betaFeedback

WorksEngineering

Thickness-Field-Excited Bimorph Piezoelectric Micromachined Ultrasonic Transducer in Submicrometer Thin-Film Lithium Niobate

Abstract

This work presents the first demonstration of a submicrometer-thick thickness-field-excited (TFE) lithium niobate (LN) flexural-mode piezoelectric micromachined ultrasonic transducer (PMUT). A 670-nm-thick bilayer periodically poled piezoelectric film (P3F) 128Y LN is used to demonstrate a suspended membrane with an area of 0.25 mm2^{2}, achieving a low resonance frequency of 33.4 kHz for airborne transduction. Its electrical measurements reveal an extracted effective electromechanical coupling efficiency (keff2k_{\mathrm{eff}}^{2}) of 3% and a moderate static capacitance C0C_0 of 17 pF. Meanwhile, mechanical measurements using a laser Doppler vibrometer (LDV) yield a high quasi-static peak out-of-plane displacement (ξstaticξ_{\mathrm{static}}) of 145 nm/V and a high corresponding open-circuit sensitivity (SOCS_{\mathrm{OC}}) of 1.4 mV/Pa in the audio band. The fabricated device is compared with state-of-the-art (SoA) airborne piezoelectric transducers with sub-100-kHz resonance frequencies, underscoring the potential of the TFE P3F LN transducer platform by simultaneously demonstrating strong actuation capability while featuring high sensitivity.

Frequency response and suspended lithium niobate transducer

Figure 12.

Introduction

Piezoelectric micromachined ultrasonic transducers (PMUTs) are frequently employed as passive components in airborne applications that require a combination of high performance, compact footprint, and low-power operation, such as range-finding and microphones [1, 2, 3, 4, 5]. Incumbent piezoelectric material platforms, e.g., lead zirconate titanate (PZT), potassium sodium niobate (KNN), and scandium aluminum nitride (ScAlN), offer distinct trade-offs in actuation and sensing, but remain limited in bidirectional performance Table I [6, 7, 8].

Manuscript received [date], revised [date], accepted [date]. This work was supported in part by the Defense Advanced Research Projects Agency (DARPA) High Operational Temperature Sensors (HOTS) project under Award HR00112420334 and in part by the National Institute on Deafness and Other Communication Disorders (NIDCD) of the National Institutes of Health (NIH) under Award R41DC022811.

V. Chulukhadze, J. Huang, Z. Liu, W. Leskowitz, T.-H. Hsu, Y. Ma, N. A. Hall, and R. Lu are with the Chandra Family Department of Electrical and Computer Engineering, The University of Texas at Austin, Austin, TX 78712 USA (e-mail: [email protected]).

U. Rawat, Y. Yu, and B. Bahr are with Kilby Labs, Texas Instruments, Dallas, TX 75243 USA.

In an effort to accurately evaluate material-level performance for flexural-mode operation, the effective electromechanical tensor components under the stress-charge piezoelectric constitutive equation system can be utilized. These parameters reflect the mixed boundary conditions of the PMUT, being clamped in-plane while free to deform out-of-plane [9, 10]. The resulting effective stress-charge

Highlights

  • This work presents the first demonstration of a thickness-field-excited (TFE) lithium niobate (LN) clamped-plate flexural-mode PMUT with a submicrometer-thick active layer.

  • Measurements reveal a high extracted effective electromechanical coupling efficiency (keff2k_{\mathrm{eff}}^{2}) of 3%, alongside high actuation and sensitivity metrics in the audio band, such as a ξstatic\xi_{\mathrm{static}} of 145 nm/V and flat-band SOCS_{\mathrm{OC}} of 1.4 mV/Pa.

  • The resulting TFE transducer exhibits high performance and integrability with a compact form factor, indicating promise as an airborne piezoelectric transducer platform.

Index Terms— Bimorph, lithium niobate, periodically poled piezoelectric film (P3F), piezoelectric micromachined ultrasonic transducers (PMUTs), thickness-field excitation

MaterialExcitationeij,effe_{ij,\mathrm{eff}} (C/m2^{2})eij,eff/(ϵii,efftan⁡δ)1/2e_{ij,\mathrm{eff}}/(\epsilon_{ii,\mathrm{eff}}\tan\delta)^{1/2} (10610^{6}(J/m3^{3})1/2^{1/2})Capacitance DensityInterface Electrode
Epitaxial PZTTFE24.05.2Very HighY
Sputtered KNNTFE10.01.7Very HighY
Sputtered AlNTFE1.03.0ModerateY
Sputtered Sc36_{36}Al64_{64}NTFE2.43.8ModerateY
X-cut LNLFE5.313.4LowN
128Y LNTFE3.69.0ModerateN

Table 1. Piezoelectric transducer FoM comparison between incumbent piezoelectric platforms, highlighting the high FoM paired with a moderate C0C_{0} density in 128Y LN.

