Lead-free ceramic ball grid array: Thermomechanical fatigue reliability

Flip-chip carriers have become the preferred solution for high-performance, application-specific integrated circuit and microprocessor devices. Typically, these are packaged in organic or ceramic ball grid array (BGA) packages, which cover a wide range of package input/output (I/O) capabilities required for high-performance devices, typically, between 300 to more than 1,600 I/O. Recently, there has been a move toward Pb-free solders as replacement alloys for standard, eutectic Sn/Pb and other Pb-based BGAs. The leading solder that has emerged from various Pb-free solder evaluations by industry and academic consortia is the Sn/Ag/Cu (SAC) alloy. One of the primary issues with changing solders is the reliability of the joints when these are subjected to thermomechanical fatigue (TMF). This evaluation has previously been conducted on SAC ceramic ball grid array (CBGA) assemblies in a 1.27-mm pitch.¹ However, with the need to shrink the I/O pitch to accommodate higher wiring density, it has become increasingly important to conduct TMF reliability assessments in a 1-mm pitch format. This paper describes such an evaluation conducted using SAC BGA assemblies. The results show that, for a 1-mm pitch, the Pb-free SAC CBGA solution provides superior reliability as compared to the standard Sn/Pb CBGA solutions. This finding is an added incentive for a new CBGA offering employing the new Pb-free, SAC, single-alloy, self-aligning system.     

Simulation of fatigue distributions for ball grid arrays by the Monte Carlo method

Any approach to qualification of advanced technologies during product development must include an assessment of variation expected in product life over the life cycle. However, testing product design options in development, to approach an optimal design is costly and time consuming. Hence, simulation of product life distributions for virtual qualification can be a valuable tool to evaluate and qualify design options. This paper presents a physics of failure-based approach to virtual qualification of advanced area array assemblies against solder fatigue failure. The approach applies Monte Carlo simulation to evaluate solder joint fatigue life distributions, given material property variations and manufacturing capabilities. Preliminary results using the simple Engelmaier model as the basis of simulations are presented. Simulation results are compared to data accumulated from two test environments and two ball grid array product types. The results reveal some of the limitations of the Engelmaier model as a basis for simulation. They also show the potential of this approach to virtual qualification for design and manufacturing capability assessment in development.     

Lead-free plastic area array BGAs and polymer stud grid arrays package reliability

Many packaging and original equipment manufacturers (OEMs) have initiated a program of introducing lead-free electronics. Although lead usage in the packaging industry is relatively small, major efforts are ensuring to eliminate lead usage. In this context, we manufactured and qualified two “green” package solutions: lead-free low-profile fine pitch ball grid array (LFBGA) and polymer stud grid arrays (PSGA^TM) area array packages.The primary source of lead in BGA is in the solder balls. Lead-free solder alloys are readily available, although there is no universal drop-in replacement identified so far for plastic BGAs. One of the most promising alloys for this purpose appears to be the eutectic Sn95.5Ag4Cu0.5, although it involves higher processing temperature. Based on a design of experiment (DOE), the best reflow profile has been determined and used. A lead-free 8×8 mm² LFBGA has been manufactured without other process or equipment changes, and has been qualified for JEDEC moisture level 3, with 245°C reflow simulations, and standard 1st level package reliability tests.An alternative to lead-free plastic BGA is the PSGA^TM. The PSGA^TM package consists of a polymer injection-molded 3D body with a location for chip mounting and polymer studs. The metallized studs, which are by nature lead-free, replace the solder balls within the BGA package. Moisture conditioning and reliability analysis was performed on 8×8 mm² PSGA^TM. The package passed JEDEC moisture conditioning level 3 as well as standard reliability tests.     

粘结材料对TCBGA封装组件弯曲可靠性的影响

采用三点弯曲试验,测试了十组使用不同粘结材料及不同粘接方式的薄基板球栅阵列封装(TCBGA)组件和一组未使用粘结材料的组件的弯曲可靠性。结果表明,这些粘结材料均不同程度地提高了组件的弯曲可靠性。不同粘接方式对弯曲可靠性影响不同,其中边沿绑定粘接方式效果最佳,相对于无粘结材料,对组件所能承受的最大主应变、位移和荷载分别增加了41.73%,40.86%和37.50%。     

Effect of the plasma cleaning process on plastic ball grid array package assembly reliability

