WorksPhysics and Astronomy
High-Q Millimeter-Wave Acoustic Resonators in Thin-Film Lithium Niobate Using Higher-Order Antisymmetric Modes
Abstract
This letter presents miniature millimeter wave (mmWave, above 30 GHz) acoustic resonators based on a single-layer thin-film lithium niobate (LN) platform. More specifically, we present high performance third-order antisymmetric (A3) mode laterally excited bulk acoustic resonators (XBAR). Compared to prior demonstrations, the proposed platform features a compact footprint due to a smaller lateral wavelength and aperture. We showcase an A3 mode device operating at 39.8 GHz with a high extracted electromechanical coupling (k^2) of 4%, a high 3-dB series resonance quality factor (Q_s) of 97, and a high 3-dB anti-resonance quality factor (Q_p) of 342, leading to a figure of merit (FoM=k^2*Q_p) of 13.7 with a footprint of 32x44 micron^2. To demonstrate frequency scalability, the piezoelectric film thickness is varied while keeping the device layout. As a result, we present a multitude of high-performance devices covering a wide frequency range of 30-50 GHz, validating the proposed XBAR design at mmWave.
Introduction
With the pervasiveness of 5G wireless communication, compact front-end signal processing components at millimeter-wave (mmWave, above 30 GHz) have garnered attention [1, 2, 3, 4]. Below 6 GHz, surface acoustic wave (SAW) and bulk acoustic wave (BAW) devices - based on piezoelectric materials such as aluminum nitride (AlN), scandium aluminum nitride (ScAlN), lithium niobate (LN), and lithium tantalate (LT) - have been instrumental in realizing miniature filters for mobile applications [5, 6, 7]. However, scaling these technologies to mmWave introduces intrinsic limitations, posing a challenge for the next generation of high-frequency mobile communication systems [8], [9].
Despite significant advances in state-of-the-art radio frequency acoustic technologies, inherent limitations to frequency scaling continue to motivate investigations into alternative platforms. Solidly mounted devices have recently achieved very high phase velocities, but they necessitate sub-100 nm feature patterning alongside expensive substrates, e.g. silicon carbide (SiC) and diamond [10], [11]. Cross-sectional Láme mode resonators exhibit high phase velocity and , yet they also require fine features [12, 13, 14]. Film bulk acoustic
wave resonators (FBAR) exhibit potential for mmWave utilizing acoustic overtones or periodically poled piezoelectric films (P3F), but they face challenges with moderate , primarily due to the large metal to piezoelectric volume ratio at the required film thickness [15].
Alternatively, recent progress in thin-film transfer technology has enabled first-order antisymmetric (A1) mode laterally excited acoustic wave resonators (XBAR) in lithium niobate (LN), without the need for a bottom electrode [4]. However, while recent mmWave XBAR prototypes have reported high electro-mechanical coupling (), they have shown moderate [16, 17, 18, 19], yielding a moderate figure of merit (FoM, , Figure 1). To address the limitations, recent efforts developed P3F LN XBARs, achieving as high as 237 at 50 GHz using bilayer LN [22, 23, 24, 25]. Nevertheless, the P3F structure introduces new fabrication limitations, particularly in maintaining precise thickness control over each layer. These implications motivate a revisit of single-layer XBARs with a new design strategy.

Figure 1. A survey of acoustic resonator performance at 10-60 GHz, including prior single-layer LN [16, 17, 18, 19], single-layer AlN and ScAlN [13, 14, 15], [20], [21], P3F LN [22, 23, 24, 25], and P3F ScAlN [26, 27, 28]. We use the obtained 3-dB to compare with previously reported FoMs, where the higher value between the self-reported and was used.
In this work, we present high-performance single-layer mmWave acoustic resonators by minimizing their footprint and maximizing the device volume-to-area ratio. We present devices spanning 30-50 GHz, covering a broad portion of the mmWave spectrum. Notably, we highlight a miniature XBAR at 39.83 GHz with a 3-dB series resonance () quality factor () of 97, a high 3-dB antiresonance () quality factor () of 342, and an extracted of 4%. Based on the extracted and 3-dB , the device achieves a high FoM of 13.8 with a compact footprint of .