The subscripts are 31 for TFE (3, 1) excitation and 11 for LFE (1, 1) excitation. Magnitudes are listed; the sign of each coefficient depends on the material and orientation. Values from [8], [13, 14, 15, 16].

piezoelectric constant (eij,effe_{ij,\mathrm{eff}}, in C/m2^{2}, where ii is the electric field direction and jj is the stress component) acts as a bidirectional figure of merit (FoM) when combined with the effective electrical permittivity (ϵii,eff\epsilon_{ii,\mathrm{eff}}) and the loss tangent (tan⁡δ\tan\delta) [11], [12]. Accordingly, the comparison in Table I exhibits concurrent limitations in incumbent piezoelectric material platforms: PZT features a large e31,effe_{31,\mathrm{eff}}, exhibiting potential for efficient actuation, but remains limited as a sensor due to its high ϵii,eff\epsilon_{ii,\mathrm{eff}} and tan⁡δ\tan\delta. On the other hand, ScAlN is well-positioned as a sensor, but features limited actuation capability, highlighting the need for an alternative PMUT platform with high bidirectional performance metrics.

State-of-the-art (SoA) PMUTs operate in the fundamental flexural mode, characterized by a contrasting dominant in-plane stress profile through the piezoelectric film thickness, requiring a multi-material stack for efficient operation. Figure 1 shows how the in-plane stress evolves from tensile to compressive along the zz-axis of the piezoelectric solid during flexural deflection, leading to internal charge cancellation in a single-layer (unimorph) piezoelectric solid. To compensate, a passive layer can be utilized to displace the neutral stress axis in an effort to reduce destructive interference. However, a portion of the material’s available performance is sacrificed as a result. As an alternative, a bilayer piezoelectric (bimorph) structure can be utilized, which has conventionally relied on an intermediate electrode layer in thickness-field-excited (TFE) devices to flip the electric field polarization between two layers of piezoelectric material, as shown in Figure 2(a) [17]. These structures harness the full potential of a given platform, at the cost of fabrication complexity and enhanced thermal and mechanical stress due to the presence of a large-area multi-material interface.

A cross-sectional view of the dominant in-plane stress profile characteristic of the flexural mode

Figure 1. A cross-sectional view of the dominant in-plane stress profile characteristic of the flexural mode, highlighting the vertical charge cancellation due to the contrasting stress profile.

Cross-sectional views of a conventional bimorph device and a TFE P3F LN piezoelectric transducer

Figure 2. Cross-sectional views of (a) a conventional bimorph transducer and (b) the proposed TFE device using P3F LN.

Intrinsic limitations to bidirectional transduction and difficulties in high-quality bimorph stack fabrication have motivated research into alternative PMUT material platforms. Single-crystal lithium niobate (LN) has shown significant potential via its strong piezoelectric response and bimorph film availability [18, 19, 20]. More specifically, recent advances in LN thin-film transfer technologies have enabled bimorph periodically poled piezoelectric film (P3F) LN with single-crystal quality at film thicknesses ranging from as little as 100 nm to bulk values [21, 22, 23, 24]. Here, constructive interference between two LN layers is achieved by flipping the relevant electromechanical coupling tensor components with either in-plane or out-of-plane rotation [25]. Figure 2(b) shows an example cross-section of a TFE P3F bimorph, optimized for robustness by removing the need for intermediate electrode layers. These developments have led to the demonstration of transferred P3F LN-based lateral-field-excited (LFE) transducers as a promising alternative to SoA PMUT platforms, showcasing high performance and resilience to temperatures as high as 800 °C [18], [19], [26]. However, the LFE mechanism showed a low static capacitance (C0C_{0}) density and employed a 20-µm-thick active layer for robustness, limiting compliance and footprint for airborne applications. A TFE P3F prototype in [21] showed how the change in the electrical excitation scheme can overcome these limitations while preserving high performance, but its thickness severely limited the compliance.

Materiali,ji,jeij,effe_{ij,\mathrm{eff}} (C/m²)εr,ii,eff\varepsilon_{r,ii,\mathrm{eff}}eij,eff/εii,effe_{ij,\mathrm{eff}}/\varepsilon_{ii,\mathrm{eff}} (GV/m)
Y-cut LN3, 13.146.97.5
36Y LN3, 13.550.77.8
128Y LN3, 13.637.610.8

Table 2. THIN-FILM LN MATERIAL PARAMETERS. Magnitudes are listed; the sign of each coefficient depends on the orientation.

This work aims to realize the potential shown by the initial bulk TFE P3F LN transducer prototype, complementing the LFE P3F LN platform by providing the first demonstration of a high-performance TFE P3F LN PMUT with a submicrometer-thick active layer, featuring a compact footprint and high performance fit for airborne applications.

Design and Simulation

Devices were designed with a submicrometer-thick (670 nm) active layer to demonstrate TFE P3F transducers with high compliance and a compact footprint, optimal for sub-100-kHz airborne applications. The process began with LN orientation selection for TFE flexural-mode operation. Among commercially available LN material cuts with high relevant electromechanical coupling tensor components (e31,effe_{31,\mathrm{eff}} and e32,effe_{32,\mathrm{eff}}), 36Y and 128Y LN stand out as the primary candidates, as shown in Table II. Here, εii,eff\varepsilon_{ii,\mathrm{eff}} and the ratio between eij,effe_{ij,\mathrm{eff}} and εii,eff\varepsilon_{ii,\mathrm{eff}} are used as secondary performance metrics, where εii,eff\varepsilon_{ii,\mathrm{eff}} describes the C0C_0-density and the ratio describes the platform’s open-circuit sensitivity, acting as another receiving performance metric. The comparison in Table II shows that while 36Y LN and 128Y LN exhibit electromechanical coupling tensor components of similar magnitude, the lower εii,eff\varepsilon_{ii,\mathrm{eff}} of 128Y LN yields a higher receiving FoM, leading to the adoption of 128Y LN for the TFE P3F LN transducer.