Interface delamination is recognized as one of the major failures of microelectronics packaging. It can result from various factors, including stresses from mismatch of adherent materials, hygrothermal stress from the release of vapor pressure of moisture during soldering reflow process, and interface material adhesion strength. The failure mechanisms are associated with cyclic loads, temperature and moisture condition as well as interface adhesion strength degradation. This paper focuses on the evaluation of plasma cleaning on PBGA assembly, including resistance to interface delamination. Two different plasma systems, powered by radio frequency (RF) and microwave (MW) energy, are studied. The optimized plasma cleaning process parameters are obtained by surface contact angle measurements. The plasma cleaning results are also verified by scanning electron microscopy (SEM) as well as physical pull and shear tests. The test vehicles are 27/spl times/27 mm 292-ball PBGAs. The results from encapsulation peel tests, die and encapsulant pull tests, bonding wire pull tests and C-Mode SAM (C-SAM) examination are presented. It is clear that an optimal plasma cleaning process can be achieved with different plasma systems. The experimental results also demonstrate that plasma cleaning has little effect on wire bonding process and die attach pull strength for given substrates and assembly materials. In all the cases, optimal plasma cleaning steps improve PBGA resistance against interface delaminations for cases where plasma cleaning is carried out before encapsulation process. Moreover, different plasma cleaning techniques would affect the assembly productivity, investment and yield. This paper demonstrates that the optimized plasma cleaning process would enhance PBGA package qualification level and improve the process yields and productivity.     

Failure analysis and stress simulation in small multichip BGAs

This paper examines one of the common modes of structural failure in multichip ball grid arrays (BGAs), determines its locations within the package structure, relates it to the stresses generated in the reliability tests under which it occurs, and by finite element simulations, determines an explanation for the failure, and finally proposes a method to avoid this failure mechanism. Several designs of multichip BGA substrates were manufactured and production silicon assembled into them. These were all 14 mm×22 mm 119 ball PBGAs. These were subjected to a set of package reliability tests, until some units failed electrical test. The failed units were analyzed and the physical location and shape of the failure was determined in many cases. From this information, the mechanical mode of failure for each unit was determined. In addition there was sufficient information in some of the analyses to provide definite suggestions as to the mechanism of failure. Meanwhile, finite element analysis was performed using simplified representations of the multichip BGAs, in order to find the locations of highest stress, and the expected modes of failure. This data was matched to the failure modes found in the physical analysis. Some novel failure analysis techniques were used to expose the damage in the failed units. A particular failure mode occurred frequently in temperature cycle, and the sites of failure were located by failure analysis. The failure was due to open circuit in the copper tracks in the top layer of the substrate caused by cracking in the solder resist directly underneath the edge of the die attach fillet. Finite element analysis was carried out and the location of the actual failures was found to be a local zone of high tensile stress in the solder resist     

Near-CSP plastic ball grid array

The mold array process plastic ball grid array continues to demonstrate progress as a cost-effective, medium to high performance electronics package. Highly manufacturable and reliable solutions to the challenges of molding up to several hundred packages at one time as well as subsequent singulation of the packages have been found. JEDEC moisture performance at up to Level 1 with 220°C reflow temperatures has been qualified for production. Thermal performance remains critically dependent on die size, package size, and motherboard construction.     

Processing induced material interactions determining the reliability of LTCC multichip modules

Metals can exhibit dendritic short-circuit growth caused by electrochemical migration in conductor–insulator structures, which may result in failures and reliability problems in microcircuits. The classical model of electrochemical migration has been well known for several decades. This process is a transport of metal ions between two metallization stripes under bias through a continuous aqueous electrolyte. Due to the electrochemical deposition at the cathode, dendrites and dendrite-like deposits are formed. Ultimately, such a deposit can lead to a short circuit in the device and can cause catastrophic failure. Recent investigations have demonstrated that not only metallic components, but also oxides from the isolating layers can take part in the formation of migrated shorts, after a chemical reduction process. Material design aspects need to clarify the correlation between material composition, processing, chemical bonding state, and electrochemical migration failure rate in isolating compounds: this is the scope of the present study.     

Estimation of warpage and thermal stress of IVHs in flip–chip ball grid arrays package by FEM

Flip–chip substrates have been developed to meet the recent technical trend. They have a small IVH (Inner Via Hole) diameter to improve electrical packaging performance. However, under thermal loading conditions, substrate warpage increases as substrate thickness decreases. Performance of FCBGA may be severely limited by substrate warpage. Furthermore, large thermal deformation induces cracks and delaminations in an IVH. It is important to understand substrate thermal deformation to improve FCBGA reliability.Thermal deformation of the FCBGA (Flip–Chip Ball Grid Array) with assembly conditions has been calculated globally by finite element analysis. And residual plastic strain of an IVH has been estimated microscopically to understand thermal stress of the IVH. Finite element method considering non-linear material model is verified with experiment on warpage to improve simulation accuracy. Also, the Taguchi method is applied to optimize FCBGA substrate design.Based on the computed results by the Taguchi method, we know core thickness in FCBGA substrate is the most determining factor for thermal deformation. The second most significant factor is the core material properties. Even though the plugging material in the IVH has little thermal deformation macroscopically with respect to the entire FCBGA substrate, the plugging material lowers the reliability of the IVH alone microscopically. In some cases depending on the plugging material, the IVH may develop some cracks.     

