Avoid these common MEMS failure mechanisms
Demand for micro electro-mechanical systems (MEMS) technology is on the rise. To service that demand with reliable products, both developers and users of MEMS devices need to know about likely failure mechanisms and how to avoid them. Stiction, electrostatic discharge (ESD), micro-contamination, and mechanical shock are key reliability failure mechanisms to understand.
An important driver for the demand of MEMS is, and will continue to be, the Internet of Things (IoT). MEMS and sensors are being used increasingly in healthcare, consumer electronics, low power applications, security, asset tracking, automotive technologies, and smart homes, to name a few, with MEMS marketing and industry analysts predicting tens of billions of shipments within the next decade. With so many interacting MEMS and sensors in the field, up-time is important and reliability is critical.
The first step in ensuring MEMS reliability is to avoid common pitfalls during the design and process development phase to assure a stronger and more reliable part-upon-marketplace introduction. And the time to market for MEMS is fast. Upon product launch, the part must meet all its datasheet specifications as well as storage, shipping, and operational environment reliability tolerances. One of the first things to be prepared to deal with is stiction.
The term stiction comes from "static friction" and it has been a factor for years in a wide variety of technologies, including suspension linkages for cars, polished glass, hard disk drives, and precision gage blocks. It occurs when two objects are initially brought into contact (Figure 1). In a MEMS device, objects that could come in contact include elements such as actuators, proof masses, and sensing fingers. Such contact may occur as part of the device's normal operation, or may unintentionally occur as the result of an external force such as a mechanical shock. Either way, however, once contact has occurred the device needs a reliable way to ensure that it can separate the surfaces again in order to keep functioning properly.
Figure 1. Surfaces are brought into contact (lateral stiction)
The primary forces that come into play upon bringing two surfaces very close together are electrostatic attraction and surface work of adhesion. The force of electrostatic attraction is proportional to 1/d2, and surface work of adhesion is proportional to 1/d3. Surface work of adhesion in MEMS is primarily due to van der Waals and hydrogen bond forces.
The two surfaces must be very close to be drawn into contact for stiction. The electrostatic force attraction distance is a function of the potential difference between the surfaces, and is typically in the micron range. Once the two surfaces are in single digit Ångstrom range, van der Waals and hydrogen bonding forces come into play. Although the latter forces are classically defined as weak interactions, they are additive and become significant in stiction.
Figure 2. Surfaces are in contact (lateral stiction)
Release from stiction is only possible if the release forces, also called restoring forces, exceed the forces that allow the surfaces to stay in contact: Frelease > Fcontact. Release forces in a MEMS device include the mechanical properties of the MEMS design (spring constant) and, when packaged in a gas or fluid, squeeze film damping.
To ensure that a device can overcome stiction as needed, the design needs to properly address all these forces. Increasing a structure's stiffness (spring contact) will increase Frelease, for instance, but there are serious trade-offs. Stiffness affects the pull-in voltage needed to move an element.
Another approach to ensuring that stiction can be overcome would be the reduction of surface work of adhesion. Such reduction traditionally been performed with both design and manufacturing methods to reduce Fcontact. A popular design method for reducing Fcontact involves reducing surface area of contact through the inclusion of stoppers or bumpers. Manufacturing methods include reducing surface area through surface roughening techniques, also called ‘nano-texturing’. In Figures 3a and 3b, diagrams represent surface texture through, for example, oxidation of polysilicon. Lower surface area of contact is the goal.
Figure 3a: Representation of smooth surface Figure 3b: Rougher surface due to oxidation
Another popular and long-used manufacturing method for reducing contact forces involves the use of anti-stiction coatings that reduce the surface work of adhesion. Early coatings such as OTS (octadecyltrichlorosilane) reduced work of adhesion to a polysilicon surface by 3 to 4 orders of magnitude. Due to growth in MEMS designs and applications, new surface coatings are always in development.
Along with the mechanical problems of stiction, MEMS are susceptible to electrical problems, such as ESD (electrostatic discharge). The generation of static charge and its transfer between two objects (discharge) will frequently occur in normal use and ESD generation is so well-known to cause havoc in semiconductor devices that entire careers have been spent in designing ESD protect circuits.
ESD can also cause failure for some MEMS. If your MEMS device is electrostatically actuated, for instance, then ESD is a likely failure mechanism for your part. ESD could cause the actuator to move beyond its intended range, possibly resulting in contact and stiction. You should test your part to the proper standards both to quantify the effect and to determine if the device will fail in the field.
Electromechanical failure due to ESD can be reduced by eliminating the electric potential differences between the MEMS element and any potential landing location. Again, this is a design decision and landing features can be designed with ESD effects in mind. Yet in severe cases where design cannot prevent the failure, protect circuitry is recommended for MEMS ESD prevention.
For high resonant-frequency devices, the structure is so stiff that ESD related motion is less likely. The timescale of the ESD pulses are on the order of nanoseconds, too short to result in significant motion. In these cases, the movement in the MEMS is primarily that of resonance and damping.
An ESD event can yield a combination of electrical and mechanical failures in MEMS devices. Joule heating effects, for instance, can cause melting of MEMS structures such as a MEMS comb finger. It can end up ‘welded’ to the ground plane as the result of ESD. (The welding mechanism in this case can appear like a stiction event.) Of course ESD damage can also be purely electrical, such as an insulator breakdown or metal lead fusing and melting.
Methods for elimination of ESD in MEMS are similar to those for semiconductor devices. ESD protective handling procedures and packing materials, for instance, can be critical to protecting an ESD-sensitive MEMS device. Design changes a developer might consider to reduce sensitivity to ESD in MEMS include wider spacing between leads and wider leads to carry higher current densities.
Testing for ESD failure is a case where industry standards for semiconductors can be applied to MEMS devices. HBM (human body model), MM (machine model), and CDM (charged device model) are the typical standard tests (see Table 1). The human body model simulates when a person touches a device. The machine model has a faster pulse and a more severe discharge from a charged machine. The charged device model is common in semiconductors. Recently, many standard organizations have obsoleted the Machine Model standards. JEP172A explains the discontinuation yet it is recommended to test MEMS to this model and use scientific methods to determine applicability of results to the MEMS and use environment.
Reference only: JESD22-A115C
ANSI/ESD Association Specification STM5.2-2012
Table 1. Three ESD Models and associated specifications
(See page 2 for more MEMS failure causes: Micro-contamination and Shock)