Prior investigations into LN-based transducers showed the importance of in-plane anisotropy for flexural-mode operation [18]. Hence, in an effort to study electrode placement, we plot the interaction between the effective in-plane electromechanical coupling tensor components (e31,effe_{31,\mathrm{eff}} and e32,effe_{32,\mathrm{eff}}), as shown in Figure 3. The analysis shows that the transverse e32,effe_{32,\mathrm{eff}} coupling coefficient is parasitic when e31,effe_{31,\mathrm{eff}} is at its peak value. Meanwhile, the inset mode shapes depict the dominant in-plane stress components during a flexural deflection, revealing significant transverse components in an isotropic, circular geometry. Hence, the in-plane anisotropy in 128Y LN necessitates a customized membrane shape designed and oriented to localize in-plane stress alongside the maximum e31,effe_{31,\mathrm{eff}} while eliminating the transverse stress antinodes. Prior studies on LFE P3F LN transducers encountered a similar symmetry in X-cut LN. They effectively utilized an elliptical membrane with the minor axis aligned with maximum e11,effe_{11,\mathrm{eff}} to minimize destructive interference [18]. The same approach can be used here by aligning the minor axis of the ellipse to 90∘90^\circ from the crystal X axis. Consequently, the top and the cross-sectional views of the thin-film TFE P3F LN transducer are shown in Figure 4(a) and (b), respectively, with the crystal axes and the key design parameters labeled. The design process closely followed that of the LFE counterpart in [18], using parametric sweeps while tracking the effective electromechanical coupling efficiency (keff2k_{\mathrm{eff}}^2) and C0C_0 defined as:

Relevant effective electromechanical coupling tensor components plotted against in-plane rotation

Figure 3. Relevant effective ee electromechanical coupling tensor components plotted against rotation from the crystal X axis (α\alpha), alongside top views of in-plane stress in a circular and an elliptical geometry. The results highlight how a circular geometry simultaneously excites an e31,effe_{31,\mathrm{eff}} of −3.6-3.6 C/m² and e32,effe_{32,\mathrm{eff}} of +1.05+1.05 C/m², while the elliptical geometry primarily harnesses e31,effe_{31,\mathrm{eff}} alone, alleviating destructive interference.

Top and cross-sectional views of the thin-film TFE P3F LN transducer

Figure 4. The (a) top and (b) cross-sectional views of the thin-film TFE P3F LN transducer, highlighting the elliptical membrane shape.

keff2=Φ2CadC0,(1)k_{\mathrm{eff}}^2 = \frac{\Phi^2 C_{\mathrm{ad}}}{C_0}, \tag*{(1)}
C0=Im⁡(Y)ω,(2)C_0 = \frac{\operatorname{Im}(Y)}{\omega}, \tag*{(2)}

where Φ\Phi is the turns ratio of the pressure-charge transformer, CadC_{\mathrm{ad}} is the acoustic compliance of the membrane, YY is the electrical admittance, and ω\omega is the angular frequency. Both Φ\Phi and CadC_{\mathrm{ad}} are extracted under a unit pressure boundary load with the electrodes short-circuited: Φ\Phi can be obtained as the short-circuit charge normalized by the volumetric displacement, and CadC_{\mathrm{ad}} is obtained by integrating the out-of-plane displacement over the membrane surface due to a unit pressure boundary load.

The proposed structure relies on a floating bottom electrode while utilizing the top electrodes for signal and ground, thereby establishing thickness-field excitation under electrostatic equilibrium, alleviating fabrication-related challenges associated with a patterned bottom electrode [27, 28] [29, 30]. This design choice limits the achievable C0C_0 of the device by placing the inner and outer electrodes in series, which is an acceptable design trade-off due to the moderate εr,ii,eff\varepsilon_{r,\mathrm{ii,eff}} in LN.

The obtained key variables, summarized in Table III, aim to maximize performance for a device with a resonant frequency following the audio band (20 Hz–20 kHz) at 30 kHz, indicating an exceedingly compact footprint of 0.25 mm2^2 for sub-100-kHz airborne operation.

tLNt_{\mathrm{LN}}WWLLD1D_1D2D_2D3D_3LeL_e
0.67 μm\mu\mathrm{m}400 μm\mu\mathrm{m}800 μm\mu\mathrm{m}37.5 μm\mu\mathrm{m}120 μm\mu\mathrm{m}690 μm\mu\mathrm{m}115 μm\mu\mathrm{m}

Table III. Chosen key design parameters for the thin-film thickness-field-excited LN transducer.

The device was simulated in vacuum using COMSOL finite-element analysis (FEA) to extract the electromechanical performance, decoupling significant acoustic effects that may arise with a compliant piezoelectric membrane. The resulting impedance and displacement plots can be seen in Figure 5(a) and (Figure 5b), with an assumed quality factor (QQ) of 50.

FEA-simulated electrical impedance and phase, and displacement sensitivity

Figure 5. FEA-simulated (a) electrical impedance and phase, and (b) displacement sensitivity for a nominal quality factor of 50. Extracted performance parameters are labeled.

