Based on microelectronic mechanical system (MEMS) processing, a large-size 2-D scanning mirror (6.5 mm in diameter) driven by electromagnetic force was designed and implemented in this paper. We fabricated the micromirror with a silicon wafer and selectively electroplated Ni film on the back of the mirror. The nickel film was magnetized in the magnetic field produced by external current coils, and created the force to drive the mirror’s angular deflection. This electromagnetically actuated micromirror effectively eliminates the ohmic heat and power loss on the mirror plate, which always occurs in the other types of electromagnetic micromirrors with the coil on the mirror plate. The resonant frequency for the scanning mirror is 674 Hz along the slow axis, and 1870 Hz along the fast axis. Furthermore, the scanning angles could achieve ±4.5° for the slow axis with 13.2 mW power consumption, and ±7.6° for the fast axis with 43.3 mW power consumption. The application of the MEMS mirror to a laser display system effectively reduces the laser speckle. With 2-D scanning of the MEMS mirror, the speckle contrast can be reduced from 18.19% to 4.58%. We demonstrated that the image quality of a laser display system could be greatly improved by the MEMS mirror.

The Micro-Electro-Mechanical-System (MEMS) micromirror has shown great advantages in Light Detection and Ranging (LiDAR) for autonomous vehicles. The equipment on vehicles is usually exposed to environmental vibration that may degrade or even destroy the flexure of the micromirror for its delicate structure. In this work, a mechanical low-pass filter (LPF) acting as a vibration isolator for a micromirror is proposed. The research starts with the evaluation of vibration influences on the micromirror by theoretical calculation and simulation. The results illustrate that mechanical load concentrates at the slow flexure of the micromirror as it is excited to resonate in second-order mode (named piston mode) in Z-direction vibration. A specific LPF for the micromirror is designed to attenuate the response to high-frequency vibration, especially around piston mode. The material of the LPF is a beryllium-copper alloy, chosen for its outstanding properties of elasticity, ductility, and fatigue resistance. To measure the mechanical load on the micromirror in practical, the on-chip piezoresistive sensor is utilized and a relevant test setup is built to validate the effect of the LPF. Micromirrors with or without the LPF are both tested under 10 g vibration in the Z-direction. The sensor output of the device with the LPF is 35.9 mV in piston mode, while the device without the LPF is 70.42 mV. The attenuation ratio is 0.51. This result demonstrates that the LPF structure can effectively reduce the stress caused by piston mode vibration.

We have developed a manufacturing process for micromirrors based on microelectromechanical systems (MEMS) technology. The process involves designing an electrostatic vertically comb-driven actuator and utilizing a self-alignment process to produce a height difference between the movable comb structure and the fixed comb structure of the micromirror. To improve the stability of the micromirror, we propose four instability models in micromirror operation with the quasi-static driving principle and structure of the micromirror considered, which can provide a basic guarantee for the performance of vertical comb actuators. This analysis pinpoints factors leading to instability, including the left and right gap of the movable comb, the torsion beams of the micromirror, and the comb-to-beams distance. Ultimately, the voltages at which device failure occurs can be determined. We successfully fabricated a one-dimensional micromirror featuring a 0.8 mm mirror diameter and a 30 μm device layer thickness. The height difference between the movable and fixed comb structures was 10 μm. The micromirror was able to achieve a static mechanical angle of 2.25° with 60 V@DC. Stable operation was observed at voltages below 60 V, in close agreement with the theoretical calculations and simulations. At the driving voltage of 80 V, we observed the longitudinal displacement movement of the comb fingers. Furthermore, at a voltage of 129 V, comb adhesion occurred, resulting in device failure. This failure voltage corresponds to the lateral torsional failure voltage.

In scanning laser projection systems, the laser modulation time is important for the projection resolution. The modulation time needs to be matched with the motion of the micromirror. For this paper, the piezoresistive sensor was integrated on the torsion beam of the micromirror to monitor the physical position of the micromirror. The feedback signal was used to generate the zero-crossing time, which was used to estimate the physical position of the resonating mirror over time. The estimated position was affected by the zero-crossing time and the error directly influenced the definition of the projected image. By reducing the impurity concentration from 3 × 10¹⁸/cm³ to 1 × 10¹⁸/cm³ and increasing shear stress on piezoresistive sensor, the sensitivity of the piezoresistive sensor increased from 4.4 mV/V° to 6.4 mV/V° and the error of the image pixel reduced from 1.5 pixels to 0.5 pixels. We demonstrated that the image quality of an Optical-Microeletromechanical Systems (MOEMS) laser projection could be improved by enhancing the sensitivity of the piezoresistive sensor.

Laser beam scanning (LBS) projection systems based on MEMS micromirrors offer advantages such as compact size, low power consumption, and vivid color performance, making them well suited for applications like AR glasses and portable projectors. Among various scanning methods, raster scanning is widely adopted; however, it suffers from artifacts such as dark bands between adjacent scanning lines and non-uniform distribution of the scanning trajectory relative to the original image. These issues degrade the overall viewing experience. In this study, we address these problems by introducing random variations to the slow-axis driving signal to alter the vertical offset of the scanning trajectories between different scan cycles. The variation is defined as an integer multiple of 1/8 of the fast-axis scanning period ($1/{\mathrm{f}}_{\mathrm{h}}$) Due to the temporal integration effect of human vision, trajectories from different cycles overlap, thereby enhancing the scanning fill factor relative to the target image area. The simulation and experimental results demonstrate that the maximum ratio of non-uniform line spacing is reduced from 7:1 to 1:1, and the modulation of the scanned display image is reduced to 0.0006—below the human eye’s contrast threshold of 0.0039 under the given experimental conditions. This method effectively addresses scanning display artifacts without requiring additional hardware modifications.