Figure 2. The cross-sectional view of the chosen acoustic stack.

Figure 5. (a) Top view of the finalized device design. The key design parameters are labeled.

Figure 3. A3 Mode dispersion analysis in 128Y LN under electrically open and short boundary conditions with air feedthrough. Calculated coupling is labeled.

Figure 6. FEA simulation using the parameters listed in Fig. 5.
Device Design
Suspended acoustic resonators were designed in a 128Y LN/amorphous silicon/sapphire stack (Fig. 2). We selected a sample with LN film thickness ranging from 110 to 160 nm to investigate device performance from 30 to 50 GHz.
Diverging from the conventional XBAR design space with long lateral wavelengths (), the devices were made compact by reducing to 3.15 m and 4 m. This choice was made by inspecting the third-order antisymmetric Lamb mode (A3) dispersion curves, enabling a trade-off between to achieve a higher (Fig. 3). While prior research indicates that a higher is preferred due to superior spurious-mode performance for A1, we specifically tailored the design for A3 operation at mmWave [29]. Utilizing the low chosen , the device footprint was reduced for compactness and to minimize surface losses in the acoustic resonator, which arise from both the electrical and the mechanical domains and significantly impact overall device performance [30].
Next, the electrode material and thickness were examined to account for their substantial impact on device performance at reduced piezoelectric film thicknesses. Aluminum (Al) was selected for its excellent conductivity and low mass-loading. The electrode thickness was analyzed using COMSOL finite element analysis (FEA). Strong mechanical damping was assigned in the electrodes, where a low of 5 was chosen as a nominal value for the metal to highlight its effects on the overall performance. On the other hand, LN was modelled as a low-loss structure. With this approach, we gained insights into relative trends in as the ratio of metal to LN decreased.
Simulations conducted with a 50% electrode duty cycle reveal that reducing the electrode thickness yields a higher impedance ratio and consequently leads to a higher overall [Fig. 4 (a)(b)]. In this work, we selected an electrode thickness of 50 nm Al. While variations in the duty cycle can further influence impedance characteristics, a 50% duty cycle was maintained while considering its effect on the feature size. Additionally, a minimized device aperture of was selected to mitigate the routing resistance from the electrodes. These design choices enhance overall while maintaining a balance between performance at and .
The top view of the resulting design can be seen in Fig. 5, where key design parameters are labeled. The FEA simulated admittance with a 4 m shows mmWave operation and a high extracted of 6.4% at 47.7 GHz (Fig. 6), which is obtained from an equivalent Butterworth-Van-Dyke (BVD) model with two motional branches.
Fabrication and Experimental Results
The fabrication steps are outlined in Fig. 7 (a). The process follows closely with [31]. A scanning electron microscope image is in Fig. 7 (b). The XBAR resonators were measured using a Keysight vector network analyzer (VNA) at an input power of −15 dBm, and calibrated with a GGB CS-5 substrate. Due to the miniature footprint, the devices exhibited a parasitic electromagnetic resonance well above their acoustic response. Accordingly, we report raw data from the calibrated VNA.
The measured admittance magnitude and phase of a representative device are in Figure 8. The device exhibited at 39.83 GHz with a 3-dB of 97 and at 40.67 GHz with a 3-dB of 342. The measurement is further fit with an mBVD model with series routing resistance () and inductance (). [25]. The mBVD fit shows good agreement with the measured data. The A3 XBAR showed a motional inductance () of 19 nH, a motional capacitance () of 0.8 fF, a motional resistance () of 18.6 Ω, series resistance () of 18.42 Ω, a series inductance () of 0.1 nH, mBVD () of 260, and an extracted of 4%. and are slightly different, due to the limited motional branches in the fit. In comparison with the SoA (Figure 1), the high achieved 3-dB , alongside the extracted , leads to a high FoM of 13.8 at 40 GHz, highlighting the effectiveness of the proposed design. To validate the extracted of A3 against potential dynamic range limitations of the VNA, we plot the open response (with probe unlanded), which confirms sufficient dynamic range margin for A3, and supports the reliability of the extraction. The higher measured of A1 over A3 suggests increased resistive, EM, and mechanical losses at higher frequencies. To clarify, Figure 4 only considers the mechanical damping, indicating that mechanical loss in the electrodes affects disproportionately compared with , which is a previously underappreciated loss mechanism at series resonance. Still,

Figure 4. FEA simulations of resonators with varying electrode thicknesses at (a) series and (b) parallel resonances, indicating the effects of thin electrodes.

