PACKAGING AND RELIABILITY GROUP RECOMMENDATIONS
Packaging and reliability challenge is a well-known potential showstopper to the growth of MEMS applications. Packaging cost is about 50 to 90% of the total cost, and reliability is always a top killer issue for every MEMS product. The working group has identified and discussed four challenges that demand in-depth, scientific research studies: functional interfaces, reliability, modeling and integration. These challenges will be discussed with a few examples of basic studies needed.
Challenges and Research Topics
Packaging (including reliability) has been and continues to be a major challenge. Packaging cost is about 50 to 90% of the total cost of the MEMS product. At the present time, custom packaging solutions and reliability qualification processes have been developed for specific products. Such a solution procedure has a potential of stalling the rapid growth of MEMS applications. MEMS packaging (including reliability) is a well-known potential "show stopper." Four challenges to packaging and reliability have been identified: f unctional interfaces, reliability, modeling and integration. They will be discussed as follows.
Functional Interfaces
Figure 1 illustrates a typical MEMS package with different functions. The package provides functional interfaces between the MEMS device and the environment. The input interfaces are controlled by the desired electrical inputs and affected by tolerated and rejected environmental influences. The output interfaces are accompanied by desired electrical outputs and some byproducts. In addition, test inputs and outputs are critical package functions to qualify a product.
The functional interfaces are directly related to the applications. Unfortunately (or fortunately), MEMS has a large number of diverse applications: microfluidics related to bio-medical applications, data storage, microsurgical instruments, RF communication, optical communication and interconnects, energy storage, display, etc. As a result, different functional interfaces needed are: optical, RF, thermal (radiation, conduction or convection), fluids (liquids or gases), mechanical (body or surface loadings) and others (e.g. radiation, magnetic, etc.).
Figure 1: A typical MEMS package and its functions
Each functional interface has its challenging requirements. Here is an example list of major requirements in a package for optical MEMS :
Similar or very different requirements exist for every functional interface mentioned above. As a result, it is impossible to develop a "standard" package to serve all the MEMS functional interfaces. In fact, using microelectronic packaging as a reference, it is clearly very challenging to develop a "standard" package to serve even one functional interface. Nevertheless, it is the opportunity for packaging researchers to study and develop packages that may serve as many functional interfaces as possible.
Here is a list of basic studies proposed by the working group:
- Modification of standard microelectronics packages to provide additional optical, RF, thermal (radiation, conduction or convection), fluids (liquids or gases), mechanical (body or surface loadings) or others (e.g., radiation, magnetic, etc.) functional interface.
- Develop a package to serve a set of functional interfaces. For example, the functional interfaces can be grouped as solid or fluid interfaces or as physical or chemical interfaces. Sometimes, we may develop a package to serve one group of functional interfaces.
- Selective membranes for chemical- or bio-MEMS.
- Hermetic sealing using gaskets, o-rings or low-temperatures brazing or welding methods.
- Non-hermetic sealing using adhesives, gaskets, o-rings, etc.
It should be noted that the list is served as examples, which can be used to simulate novel proposals. The list is not comprehensive and should not be used as the guidelines.
Reliability
Stiction, fracture and fatigue, mechanical wear with respect to frequency and humidity, and shock and vibration effects are the major causes of MEMS failures. For example, a potential reliability problem for Texas Instrument's digital mirrors is that the mirror might get stuck by particle contamination, surface residue and capillary condensation. Another example is Sandia Labs' driver gear that might suffer a pinhole wear problem after 105 cycles.
There are two approaches to develop reliable MEMS. One is the reliability assurance by testing existing structures and the other is by processes/materials development. For the reliability test, the problem is three-fold: 1) the new failure mechanisms are poorly understood and modeled, 2) methods (tests) for accelerating these new failure mechanisms are not defined or understood, and 3) for a given life cycle environment we have to be able to figure out which failure mechanisms are relevant (and which are not) in order to design an accelerated test that actually accelerates relevant mechanisms (versus accelerating mechanisms that are not relevant). In many cases, different failure modes may demand conflicting requirements. For example, humidity may not be desirable since it might cause stiction problems; however, a certain humidity level is good to reduce the wear.
For the processes/materials development, non-hermetic environment and the control of friction and stiction are critical considerations. For example, getters are important materials being used to control the low temperature moisture, high temperature moisture, and micro-particles. Another example is the development of nano-technologies to coat and protect the device for reliable MEMS.
