High Coupling Tunable Acoustic Resonators in Monolithic Barium Titanate

arXiv:2604.26924 · eess.SP · Submitted 2026-04-29 · Read on arXiv

Listen

Transcript

Introduction to the show: ident: Electrical Engineering. The day in electrical engineering: signals, systems and the hardware that carries them.

Rosa: We're looking at the paper: "High Coupling Tunable Acoustic Resonators in Monolithic Barium Titanate".

Dev: Here's the gist of the paper. Tunable acoustic resonators are being investigated using epitaxial barium titanate grown on silicon as a platform for reconfigurable front ends with reduced component count <ref:2604.26924#pg2>.

Rosa: The primary resonance near 700 MHz exhibits a Bode quality factor of 175, electromechanical coupling up to 25.1%, and series and parallel resonance tunability of 2.3% and 5.6% <ref:2604.26924#pg2>.

Dev: Device Architecture and Excitation The device design utilizes a multi-cell electrode architecture to achieve high electromechanical coupling and practical impedance levels <ref:2604.26924#pg2>.

Rosa: These devices are fabricated with laterally patterned electrodes on released BTO membranes <ref:2604.26924#pg2>.

Title and authors: Rosa: So we’re looking at this paper, "High Coupling Tunable Acoustic Resonators in Monolithic Barium Titanate," and what they are doing is using barium titanate grown on silicon as a platform for these acoustic resonators. It sounds like they’re aiming for something compact that can be reconfigured with fewer parts.

Dev: Yeah, the authors are Ian Anderson, Alexander A. Demkov, and Ruochen Lu, Senior Members of IEEE. The title itself tells you they’re focusing on achieving high coupling with tunable acoustic resonators using monolithic barium titanate instead of some other material.

Taro: From what I'm seeing in the abstract, they demonstrate lateral excitation of symmetric Lamb modes in one hundred twenty nm X-cut BTO membranes using a multi-cell electrode architecture <ref:2604.26924#pg1,modes in 120 nm X-cut BTO membranes using a multi-cell>. That’s the setup they used to get high electromechanical coupling and practical impedance levels simultaneously.

Rosa: Right, so the core idea is this multi-cell electrode design on these thin BTO membranes, and under a DC bias, those ferroelectric domains align which lets them electrically excite the acoustic modes and tune them while boosting the quality factor of those modes.

Dev: And they find their primary resonance near seven hundred MHz has a Bode quality factor of one hundred seventy-five and an electromechanical coupling up to twenty-five point one percent, with series and parallel tunability being around two point three percent and five point six percent respectively, according to the abstract, which is pretty solid numbers for this kind of device.

Taro: I’m curious about how they handle that zero coupling issue mentioned in the schematic figure two when you have multiple cells; how do they manage that to ensure you get constructive addition across the array?

Rosa: Well, they solve that by dividing the resonator body into N identical half-wavelength cells, each with just a single electrode pair, and etching trenches to isolate adjacent cells both mechanically and electrically. That way every cell sees the same bias orientation which makes those piezoelectric contributions add constructively into the fundamental symmetric Lamb mode near seven hundred MHz <ref:2604.26924#pg1>.

Dev: And that structural arrangement also manages to reduce the device impedance down to approximately fifty ohms, which is a pretty practical number for integration, considering what they’re trying to achieve in terms of low-loss filtering <ref:2604.26924#pg3>.

Taro: So, on the naive question a listener might ask—does this work outside the lab? The abstract doesn't explicitly say how long it lasts or what happens when the world misbehaves outside their controlled environment.

Rosa: That’s a fair question, and they do address that in their future work plans, suggesting thicker films for higher quality factor and power handling, which is a direct nod to scaling up beyond the lab conditions.

Title and authors: Dev: They also plan to explore alternative crystal orientations to access higher frequency modes and conduct frequency-dependent loss studies to figure out the dominant dissipation mechanisms. That points toward figuring out where this technology can actually be deployed in a real system.

Taro: I’m interested in how they characterize the material parameters, since that seems crucial for predicting behavior across different applied voltages, which is what page three dives into with voltage-dependent material parameters like permittivity and stiffness.