Scanning electron microscopy study of the cross section of intermetallic compound in the solder ball on the plastic ball grid array package following reliability tests

The 63Sn-37Pb solder ball (φ=300 μm) was attached to gold-nickel-plated plastic ball grid array (PBGA) substrates, with gold and nickel thicknesses of 0.6 μm and 7 μm, respectively. The thickness of the intermetallic compound (IMC) in solder balls was measured following each instance of infrared (IR) reflow (90 sec at 230 °C), level II preconditioning, a pressure cooking test (for 96 h or 168 h), and a temperature cycle test (with 500 or 1,000 cycles). Scanning electron microscopy (SEM) was used to identify the cross-section sites of the solder balls following testing. Following all the reliability tests, the IMC demonstrated that an IMC thickness exceeding 5 μm will reduce the solder ball shear strength owing to diffusion of Ni into the solder balls.     

Reliability of plastic ball grid array package

The reliability of plastic ball grid array (PBGA) package is studied for different materials. The reliability of the PBGA packages using conventional Bismaleimide-Triazine type PWB and our original product PWB that is made of high Tg epoxy resin is evaluated. The PBGA package using our original PWB has a feature of lower warpage for the package, and has similar performance regarding the thermal cycling stability and the endurance during reflow soldering as compared with the PBGA using the conventional PWB. The endurance during reflow soldering for each PBGA is JEDEC STD Method A112 level 3. In order to improve the endurance during reflow soldering, not only PWB materials but also other factors are investigated. As a result, the molding compound with the property of low moisture absorption and the die attach material with the properties of high adhesion strength and fracture strength are effective to improve the endurance property during reflow soldering. The package crack mechanism during reflow soldering is briefly described as follows     

Measurement of water absorption in ball grid array package

The plastic ball grid array (BGA) package has poor resistance to popcorn cracking, which occurs when high temperatures involved in soldering cause water vapor to expand rapidly. Popcorn cracking occurs at die attach paste, and therefore water absorption and desorption occurring in the vicinity of die attach paste must be studied. We examine the mechanism of popcorn cracking in a BGA, particularly from the aspect of water absorption distribution. Water absorption was simulated by use of deuterium oxide, because the absorption performance of deuterium oxide approximates that of water. Deuterium oxide absorption distribution was measured by time of flight secondary ion mass spectroscopy (ToF-SIMS). We found that the water is absorbed mainly through the upper side of molded portion of the BGA package, and that absorption through the substrate is small. A BGA substrate has a laminated structure, and therefore water cannot penetrate the substrate first. On the basis of the results obtained in our study, we designed a BGA package system that is not prone to popcorn cracking     

Accurate measurement of ground-ring inductance in ball grid arraypackage

A new technique has been developed to measure the ground-ring inductance in a ball grid array (BGA) package. A simple parallel LC circuit is used to model the ground-ring parasitics at frequencies up to 1 GHz. After connecting an SMA connector to the ground ring of the BGA package, a network analyser can be used to measure the reflection coefficient (S₁₁) up to 1 GHz from which the ground-ring inductance can be extracted. It has been found that the ground-ring inductance depends very strongly on the phase of S₁₁. This leads to the advantage of excellent accuracy for the extracted quantities. The experimental ground-ring inductance data for a variety of BGA packages are verified by Ansoft simulation results     

Plasma cleaning of plastic ball grid array package

An investigation of O₂, Ar and Ar/H₂ plasma cleaning was carried out on plastic ball grid array (PBGA) substrates to study its effects on surface cleanliness, wire bondability and molding compound/solder mask adhesion. Optimization of the plasma cleaning process parameters was achieved using the contact angle method and verified by auger electron spectroscopy (AES), X-ray photoelectron spectroscopy (XPS) and wedge pull tests. It was found that both the wedge bond quality and moisture sensitivity of a 225 I/O PBGA package were improved after plasma cleaning. Furthermore, atomic force microscopy (AFM) characterization and XPS analysis revealed that the solder mask has undergone plasma-induced surface modification. Cross-contamination of Au and F traces on the solder mask that has occurred during plasma cleaning was identified by XPS. This study has demonstrated the benefits and consequences of plasma cleaning for a PBGA package.     

