Micromachined Accelerometers with Sub-µg/√Hz Noise Floor: A Review

Huafeng Liu, Chen Wang 0040, Fang Chen, Yuan Wang, Sina Sadeghpour, Chenxi Wang, Mathieu Baijot, Rui Esteves, Chun Zhao, Jian Bai, Huafeng Liu 0002, Michael KraftSensors 2020 · DOI

TL;DR

This review paper surveys the cutting-edge research in micromachined accelerometers designed to achieve an ultra-low noise floor below 1 µg/√Hz. It highlights various innovative design strategies, transduction methods, and circuit techniques that are crucial for applications demanding extreme precision, such as seismic monitoring and inertial navigation.

What Problem Does This Paper Address?

High-resolution accelerometers are essential for a wide range of scientific and technological applications, including inertial navigation, earthquake detection, and geophysical sensing. While traditional macroscopic accelerometers have historically met these demands, they are often bulky, expensive, and power-hungry. The advent of Micro-Electro-Mechanical Systems (MEMS) technology has enabled the development of miniaturized, low-cost, and low-power accelerometers. However, achieving the extremely high sensitivity required for certain applications, specifically a noise floor below 1 µg/√Hz, presents significant challenges. This paper addresses the ongoing research and development efforts aimed at overcoming these challenges and achieving this benchmark in MEMS accelerometers.

Methods

The paper categorizes the methods for reducing the noise floor in MEMS accelerometers into five main approaches:

  • Low Spring Constant: This involves designing mechanical structures with very compliant springs to increase sensitivity. Techniques like Geometric Anti-Spring (GAS) and the use of materials with low Young's modulus (e.g., parylene) are discussed.
  • Large Proof Mass: Increasing the mass that experiences acceleration directly reduces thermal noise (TNEA). This is often achieved through bulk micromachining or using high-density materials like gold.
  • High Quality Factor (Q): A higher Q factor in the resonant system reduces TNEA. Vacuum packaging is a primary method to achieve this, although it can introduce complexities like long settling times.
  • Low Noise Interface Circuit: The electronic circuitry that reads the sensor's output can be a significant source of noise. This section reviews advancements in low-noise ASICs and interface circuits.
  • High Scale Factor Transduction Schemes: Novel sensing methods that inherently provide a larger output signal for a given acceleration input are explored. This includes optical sensing (e.g., grating interference, Fabry-Pérot interferometry), resonant sensing (frequency or amplitude readout), tunneling, and electrochemical sensing.

Datasets

Not discussed in the paper. This is a review paper summarizing existing research, not presenting new experimental data or benchmarks.

Key Insights and Ideas

  • Multi-pronged Approach is Key: Achieving sub-µg/√Hz noise floors rarely relies on a single technique. Most successful designs combine several strategies, such as a large proof mass and a low spring constant and a low-noise interface circuit.
  • Trade-offs are Inherent: While high Q reduces thermal noise, it can lead to stability issues and require complex closed-loop systems, which can introduce their own noise. Similarly, low resonant frequencies increase sensitivity but limit bandwidth and dynamic range.
  • Optical Sensing Shows Promise: Optical MEMS accelerometers, leveraging phenomena like grating interference and Fabry-Pérot cavities, offer a path to high sensitivity by avoiding some of the inherent electrical noise limitations of capacitive or piezoresistive methods.
  • Resonant Sensors Offer High Sensitivity: Resonant accelerometers, particularly those using frequency readout, can achieve high sensitivity and are less susceptible to environmental factors like temperature fluctuations.
  • The "Holy Grail" of Stability: For applications like gravity surveys, long-term stability over days, weeks, or months is as critical as low noise. This remains a significant challenge for MEMS devices.

Key Results

The paper reviews numerous MEMS accelerometers that have achieved noise floors below 1 µg/√Hz. Specific examples include:

  • Boom and Kamp et al. [16]: Achieved a noise floor below 2 ng/√Hz at 28.1 Hz using Geometric Anti-Spring (GAS) technology.
  • Pike et al. [31]: Reported a noise floor of 0.25 ng/√Hz in a frequency band from 0.1 to 10 Hz, utilizing a large proof mass and optimized design.
  • Williams et al. [74]: Demonstrated a noise floor lower than 3 ng/√Hz between 0.1 Hz and 100 Hz using optical micro-grating sensing.
  • Deng et al. [124]: Developed a MEMS-based electrochemical accelerometer with a noise floor of 3.2 ng/√Hz at 0.02 Hz.
  • Pandit and Seshia et al. [112]: Achieved a bias stability of 2.96 µg and a noise floor of 3 µg/√Hz using differential mode localization in a WCR accelerometer.

The paper also summarizes commercial accelerometers that have achieved sub-µg/√Hz noise floors, often by combining large proof masses, low-noise interface circuits, and vacuum packaging.

Limitations and Future Work

The authors identify several limitations and suggest future research directions:

  • Integration and Miniaturization: While many advanced sensing schemes show promise, their integration into compact, robust systems, especially for harsh environments, remains a challenge.
  • Long-Term Stability: Achieving the stability required for applications like gravity surveys (signals down to 10⁻⁴ Hz) is still a significant hurdle for MEMS devices, often requiring performance comparable to commercial macroscopic accelerometers.
  • Novel Materials and Fabrication: Further exploration of non-silicon materials and novel fabrication techniques is needed to optimize mechanical designs and circuit technologies.
  • Combined Approaches: The paper emphasizes the trend towards combining multiple techniques (e.g., GAS with low-noise electronics) to achieve the desired performance.
  • Electrostatically Levitated Accelerometers: While theoretically capable of very low spring constants, these devices have not yet demonstrated noise floors below 1 µg/√Hz, but hold significant future potential.

Summary updated

BibTeX
@article{liu2020micromachined,
  author = {Huafeng Liu and Chen Wang and Fang Chen and Yuan Wang and Sina Sadeghpour and Chenxi Wang and Mathieu Baijot and Rui Esteves and Chun Zhao and Jian Bai and Huafeng Liu and Michael Kraft},
  title = {{Micromachined Accelerometers with Sub-µg/√Hz Noise Floor: A Review}},
  journal = {Sensors},
  year = {2020},
  doi = {10.3390/s20144054},
}

Loading related content...

Figure