Cladding Layer Enhanced GHz Bulk Acoustic Wave Resonance in Sodium Niobate Thin Films on Silicon

Technical Summary: Cladding Layer Enhanced GHz Bulk Acoustic Wave Resonance in Sodium Niobate Thin Films on Silicon

Problem Statement
The development of lead-free Bulk Acoustic Wave Resonators (BAWR) for next-generation 5G/6G telecommunications faces significant hurdles when utilizing Sodium Niobate (NaNbO3 or NNO) on silicon substrates. While NNO offers a high electromechanical coupling factor, its integration is plagued by two primary issues:

1. Leakage Current: Physical vapor deposition (PVD) often leads to sodium deficiency and incomplete ionization of Nb5+, creating a mixture of Nb4+/Nb5+ states. This results in unintentional hole doping, reduced bandgaps, and small polaron hopping leakage currents, which degrade resonator performance.
2. Thermal Strain and Cracking: There is a substantial mismatch in thermal expansion coefficients between NNO (1.6×10⁻⁵) and silicon (2.6×10⁻⁶). Upon cooling from growth temperatures, this induces large thermal tensile strain, causing ubiquitous cracking in NNO films. These cracks lead to non-uniform vibration, high electrical loss, and low breakdown voltage limits.

Methodology
The authors fabricated NNO-based heterostructures on Si(001) using two distinct buffer layer strategies to bridge the lattice mismatch and manage strain:

• ZCL-route: Utilizes a ZrO2-CeO2-La0.5Sr0.5Co0.5O3 (LSCO) tri-layer buffer. This route employs a 45° in-plane rotation of the perovskite lattice relative to the silicon substrate to accommodate lattice mismatch.
• ZPS-route: Integrates a crystalline Pt(001) bottom electrode sputtered directly onto ZrO2, avoiding the 45° rotation.

To address the leakage and strain issues, the authors introduced a cladding-layer strategy. They sandwiched the 250 nm NNO piezoelectric layer between thin (20 nm) wide-bandgap insulator (WBI) cladding layers, specifically DyScO3 (DSO) or LaAlO3 (LAO). These claddings were selected for their wide bandgaps (~5.6 eV) to suppress leakage and their distinct lattice parameters to manipulate the strain state of the NNO. The films were grown via Pulsed Laser Deposition (PLD) and DC sputtering, followed by high-temperature annealing in oxygen to remove vacancies.

Key Results

• Strain Engineering and Phase Control: The insertion of cladding layers successfully mitigated thermal tensile strain and prevented film cracking. X-ray diffraction (XRD) and Transmission Electron Microscopy (TEM) revealed that the cladding layers exerted significant in-plane strain on the thick NNO film.
  • The LAO/NNO/LAO (LNL) sandwich resulted in a c-axis lattice parameter of 3.93 Å, promoting a vertically distorted tetragonal phase.
  • The DSO/NNO/DSO (DND) sandwich resulted in a c-axis of 3.89 Å.
• Electrical Performance:
  • Leakage Mitigation: The WBI claddings effectively compensated for the large leakage current inherent in pure NNO films, allowing for higher DC bias application without breakdown.
  • Resonance Signals: Devices without claddings exhibited weak resonance (S11 ~ -0.66 dB) or broad, low-Q peaks due to cracks and high electrical loss. In contrast, the cladded devices showed significantly improved resonance signals. The LNL structure achieved an S11 of ~-5.9 dB and Z11 of ~8.4 dB-Ω at ~3.4 GHz.
  • Electromechanical Coupling: Both DND and LNL structures demonstrated high effective electromechanical coupling factors ($k_{eff}^2$) of 29.8% and 31.3%, respectively.
  • Piezoelectric Coefficient: Laser Doppler Vibrometry (LDV) measurements showed an effective piezoelectric coefficient ($d_{33,eff}$) of ~20 pm/V for the LNL structure, comparable to NNO grown on single-crystal substrates, while the DND structure showed only ~6 pm/V.
• Remanence and Ferroelectricity: The LNL structure exhibited a "negative remanence" signal and hysteretic butterfly loops, indicating the presence of out-of-plane ferroelectric polarization. The DND structure lacked these features, suggesting a lack of out-of-plane polarization.

Significance and Claims
The paper claims that the cladding-layer strategy is a critical enabler for high-performance, lead-free NNO resonators on silicon. By sandwiching NNO between wide-bandgap insulators, the authors successfully:

1. Countered Thermal Tensile Strain: Preventing the formation of cracks that typically plague NNO-on-Si films.
2. Suppressed Leakage Current: Enabling the application of necessary DC bias voltages for polarization.
3. Engineered Crystal Phase: Demonstrated that thin cladding layers can promote a vertically aligned tetragonal phase in NNO, which is essential for strong BAWR signals relying on $d_{33}$.

The authors note that while the current devices function as Solidly Mounted Resonators (SMR-BAW) with low Q-factors (approx. 10) due to the lack of a released acoustic cavity, the achievement of high coupling factors and strong signals on silicon validates the material system for future acoustic filter applications. They posit that with the future creation of acoustic cavities (via substrate release), the Q-factor and signal strength could approach industry standards for broadband internet filters. The work aligns with environmental goals by offering a viable lead-free alternative to traditional piezoelectric materials.