An improved understanding of major packaging process elements - coupled with the appropriate perspective on manufacturing techniques - reduces product failure and risk management for MEMS packaging. Two technologies that can play a role in better understanding of the analysis and prediction of packaging design failures are microsystems simulation software and materials data information systems.
Failure analysis of MEMS devices is like finding the proverbial “needle in a haystack.” Software for device system simulation complemented by materials data information systems is positioned to help.
Micro-electro-mechanical systems (MEMS) involve the integration of mechanical elements, sensors, actuators, and electronics on a common silicon substrate through microfabrication technology.
MEMS devices, which are found everywhere around us (the average car has five), are “invisible” to the average consumer. Commercialization success occurs when a MEMS device revolutionizes a product category (such as automotive airbags) or becomes an enabling technology. Product development trends are driving the use of multilayer, monolithic systems with reduced component feature size. Thus, physical defects responsible for device failure are decreasing in size - the forensic task is similar to looking for “smaller needles in larger haystacks.”
Failure analysis often concerns packaging design, because it is through packaging that MEMS devices interface with the macro world and with harsh environments. Specific aspects of the packaging process that must be addressed include the ability to package MEMS devices alongside electronic circuits as a single unit (monolithic systems), automating the entire process, and refining knowledge of the “ilities” of the process (reliability, maintainability, survivability, etc.). The process to enclose MEMS devices is often naively believed to be as low risk as that of traditional IC manufacturing techniques, although the latter is high tech, commonly automated, and used to produce large numbers of high-volume IC-enabled devices for consumer markets. An improved understanding of major packaging process elements—coupled with the appropriate perspective on manufacturing techniques—reduces product failure and risk management for MEMS packaging. Here, I will briefly highlight two technologies that can play a role in better understanding of the analysis and prediction of packaging design failures: Microsystems simulation software and materials data information systems.
Microsystems Simulation Challenges
Recent advances in microsystems simulation software (finite-element computer-aided engineering and system modeling software) may surprise failure analysts who have not been monitoring the technology. Products such as MEMS-Pro (Soft-MEMS, Los Gatos, Calif.) and CoventorWare (Coventor, Cary, N.C.), for example, are increasingly robust, powerful, and interdisciplinary in scope than tools available as recently as four years ago. Of significance is that the device analyst does not have to be a computer scientist to engage simulation tools in his or her daily work. And, on the horizon are computational subroutine capabilities that can impact packaging failure analysis modeling and simulation. In the broadest sense, “simulation” involves the use of computer software to model “real-world” devices, systems, and processes. Simulations provide virtual device testing and experimentation based on somewhat simplified views of the world; our ability to model reality assumes the underlying theoretical logic is representative of the physics of interest. In the past, desktop computer processing power was the key limiting factor to computationally intensive grid generation and algorithm execution. More recently, however, it is the challenge to acquire accurate material data, life data models, and integrated materials compatibility information that limits our ability to “model reality.”
Many MEMS “simulations” are actually a collection of several programs. For example, the MEMS-Pro system has extensions to the ANSYS, Inc. finite-element analysis program that can directly import, say, MEMS mask layout information from AutoCAD (Autodesk, Inc., San Rafael, Calif.). MEMS mask layout files can be exported from the MEMS-Pro layout editor to mask makers and MEMS fabrication foundries in drawing exchange format (among others). So, the challenge in MEMS simulation is to understand the complete suite of software tools involved in each stage of the MEMS design process, and typically, there are at least a few programs for which a designer may already be a user (as in the example of AutoCAD for mask design).
Every major software vendor has some form of user-programmable interface, allowing for the customization and extension of the base software command and function set. It is this area where the opportunity exists to extend simulation capabilities to perform life studies based on unique materials information.