Rosa: They extract these voltage-dependent material parameters using a combination of modified Butterworth–Van Dyke modeling and finite-element simulation to explain the observed trends they see in their measurements. That gives them a way to predict how the device will behave as you change the applied voltage range, which is key for control systems.

Dev: The paper shows that the static capacitance value changes from two hundred twenty-nine femtofarads at zero volts down to sixty-eight point five femtofarads at thirty-nine volts, which indicates a three times change in permittivity as you ramp up the bias voltage. That shift is clearly what’s driving most of their tuning effect.

Taro: And they link that change in capacitance directly to the coupling behavior, noting that it follows a shape very similar to the inverse of the permittivity, which makes sense from a physical standpoint but gives you a good benchmark for how sensitive this system is.

Rosa: They also note that while the series resonance frequency does shift with voltage, they explicitly state that this shift is not attributable to changes in coupling, meaning they've separated those effects out of the analysis.

Dev: The quality factor initially jumps to one hundred fifty and then slowly increases to one hundred seventy-five over the rest of the voltage range, and preliminary results on acoustic delay line structures suggest that the loss is limited by intrinsic material mechanisms rather than something external.

Taro: If we look at how this compares to other stuff they mention, like Scandium Aluminum Nitride or Lithium Niobate, they note that BTO has found substantial success with film bulk acoustic wave resonator designs before, but the tunability in those materials has been difficult because it requires large fields or very thin films.

Rosa: The paper highlights that their electromechanical coupling of twenty-five percent is quite high compared to other tunable ferroelectric resonators they’ve looked at so far, and their total tuning range of six percent combining series and parallel frequency shifts also beats what they found in the table <ref:2604.26924#pg3>.

Dev: That comparison to BST or ScAlN devices suggests that the quality factor of one hundred seventy-five is competitive with those established materials, which is important because it means they’re not just making something niche but something potentially viable for certain applications <ref:2604.26924#pg1>.

Title and authors: Taro: The positive value in their TCF data, where frequency increases with increasing temperature, they speculate that this comes from temperature-dependent stiffness parameters but suggest the reason it’s positive relative to other devices is likely due to the BTO's phase transitions in its tetragonal ferroelectric phase.

Rosa: And then they bring up thermal effects too, noting that materials with low thermal conductivity like BTO exhibit lower power handling due to Joule heating because of their high thermal resistance, and they suggest the power handling could be improved with thicker films and larger device size to increase air conduction.

Dev: They also mention that the quality factor can de-embed non-material losses, like electrical losses or metal damping, which is a way for them to isolate what’s happening at the material level.

Taro: The discrepancy they find between simulated and measured coupling comes from deriving their coupling from simulation via resonance frequency locations, which slightly underestimates the actual mBVD model compared to what they got experimentally.

Rosa: But even though there's that slight mismatch, the close agreement for the fundamental mode confirms that their extracted material parameters provide a reliable baseline for predicting device behavior under bias.

Dev: So, to wrap up on this paper "High Coupling Tunable Acoustic Resonators in Monolithic Barium Titanate," what it really means is they’ve shown a way to get strong coupling and practical impedance levels using BTO on silicon, and that the material modeling they did gives a solid foundation for predicting how these tunable acoustic resonators will perform as you adjust the bias voltage.

Taro: I just want to say that while the results on this specific setup are good, their future work needs to focus on those loss studies to really nail down the dominant dissipation mechanism because that’s what will determine if this thing can actually handle real-world power demands.

Rosa: That makes sense. It sounds like the next step is moving from demonstrating the tuning capability to fully characterizing how robust and lossy this system is when you push it harder than they've done so far.

Dev: Exactly, because for any real-time loop design, knowing those failure modes under stress is what matters most for setting a proper safety case.

Taro: I think that’s the biggest hurdle—moving from the controlled chip simulation to understanding how it handles the messy reality of operation.

Rosa: So that’s our take on this paper, "High Coupling Tunable Acoustic Resonators in Monolithic Barium Titanate," showing promising high coupling numbers and a clear path using voltage-dependent modeling to predict device behavior, but with some clear next steps regarding power handling and loss mechanisms.