Figure 8. Measured wide-band response alongside an mBVD fit accounting for routing resistance and inductance. Note that in high for A1 mode is not realistic, as the impedance is too high for VNA testing. On the other hand, the for A3 is within the VNA dynamic range.

Figure 7. (a) The fabrication flow for this work. (b) A Scanning electron microscope (SEM) image of the fabricated device.
isolating a single dominant loss mechanism at mmWave remains difficult due to the combined influence of resistive, EM, and surface effects [22], [30], [32]. Addressing this interplay will be crucial to future high- resonator designs. Additionally, it is notable that the extracted here is significantly smaller than the perceived directly calculated from , due to the EM self-resonance [25]. Accordingly, we report only the extracted values from fitting to avoid unnecessary overclaims. Measurements with the same device layout across 30-50 GHz consistently show high performance (Figure 9). Across this frequency span, ranged from 33 to 102, and from 210 to 464. decreased linearly with frequency, attributed to the increasing acoustic and dielectric loss at higher frequencies. Conversely, was predominantly limited by the resistive and mechanical loss arising from the thin Al electrodes. Hence, based on the extracted and 3-dB , we obtain FoMs ranging from 8 to 22.3 within 30-50 GHz (Figure 1).

Figure 9. (a) Measured 3-dB and , indicating high performance from 35-50 GHz. (b) Extracted from mBVD.
Table 1 benchmarks the proposed mmWave resonators against prior single-layer LN XBARs [16, 17, 18, 19], [33]. Our devices significantly exceed state-of-the-art XBARs in terms of the product, achieving up to while maintaining a compact normalized footprint to achieve 50 Ω at . These improvements primarily result from key innovations: a compact resonator design thanks to a reduced combined with a minimized transverse aperture; and thin (50 nm) electrodes to lower mechanical loss. In tandem, these design enhancements allow resonators to achieve a high at mmWave frequencies. Future mmWave LN XBAR iterations will focus on systematically optimizing key device parameters, including electrode thickness, metal composition and quality, the electrode duty cycle, and the anchor geometry. Additionally, we will explore post-processing techniques such as annealing for further potential reduction in loss. Future device scaling for practical filter applications will focus on shorter aperture lengths and optimized electrode arrays, accompanied by thicker buslines to minimize routing resistance and improve .
| Ref. | Lateral Wave. (μm) | Electrode Thick. (nm) | Normalized Footprint (μm²) | (GHz) | 3-dB | 3-dB | ||
| [33] | 4 | 50 | 3162 | 4 | 28% | 400 | 100 | 1.6E12 |
| [16] | 8 | 350 | 1377 | 57 | 5.5% | 60 | 60 | 3.4E12 |
| [17] | 20 | 50 | 1920 | 12.9 | 3.8% | 282 | 214 | 3.7E12 |
| [18] | 13 | 60 | 2028 | 12.8 | 3.7% | 224 | 224 | 2.9E12 |
| [19] | 34 | 310 | 36000 | 33.7 | 3.7% | 67 | 67 | 2.3E12 |
| This Work | 4 | 50 | 857 | 29.8 | 4.8% | 102 | 464 | 1.4E13 |
| This Work | 3.2 | 50 | 626 | 36.8 | 4.8% | 33 | 360 | 1.3E13 |
| This Work | 4 | 50 | 684 | 40 | 4% | 96 | 342 | 1.4E13 |
| This Work | 3.2 | 50 | 564 | 48.8 | 3.8% | 58 | 210 | 1.0E13 |
Table 1. Comparison to Prior Single-Layer LN XBARs. *Impedance at series resonance normalized to 50 Ω. ** is the extracted value from fitting. *** is selected as the maximum between reported and .