Integrated circuit ceramic ball grid array package antenna

The recent advances in such highly integrated RF transceivers as radio system-on-chip and radio system-in-package have called for the parallel development of compact and efficient antennas. This paper addresses the development of a new type of dielectric chip antenna known as integrated circuit package antenna (ICPA) for highly integrated RF transceivers. A compact ICPA of this type has, for the first time, been designed and fabricated in a ceramic ball grid array (CBGA) package format. The novel ICPA, except economical advantage of mass production and automatic assembly, has potential benefit to the system-level board miniaturization and the system-level manufacturing facilitation. The simulated and measured antenna performance of the ICPA is presented. The effects of the different physical parts of the ICPA on the antenna performance are investigated. Results show that the ICPA achieved impedance bandwidth of 4.1% and radiation efficiency of 72%, and gain of 4.8 dBi at 5.715 GHz.     

Frequency stabilization of power-combining grid oscillator arrays

In this paper, we report the results of phase locking of grid oscillator arrays. First, a voltage-controlled grid oscillator array with a center frequency of 4.7 GHz and with a 300-MHz electric tuning range was locked to a frequency synthesizer through a phase-locked loop. Second, a 4 × 4 and a 6 × 6 grid oscillator arrays were locked by way of the injection locking. In both methods, a simple loop antenna mounted on the reflection mirror was used for taking/injecting signals from/to the array. Results show that the phase noise performance is improved significantly in the locked oscillator arrays     

Compact conduction model of ball grid array packages

This work focuses on the application of compact conduction model (CCM) creation approach to ball grid array (BGA) packages under steady state conditions. The procedure of creating CCM from detailed BGA model is demonstrated. The authors have imposed realistic boundary condition scenarios that are likely to test the boundary condition independence necessary for the compact modeling approach. Results showed acceptable agreement in die temperature and heat flow predictions from simulations using detailed BGA models and their CCM equivalents.     

Impact of the number of chips on the reliability of the solder balls for wire-bonded stacked-chip ball grid array packages

This investigation examines how the number of chips affects the reliability of solder balls for wire-bonded stacked-chip ball grid array packages under thermal cycling tests. The studied objects were packages with one, two, three and four stacked chips. Three-dimensional finite element analysis was utilized to simulate the stress/strain behavior of all studied packages. Two kinds of properties of 63Sn/37Pb eutectic solder were employed individually in the finite element analyses. One property of the solder was assumed to exhibit the elastic–plastic–creep behavior. Temperature-dependent stress/strain curves and Norton’s steady creep equation were used in the analysis. Another property of the solder governed by the Anand’s viscoplastic model was also employed to describe the behavior of solder balls. The simulation results in the elastic–plastic–creep analyses and viscoplastic analyses reveal that the von Mises stress, the non-linear strain, and the inelastic strain energy density of the critical solder balls increase with the number of stacked chips, but the increments become gradually stable as the number of chips increases. Three fatigue life prediction models—Darveaux’s model, the modified Coffin–Manson model and the creep-fatigue model—were applied to evaluate the fatigue life of the studied packages. Prediction results indicate that the fatigue life of the solder balls decreases as the number of stacked chips increases, and the decrease in predicted life shows stable behavior as the number of chips increases. The stable trend is consistent with experimental observation in the thermal cycling tests. By comparing with the experimental data, it is shown that the Darveaux’s model gives better prediction than the other two models.     

Three-dimensional inspection of ball grid array using laser visionsystem

An inspection process of the ball grid array (BGA) is proposed using a laser vision system which provides range image. This study suggests an inspection algorithm using the range images for evaluating the height of the solder ball, coplanarity and lead pitch, which are major factors in the BGA surface mounting technology. A feature-based algorithm was used in implementing region segmentation of range images. By deciding the BGA orientation using plane modeling, the errors of the range finding system were compensated and the reference of the height of the solder ball was found. Also, a method for measuring the height of the solder ball using free-form surface modeling was proposed. And, from the measured heights, the pitches and coplanarity were calculated     

Near-field and far-field prediction of quasi-optical grid arrays

Field profiles, equivalent isotropic radiated power (EIRP), beamwidth, and sidelobe levels of a quasi-optical power-combining grid amplifier structure are obtained using a whole-field electromagnetic analysis of a finite structure. This is efficiently obtained using a mixed spectral- and spatial-domain method-of-moments technique. It is seen that edge effects can have a significant effect on system performance