Anisotropic Conductive Adhesive Joints After the Thermocompression

Flexible interconnects are needed that meet assembly requirements for future applications in flexible consumer electronics products. This work investigates the electrical contact mechanism of ultrathin chip-on-flex (UTCOF) package using anisotropic conductive adhesive (ACA), which is highly flexible. In this paper, a 3-D nonlinear finite element (FE) model, which integrates analytical models of ACA joints and the thermal-mechanical behaviors of the UTCOF, is presented. The model is then applied to simulate the electrical contact mechanism for various ACA joints after thermocompression. Moreover, a “death-birth” meshing scheme is utilized to determine the effect of ACA resin temperature on contact resistance of the ACA joints. Multiple particle models are also generated using the ANSYS program. To validate the suitability of the proposed FE analytical model, contact resistance is measured to determine the bonding quality for 80-μm -pitch UTCOF test samples. The interfaces between the silicon chip and substrate for samples bonded under different bonding pressures are observed using cross-sectional scanning electron microscopy. The contact resistances obtained from the 3-D FE models were in good agreement with experimental results and can be used to predict the effects of thermocompression loading, pressure unloading, and cooling to room temperature on electrical contact mechanism in ACA joints after thermocompression. Generally, the gold bump and the compliant bump joined with multiple conductive particles reached stable contact resistance by capturing five to nine conductive particles. Contact resistance of an ACA-bonded UTCOF can be estimated using a proposed empirical model equation, particularly under loading conditions of L = 40%-60%. Overall, this work improves the design of flip chips for flexible electronics and the understanding of the electrical contact mechanism for ACA joints.

Thermal Chip Placement in MCMs Using a Novel Hybrid Optimization Algorithm

This paper reports on enhancing the thermal performance of multiple-chip modules (MCMs) that contain a number of chips under natural convection by optimizing the chip placement layout. To attain this goal, an innovative hybrid optimization approach (HOA) incorporating a genetic algorithm (GA) into an algorithm based on a response surface method (RSM) is introduced for improving the performance of both these algorithms. The GA in the proposed HOA is responsible for not only evolving the population toward better fitness value but also, based on the newly evolved populations at each GA generation, for continuously updating the RS mathematical model for better approximation of the chip junction temperature. The sum of the mathematical expressions representing the total system temperature defines the objective of the optimization problems. For each GA generation, a constrained quadratic optimization subproblem is formed, based on the newly updated approximate RS mathematical model as the objective function along with the specified constraints. The solution of the optimization subproblem is sought through a mathematical programming model. As the genetic RS optimization progresses, a sequence of approximate solutions associated with the continuously updated RS mathematical models is constructed. The iterative process continues until convergence of the approximate solutions is attained. To demonstrate the effectiveness of the developed algorithm, several thermal design problems associated with two types of MCMs with equal/unequal power are performed. The obtained results are compared with those derived using two conventional approaches-the GA-based and the RSM-based optimization techniques. Results show that the developed algorithm can provide good optimal solutions with much less computational effort, and the larger the scale of the design problems, the more significant the improvement in the computation cost.

Epitaxial growth of Cu thin films on atomically cleaned (111)Si at room temperature

Abstract Epitaxial Cu thin films have been grown on (111)Si at room temperature in an ultrahigh vacuum environment. Plan-view and cross-sectional transmission electron microscopy revealed that both aligned and twinned epitaxy were present. An interface compound, about ten atomic layers in thickness, was observed to be present at the Cu Si interface. From atomic image and diffraction analysis, the intermediate layer was identified as ζ phase, which is of h.c.p. CuSix structure with x = 11.2−14 at.%. Interfacial dislocations at the silicide/Si interface were identified as edge type, with 1 2 〈110〉 Burgers vectors. The average spacing of the dislocations was measured to be 1.4 nm, which correlates well with a 15% mismatch at the silicide/Si interface.