The device achieves a high keff2k_{\mathrm{eff}}^2 of 7% alongside a C0C_0 of 17.2 pF and a high peak out-of-plane displacement (ξpeak\xi_{\mathrm{peak}}) of 19.5 μ\mum/V at 30.4 kHz. In order to decouple resonant effects for flat-band operation, the quasi-static peak out-of-plane displacement (ξstatic\xi_{\mathrm{static}}) can be evaluated as close to DC as possible, yielding 330 nm/V at 1 kHz, indicating high film compliance in combination with a high keff2k_{\mathrm{eff}}^2. The simulated device demonstrates an excellent keff2k_{\mathrm{eff}}^2 alongside a high ξstatic\xi_{\mathrm{static}} in a compact footprint of 0.25 mm2^2 while achieving a moderate C0C_0 of 17.2 pF, in spite of the floating bottom electrode. Hence, the thin-film TFE P3F LN platform features high performance on par with its LFE counterpart and shows good integration capability via its C0C_0-density.

Fabrication and Measurement

The transducers were fabricated using a 670-nm-thick 128Y P3F LN wafer, provided by NGK Corporation. The fabrication began by evaporating 50 nm of gold (Au) top electrodes with a 10-nm-thick chromium (Cr) adhesion layer, introducing minimal mass loading while featuring high conductivity. Next, silicon dioxide (SiO2_2) was deposited on the top and back sides of the wafer using low-temperature plasma-enhanced chemical vapor deposition (PECVD) at 100 °C to act as a protective layer for the active material as well as the substrate during the deep reactive ion etching (DRIE) process. The thicknesses were 200 nm and 1 μ\mum for the top and the bottom layers, respectively. The etched trench and its alignment with the top electrodes were evaluated optically due to LN’s transparency. Once a satisfactory result was reached, 50 nm Au alongside a 10-nm Cr adhesion layer was deposited via electron-beam evaporation into etched Si trenches to form the bottom electrode. A buffered oxide etch (BOE) was used to clean the remaining SiO2_2 residue. The fabrication flow and an optical image of the device can be seen in Figure 6(a) and (Figure 6b).

Fabrication flow and optical image of the thin-film TFE transducer

Figure 6. (a) The fabrication flow of the thin-film TFE transducer. (b) An optical image of the resulting device.

The fabricated device was measured with a Zurich MFIA impedance analyzer and a Polytec laser Doppler vibrometer (LDV), with the results shown in Figure 7 and Figure 8, respectively.

Measured impedance and phase of the TFE P3F LN transducer with BVD equivalent circuit model fit

Figure 7. Measured (a) impedance and (b) phase of the TFE P3F LN transducer on a grounded elevated plate, alongside its BVD equivalent circuit model fit (inset).

LDV-measured out-of-plane displacement and phase of the transducer

Figure 8. The LDV-measured out-of-plane displacement and phase of the transducer on an elevated, grounded metallic plate, showcasing a high transmit efficiency with a ξpeak\xi_{\mathrm{peak}} of 8.5 µm/V and ξstatic\xi_{\mathrm{static}} of 145 nm/V.

The measurements were taken in air on an elevated, grounded metallic pedestal with a cutout section under the device in an effort to minimize the influence of the back cavity on the transducer. An equivalent Butterworth–Van Dyke (BVD) circuit model was utilized to extract device performance, shown as an inset image in Figure 7(b), where the motional components are parametrized such that the extracted keff2k_{\mathrm{eff}}^{2} follows a definition identical to (1). The resulting data revealed a well-defined resonance at 33.4 kHz with an extracted keff2k_{\mathrm{eff}}^{2} of 3% and a C0C_{0} of 17 pF, in line with the simulated value from Figure 5. The LDV measurement in Figure 8 excites the device with a 50-mV periodic chirp, confirming fundamental flexural-mode operation and yielding a high ξpeak\xi_{\mathrm{peak}} of 8.5 µm/V at resonance. The audio band can be analyzed for ξstatic\xi_{\mathrm{static}}; however, the response near DC contains significant artifacts from the chirped signal, and ξstatic\xi_{\mathrm{static}} is therefore obtained at 5 kHz, which is still in the flat band. The measurement yields a high ξstatic\xi_{\mathrm{static}} of 145 nm/V, which will be used to evaluate device performance within the audio band.

In tandem, the measured results indicate that the device obtains an excellent transmission efficiency and exhibits a low resonance frequency ideal for ultrasonic and audio-band applications while featuring an exceedingly compact footprint of 0.25 mm2^{2}. Moreover, the suspended film consists of only a 670-nm-thick active layer, demonstrating the ability of transferred thin-film P3F LN to realize large-area membranes. Regardless, the presented data warrants further discussion, as important metrics such as receiving sensitivity, acoustic effects, and simulation discrepancies remain to be investigated.