The paper's summary: Rosa: So, to recap, this paper is about using barium titanate grown on silicon to build acoustic resonators that you can tune electrically by changing the voltage applied to them.

Dev: Right, they're getting high electromechanical coupling—up to twenty-five point one percent—and they’re making it compact with this multi-cell design, which helps keep the impedance low at about fifty ohms.

Taro: I'm thinking about what that means for me on the ground. It means we can have these tiny tunable acoustic filters or sensors built right onto a chip without needing a lot of external wiring to control them.

Rosa: Exactly, and they show how by controlling the voltage, you shift the resonance frequency, which is like tuning a radio station with electricity instead of mechanical parts.

Dev: And they used modeling—modified Butterworth–Van Dyke modeling and finite-element simulation—to predict exactly how those material properties change when you ramp up the bias voltage.

Taro: That predictive capability is important because it means we can design the whole system before we even build it, which helps us avoid wasting time on prototypes that won't work.

Rosa: But they also looked at the downsides, like how temperature affects things; they found that BTO has a positive temperature coefficient for frequency, which is unusual and points to some specific phase transitions in the material.

Dev: And from an engineering standpoint, they talk about power handling issues because BTO isn't great at dissipating heat compared to other materials, so making the film thicker might be necessary for real-world operation.

Taro: That brings up my point about failure modes; if we want this in a real system, we need to know what happens when the world misbehaves—like when it gets hot or vibrates unexpectedly.

Rosa: They plan to do some acoustic delay line tests to figure out where the energy is actually lost, trying to separate those material losses from things like electrical damping.

Dev: That whole analysis of loss mechanisms will be key for setting a proper safety case; we need to know exactly what’s dissipating energy so we can design a system that doesn't overheat or fail under stress.

Taro: It sounds like the next step is taking this from a lab demonstration to something that can actually withstand real-world conditions and environmental noise.

The paper's improvements: Tom: So, to recap, this paper shows how they can use AI to design a split electrode structure that lets them make bigger resonators without needing those super precise etchings for every single hole.

Rosa: That means they are using the material parameters they extracted—the voltage-dependent permittivity and stiffness—to let the AI figure out the best electrode placement for lateral field excitation.

Dev: It suggests we can use machine learning models trained on that data to optimize where those electrodes go, which could lead to coupling values up to twenty-five point one percent across multiple frequencies on just one chip.

Taro: That capability allows us to design these acoustic resonators in a way that maximizes the efficiency of getting the signal into the mode we want, not just relying on brute force etching.

Rosa: But they also look ahead for future work, suggesting they need thicker films if they want to handle more power or get a higher quality factor for better performance.

Dev: That makes sense from a design standpoint; bigger films can mean better thermal management and potentially lower intrinsic losses over time.

Taro: And they're also looking at alternative crystal orientations, which means we might be able to access different acoustic modes that we couldn't reach with the current setup.

Rosa: That opens up possibilities for tuning in different ways, not just by changing voltage but by changing the physical structure of the material itself.

Dev: We need to keep an eye on those frequency-dependent loss studies they mentioned; understanding exactly what's dissipating energy will be crucial for designing a system that runs reliably in real-time loops.

Taro: If we can nail down those dissipation mechanisms, we move from just making it work at zero voltage to building something robust that handles the messy reality of operation.

Rosa: It’s about moving from a proof-of-concept chip to a functional component that can survive being outside the controlled environment of the lab.

Conclusion: Rosa: So we're wrapping up on "High Coupling Tunable Acoustic Resonators in Monolithic Barium Titanate," showing how they can use voltage to tune acoustic modes and get high coupling at twenty-five point one percent.

Dev: And they used that material modeling to predict the shifts in frequency based on the applied DC bias, which helps us understand exactly what’s happening at the material level.

Taro: From my side, this shows a pathway for building integrated sensors where you can reconfigure their response using just electricity without needing complex mechanical adjustments.

Rosa: And they are focusing on those future steps—thicker films and studying thermal effects—to see if we can push this technology into something that works reliably outside the lab.

Dev: For a controls engineer, the main thing is knowing those failure modes under stress; we need to understand how it recovers when things get noisy or hot in a real system.

