Testing the MEMS revolution

TEMPE, AZ—Microelectromechanical systems (MEMS) technology creates unprecedented challenges throughout the design and manufacturing process for any company brave enough to use it. Not only does test see its own unique MEMS challenges, but the challenges MEMS presents to other activities such as packaging translate into new test challenges as well. For Motorola's Sensor Products Division (SPD) here, the biggest challenge turned out to be coming up with a tester capable of the high throughput needed for production test of devices for automotive applications.

Motorola SPD employs a team of test-development engineers and technicians, led by Theresa Maudie, to link standard testers and handlers with custom environmental-interface hardware. Here, Maudie is shown in front of a production handler for accelerometer testing.

"Our vision is to become the pre-eminent supplier of silicon-based sensor systems," says Demetre Kondylis, director of inertial operations at SPD. "Our core competencies to realize this vision include our MEMS capability in this division, as well as the resources the rest of Motorola's Semiconductor Product Sector has to offer."

Those resources include the ability to integrate the various elements into a system, digital and analog IC design and manufacturing, CMOS-fabrication capability, device-packaging expertise, and test.

MEMS technology is revolutionizing sensor applications. MEMS-based sensors are generally orders of magnitude smaller than their more conventional counterparts. They also consume miniscule amounts of power, can be much more reliable, and can have much longer service lives. In addition, manufacturing methods lifted from the semiconductor IC industry can mass-produce MEMS-based sensors cheaply in enormous quantities.

So, what's the problem?

The problem is that MEMS developers have to solve every problem semiconductor engineers have wrestled with over the decades, plus a whole slew of new ones. Conventional IC designers just have to worry about their creations' electrical performance. MEMS designers have to worry about electrical performance as well as their devices' reactions to physical and environmental inputs.

"There aren't too many other semiconductors where you actually have to expose the silicon to the environment," says Dave Monk, SPD engineering manager, "Not only do you have guys that are spending time doing electrical modeling, but we have materials experts and chemical engineers who are working to make the device reliable when exposed to a full array of harsh environments. What happens if this device is put into gasoline or oil? If we're making automotive pressure sensors, chances are the device will have to deal with really tough environments in the field."

The implications for test are obvious. If you're making a pressure sensor, for example, you need to allow the fluid that you're measuring the pressure of into the package, making sure it reaches the transducer but stays away from everything else.

The sensing element of a MEMS pressure transducer is a silicon membrane that may be on the order of a half a millimeter across and 1 to 3 microns thick (see "Making a robust MEMS pressure sensor "). The fluid being measured presses on one side, while a reference pressure exists on the other side. The package has to ensure that the measured fluid can't somehow leak over to the reference side—even if the fluid is just air. Nor can either fluid be allowed to reach integrated electronic components elsewhere on the chip whose characteristics might be affected by such contact.

Designers have to make sure the package seals will work correctly and will continue to hold in the face of environmental challenges, such as temperature cycling, shock, and vibration, over the device's 15-year life. Engineers have to set up test challenges to make sure the package will work the way the designers intended.

Motorola's SPD concentrates on developing sensors for two physical parameters: pressure and acceleration. Yet, engineers and scientists can use such sensors for a variety of applications beyond measuring those basic parameters.

"Inertial sensors measure acceleration or deceleration," Monk points out, "but they can also measure shock or vibration or inclines, so there are several variations to acceleration or deceleration. With a pressure sensor, you can also measure liquid level and flow, while pressure is still the physical parameter."

Figure 1. The TPM system chip set includes a sensing module that communicates via radio signals with a dash-mounted received module. The receiver module also incorporates the functions of another automotive wireless system—remote keyless entry (RKE).

The TPM system helps drivers maintain proper tire pressure by notifying them when tire pressure is not at the optimal level. Proper tire inflation decreases tread wear, improves gas mileage, and improves vehicle-handling safety.

What gives TPM its high unit-sales growth potential is the Transportation Recall Enhancement, Accountability, and Documentation (TREAD) Act of 2000 (see "Automotive apps create market for electronics "). The TREAD Act mandates that automotive manufacturers phase TPM systems into their new vehicles starting in November 2003 with 10% of the fleet and ramp up to 100% of the fleet by October of 2006. The total market for TPM systems will grow from approximately 4 million units in 2003 to more than 40 million units by the end of the decade.

Of course, a market surge that powerful carries with it some powerful side effects. On top of the general MEMS challenges, Motorola has to deal with stringent demands specific to the automotive industry.

