Microfabrication has become a key technology for sensor miniaturization. The ability to reduce the size of sensing elements through the use of standard semiconductor manufacturing techniques allows for significant size reduction. Integral signal processing along the sensing element further enhances the opportunity to reduce system size, eliminating the need for additional pins linked to external devices. The choice of microfabrication process technology can also determine the limitations of miniaturization, but this is generally determined by the sensor type. Piezoelectric micromechanical elements used for pressure sensing offer a scale ratio compared to fabricating substrates from CMOS silicon, such as on-surface diaphragms, but can provide higher performance.
Figure 1: Surface micromachining of sensing elements. Sensors also mitigate the impact of calibration on sensitivity and performance issues, although this can be alleviated through innovative design of the element structure. Such innovative designs can also be used to integrate multiple elements into a single sensor. This is most evident in 6-axis accelerometers, where carefully designed elements can provide motion data as well as data from multiple axes. This also helps in miniaturizing systems, with one device replacing multiple sensors. However, this move towards integration and miniaturization comes at a cost, typically in noise and dynamic range. For example, adding filtering and signal processing to identify composite sensing elements with different forces and filtering out noise from different elements can result in a slower response time and a limited dynamic range. While this may be acceptable for small, space-constrained applications such as wearable electronics, it may not be suitable for small, unmanned aerial vehicles (UAVs) with similar space constraints but higher accuracy requirements.
Plasma treatment
The Kionix KXCJ9 is a 3-axis silicon micromechanical accelerometer with a range of +/-2g, ±4g, or ±8g. The sensing element is constructed using Kionix's proprietary plasma micromachining technology. Accelerometer sensing is based on the principle of differential capacitance generated by motion caused by process variations, temperature, and environmental stress, which employs a common-mode release design from the sensing element to reduce the occurrence of erroneous acceleration due to process variations, temperature, and environmental stress. The sensing element is hermetically sealed within a second silicon wafer cap using a glass frit, rather than at the wafer level, which helps reduce the overall size of the device.
Figure 2: The Kionix KXCJ9 silicon accelerometer is housed within a hermetically sealed cavity created at the wafer level from a second-mounted wafer. The sensing element is mounted on a separate signal processing ASIC to further reduce the overall sensor size, resulting in a 3×3×0.9 mm LGA plastic package operating from a 1.8V to 3.6V DC power supply. A regulator is used to maintain a constant internal operating voltage across the input supply voltage range. This stable operating characteristic across the input supply voltage range eliminates any errors from supply variations and eliminates the need for external power supply regulation, further reducing the overall design size.
Pressure sensor
STMicroelectronics' LPS25H is an ultra-compact absolute pressure sensor that utilizes a proprietary, in-process monolithic piezoelectric resistor element developed by STMicroelectronics. This is achieved by suspending a membrane within a single monocrystalline silicon substrate, creating an element significantly smaller than those built using conventional methods for silicon microfilms. Membrane breakage is prevented by internal mechanical stops. This provides a range of 260–1260 hPa for measuring absolute pressure and can be applied to sports watches and weather stations.
Figure 3: The LPS25H is housed in a 10-pin cavity-through LGA package (HCLGA) measuring 2.5 × 2.5 × 1 mm, concealing external pressure that can reach the piezoelectric sensing element. This interface uses standard CMOS processes, allowing highly integrated designs to be tailored to better match the characteristics of the sensing element's dedicated circuitry. The entire measurement chain is comprised of a low-noise amplifier where the resistive imbalance of the MEMS sensor (pressure, temperature) is converted to an analog voltage, provided to the user via an embedded 24-bit analog-to-digital converter (ADC) with an output data rate (ODR) selectable from 1 Hz to 25 Hz, single-trigger option. The LPS25H is housed in a 10-pin cavity-through LGA package (HCLGA) measuring 2.5 × 2.5 × 1 mm, concealing external pressure that can reach the sensing element. The use of piezoelectric elements allows it to operate over a temperature range from -30°C to +105°C.
Combined sensor
Combining data from multiple axes of a micromachined accelerometer is often used to alert the main processor that the device is about to be activated, but this approach is also increasingly becoming part of user interfaces that support gesture interfaces. Using combined data allows system designers to further miniaturize devices by eliminating interface technologies such as keyboards and buttons. The Freescale MMA8451Q is a three-axis, capacitive accelerometer used in bulk micromachined machining, utilizing all three axes to provide the necessary data in a 3×3×1 mm QFN package with 14-bit resolution. Motion detection can analyze changes in static acceleration or faster jolts. For example, if the object to be detected is spinning yarn, all three axes will be enabled with a threshold detection exceeding 2 grams. When acceleration exceeds the set threshold, whether as a rapid jitter or a slow tilt, motion can trigger an interrupt, depending on the configured event threshold and timing value. This condition will need to occur for at least 100 milliseconds to ensure the event is not just noise. The timing value is set via a configurable debounce counter, which acts as a filter to determine if a condition exists within a set of configurable times (i.e., 100 milliseconds or longer). There is also directional data provided in the source register to detect the direction of motion. This is useful for applications such as directional shaking or flicking, which helps with algorithms regarding various gesture detections. It provides the opportunity to obtain both low-pass filtered data and high-pass filtered data, minimizing the need for thud detection and enabling faster data analysis. Acceleration data is passed through a high-pass filter, eliminating offset (DC) and low frequencies. The cutoff frequency of the high-pass filter can be set by the user to four different frequencies depending on the ODR. Higher cutoff frequencies ensure that DC data or slower movement data will be filtered out, allowing only higher frequencies to pass. An embedded transient detection function, using high-pass filtered data, allows the user to set thresholds and debounce counters. The transient detection function uses the same method of motion detection by bypassing the high-pass filter, which provides greater flexibility to cover different methods the accelerometer can be used in a variety of customer designs. Many applications use accelerometer static acceleration readings by measuring acceleration changes purely from gravity to measure device tilt. This is thanks to the acceleration data being filtered with a low-pass filter where high-frequency data is considered noise. However, there are also many applications where the accelerometer must analyze dynamic acceleration. Interface functions, such as tap detection, flicking, shaking, and step counting, are based on the analysis of acceleration changes. This is simpler to interpret after removing the static components of dynamic acceleration data. The MMA8451Q has embedded multiple customizable timers for setting pulse width and time delay between pulses, supporting single/dual and directional tap detection. Programmable thresholds and tap detection for all three axes are configurable to run through a high-pass filter, and also through a low-pass filter, providing more customized and tunable tap detection schemes. The status register provides information on detected events, including updates to the direction of the axis and the tap.
in conclusion
Different microfabrication techniques offer designers a variety of optimization options for miniaturizing systems. These range from integrating signal processing to reduce the overall size of sensors that combine multiple data streams, to eliminating other, larger devices in the system, and another way to reduce the overall system size. However, the choice of underlying sensor technology can limit the overall range and device performance, so careful consideration is needed along the way to miniaturization.