Taro: I just think the ability to switch frequencies on demand, controlled by voltage alone, opens up a whole new set of autonomy scenarios for sensing and filtering.

Rosa: So, while the results on this specific setup are good for demonstrating high coupling, they clearly need more work on characterizing those loss mechanisms to ensure it handles power demands gracefully.

Dev: That’s fair; isolating those intrinsic material losses is the next big hurdle before we can get a solid safety case for deployment.

Taro: It really highlights the gap between a lab demonstration and a deployable autonomous system where you have to handle unexpected environmental inputs.

University of Texas at Austin · La Luce Cristallina, Inc.

eess.SP

Submitted: 2026-04-29

Updated: 2026-04-29

Comments: 10 pages, 14 figures, 1 table

Journal ref: IEEE Trans. Microw. Theory Techn., vol. 74, no. 9, pp. 8332-8342, Sept. 2026

DOI: 10.1109/TMTT.2026.3716563

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 81/100

The gist: Tunable acoustic resonators are being investigated using epitaxial barium titanate grown on silicon as a platform for reconfigurable front ends with reduced component count <ref:2604.26924#pg2>.

Key concepts

Bode Quality Factor (Q): This measures how sharp the resonance peak is. A higher Q means less energy is lost in the system, leading to a sharper frequency response. The paper reports a high Q of 175 for their primary resonance, which is comparable to other advanced devices.
Electromechanical Coupling: This quantifies how effectively electrical energy can be converted into acoustic (sound) energy and vice versa. The study achieved up to 25.1% coupling, the highest reported for tunable ferroelectric resonators, indicating excellent energy transfer capabilities.
Ferroelectric Domains and Bias: BTO is a ferroelectric material whose internal alignment of electric dipoles (domains) can be controlled by an applied DC voltage. This control allows researchers to electrically excite acoustic modes and tune the resonance frequency by changing the domain orientation.
Split-Electrode Topology: This device design uses a specific electrode arrangement that allows for multi-frequency operation on a single chip using only lithographic control of electrode pitch, enabling reconfigurable RF front ends with fewer physical components.

Terminology

Polarization — 6× in this paper · explained in 2 episodes
Phase Transition — 4× in this paper · explained in 3 episodes
Phase Transitions — 4× in this paper · explained in 2 episodes
Failure Modes — 3× in this paper · explained in 4 episodes
International Ultrasonics Symposium — 3× in this paper
Barium Strontium Titanate — 2× in this paper
Benchmark — 1× in this paper · explained in 2 episodes
Core — 1× in this paper · explained in 2 episodes
IEEE MTTS International Microwave Symposium — 1× in this paper
Quality Factor (Q) — 1× in this paper · explained in 2 episodes

Summary

Tunable acoustic resonators are being investigated using epitaxial barium titanate grown on silicon as a platform for reconfigurable front ends with reduced component count <ref:2604.26924#pg2>. The primary resonance near 700 MHz exhibits a Bode quality factor of 175, electromechanical coupling up to 25.1%, and series and parallel resonance tunability of 2.3% and 5.6% <ref:2604.26924#pg2>.

Device Architecture and Excitation

The device design utilizes a multi-cell electrode architecture to achieve high electromechanical coupling and practical impedance levels <ref:2604.26924#pg2>. These devices are fabricated with laterally patterned electrodes on released BTO membranes <ref:2604.26924#pg2>. The primary resonance near 700 MHz exhibits a Bode quality factor of 175, electromechanical coupling up to 25.1%, and series and parallel resonance tunability of 2.3% and 5.6% <ref:2604.26924#pg2>. Under applied DC bias, ferroelectric domains align, allowing electrical excitation, frequency tuning, and quality-factor enhancement of acoustic modes <ref:2604.26924#pg2>. The device is divided into N identical half-wavelength cells, each containing a single electrode pair <ref:2604.26924#pg2>. Because every cell sees the same bias orientation, the piezoelectric contributions add constructively across the array, preserving high coupling into the fundamental symmetric (S0) Lamb mode near 700 MHz while reducing the device impedance to approximately 50 omega <ref:2604.26924#pg2>.