Conclusion
In this work, we have demonstrated miniature, high-performance A3 XBARs from 30 to 50 GHz. The compact structure outperforms the SoA, highlighting the frequency scalability. This alternative approach tailors the device design to achieve a high at mmWave. With continued optimization, the A3 mode XBAR platform is poised to introduce low-loss acoustic resonators at mmWave frequencies.
References
- [1]A. Hagelauer et al., “From Microwave Acoustic Filters to Millimeter-Wave Operation and New Applications,” IEEE J. Microw., vol. 3, no. 1, pp. 484–508, Jan. 2023, doi: 10.1109/JMW.2022.3226415.DOI
- [2]U. Gustavsson et al., “Implementation Challenges and Opportunities in Beyond-5G and 6G Communication,” IEEE J. Microw., vol. 1, no. 1, pp. 86–100, Jan. 2021, doi: 10.1109/JMW.2020.3034648.DOI
- [3]S. Mahon, “The 5G Effect on RF Filter Technologies,” IEEE Trans. Semicond. Manuf., vol. 30, no. 4, pp. 494–499, Nov. 2017, doi: 10.1109/TSM.2017.2757879.DOI
- [4]R. Lu and S. Gong, “RF acoustic microsystems based on suspended lithium niobate thin films: advances and outlook,” J. Micromechanics Microengineering, vol. 31, no. 11, p. 114001, Nov. 2021, doi: 10.1088/1361-6439/ac288f.DOI
- [5]E. Guerrero, A. Gimenez, W. Heeren, and M. Al-Joumayly, “BAW Filter Trends for 5G and Beyond,” in 2024 IEEE International Microwave Filter Workshop (IMFW), Feb. 2024, pp. 75–77. doi: 10.1109/IMFW59690.2024.10477135.DOI
- [6]Y. Yang, R. Lu, and S. Gong, “Scaling Acoustic Filters Towards 5G,” in 2018 IEEE International Electron Devices Meeting (IEDM), Dec. 2018, p. 39.6.1-39.6.4. doi: 10.1109/IEDM.2018.8614699.DOI
- [7]J. P. Smith and V. Hinson-Smith, “The new era of SAW devices,” Anal. Chem., vol. 78, no. 11, pp. 3505–3507, Jun. 2006, doi: 10.1021/ac0694122.DOI
- [8]V. V. Felmetsger, P. N. Laptev, and R. J. Graham, “Deposition of ultrathin AlN films for high frequency electroacoustic devices,” J. Vac. Sci. Technol. A, vol. 29, no. 2, p. 021014, Feb. 2011, doi: 10.1116/1.3554718.DOI
- [9]E. Iborra, J. Olivares, M. Clement, L. Vergara, A. Sanz-Hervás, and J. Sangrador, “Piezoelectric properties and residual stress of sputtered AlN thin films for MEMS applications,” Sens. Actuators Phys., vol. 115, no. 2, pp. 501–507, Sep. 2004, doi: 10.1016/j.sna.2004.03.053.DOI
- [10]J. Campbell et al., “21.4 GHz Surface Acoustic Wave Resonator with 11,400 m/s Phase Velocity in Thin-Film Lithium Niobate on Silicon Carbide,” in 2024 IEEE Ultrasonics, Ferroelectrics, and Frequency Control Joint Symposium (UFFC-JS), Sep. 2024, pp. 1–4. doi: 10.1109/UFFC-JS60046.2024.10793880.