Discussion

Sensitivity

Using the LDV measurement in Figure 8 and the BVD model parameters from Figure 7, both the charge and voltage sensitivity of the transducer can be extrapolated. Charge sensitivity (SSCS_{\mathrm{SC}}, in C/Pa) can be directly obtained through the measured two-dimensional out-of-plane displacement mode shape as follows:

SSC=∬Aw dA,(3)S_{\mathrm{SC}} = \iint_{A} w\,\mathrm{d}A, \tag*{(3)}

where ww is the out-of-plane particle displacement per volt, measured as close to DC as possible (here, 5 kHz) to stay within the flat band well below resonance, and AA is the membrane area. Hence, (3) shows that the volumetric displacement per volt is equivalent to the stored charge per unit pressure, which stems from the inherent reciprocity within the piezoelectric constitutive equation system [31], [32]. Here, analysis at 5 kHz yields a moderate SSCS_{\mathrm{SC}} of 24.7 fC/Pa, which can be used as an effective comparison metric against other airborne transducers. The open-circuit sensitivity (SOCS_{\mathrm{OC}}, in V/Pa) can be obtained by dividing SSCS_{\mathrm{SC}} by the free capacitance of the system, found as Cm+C0C_{\mathrm{m}} + C_{0} from Figure 7(b) [32]. Accordingly, SOCS_{\mathrm{OC}} at 5 kHz is 1.4 mV/Pa, indicating a high voltage sensitivity within the flat band. Regardless, these metrics are strongly influenced by both acoustic and system-level effects, such as the back-cavity compliance (CbC_{\mathrm{b}}) and input capacitance of an amplifier [33]. While the system-level integration of the device is out of the scope of this work, the influence of the back cavity on both actuating and sensing performance will be discussed next.

Cavity Effects

In an effort to investigate back-cavity-induced acoustic effects, the elevated pedestal was exchanged for a flat metallic substrate. The transducer showed a significant change in performance and resonant behavior, with the comparison shown in Figure 9 and Figure 10. The measured impedance indicates a frequency upshift due to CbC_{\mathrm{b}}, while also showing a reduced keff2k_{\mathrm{eff}}^{2}, QQ, and lowered C0C_{0} when fit with the BVD model from Figure 7(b). On the other hand, the LDV measurement shows both a reduced ξpeak\xi_{\mathrm{peak}} and a reduced ξstatic\xi_{\mathrm{static}}, indicating the lower overall compliance of the system. These results can be used to extrapolate the receiving performance with the influence of the back cavity present; however, an important distinction must be made: the BVD fit of the electrical measurement in Figure 9 lumps CbC_b into CmC_m, although CbC_b does not contribute to the piezoelectric effect, and this approach can significantly underestimate CmC_m, and hence keff2k_{\mathrm{eff}}^2, when in series with CbC_b of comparable value. The LDV measurement can be analyzed to extract sensitivity metrics using (3). Accordingly, SSCS_{\mathrm{SC}} with the back-cavity effect present is found to be 12.5 fC/Pa at 5 kHz, while SOCS_{\mathrm{OC}} is 0.78 mV/Pa at the same frequency, presenting a twofold reduction from the original measurement. Hence, the back cavity plays a critical role in determining the transducer’s performance when the membrane compliance is sufficiently high. Future design iterations must aim to either minimize its effect by enhancing CbC_b compared to the membrane compliance, or closely match them [4] [33].

Impedance measurement of the device with the back-cavity effect included, compared with the measurements taken on an elevated pedestal

Figure 9. The impedance measurement of the device with the back-cavity effect included, compared with the measurements taken on an elevated pedestal, highlighting the frequency offset as well as changing damping and electromechanical performance.

LDV-measured out-of-plane displacement response of the transducer with the back-cavity effect present, compared with the original measurement

Figure 10. LDV-measured out-of-plane displacement response of the transducer with the back-cavity effect present, compared with the original measurement, indicating a reduced ξpeak\xi_{\mathrm{peak}} as well as a reduced ξstatic\xi_{\mathrm{static}}.

Materialfsf_s (kHz)Area (mm2^2)Thickness#^{\#} (µm)ξstatic\xi_{\mathrm{static}} (nm/V)C0C_0 (pF)SSCS_{\mathrm{SC}} (fC/Pa)SOCS_{\mathrm{OC}} (mV/Pa)
PZT [1]401.298.4430†430^{\dagger}1200960.08†0.08^{\dagger}
ScAlN [34]47.92.5410.63325682‡82^{\ddagger}0.32†0.32^{\dagger}
ScAlN [35]30.70.531.3–10738‡38^{\ddagger}0.35
This Work33.40.250.7914517251.4

Table 4. SURVEY OF SOA SUB-100-KHZ AIR-COUPLED PIEZOELECTRIC TRANSDUCERS, HIGHLIGHTING THE PERFORMANCE AND THE COMPACT FOOTPRINT OF THE PRESENTED DEVICE. #^{\#}Includes the electrode thickness. ∗^{*}In the flat band. †^{\dagger}Resonant value divided by the reported QQ. ‡SSC=SOCC^{\ddagger}S_{\mathrm{SC}}=S_{\mathrm{OC}}C with the measured capacitance.

Comparison to Prior Works

The extracted device performance is compared to SoA airborne piezoelectric transducers with a resonance frequency below 100 kHz in Table IV [1] [34, 35]. The comparison focuses on evaluating the balance between ξstatic\xi_{\mathrm{static}} and audio-band sensitivity metrics. The comparison highlights that the presented LN device is the most compact of the surveyed flexural-mode clamped membrane transducers with a sub-100-kHz resonance frequency, while also featuring the highest SOCS_{\mathrm{OC}} alongside a moderate ξstatic\xi_{\mathrm{static}} comparable to high-transmit-efficiency material platforms, such as PZT. This performance is achieved with the combination of the submicrometer-thick bimorph active layer and the achieved keff2k_{\mathrm{eff}}^2, facilitating a high volumetric displacement in the flat band, underscoring the potential of the TFE P3F LN transducer platform.

