The pressure to reduce design cycle time and minimize testing and
validation necessitates improved mechanical and reliability modeling
and electrical and thermal simulation tools.
The electronics industry faces constant pressure to reduce design and process cycle times as well as costs, despite the fact that products are becoming more complex. In this environment, modeling and simulation offer considerable advantages to manufacturers by allowing rapid evaluation of “what-if” scenarios without building expensive, complex hardware. Simulations can be used to determine performance/cost trade-offs and gain insight for the next product/process design. They can also reduce the amount of testing required and help manufacturers do things right the first time.
The 2007 iNEMI Roadmap discusses issues relating to modeling and simulation and the challenges posed by current and future products. This article discusses mechanical and reliability modeling, thermal and thermo-fluid simulation, and electrical simulation, along with some of the particular challenges posed by RF/wireless products and embedded passives.
Reliability Modeling
Mechanical reliability is a key focus for all product sectors covered by the iNEMI Roadmap (automotive, defense/aerospace, office/large business/communication systems and portable/consumer). It is an area that was flagged as needing attention in the 2004 Roadmap, and this trend continues in 2007. The need has increased in severity with the recognition that stiffer lead-free solder joints are more susceptible to mechanical failure. Methodologies for predicting the solder joint fatigue life for various package families, designs and application environments need to be re-examined, especially for lead-free solders. The ability to predict the formation and effect of intermetallic on solder joint reliability is required both for lead and lead-free solder systems.
As in the 2004 Roadmap, interfacial delamination, moisture modeling and solder joint reliability continue to be areas of concern and represent instances where predictive modeling could help industry ensure greater product reliability. However, modeling capabilities are still lacking in all of these areas.
Although the increased use of wafer-level packaging makes knowledge of interfacial delamination crucial, there continues to be deficiencies in industry’s ability to predict the nucleation of cracks and their subsequent propagation under static and cyclic loads.
Today, moisture performance of a package is typically predicted through build and test. Modeling such factors as diffusion of moisture, transient thermal analyses and associated stress, interfacial stress due to moisture desorption during reflow and its effect on interfacial crack propagation could significantly reduce cycle time.
There are several other areas where modeling could help shorten cycle times and improve first-pass success, some of which are discussed below and summarized in Table 1 [PDF format].
Wafer-Level and 3D Packaging. The need for low-cost, high-I/O packaging is driving the adoption of wafer-level and 3D packaging. Advances have been made in package development and reliability characterization, and development of these areas could greatly benefit from simulation and modeling. Issues of interest include characterization of interface and thermal management in 3D packages; their mechanical reliability, effect and choice of stress compensation layers in wafer level packaging (WLP); reliability of vertically stacked microvias in 3D packaging; and thermal, thermo-mechanical and electrical characterization of conductive and non-conductive epoxies used in these technologies. As the number of stacks increases in a 3D package, the die becomes thinner, down to 50 microns in 2006. Thus, the effect of die thickness on die reliability (propensity to cracking under induced stresses) and the limitations of such reliability should be understood thoroughly.
Very High Temperature Applications. The automotive sector is anticipating higher operating temperatures that will most likely require development of new packaging materials. As these new materials are implemented, manufacturers must understand the effect of combined vibration and temperature on mechanical reliability, and also understand the physics of failure for electronic assemblies made with these new materials.
Drop/Impact Stresses. The drop/impact analysis algorithms currently used are not fast enough to treat large and complex portable systems. Research in the area of computational mechanics is needed to develop stable and faster algorithms for reduced cycle time. Since the results of the analyses have to be validated, there is a need to develop experimental methods for the same. Because large strain rates are involved, it is necessary to characterize mechanical properties of the materials set at such high strain rates so that accurate inputs are available for analysis.
Thermal and Thermo-Fluid Simulation
The key computational issues relating to thermal and thermo-fluid simulation are still buoyancy-radiation coupling and fast algorithms to model complex heatsinks and transient power changes (Table 2 [PDF format]).
The increased power densities and vastly increased complexity of products such as those found in office, large business, communication and military systems makes it imperative to resolve the issue of predicting and analyzing hot spots due to thousands of connections/contacts with varying amounts of currents. The key issue is the ability to analyze thousands of connections at the system level.
Two critical, assembly-related problems were identified in the office/large business/communication systems product sector. These problems relate to underfill flow and simulation of solder-joint melting and solidification during assembly – both primarily being driven by the proliferation of flip-chip packages. The same concerns plague the defense/aerospace sector. Some algorithms exist for solving both the problems, but no codes exist that are user-friendly and designed specifically to solve these problems. Also, convergence of solutions is typically extremely slow. It is possible that, with the emergence of wafer-level packages, this need may go away.
Radiation heat transfer in non-rectangular geometries could be beneficial for under-the-hood automotive applications. Research is available at the university level, but user-friendly codes are needed for industrial applications. Not much progress has been made in this area.
Electrical Simulations
For electrical simulation and modeling, the main issues are with the interfaces between design tools and electrical modeling tools, and not the fundamental understanding of electrical phenomena. Some key issues are: 1) handling the complexity of large business systems and communication infrastructure products, 2) the ability to handle entire systems for these types of high-performance products, and 3) faster simulations driven by shortening effective design times in the portable/consumer sector.