Here is a list of basic studies proposed by the working group:
- Moisture control in non-hermetic packages: 1) "metal" gasket/sealing with compliance, 2) active control of atmosphere, and 3) design and selection of getters
- Stiction: 1) low surface energy films, 2) quantitative understanding of stiction, geometric and surface roughness effects, effects of surface microstructure and chemistry, nano-scale coating, and accelerated testing for stiction
- Accelerated testing (ALT): 1) identification, understanding and modeling of failure mechanisms: stiction, wear… 2) damage accumulation of devices, and 3) material properties and fatigue behavior in micro-scale structured materials
- Qualification: 1) application specific methodology for accelerated qualification (85/85 good test?), 2) MEMS specific testing methodology and equipment, and 3) burn-in pressure w.r.t. functional interfaces
Again, the list is not comprehensive and it is served as examples to simulate novel proposals.
Modeling
Package is usually an integral part of the device. Both device and package have to be designed at the same time. In order to have a one-pass design, physical and semi-empirical models have to be developed. With the diverse applications, a MEMS CAD tool might need to cover every engineering discipline: electrical, thermal, mechanical, optical, electromagnetic wave, and chemical. How to integrate all the existing tools with innovative interface solutions will be challenging. In addition, how to design MEMS for reliability will be as important as the aforementioned tests and new processes/materials.
The state-of-the-art CAD tools are being developed to conduct integrated analysis with the consideration of electrical, thermal, mechanical, optical, and electromagnetic wave performance. Such an integrated analysis is very challenging when the device and the package become complicated. Furthermore, the growth of fluidic- and bio-MEMS demand efficient modeling to cover fluid and bio-chemical phenomena. Such modeling needs many basic studies.
Here is a list of basic studies proposed by the working group:
- Materials properties: 1) temperature and strain rate dependent, and 2) aging
- Experimental techniques: in-situ determination of properties
- Failure modes and mechanisms: physics of failures, e.g. optical misalignment or permanent aging
- Integration of different models including those for bio-medical devices and functions
- Stiction modeling
- CAD tools methodology and integration
- Multi-component and multi-phase modeling for environmental control with or without getters
- Micro-scale transport phenomena
Integration
As indicated in the Functional Interface challenge, MEMS packaging and reliability is strongly related to applications. In addition, packaging and reliability is also strongly related the device fabrication. For every MEMS product, there is always an integration issue needs to be considered: where and how to integrate the fabrication and the packaging processes? Such an integration consideration also provides us an opportunity to create new concepts or technologies for low-cost, high-performance MEMS.
For example, wafer-level packaging can be completed in the same fabrication facility; it may eliminate a packaging step. Such a packaging approach would result in low-cost and very compact MEMS and is the main development target for most of MEMS packages being manufactured today. On the other hand, packaging technologies can be used to fabricate MEMS devices; it may again eliminate a packaging step since the fabrication is the packaging step. Flexible circuit board technologies have been used to develop paper movers and RF MEMS switches. Co-fired ceramics technologies are very popular to develop bio-MEMS and high-temperature MEMS.
In addition, fabrication and packaging technologies can be integrated to form new MESM. For example, solder technologies have been developed to self-assemble MEMS. The combination of the planar fabrication and the solder self-alignment enables us to develop three-dimensional, complex MEMS without demanding complicated fabrication processes.
Here is a list of basic studies proposed by the committee:
- Incompatible fabrication, packaging and testing: 1) PCB for MEMS fabrication, 2) ceramics for MEMS fabrication, 3) wafer-level packaging including release, lubrication, test and seal, and 4) novel dicing
- Interconnects: 3-D stacking of MEMS devices
Other Studies
The above challenges cannot cover all the issues, and the proposed list of studies cannot reflect the exciting opportunities for researchers interested in packaging and reliability. Therefore, we have generated additional list of proposed studies:
- Restricted motion packaging: A major feature of MEMS is that it moves. As a result, it may be necessary to use package to guide the MEMS movement along a particular direction.
- Nano packaging: Nano-devices are being demonstrated for many new applications. How can we package and interconnect them?
- Self-assembly using DNA, surface tension, eletromagenetics: In the scales of m, there are many other self-assembly mechanisms in addition to soldering. Which ones should be used and how?
- Givers (lubricants): Getters are to get the undesirable materials out of the inside environment in a package. But, sometimes, we may want some desirable materials inside. Can we produce a giver to release lubricant at a desirable level?
- Automatic test pattern generation during device design: Testing is a major cost issue. It is important to design MEMS with testing as one of the major design considerations. It is possible the test inputs/outputs need to be generated while designing.
- BIST for MEMS : Test pattern generation might not be cost-effective when the test becomes a complicated issue. Built-in self-test (BIST) is used widely to solve this problem for VLSI integrated circuits and sensors. It is expected to play a key role for actuators. Can we have a MEMS design with BIST?
From this list, it is clear that there are many packaging and reliability issues outside the four challenges mentioned above. The area of packaging and reliability is a potential showstopper to the growth of MEMS applications. On the other hand, it also provides us with an opportunity to make an impact. The committee does expect a significant growth of research studies focusing on MEMS packaging and reliability. Such a growth will be essential to keep US industry competitive.