Flexible No-Drift Data Glove Using Ultrathin Silicon for the Metaverse
Academic Background
With the rapid development of the Metaverse, human-machine interface (HMI) technology has become a key link between virtual space and human users. Among these, gesture recognition technology is particularly important in the Metaverse, especially the precise capture of finger movements. Traditional bending sensors used for gesture recognition are typically based on polymer materials, such as rubber and adhesives. However, due to the viscoelasticity of polymers, these sensors often suffer from signal drift during long-term use. This drift causes the sensor’s output to change over time, reducing its reliability and long-term stability. Therefore, developing a flexible bending sensor without signal drift has become a significant challenge.
This study aims to address this issue by using ultrathin silicon material and a novel direct bonding technique to design an elastic bending sensor with no signal drift. This sensor not only exhibits high sensitivity and long lifespan but also maintains stable performance after multiple bending cycles, making it suitable for applications such as gesture recognition in the Metaverse and robotic control.
Source of the Paper
This research was jointly conducted by a team from The University of Tokyo, RIKEN (the Institute of Physical and Chemical Research in Japan), and the National Institute of Advanced Industrial Science and Technology (AIST). The main authors include Seiichi Takamatsu, Masahito Takakuwa, Kenjiro Fukuda, and others. The paper was published in 2025 in the journal Device, titled “Flexible No-Drift Data Glove Using Ultrathin Silicon for the Metaverse.”
Research Process
1. Design and Fabrication of the Ultrathin Silicon Sensor
The research team first used microelectromechanical system (MEMS) technology to fabricate an ultrathin silicon sensor with a thickness of only 5 micrometers. The specific steps are as follows:
- A device called a “deep etching machine” was used to thin the silicon wafer, forming an ultrathin silicon layer.
- A 150-nanometer-thick piezoresistive layer was formed on the silicon surface through phosphorus ion implantation and annealing.
- Cr and Au materials were deposited and patterned on the piezoresistive layer to form electrodes.
- The ultrathin silicon sensor was separated from the silicon substrate using a vacuum-assisted lift-off technique, ensuring sufficient flexibility.
2. Water-Vapor-Plasma-Assisted Au–Au Direct Bonding
To eliminate the viscoelastic effects of adhesives, the research team developed a novel direct bonding technique. The process is as follows:
- Au wiring was deposited and patterned on a 2-micrometer-thick parylene film.
- Water vapor plasma treatment was used to activate the Au surfaces, enhancing surface activity.
- The Au electrode of the ultrathin silicon sensor and the Au wiring on the parylene film were manually aligned and bonded at room temperature.
- The bonded samples were annealed at 200°C to further enhance the bonding strength.
3. Performance Testing of the Bending Sensor
The research team conducted multiple tests on the newly designed sensor, including bending sensitivity, signal drift, and cyclic bending stability:
- Bending Sensitivity Test: The sensor was wound around cylinders of different curvatures, and the resistance change relative to curvature was measured. The results showed that the new sensor had a sensitivity of 0.0712 (1/mm) and a linearity (coefficient of determination) of 0.99, demonstrating high sensitivity and linearity.
- Signal Drift Test: The sensor was fixed on a cylinder with a curvature of 0.4 mm⁻¹, and its output was measured over time. The results showed that the new sensor exhibited no drift over 1,500 seconds, whereas the output of the traditional sensor with adhesive decreased by 15%.
- Cyclic Bending Test: The sensor underwent 10,000 bending cycles, and the results showed no performance degradation in the new sensor, indicating excellent long-term stability.
Research Findings and Conclusions
The study revealed that the bending sensor based on ultrathin silicon and Au–Au direct bonding has the following features:
- No Signal Drift: By eliminating the viscoelastic effects of adhesives and polymer substrates, the sensor exhibited stable output during long-term use.
- High Sensitivity and Linearity: The sensor’s resistance change showed a linear relationship with curvature, with a high sensitivity of 0.0712 (1/mm).
- Long-Term Stability: The sensor showed no performance degradation after 10,000 bending cycles, making it suitable for long-term applications.
This new sensor is not only suitable for gesture recognition in the Metaverse but also can be applied in robotic joint control, wearable devices, and other fields. The establishment of a mechanical model also provides a theoretical framework for the design of other elastic sensors.
Research Highlights
- Innovative Material and Structural Design: For the first time, ultrathin silicon was combined with water-vapor-plasma-assisted Au–Au direct bonding to completely eliminate viscoelastic issues.
- High-Performance Characteristics: The new sensor outperformed traditional sensors in sensitivity, linearity, and long-term stability.
- Broad Application Prospects: This research provides significant technological support for the Metaverse, robotics, and wearable device fields.
Other Valuable Information
The research team also demonstrated the sensor’s application in virtual reality (VR). By integrating the sensor into a data glove, they successfully achieved real-time mapping of gestures in virtual space, showcasing its potential in practical applications.
This paper offers new design ideas and experimental methods for the field of flexible sensors, holding significant scientific value and application potential.