The single-crystal PZT transducer in [1] obtains a high flat-band SSCS_{\mathrm{SC}} of 96 fC/Pa owing to a C0C_0 of 1200 pF, which also limits its SOCS_{\mathrm{OC}} to 0.08 mV/Pa, while featuring a large ξstatic\xi_{\mathrm{static}} of 430 nm/V. On the other hand, a clamped 30% ScAlN membrane from [34] simultaneously obtains a high flat-band SSCS_{\mathrm{SC}} of 82 fC/Pa and a high SOCS_{\mathrm{OC}} of 0.32 mV/Pa, but shows a limitation in its ξstatic\xi_{\mathrm{static}}. Its 9.6% ScAlN bimorph counterpart in [35] obtains similar performance despite its lower scandium alloy incorporation, yielding an SOCS_{\mathrm{OC}} of 38 fC/Pa and SOCS_{\mathrm{OC}} of 0.35 mV/Pa. In comparison with these devices, the TFE P3F LN PMUT strikes a balance between transmitting and receiving performance: the measured ξstatic\xi_{\mathrm{static}} is comparable with PZT and significantly higher than ScAlN. Its SSCS_{\mathrm{SC}} is competitive with both despite the smaller C0C_0 of the LN transducer, while its flat-band SOCS_{\mathrm{OC}} is the highest. Importantly, the SSCS_{\mathrm{SC}} can be further enhanced by exchanging the floating bottom electrode architecture for its grounded version, maximizing C0C_0 density.

The presented TFE P3F LN results highlight the platform’s ability to realize large-area membranes with a high keff2k_{\mathrm{eff}}^2, enabling a high SOCS_{\mathrm{OC}} alongside a high ξstatic\xi_{\mathrm{static}}. Compared with prior demonstrations of P3F LN transducers, summarized in Table V, this implementation introduces an additional application space. The LFE P3F LN PMUT in [19] shows a balance between robustness and performance, utilizing a 2-µm bimorph for an LFE P3F transducer with a small footprint fit for array applications in the MHz regime. A similar LFE P3F LN transducer in [18] scaled the film thickness to 20 µm, featuring robustness to extreme environmental conditions. The TFE P3F LN prototype in [21] showed promise for airborne applications, which was realized in this work using a 670-nm-thick P3F active layer. With this addition, the P3F LN platform demonstrates exceptional flexibility with its ability to realize high-quality films for piezoelectric transducers with thicknesses ranging from submicrometer to bulk values.

WorkMaterialInterlayer electrodeExcitationtactt_{\mathrm{act}} (µm)fsf_s (kHz)Area (mm2^2)keff2k_{\mathrm{eff}}^2 (%)
[19]X LNNLFE212000.043.7
[18]X LNNLFE207750.286.4
[21]36Y LNNTFE30011523.05.0
This Work128Y LNNTFE0.6733.40.253.0

Table V. Survey of prior P3F LN-based flexural-mode piezoelectric transducers, highlighting the submicrometer film thickness of the presented implementation, and underscoring the platform’s viability across frequencies and applications.

While the fabricated device exhibits high performance compared with incumbent airborne PMUTs, the analysis of the obtained measurements against the idealized simulation in Figure 5 shows that the device has not yet realized its full potential. The comparison between Figure 5 and Figure 7 shows that the measured device obtained a lower keff2k_{\mathrm{eff}}^2 than predicted, leading to a significantly reduced ξpeak\xi_{\mathrm{peak}} and ξstatic\xi_{\mathrm{static}}. At the same time, the device exhibited a resonance frequency approximately 3 kHz higher than in the simulation. The LDV-measured mode shape offers a direct explanation when compared with its simulated version, shown in Figure 11. The contrast between the ideal simulated mode shape of a clamped membrane and the measured data indicates that the fabricated LN device experiences a fundamental shift in boundary conditions, demonstrating a flexural mode shape with a significantly wider displacement distribution in the flat band, and a narrower shape at resonance. This discrepancy between simulation and measurement describes a complex effect that has a profound influence on device design and performance and will be the primary focus of future studies on thin-film P3F LN-based transducers.

Comparison of simulated and measured flexural mode shapes

Figure 11. Comparison of simulated and measured flexural mode shapes on a cut line defined on the minor axis of the ellipse in the flat band and at resonance, highlighting that the fabricated device does not follow the conventional mode shape of a clamped plate. Static data reveal a wider displacement distribution than expected, while the resonance data indicate the opposite effect, indicating a major shift in boundary conditions.

Conclusion

This work presents the first demonstration of a P3F LN-based submicrometer-thick, clamped-plate piezoelectric transducer operating at sub-100-kHz frequencies for airborne applications. The device achieves an exceedingly compact footprint due to the high film compliance achieved by the large-area, thin-film membrane, which consequently leads to high electromechanical performance in conjunction with the high ϵ31,eff\epsilon_{31,\mathrm{eff}} in single-crystal LN. As a result, the fabricated device demonstrates a keff2k_{\mathrm{eff}}^2 of 3% with a C0C_0 of 17 pF, a resonance frequency of 33.4 kHz, and yields a high ξstatic\xi_{\mathrm{static}}, as well as a moderate extracted flat-band SSCS_{\mathrm{SC}} and a high flat-band SOCS_{\mathrm{OC}}. Compared with SoA airborne piezoelectric transducers, the LN device achieves a comparable ξstatic\xi_{\mathrm{static}} while showcasing a significantly higher SOCS_{\mathrm{OC}}, positioning LN as a SoA sensor and bidirectional transducer platform. Importantly, the presented study underscores further potential of the TFE P3F LN transducer platform by pinpointing discrepancies between the simulation and the measured results, outlining clear next steps in order to realize the potential performance of the P3F LN piezoelectric transducer platform.

