Lattice Distortion and Dielectric Fatigue in Piezoelectric Micromachined Ultrasonic Transducers (pMUTs) for Sub-Dermal Drug Delivery

The Direct and Converse Piezoelectric Effects

The piezoelectric effect is the fundamental physical mechanism enabling pMUTs to convert electrical signals into mechanical motion. The direct piezoelectric effect describes the generation of electric charge when mechanical stress is applied to certain crystalline materials; the converse piezoelectric effect (also called the inverse piezoelectric effect) describes mechanical deformation when an electric field is applied. In thin-film lead zirconate titanate (PZT), both effects operate simultaneously during device operation, making PZT the material of choice for piezoelectric micromachined ultrasonic transducers.

The direct effect is quantified by the piezoelectric charge coefficient d, with units of coulombs per newton (C/N) or picometers per volt (pm/V). For a PZT film with composition near the morphotropic phase boundary (MPB)—typically around 52% PbZr and 48% PbTi—the d₃₃ coefficient (strain in the poling direction per unit electric field) can reach 600–700 pC/V, significantly higher than single-crystal quartz (~2 pC/V). This high coupling coefficient is why PZT dominates commercial piezoelectric applications.

The converse effect is equally important for pMUT operation. When a voltage is applied across a PZT thin film, the material experiences strain proportional to the applied field. This electromechanical coupling is expressed through the piezoelectric voltage coefficient g, defined as electric field per unit stress (V·m/N). The relationship between d and g is mediated by the elastic compliance tensor and permittivity, making the material's elastic properties critical to understanding pMUT performance.

Ferroelectric Domains and Polarization

PZT is a ferroelectric material, meaning it possesses spontaneous electric polarization that can be reversed by applying an external electric field. This polarization exists at the nanoscale through the alignment of electric dipole moments within the crystal lattice. However, bulk PZT does not exhibit a single, uniform polarization direction; instead, it consists of many ferroelectric domains—regions of uniform polarization separated by domain walls.

Each domain represents a different crystallographic orientation of the spontaneous polarization vector. In an unpoled PZT thin film, domains are randomly oriented, resulting in zero net polarization and minimal piezoelectric response. Poling is the process of applying a strong DC electric field (typically 100–200 kV/cm for thin films) to align domains preferentially along the field direction. This alignment is not permanent at room temperature; rather, it represents a metastable state that can be gradually randomized by thermal cycling, mechanical vibration, or repeated electrical cycling.

The domain structure in thin-film PZT is constrained by the film's thickness, grain size, and interface conditions. In typical pMUT films (0.5–2 μm thick), the domain size ranges from 50 nm to several hundred nanometers, with multiple domains often present within individual crystalline grains. The grain boundaries act as obstacles to domain-wall motion, a phenomenon known as domain-wall pinning. Pinning sites arise from structural defects, impurities, and strain gradients at grain boundaries.

Domain Switching Under Cyclic Electric Fields

When a pMUT is driven at high frequencies (typically 100 kHz to 10 MHz for sub-dermal applications), the applied electric field alternates rapidly. At low field amplitudes, domains respond elastically to the field oscillation without switching—this is the linear piezoelectric regime. However, at higher field amplitudes, some domains overcome the energy barriers at pinning sites and switch orientation, contributing additional strain. This is the nonlinear piezoelectric regime, characterized by hysteretic behavior in the strain-field relationship.

Repeated domain switching under cyclic loading causes fatigue in the piezoelectric response. The mechanisms include:

• Domain-wall pinning strengthening: As domains switch repeatedly, they interact with pinning sites in ways that increase the energy barrier for future switching, progressively reducing the material's ability to generate strain.
• Accumulation of defects: Each domain-switching event involves local lattice distortion. Over millions of cycles, these distortions accumulate, creating new pinning sites and degrading the overall electromechanical coupling.
• Oxygen-vacancy migration: In PZT thin films, oxygen vacancies can migrate under applied electric fields, creating space-charge regions that modify the local electric field and inhibit domain switching.

For sub-dermal drug-delivery pMUTs operating continuously at high duty cycles (>50% on-time), fatigue manifests as a gradual reduction in acoustic output over hours to days of operation, even when the electrical input remains constant.

Electromechanical Coupling Coefficients

The electromechanical coupling factor k quantifies the efficiency with which electrical energy is converted to mechanical energy (or vice versa). It is defined as:

k² = (energy converted) / (total electrical energy input)

For a thin-film PZT pMUT, the relevant coupling factor is typically k₃₃ (coupling between electric field and strain in the poling direction) or k₃₁ (coupling between electric field perpendicular to the film and in-plane strain). Values for PZT near the MPB range from 0.65 to 0.75, meaning 65–75% of electrical energy is converted to mechanical work.

