TL;DR
This paper introduces a novel wafer-level packaging technique for micro resonant pressure sensors that combines a modified tri-layer anodic bonding method with an integrated titanium (Ti) getter. This approach significantly improves the quality factor (Q) of the resonators and ensures long-term vacuum stability, crucial for high-performance pressure sensing applications.
What Problem Does This Paper Address?
Micro resonant pressure sensors offer high accuracy and long-term stability, making them vital for applications like aerospace and weather monitoring. A key factor in their performance is maintaining a high quality factor (Q), which is heavily influenced by the degree of vacuum within the sensor package. Air damping dissipates kinetic energy, reducing the Q value. Achieving and sustaining a high vacuum level over time is challenging due to gas leakage and outgassing from packaging materials. Existing wafer-level packaging methods for multi-layer resonant sensors, like single-step anodic bonding, often suffer from poor bonding quality and complex processes. This paper aims to address these limitations by developing a more robust and reliable wafer-level packaging solution that enhances vacuum integrity and sensor performance.
Methods
The core innovation lies in a "cross-layer anodic bonding" technique. This method addresses the challenges of traditional tri-layer anodic bonding by first performing silicon-to-silicon direct bonding (SDB) to create a double-layer structure. Subsequently, aluminum (Al) is sputtered onto this combined wafer, acting as an electrical interconnection. This Al layer effectively "shorts out" the insulating silicon dioxide (SiO2) layer, enabling a more uniform and higher-quality anodic bonding with the glass cap layer. This process is designed to achieve a good bonding quality similar to standard glass-silicon (GS) anodic bonding, but for a tri-layer structure.
To further enhance vacuum stability, a titanium (Ti) getter is integrated into the glass cap groove. The paper details an optimization process for the Ti getter, including:
- Sputtering Parameter Optimization: Investigating the effects of pressure and sputtering power on the Ti film's morphology, surface roughness, and specific surface area using SEM and BET measurements.
- Activation Process Optimization: Determining the optimal activation temperature and time for the Ti getter by measuring the oxygen content using EDX after exposure to atmospheric conditions. This activation process is crucial to remove the passivation layer and expose the getter's active surface.
The wafer-level packaging process involves SDB, Al sputtering, cross-layer anodic bonding, and Ti getter deposition. Finally, a Q test system was set up to evaluate the performance of the packaged sensors.
Datasets
Not discussed in the paper. The paper focuses on experimental fabrication and characterization of the sensor package.
Key Insights and Ideas
- "Shorting Out" Principle for Tri-Layer Anodic Bonding: The cross-layer anodic bonding, by sputtering Al to create an electrical interconnection, effectively bypasses the insulating SiO2 layer. This is a crucial insight that simplifies the tri-layer bonding process and improves its quality, overcoming the "voltage division" issue in single-step methods.
- Ti Getter Optimization is Critical: The paper highlights that simply adding a getter isn't enough. Optimizing sputtering parameters (pressure, power) and activation conditions (temperature, time) is essential to maximize the getter's surface area and absorption capacity, ensuring long-term vacuum maintenance.
- Long-Term Stability Demonstrated: The study provides compelling evidence of the package's long-term stability by re-testing samples after 1.5 years, showing that the Q factor remained largely constant, indicating the getter's effectiveness and the package's hermeticity.
- Tensile Strength Improvement: The cross-layer anodic bonding method demonstrated significantly higher tensile strength (25 MPa) compared to single-step bonding (<9 MPa), indicating a more robust and reliable bond.
Key Results
- High Q Factor Achieved: Over 73% of the tested resonators exhibited a Q value greater than 10,000. Nearly one-third of the devices achieved Q values exceeding 20,000, demonstrating the effectiveness of the packaging in maintaining high Q.
- Hermetic Package: The wafer-level packaging process achieved 100% hermeticity, confirmed by visual inspection of the entire wafer.
- Long-Term Q Stability: After 1.5 years, four selected samples showed minimal degradation in their Q values, with two remaining virtually unchanged and two showing slight decreases, attributed to potential getter saturation.
- Improved Bonding Strength: The cross-layer anodic bonding achieved a tensile strength of 25 MPa, significantly higher than the <9 MPa achieved with single-step bonding.
- Getter Performance: Optimized Ti getter preparation resulted in a specific surface area of 50.8695 m²/g and a thickness of approximately 300-500 nm. Optimal activation was achieved at 600 °C for 60 minutes or more.
Limitations and Future Work
- Getter Saturation: While the getter demonstrated long-term effectiveness, a slight decrease in Q for some samples after 1.5 years suggests potential getter saturation. Future work should focus on improving getter preparation to increase specific surface area and adsorption capacity for even longer inspiration times.
- Yield Optimization: The authors acknowledge that the proposed packaging process, while feasible, still requires further optimization to improve the overall yield of high-Q sensors.
- Getter Deflation: The slight decrease in Q for some samples was interpreted as possibly caused by getter deflation, indicating that further research into getter material stability and integration might be beneficial.
Summary updated