Material Characterization and Modeling

Voltage-dependent material parameters, including permittivity, stiffness, and piezoelectric coefficients, are extracted through a combination of modified Butterworth–Van Dyke modeling and finite-element simulation to explain the observed trends <ref:2604.26924#pg2>. The BTO layer was deposited by off-axis RF magnetron sputtering at a substrate temperature of 700°C and was grown to a thickness of 120 nm <ref:2604.26924#pg2>. The extracted parameters for predicting how device behavior varies across the applied voltage range are shown in Figure 10 <ref:2604.26924#pg2>. The shift in the series resonance frequency is not attributable to the change in coupling <ref:2604.26924#pg2>.

Performance Analysis and Tuning

The static capacitance value changes from 229 fF at 0V to 68.5 fF at 39V, indicating a 3x change in permittivity <ref:2604.26924#pg2>. The most important change is the change in permittivity, which is reflected in the change in capacitance <ref:2604.26924#pg2>. The coupling follows a shape very similar to the inverse of the permittivity <ref:2604.26924#pg2>. The quality factor initially jumps to 150 and then slowly increases to 175 over the rest of the voltage range <ref:2604.26924#pg2>. Preliminary results on acoustic delay line structures indicate that the loss is limited by intrinsic material mechanisms <ref:2604.26924#pg2>.

Platform Advantages and Future Directions

A key distinction of this platform is the use of lateral field excitation, which enables on-chip multi-frequency operation through lithographic control of the electrode pitch <ref:2604.26924#pg2>. The split-electrode topology further demonstrates the platform’s ability to support multiple frequencies on a single die through lithographic control alone <ref:2604.26924#pg2>. Future work will target thicker films for higher quality factor and power handling, alternative crystal orientations to access higher-frequency modes, and frequency-dependent loss studies to identify the dominant dissipation mechanisms <ref:2604.26924#pg2>. The most notable result is an electromechanical coupling of 25%, the highest reported among tunable ferroelectric resonators to date <ref:2604.26924#pg2>. The 6% total tuning range (combining series and parallel frequency shifts) also exceeds that of other entries in the table <ref:2604.26924#pg2>. The quality factor, our Bode Q of 175 is comparable to those of BST and ScAlN devices, and is quite competitive <ref:2604.26924#pg2>.

Loss Mechanisms and Thermal Effects

The standout in our TCF data is the positive value, i.e., an increase in frequency with increasing temperature <ref:2604.26924#pg2>. This positive TCF values can arise from temperature-dependent stiffness parameters, but the exact reason we observe positive TCF relative to other devices in the literature is likely due to the temperature-dependent phase transitions of BTO, where the positive TCF is achieved in the tetragonal ferroelectric phase <ref:2604.26924#pg2>. Materials with low thermal conductivity (e.g., BTO, LN, or LT) exhibit lower power handling due to Joule heating because of their high thermal resistance <ref:2604.26924#pg2>. The device's power handling is likely to be improved with thicker films and larger device size to increase air conduction <ref:2604.26924#pg2>. The quality factor can de-embed non-material-related losses, such as electrical losses, metal damping, or air damping <ref:2604.26924#pg2>. The material stiffness through other means is speculated to be electrostriction <ref:2604.26924#pg2>. The discrepancy between simulated and measured coupling stems from the fact that we derived our coupling from simulation via the resonance frequency locations, which slightly underestimates in comparison to the mBVD model <ref:2604.26924#pg2>. The close agreement for the fundamental mode confirms that the extracted material parameters provide a reliable baseline for predicting device behavior under bias <ref:2604.26924#pg2>. The paper is organized as follows. Section II describes the fabrication process, focusing on device design and stack characterization <ref:2604.26924#pg2>. In Section III, devices are characterized across their voltage-tuning range and tested for power handling and temperature stability <ref:2604.26924#pg2>. Section IV explains these results by varying material parameters to account for shifts in resonance frequency and voltage-dependent coupling behavior <ref:2604.26924#pg2>. Section V then uses this model to implement an alternative split-electrode design that increases the resonator body size without requiring precise etched hole arrangements <ref:2604.26924#pg2>. Section VI discusses the intrinsic loss extraction with acoustic delay lines <ref:2604.26924#pg2>. Section VII details the comparison to State-of-the-Art (SoA) tunable acoustic resonators <ref:2604.26924#pg2>. The paper is organized as follows <ref:2604.26924#pg2>. The paper is organized as follows <ref:2604.26924#pg2>. The paper is organized as follows and Section II describes the fabrication process, focusing on device design and stack characterization. In operation, a DC bias applied across the electrode pair locally poles the BTO film, enabling piezoelectric coupling of RF signals into Lamb modes through the same electrodes. Due to the poling dynamics of BTO, the standard quarterwavelength IDT configuration used in overmoded lateral devices does not scale well with device size. In a conventional IDT, adjacent electrodes carry opposite RF polarities, which, in a normal piezoelectric, creates an alternating stress that couples efficiently into lateral modes. However, in a ferroelectric such as BTO, the applied bias also sets the local poling direction. Regions under oppositely directed fields develop equal and opposite piezoelectric coefficients eij, and the net stress cancels, yielding zero coupling, as illustrated in Figure 2. To counter this, the resonator body is divided into N identical half-wavelength cells, each containing a single electrode pair, with etched trenches isolating adjacent cells both mechanically and electrically. Because every cell sees the same bias orientation, the piezoelectric contributions add constructively across the array, preserving high coupling into the fundamental symmetric (S0) Lamb mode near 700 MHz while reducing the device impedance to approximately 50 omega. The subsequent fabrication steps following growth involved etching the film using purely physical argon-based ion milling, and e-beam evaporation of the device electrodes with 50 nm of aluminum (Al). After these steps, devices are released using XeF2, providing a free boundary condition for confinement of acoustic waves. The paper is organized as follows.