Simulation is gaining traction as a complement to microsystems design due to the relative advantages of its low cost, high level of availability to users, and the effective cost education it offers in comparison to an iterative laboratory “make-it-and-break-it” fabrication model. Although software on the market today is becoming increasingly specialized and applicable to microsystems design analysis, there still remains work to be done on subroutines for microsystems processes (e.g., etching) and materials properties in the context of those processing steps (e.g., borosilicate etch rates). For this reason, we stress the complementary (not competitive) role of simulation in the laboratory. Tools designed exclusively for failure analysis are available (FASTAdvise, Sandia National Laboratories), but the focus on integrated circuit design is somewhat different than for microsystems actuator and sensor devices.
For MEMS simulation:
• Significant advances in simulation robustness, ease of use, and analysis platform compatibility have been made in the past few years.
• Life, creep, and long-term effects remain heuristic/statistical; materials properties over time are conspicuously absent today.
• Experimental techniques remain essential adjuncts to setting baselines for initial calibration of calculation results.
Microsystems Materials Information Database
Materials engineers and device designers need materials reference information describing Microsystems properties and processes. Such information is often found scattered among an array of journals, documents, spreadsheets, and databases, representing a challenge to the designer who then has to qualify, aggregate, review, and format the materials information for subsequent use.
When this process is complete, how can the results be shared easily across the organization or then interfaced with analysis software? A materials information database can be of assistance in four important ways:
• Improved content integration: Data from many different sources and formats are blended into a relational database format, simplifying transfer of information to simulation software or deployed to other users though summary data reports.
• Reduced maintenance cost: Aggregating and centralizing the information reduces the cost of maintaining qualified information (and filtering out unqualified information) for engineers and designers.
• Simplified data comparison: When materials information is in a database, data manipulation becomes quite straightforward. Refined data searches, materials property comparisons, and cross referencing manufacturers’ materials reference information become fast and easy with database technology. Web-enabled user interfaces allow information access from anywhere an internet connection exists.
• Education: The role of a materials information database as an educational tool cannot be overestimated. The ability to link specific materials and processes to case studies in design, selection, and use of materials is a powerful way for engineers to learn the “strides and stumbles” experienced by others, thereby improving the chance that costly errors are avoided.
ASM International has collaborated with Granta Design Ltd. to launch a database information system providing the screening, selection, and sourcing of materials information data. Relevant to the simulation task is export of materials data in a format that can then be immediately imported into modeling software. The ASM International MEMS materials database for packaging represents a new source for the mechanical, physical, and chemical proper-ties of materials appropriate to the range of MEMS packaging challenges facing device designers. This relational database comprises thousands of hours of data acquisition, peer review, and verification by specialists.
It includes a comprehensive presentation of interrelated information, all fully traceable to sources. This includes citations to published literature, manufacturers’ data sheets, and websites.
The materials database can help an analyst:
• Save hours and money on unnecessary exploratory screening and investigation
• Identify materials that optimally satisfy various mechanical and performance design requirements
• Select candidate materials based on demonstrated successful application and compatibility with design parameters
• Identify previously successful packaging designs that use a specific material
• Create reports that compare materials and processes per user-defined criteria
ASM International conducted several focus group studies on how the database would be used “in practice” and was surprised at how popular the case studies were with users. It turns out that many designers are eager to learn more about “lessons learned” in product design, and case studies provide many such examples.
At this stage, ASM International has released Phase 1 of the three-phase database development effort:
• Phase 1 (recently launched) provides general data on microsystems packaging processes and materials.
• Phase 2 (under development) will focus on specific microsystems materials compatibility and life characterization.
• Phase 3 will extend the materials characterization effort into the nanotechnology domain.
It is stressed that the coupling of the database to failure analysis simulation is a major focus of the Phase 2 effort already underway. ASM International welcomes participation and interaction with the failure analysis community on this important aspect of the project.
Whether the need is for materials data selection and comparison, organizational deployment of recommended materials, or simply as an educational resource, database technologies have proven extraordinarily useful in identifying, screening, and selecting materials, processes, and techniques appropriate to MEMS packaging. Further, with the idea that the purpose of simulation is “insight, not numbers,” ASM International believes that expanded materials life information could provide a new level of productivity for users involved in forensic analysis of MEMS devices.