Some areas that need attention are electromagnetic simulations, signal integrity verification (SIV), power ground network simulation and optimization, system-level power distribution and noise analysis, electromagnetic coupling (emc)/electromagnetic interference (EMI) verification, embedded passive design and modeling, chip/package co-design, mixed signal design and analysis for system-in-package/system-on- package (SIP/SOP), and the ability to model entire systems.
At the system level, accurate, efficient and rapid modeling of system-level power distribution noise has become a critical issue. Power supply noise represents one of the largest bottlenecks in the design of high-speed systems.
RF/Wireless Products and Embedded Passives
Since the last roadmap we have not only experienced continued digital convergence in products, but also the proliferation of wireless communication between digital products. RF circuits used for wireless communications require more detailed simulation – such as linear S-parameters and nonlinear harmonic-balance – than digital ICs. And, unlike digital ICs, there is little replication of functional blocks in the RF sections of wireless consumer products such as cell phones and pagers.
Existing CAD tools do not allow complete S-parameter, harmonic-balance or similar calculations to be done for embedded passives used in wireless products. As a result, it is often necessary to go through numerous iterations of prototype design, fabrication and testing. The next-generation terminals, with full Internet access, will depend on drastically increased use of embedded passive components to achieve the required functionality in a hand-held appliance. This is a prime example of the widening gap between product complexity and the ability of modeling and simulation tools to keep pace.
With the increasing complexity of high pin count packages, and the pressure to produce these at lower cost, the issue becomes how to predict the electrical performance of the package in terms of simultaneous switching noise (SSN), crosstalk and return path. There is a need for integrated and user-friendly tools that can predict this information.
High-Frequency RF Designs
Besides the overall methodology issues involved with RF and embedded passives, designers and engineers are struggling with several issues, especially in the area of high-frequency RF design:
- Antenna effect of wire bonds, vias and stubs on PBGA type packages – need user-friendly tools and increased accuracy, especially for curved shapes
- Electromagnetic coupling between the lines, vias – need efficient tools with fast convergence time
- Analog vs. digital ground plane isolation (which is a complex problem) – methodology is unknown, no tools exist to address mixed signal design
- Coupling of digital noise to RF portion of the package (also a complex problem) – no tool exists that can analyze the coupling between the digital and analog parts on the package/board
- Inductor design with high Q-factor (i.e., complex geometries) – spiral inductors, no robust modeling tool; faster solutions, convergence problems
- Dielectric constant and loss-tangent of the substrate material property function of frequency – methodology is not clear, measurement tools are expensive
- Design interfaces – need robust interfaces to transfer appropriate data
Mindset Issues
There is a common impression that sophisticated tools and interfaces alone will solve all the simulation problems. On the contrary, to have effective results in the fastest time requires experts who are not only well versed in the simulation tools and methodologies, but also in the subject matter they are trying to simulate. Experts must be able to determine what phenomena can be modeled and what must be tested, and to identify the requirements for model verification.
Another common mindset is that modeling is a cure-all and does not require expensive hardware. However, in reality, verification is a key element of a successful model, which requires testing facilities. Also, many believe that modeling can be used for prediction only, but it could be very effective in gaining insight into complex systems without requiring too much accuracy or studying “what if” qualitative trends. It is then possible to use these insights for development of future product.
System-Level Issues
As the electronics manufacturing arena is becoming increasingly competitive, science-based techniques, such as simulation that can effectively use and streamline information, are becoming critical. Today, OEMs see EMS providers as their virtual manufacturing environment. As OEMs are introducing new products more frequently, it is extremely important that a product is introduced to market within the appropriate time window. New technologies are being introduced more rapidly, driving time-to-market and time-to-volume into shorter cycle requirements. NPI is often on the critical path for most companies. The automatic introduction of a new product from conceptual design to the shop floor requires complex computational tools that must be accurate and fast. This necessitates the need to move from the serial approach of design and manufacturing to concurrent “design and manufacturing” methodology.
In the world of virtual corporations, companies are a mix of external suppliers, transportation resources, assemblers, warehousing firms and retailers, often in a modular-manufacturing environment. Simulation capabilities extend from the product level to the supply chain level, allowing the testing of numerous what-if scenarios relating to such issues as outsourcing, consolidating vendors, planning or implementing e-business. The importance of a reliable supply chain is enhanced by reductions in the time that elapses between the order from the customer and the delivery of products.
This article has summarized some of the modeling and simulation technology needs identified in the 2007 iNEMI Roadmap. However, there is a major business issue also impacting modeling and simulation. Over the last four years as the supply chain has become more complex, design for manufacturing tools have not kept pace with other tools. One major reason for this slippage is that the EMS companies – who would most benefit from improved tools – are not attractive customers because of their low margins. Tool providers have focused on the major semiconductor and OEM firms as a more attractive market. The industry will need to find a solution to this dilemma to avoid slowing the introduction of new products and to continue the reduction of product cost. PCD&M
For more information about the iNEMI Roadmap, go to inemi.org/cms/roadmapping/2007_iNEMI_Roadmap.html
Dr. S.B. Park is an assistant professor in the Thomas J. Watson School of Engineering and Applied Sciences at State University of New York at Binghamton; This email address is being protected from spambots. You need JavaScript enabled to view it..