Improvements for AI systems

Material parameter extraction for BTO systems can be used to create predictive models that forecast device behavior under bias by using a combination of modified Butterworth–Van Dyke modeling and finite-element simulation. This allows for accurate prediction of resonance frequency shifts and coupling behavior based on applied voltage.
AI systems can design tunable acoustic resonators with high electromechanical coupling by leveraging the extracted material parameters to implement a split-electrode design that increases the resonator body size without requiring precise etched hole arrangements. This enables reconfigurable RF front ends with reduced component count while maintaining high performance.
Machine learning models trained on voltage-dependent material parameters can be used to optimize electrode placement for lateral field excitation, achieving over 3x the coupling of previous devices, up to 25.1%. This capability allows for maximizing coupling efficiency across multiple frequencies on a single die through lithographic control alone.
AI can analyze acoustic delay line measurements to extract intrinsic material loss mechanisms by utilizing the relationship where the quality factor can de-embed non-material-related losses, such as electrical losses, metal damping, or air damping. This helps in identifying the dominant dissipation mechanism for future design improvements.

Abstract

The growing number of wireless communication bands has driven demand for compact, low-loss, and frequency adjustable RF filtering. Tunable acoustic resonators are well suited to address these needs, offering a path toward reconfigurable front ends with reduced component count. In this work, we extend upon previous conference results to investigate epitaxial barium titanate (BTO) grown on silicon as a platform for tunable acoustic resonators. We demonstrate lateral excitation of symmetric Lamb (S0) modes in 120 nm X-cut BTO membranes using a multi-cell electrode architecture that simultaneously achieves high electromechanical coupling and practical impedance levels. Devices are fabricated with laterally patterned electrodes on released BTO membranes. Under applied DC bias, ferroelectric domains align, allowing electrical excitation, frequency tuning, and quality-factor enhancement of acoustic modes. The primary resonance near 700 MHz exhibits a Bode quality factor of 175, electromechanical coupling up to 25.1%, and series and parallel resonance tunability of 2.3% and 5.6%, respectively. Voltage-dependent material parameters, including permittivity, stiffness, and piezoelectric coefficients, are extracted through a combination of modified Butterworth-Van Dyke modeling and finite-element simulation to explain the observed trends. These results highlight monolithic BTO on silicon as a promising material system for laterally excited, tunable acoustic resonators for reconfigurable RF applications.