Acknowledgment

The authors would like to thank Dr. Todd Bauer for helpful discussions.

References

  1. [1]G.-L. Luo, Y. Kusano, and D. A. Horsley, “Airborne Piezoelectric Micromachined Ultrasonic Transducers for Long-Range Detection,” Journal of Microelectromechanical Systems, vol. 30, no. 1, pp. 81–89, Feb. 2021.
  2. [2]Y. Peng, H. Liu, C.-M. Chen, W. Yue, M. Teng, P.-C. Tsao, S. Umezawa, S. Ikeuchi, Y. Aida, and L. Lin, “9-Meter-Long 3d Ultrasonic Objects Detection via Packaged Lithium-Niobate PMUTs,” in 2024 IEEE 37th International Conference on Micro Electro Mechanical Systems (MEMS), Jan. 2024, pp. 124–127, iSSN: 2160-1968.
  3. [3]F. Xia, Y. Peng, S. Pala, R. Arakawa, W. Yue, P.-C. Tsao, C.-M. Chen, H. Liu, M. Teng, J. H. Park, and L. Lin, “High-SPL and Low-Driving-Voltage pMUTs by Sputtered Potassium Sodium Niobate,” in 2023 IEEE 36th International Conference on Micro Electro Mechanical Systems (MEMS). Munich, Germany: IEEE, Jan. 2023, pp. 135–138.DOI
  4. [4]S. Horowitz, T. Nishida, L. Cattfasta, and M. Sheplak, “Development of a micromachined piezoelectric microphone for aeroacoustics applications,” The Journal of the Acoustical Society of America, vol. 122, no. 6, pp. 3428–3436, Dec. 2007.DOI
  5. [5]M. D. Williams, B. A. Griffin, T. N. Reagan, J. R. Underbrink, and M. Sheplak, “An AlN MEMS Piezoelectric Microphone for Aeroacoustic Applications,” Journal of Microelectromechanical Systems, vol. 21, no. 2, pp. 270–283, Apr. 2012, conference Name: Journal of Microelectromechanical Systems.DOI
  6. [6]B. Belgacem, F. Calame, and P. Muralt, “Piezoelectric Micromachined Ultrasonic Transducers based on PZT films,” in 2006 15th ieee international symposium on the applications of ferroelectrics, Jul. 2006, pp. 326–331, iSSN: 2375-0448.
  7. [7]M. A. Caro, S. Zhang, T. Riekkinen, M. Ylilammi, M. A. Moram, O. Lopez-Acevedo, J. Molarius, and T. Laurila, “Piezoelectric coefficients and spontaneous polarization of ScAlN,” Journal of Physics: Condensed Matter, vol. 27, no. 24, p. 245901, May 2015, publisher: IOP Publishing.DOI
  8. [8]C. Shi, J. Ma, J. Wu, K. Chen, and B. Wu, “Enhanced electrical properties in KNN-based ternary ceramics with high Curie temperature,” Journal of Materials Science: Materials in Electronics, vol. 31, no. 4, pp. 2809–2816, Feb. 2020.DOI
  9. [9]M.-A. Dubois and P. Muralt, “Measurement of the effective transverse piezoelectric coefficient e31,f of AlN and Pb(Zrx,Ti1−x)O3 thin films,” Sensors and Actuators A: Physical, vol. 77, no. 2, pp. 106–112, Oct. 1999.
  10. [10]S. Mertin, V. Pashchenko, F. Parsapour, C. Nyffeler, C. S. Sandu, B. Heinz, O. Rattunde, G. Christmann, M.-A. Dubois, and P. Muralt, “Enhanced piezoelectric properties of c-axis textured aluminium scandium nitride thin films with high scandium content: Influence of intrinsic stress and sputtering parameters,” in 2017 IEEE International Ultrasonics Symposium (IUS). Washington, DC: IEEE, Sep. 2017, pp. 1–4.
  11. [11]P. Muralt, “Which is the best thin film piezoelectric material?” in 2017 IEEE International Ultrasonics Symposium (IUS). Washington, DC: IEEE, Sep. 2017, pp. 1–3.
  12. [12]R. Lu, M. Breen, A. E. Hassanien, Y. Yang, and S. Gong, “A Piezoelectric Micromachined Ultrasonic Transducer Using Thin-Film Lithium Niobate,” Journal of Microelectromechanical Systems, vol. 29, no. 6, pp. 1412–1414, Dec. 2020, conference Name: Journal of Microelectromechanical Systems.DOI
  13. [13]L. Qiao, G. Li, H. Tao, J. Wu, Z. Xu, and F. Li, “Full characterization for material constants of a promising KNN-based lead-free piezoelectric ceramic,” Ceramics International, vol. 46, no. 5, pp. 5641–5644, Apr. 2020.DOI
  14. [14]M. A. Caro, S. Zhang, M. Ylilammi, T. Riekkinen, M. A. Moram, O. Lopez-Acevedo, J. Molarius, and T. Laurila, “Erratum: Piezoelectric coefficients and spontaneous polarization of ScAlN (2015 J. Phys. Condens. Matter 27 245901),” Journal of Physics: Condensed Matter, vol. 27, no. 27, p. 279602, Jul. 2015.
