Infineon Technologies' first microelectromechanical system was a pressure sensor developed for its Automotive, Industrial and Multimarketing group. Now well known for its MEMS side-air-bag and tire-pressure sensors, Infineon AIM has branched out into inertia sensors, mainly for rollover detection and response. In 2003, Infineon acquired Norway's SensoNor AS to leverage that company's 20 years of MEMS experience. And last year, Infineon announced its first consumer electronics MEMS sensor, a silicon microphone that is the first of a planned family of consumer sensors aimed at cell phones, global positioning systems, laptops and even toys. Torsten Fischer, director of sensors and controls at Infineon, spoke with EE Times contributor R. Colin Johnson about Infineon's MEMS activity, both past and future.

EE Times: When did Infineon start its MEMS development efforts?

Torsten Fischer: In the late 1980s, with an initial focus on bulk-micromachining pressure sensors. We were still Siemens Semiconductor; the project supported Siemens Automotive.

EE Times: Isn't bulk micro- machining of single-crystal silicon chips expensive compared with surface micromachining of polysilicon mechanical elements?

Fischer: Cost depends on the size of the die, the cost per wafer and other factors, but for some applications bulk micromachining is the most cost-effective solution.

For instance, if the MEMS area is rather large and the interface ASIC requires a highly integrated CMOS technology, a two-chip solution using bulk micromachining and an optimized technology for the ASIC is the better choice. Further, some requirements, like robustness toward harsh environments, might favor bulk micromachining, as it is the case with our tire-pressure-monitoring sensors.

On the other side, by using a surface- micromachining solution, we have created single-chip pressure sensors that offer two sensor arrays as well as two reference arrays of 42 sensors. [Also on chip] is the circuitry for A/D conversion, temperature compensation, self-diagnosis and voltage regulation, and a state machine to enable different digital protocols. These features increased reliability and made our sensors more intelligent. And because the signal was digital, we were able to make it more immune to electrical interference.

EE Times: When did you first start investigating surface micromachining?

Fischer: In the late 1990s, we began adding surface micromachining to bulk micromachining for its cost as well as reliability advantages in many application areas. The one project that convinced us to focus on surface micromachining was a pressure sensor for side air bags. Here, we could design a sensor that allowed the sensor system supplier not only to sense impact but also to sense the pressure in the door. That lets the system discriminate about three times faster between an accident and, for example, a child's ball hitting the door.

EE Times: What process were you using then?

Fischer: We were using 0.8-micron BiCMOS technology, but since then we've moved to 0.5 micron. We have moved our bulk micromachining from 4-inch to 6-inch wafers while moving the surface-micromachining sensors to 8-inch wafers. We have also significantly shrunk the size of our sensors and the ASIC portion of the surface-micromachining products by going to finer design rules.

EE Times: When did you become interested in acquiring SensoNor AS?

Fischer: The pending legislation in 2001-2003 to mandate tire-pressure sensing on all vehicles in the United States after the Firestone incidents [when deaths and injuries related to defective tires resulted in a recall] inspired us to make the acquisition.

EE Times: What advantage did SensoNor offer?

Fischer: There are many special requirements for tire-pressure sensing, because the environment is much harsher than for simple barometric-pressure sensing. So rather than go through a longer learning curve, we acquired SensoNor, which already had 20 years of experience in bulk micromachining and was one of the leading companies in tire-pressure sensing.

Our colleagues in Norway had been able to solve the harsh environment problems inside a tire with a three-layer sandwich approach. First, we process the front side of the wafer to create the piezoresistive elements that convert the mechanical motion and pressure into electrical signals, as well as the bond pads. This is done in an Infineon standard-logic fab focusing on automotive products. Next, we etch through the backside of the chip to create the mechanical element that senses pressure and acceleration (motion of the tire). Then we put wafer caps on both the top and bottom of the chip. In this way, air does not touch the top side of the chip, which is completely covered in glass. On the backside of the glass cap, we make a small hole to let air in. That enables us to make a pressure sensor that can withstand extremely harsh environments. The later steps are all performed in a MEMS-specific fab in Norway.

EE Times: That sensor was a single-chip solution?

Fischer: We use [one-chip solutions] for barometric-pressure, manifold-air-pressure and side-air-bag deployment sensors, but for tire pressure the circuitry has to include further enhanced logic features, embedded memory and RF functionality. Since that requires a 0.18-µm CMOS technology, we decided to add to the sensor a second chip that includes a microprocessor and a wireless communications module .

EE Times: Is your silicon microphone a single chip?

Fischer:The circuitry in principle would allow a single-chip solution, as it is a surface-micromachining device. However, as our cell phone customers each have their own particular interface requirements, we have a standard sensor chip for all customers but use a second, custom ASIC for the individual circuitry.

EE Times: Did your silicon microphone come from the same group as your automotive sensors?

Fischer: From the development side, yes. But since the market for silicon microphones is mainly cell phones, we have to adopt a different approach on the production side. Automotive has extremely stringent quality requirements, besides being cost-driven, whereas consumer applications are even more cost-driven, with a somewhat lower emphasis on quality.

We considered having two lines with different levels of quality, but instead we decided to just use separate packaging fabs and test lines for consumer and automotive. For our tire-pressure sensors, we fabricate the first nine layers in our fab in Villach [Austria], then ship the wafers to our SensoNor fab in Norway for micromachining the MEMS element. Then they go to Malaysia for packaging and testing. But we ship the silicon microphone wafers directly from Villach to Malaysia, where we use a different packaging line than the one we use for automotive sensors .

EE Times: How does the development cycle differ for automotive and consumer applications?

Fischer: We have a three- to five-year cycle for automotive, but only about nine months for cell-phone microphones and other consumer applications.

EE Times: Do you have any design wins yet for your silicon microphone?

Fischer: We have samples and demonstrators available and see a strong interest in the market to get replacements to the current electret [capacitor] microphones. We are not allowed to mention any customer names right now, but we are very confident that we will see initial business in the very near future.

EE Times: What other consumer applications are you working on?

Fischer:[One is] barometric-pressure sensors for handheld applications such as GPS systems or even laptops.The automotive market has been the driver for growth in the MEMS for the last decade, but future growth will mainly come from the consumer area.