Motorola SPD's TPM system uses the company's MEMS pressure-sensing technology to monitor tire pressure directly inside the wheel. It consists of two components: A remote sensing module (RSM) mounted in the wheel senses the air pressure in the tire and sends its report wirelessly to a receiver, which is usually mounted in the dashboard.

Originally, members of the design team thought that they might need a receiver inside each wheel well to individually communicate with each tire, and another receiver to monitor the spare. They found, however, that one receiver could handle the communications with all five RSMs.

Each RSM incorporates an MPXY8020A MEMS-based absolute-pressure sensor and an MC68-HC908RF2 8-bit flash microcontroller unit, which has an RF transmitter built in. The dash-mounted receiver module contains a 16-bit MC9S12DP256 flash microcontroller with an integrated MC33591 RF receiver, demodulator, and decoder.

The receiver module also serves the vehicle's remote keyless entry (RKE) system, eliminating the need for separate wireless units to serve the two functions.

Motorola's TPM RSM incorporates system-on-chip (SOC) technology. Each chip has to go through the usual electrical testing required for any mixed-signal IC or module.

Because the TPM RSM is intended for the automotive industry, it also must pass rigorous environmental and reliability tests. Finally, it has to pass functional tests to make sure it measures pressure at the required sensitivity and accuracy.

"The mechanical side is a big part in testing the pressure and inertial sensors," says Theresa Maudie, test-development manager for SPD. "It's probably the biggest challenge. I'm probably the only Motorola test-development manager that's a mechanical engineer. That tells you where the challenges are!"

To deal with these mechanical challenges, SPD partners with an outside test-handler manufacturer, who develops semi-custom systems that the company can use for both development and production testing. It hasn't always been that way, though.

"In the old days," Maudie recalls, "we had to have great technicians who could do anything with their hands. They had to do some real hands-on construction with rubber bands, straws, and sticks. We had to have turning test heads, where you put a device in one side, and it would rotate around to the test position and something would close to seal around the pressure port. It was just a real simple test structure early on."

The next generation was more like a custom system. As the MEMS business grew, and as Motorola captured more of the business, there was a need for higher capacity. The biggest challenge was how to cost-effectively increase capacity. There really weren't any testers to buy off the shelf, and a typical semiconductor test house wasn't interested in building a tester for MEMS-based sensors because the volumes were still too small. Motorola's growing pains came from having a volume too large for rack and stack but too small to get the attention of manufacturers who could build systems for high-volume capability.

Motorola's solution was to build custom, second-generation test systems that could hold several devices at a time in a test fixture and provide an automated means of movement to the test location with the ability to temperature-condition the devices. These fixtures increased throughput by temperature-conditioning the parts in parallel. The system would run through the test process with less intervention than the first-generation systems.

SPD's MEMS business is big enough now to attract the interest of outside partners to build a new generation of MEMS-sensor testers that can achieve high-volume-production throughput levels. Motorola knows that its expertise is in making sensors, not in making test systems, so the company has been willing to share some of the MEMS-tester intellectual property to help it get a better test system for a lower cost.

"We knew that in order to really have this be cost-effective, we couldn't have exclusivity," Maudie reports. "We're now working with Multitest, in Rosenheim, Germany. We worked with them on a very early test line, so we had that experience."

The partners pooled Motorola's MEMS-testing experience with Multitest's production-tester expertise to design a system capable of achieving large-volume production rates. The goal was to use a combinational approach that would make use of standard handler and automatic test equipment components, because such an approach would offer the lowest risk.

"Our MEMS parts," says Maudie, "look similar to typical ICs" and therefore are compatible with standard handling equipment. The challenge, she says, was to find a way to apply the appropriate physical stimulus—pressure or acceleration—while being able to measure electrical responses.

The resulting platform tests devices in parallel and looks much more like the device handlers most engineers are used to seeing. Devices are secured in the nest where a locking mechanism clamps them in place to ensure proper electrical contact and proper transfer of the physical stimulus.

Conventional device-ATE instrumentation sources and measures the signals more rapidly than a rack-and-stack system could. The unit incorporates temperature control to warm or cool the devices as needed during the test.

With this third-generation tester technology, Motorola's Sensor Products Division will be able to produce MEMS-based inertial and tire-pressure monitoring devices in the quantities that its consumer and automobile-industry customers will require.