  15. [15]N. Wolff, M. R. Islam, L. Kirste, S. Fichtner, F. Lofink, A. Žukauskaitė, and L. Kienle, “All–xScxN Thin Films at High Temperatures: Sc-Dependent Instability and Anomalous Thermal Expansion,” Micromachines, vol. 13, no. 8, p. 1282, Aug. 2022.
  16. [16]J. Stregue, T. Aubert, N. Kokanyan, F. Bartoli, A. Taguett, V. Polewczyk, E. Kokanyan, S. Hage-Ali, P. Boulet, and O. Elmazria, “Stoichiometric Lithium Niobate Crystals: Towards Identifiable Wireless Surface Acoustic Wave Sensors Operable up to 600 °C,” IEEE Sensors Letters, vol. 3, no. 4, pp. 1–4, Apr. 2019.
  17. [17]S. Akhbari, F. Sammoura, B. Eovino, C. Yang, and L. Lin, “Bimorph Piezoelectric Micromachined Ultrasonic Transducers,” Journal of Microelectromechanical Systems, vol. 25, no. 2, pp. 326–336, Apr. 2016.DOI
  18. [18]V. Chulukhadze, Z. Liu, Z. Yao, L. Matto, T.-H. Hsu, N. Ravi, X. Niu, M. E. Liao, M. S. Goorsky, N. Hall, and R. Lu, “Bimorph Lithium Niobate Piezoelectric Micromachined Ultrasonic Transducers,” IEEE Transactions on Ultrasonics, pp. 1–1, 2026.DOI
  19. [19]X. Zhao, M. Pertijs, and T. Manzaneque, “Piezoelectric Micromachined Ultrasonic Transducer (PMUT) Based on Bilayer X-Cut Lithium Niobate,” in 2025 23rd International Conference on Solid-State Sensors, Actuators and Microsystems (Transducers), Jun. 2025, pp. 530–533, iSSN: 2167-0021.DOI
  20. [20]Z. Yao, V. Chuluhadze, Z. Liu, X. Niu, T.-H. Hsu, B. Kim, N. Hall, and R. Lu, “Bimorph Lithium Niobate Piezoelectric Micromachined Ultrasonic Transducer,” in 2025 IEEE International Ultrasonics Symposium (IUS), Sep. 2025, pp. 1–3, iSSN: 1948-5727.arxiv.org/abs/2509.00600
  21. [21]V. Chulukhadze, Z. Liu, L. Matto, Z. Yao, M. Liao, T.-H. Hsu, M. S. Goorsky, N. Hall, and R. Lu, “Thickness-Field Excited Bimorph Lithium Niobate Piezoelectric Transducer,” Journal of Microelectromechanical Systems, vol. 35, no. 5, pp. 762–764, Oct. 2026.DOI
  22. [22]J. Kramer, V. Chulukhadze, K. Huynh, O. Barrera, M. Liao, S. Cho, L. Matto, M. S. Goorsky, and R. Lu, “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, iSSN: 2327-1949.DOI
  23. [23]J. Kramer, B. T. Bosworth, L. Matto, N. R. Jungwirth, O. Barrera, F. Bergmann, S. Cho, V. Chulukhadze, M. Goorsky, N. D. Orloff, and R. Lu, “Acoustic resonators above 100 GHz,” Applied Physics Letters, vol. 127, no. 1, Jul. 2025, publisher: AIP Publishing.DOI
  24. [24]R. Lu, Y. Yang, S. Link, and S. Gong, “Enabling Higher Order Lamb Wave Acoustic Devices With Complementarily Oriented Piezoelectric Thin Films,” Journal of Microelectromechanical Systems, vol. 29, no. 5, pp. 1332–1346, Oct. 2020.DOI
  25. [25]J. Kramer and R. Lu, “A Generalized Acoustic Framework for Multilayer Piezoelectric Platforms,” IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 72, no. 9, pp. 1302–1311, Sep. 2025.DOI
  26. [26]Z. Liu, V. Chulukhadze, Y.-C. Wang, Z. Yao, L. Matto, T.-H. Hsu, N. Ravi, X. Niu, M. E. Liao, M. S. Goorsky, R. Lu, and N. Hall, “High-temperature characterization of bimorph X-cut lithium niobate piezoelectric micromachined ultrasonic transducers up to 1000 °C,” APL Electronic Devices, vol. 2, no. 4, p. 046102, Dec. 2026.DOI
  27. [27]V. Chulukhadze, Y. Wang, L. Matto, M. E. Liao, I. Anderson, J. Kramer, S. Cho, M. S. Goorsky, and R. Lu, “Toward Miniature High-Coupling Lithium Niobate Thin-Film Bulk Acoustic Wave Resonators at Millimeter Wave,” IEEE Transactions on Electron Devices, vol. 73, no. 2, pp. 988–994, Feb. 2026.
  28. [28]V. Chulukhadze, Y. Wang, I. Anderson, J. Kramer, S. Cho, and R. Lu, “Miniature High-Coupling Lithium Niobate Thin Film Bulk AcousticDOI

Paper details

vlnc:W_br5xssea
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 5

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Engineering

Not recorded for this paper

View statement
Looking for related papers…
Contents