- [11]J. Campbell et al., “52-GHz Surface Acoustic Wave Resonators in Thin-Film Lithium Niobate on Silicon Carbide,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 72, no. 2, pp. 275–282, Feb. 2025, doi: 10.1109/TUFFC.2024.3522042.DOI
- [12]I. Anderson et al., “Low-Loss Higher-Order Cross-Sectional Lamé Mode SAW Devices in 10-20 GHz Range,” in 2024 IEEE Ultrasonics, Ferroelectrics, and Frequency Control Joint Symposium (UFFC-JS), Sep. 2024, pp. 1–4. doi: 10.1109/UFFC-JS60046.2024.10794098.
- [13]G. Giribaldi, L. Colombo, and M. Rinaldi, “6–20 GHz 30% ScAlN Lateral Field-Excited Cross-Sectional Lamé Mode Resonators for Future Mobile RF Front Ends,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 70, no. 10, pp. 1201–1212, Oct. 2023, doi: 10.1109/TUFFC.2023.3312913.DOI
- [14]G. Giribaldi, L. Colombo, P. Simeoni, and M. Rinaldi, “Compact and wideband nanoacoustic pass-band filters for future 5G and 6G cellular radios,” Nat. Commun., vol. 15, no. 1, p. 304, Jan. 2024, doi: 10.1038/s41467-023-44038-9.DOI
- [15]Z. Schaffer, P. Simeoni, and G. Piazza, “33 GHz Overmoded Bulk Acoustic Resonator,” IEEE Microw. Wirel. Compon. Lett., vol. 32, no. 6, pp. 656–659, Jun. 2022, doi: 10.1109/LMWC.2022.3166682.
- [16]J. Kramer et al., “57 GHz Acoustic Resonator with k2 of 7.3 % and Q of 56 in Thin-Film Lithium Niobate,” in 2022 International Electron Devices Meeting (IEDM), Dec. 2022, p. 16.4.1-16.4.4. doi: 10.1109/IEDM45625.2022.10019391.DOI
- [17]Y. Yang, R. Lu, L. Gao, and S. Gong, “10–60-GHz Electromechanical Resonators Using Thin-Film Lithium Niobate,” IEEE Trans. Microw. Theory Techn., vol. 68, no. 12, pp. 5211–5220, Dec. 2020, doi: 10.1109/TMTT.2020.30207694.DOI
- [18]Y. Yang, R. Lu, T. Manzaneque, and S. Gong, “Toward Ka Band Acoustics: Lithium Niobate Asymmetrical Mode Piezoelectric MEMS Resonators,” in 2018 IEEE International Frequency Control Symposium (IFCS), May 2018, pp. 1–5. doi: 10.1109/IFCS.2018.8597475.DOI
- [19]S. Link, R. Lu, Y. Yang, A. E. Hassanien, and S. Gong, “An A1 Mode Resonator at 12 GHz using 160nm Lithium Niobate Suspended Thin Film,” in 2021 IEEE International Ultrasonics Symposium (IUS), Sep. 2021, pp. 1–4. doi: 10.1109/IUS52206.2021.9593341.DOI
- [20]M. Park, Z. Hao, D. G. Kim, A. Clark, R. Dargis, and A. Ansari, “A 10 GHz Single-Crystalline Scandium-Doped Aluminum Nitride Lamb-Wave Resonator,” in 2019 20th International Conference on Solid-State Sensors, Actuators and Microsystems & Eurosensors XXXIII (TRANSDUCERS & EUROSENSORS XXXIII), Jun. 2019, pp. 450–453. doi: 10.1109/TRANSDUCERS.2019.8888374.DOI
- [21]D. Mo, S. Dabas, S. Rassay, and R. Tabrizian, “Complementary-Switchable Dual-Mode SHF Scandium Aluminum Nitride BAW Resonator,” IEEE Trans. Electron Devices, vol. 69, no. 8, pp. 4624–4631, Aug. 2022, doi: 10.1109/TED.2022.3183963.DOI
- [22]J. Zheng, Z. Ren, J. Xu, X. Liu, F. Qian, and Y. Yang, “Periodically Poled Piezoelectric Single-Layered and Multilayered Lithium Niobate for Thickened High-Order Lamb Wave Acoustic Devices,” IEEE Trans. Microw. Theory Techn., pp. 1–11, 2025, doi: 10.1109/TMTT.2025.3567329.DOI
- [23]R. Lu, Y. Yang, S. Link, and S. Gong, “Enabling Higher Order Lamb Wave Acoustic Devices With Complementarily Oriented Piezoelectric Thin Films,” J. Microelectromechanical Syst., vol. 29, no. 5, pp. 1332–1346, Oct. 2020, doi: 10.1109/JMEMS.2020.3007590.DOI
- [24]J. Kramer et al., “Thin-Film Lithium Niobate Acoustic Resonator with High Q of 237 and k2 of 5.1% at 50.74 GHz,” in 2023 Joint Conference of the European Frequency and Time Forum and IEEE International Frequency Control Symposium (EFTF/IFCS), May 2023, pp. 1–4. doi: 10.1109/EFTF/IFCS57587.2023.10272149.DOI
- [25]J. Kramer et al., “Acoustic resonators above 100 GHz,” Appl. Phys. Lett., vol. 127, no. 1, Jul. 2025, doi: 10.1063/5.0275691.
- [26]S. Nam, W. Peng, P. Wang, D. Wang, Z. Mi, and A. Mortazawi, “A mm-Wave Trilayer AlN/ScAlN/AlN Higher Order Mode FBAR,” IEEE Microw. Wirel. Technol. Lett., vol. 33, no. 6, pp. 803–806, Jun. 2023, doi: 10.1109/LMWT.2023.3271865.DOI
- [27]Izhar et al., “A High Quality Factor, 19-GHz Periodically Poled AlScN BAW Resonator Fabricated in a Commercial XBAW Process,” IEEE Trans. Electron Devices, vol. 71, no. 9, pp. 5630–5637, Sep. 2024, doi: 10.1109/TED.2024.3435175.DOI
- [28]Izhar et al., “Periodically poled aluminum scandium nitride bulk acoustic wave resonators and filters for communications in the 6G era,” Microsyst. Nanoeng., vol. 11, no. 1, p. 19, Jan. 2025, doi: 10.1038/s41378-024-00857-4.DOI
- [29]R. Lu, Y. Yang, M.-H. Li, M. Breen, and S. Gong, “5-GHz Antisymmetric Mode Acoustic Delay Lines in Lithium Niobate Thin Film,” IEEE Trans. Microw. Theory Tech., vol. 68, no. 2, pp. 573–589, Feb. 2020, doi: 10.1109/TMTT.2019.2949808.DOI
- [30]M. Weinberg, R. Candler, S. Chandorkar, J. Varsanik, T. Kenny, and A. Duwel, “Energy loss in MEMS resonators and the impact on inertial and RF devices,” in TRANSDUCERS 2009 - 2009 International Solid-State Sensors, Actuators and Microsystems Conference, Jun. 2009, pp. 688–695. doi: 10.1109/SENSOR.2009.5285418.
- [31]V. Chulukhadze et al., “Frequency Scaling Millimeter Wave Acoustic Resonators using Ion Beam Trimmed Lithium Niobate,” in 2023 Joint Conference of the European Frequency and Time Forum and IEEE International Frequency Control Symposium (EFTF/IFCS), May 2023, pp. 1–4. doi: 10.1109/EFTF/IFCS57587.2023.10272038.DOI
- [32]L. Colombo et al., “Characterization of Acoustic Losses in Interdigitated VHF to mmWave Piezoelectric M/NEMS Resonators,” in 2024 IEEE Ultrasonics, Ferroelectrics, and Frequency Control Joint Symposium (UFFC-JS), Sep. 2024, pp. 1–4. doi: 10.1109/UFFC-JS60046.2024.10794128.DOI
- [33]Y. Yang, R. Lu, L. Gao, and S. Gong, “4.5 GHz Lithium Niobate MEMS Filters with 10% Fractional Bandwidth for 5G Front-ends,” J. Microelectromechanical Syst., vol. 28, no. 4, pp. 575–577, Aug. 2019, doi: 10.1109/JMEMS.2019.2922935.DOI