In practical pMUT devices, the effective coupling is further reduced by parasitic effects: acoustic impedance mismatch between the PZT film and surrounding materials, mechanical losses in the substrate, and electrical losses in the electrodes. The acoustic impedance of PZT (~35 MRayl) is much higher than that of human skin (~1.5 MRayl), requiring careful acoustic design to minimize reflection losses at interfaces.

pMUT Device Architecture and Fabrication

A typical pMUT for sub-dermal drug delivery consists of a thin-film PZT membrane suspended over a cavity etched into a silicon substrate. The basic stack structure, from bottom to top, includes: silicon substrate, bottom electrode (typically platinum or titanium/platinum), PZT thin film (0.5–2 μm), top electrode, and a protective passivation layer. The membrane dimensions are typically 100–500 μm in lateral extent, with the thickness optimized for resonant operation at the target frequency.

The PZT thin film is deposited using sol-gel or sputtering methods. Sol-gel deposition involves chemical synthesis of a precursor solution, spin-coating onto the substrate, and thermal annealing to crystallize the PZT phase. This method allows excellent control of stoichiometry and produces highly uniform films, but requires careful management of organic residues and pyrolysis byproducts. Sputtering uses physical bombardment to eject material from a ceramic PZT target, offering faster deposition but potentially introducing defects and requiring higher annealing temperatures.

After PZT deposition, the film is patterned using photolithography and reactive-ion etching (RIE) to define individual transducer elements. RIE uses energetic ions (typically Cl₂, CF₄, or BCl₃ chemistry for PZT) to selectively remove material. The etching process introduces surface damage, including amorphous layers, subsurface cracks, and chemical residues. These defects act as nucleation sites for fatigue-related degradation.

Array Geometry and Acoustic Field Shaping

Individual pMUT elements are arranged in 1D linear arrays or 2D matrix arrays to achieve acoustic focusing and beam steering. For sub-dermal delivery, 1D arrays of 32–128 elements are common, with element spacing of 200–500 μm. The array aperture (total lateral dimension) typically ranges from 5–50 mm, depending on the target skin area and desired acoustic pressure profile.

The acoustic field produced by an array is the coherent superposition of fields from individual elements. When all elements are driven in phase at the same frequency, they produce a focused beam with a main lobe and side lobes. The acoustic pressure amplitude at the focus is proportional to the number of elements and the individual element displacement. For sub-dermal delivery, typical acoustic pressures at the skin surface range from 0.5 to 5 MPa (peak-to-peak), with pressures of 2–3 MPa being typical for efficient transdermal transport of large-molecule therapeutics.

The focal depth depends on the array aperture and operating frequency. Higher frequencies produce tighter focusing but greater attenuation in tissue. For sub-dermal delivery, operating frequencies of 1–5 MHz are common, balancing the need for sufficient skin penetration (typically 0.5–2 mm depth for dermal drug deposition) with practical array dimensions.

Acoustic Output Requirements and Transducer Efficiency

The acoustic power output of a pMUT array is determined by:

P_acoustic = (1/2) × Z × A × (v)²

where Z is the acoustic impedance of the medium, A is the radiating area, and v is the particle velocity at the surface. For a 10 mm × 10 mm array operating at 2 MHz with surface velocity of 1 m/s, the acoustic power output is approximately 15 mW. However, only a fraction of this power is transmitted into the skin; the remainder is reflected due to acoustic impedance mismatch.

The acoustic transmission coefficient at a PZT-skin interface is approximately:

T = 4 × Z_PZT × Z_skin / (Z_PZT + Z_skin)²

With Z_PZT ≈ 35 MRayl and Z_skin ≈ 1.5 MRayl, the transmission coefficient is only ~0.35, meaning 65% of acoustic energy is reflected. This impedance mismatch is a fundamental limitation of acoustic delivery systems and motivates the use of acoustic matching layers—thin films with intermediate acoustic impedance placed between the PZT and skin.

Matching layers are typically made of polymer composites or ceramic materials with impedance in the range 3–10 MRayl. A quarter-wave matching layer (thickness = wavelength/4 at the operating frequency) can increase transmission to >90%. However, matching layers introduce mechanical compliance that changes the resonant frequency and can degrade under repeated acoustic cycling.

Membrane Mechanics and Resonance

The pMUT membrane operates as a vibrating plate with resonant frequencies determined by its geometry and material properties. For a circular membrane of radius a, thickness h, and Young's modulus E, the fundamental resonant frequency is:

f₀ ≈ (λ₁² / (2π a²)) × √(E × h² / (12 × ρ × (1 - ν²)))

where λ₁ ≈ 10.21 for a clamped circular membrane, ρ is density, and ν is Poisson's ratio. For a 200 μm radius PZT membrane, 1 μm thick, the resonant frequency is typically 2–4 MHz.

Operation at resonance maximizes displacement amplitude and acoustic output for a given electrical input voltage. However, resonant operation also maximizes mechanical strain in the membrane, accelerating fatigue-related degradation. Off-resonance operation (e.g., at 0.8 × f₀) reduces acoustic output but may extend device lifetime.

Passivation and Interface Engineering

The top surface of the pZT film and electrodes must be protected from environmental moisture, ionic contamination, and chemical attack. Passivation layers, typically made of silicon dioxide (SiO₂) or silicon nitride (Si₃N₄), are deposited after patterning. These layers prevent water ingress, which would otherwise cause polarization fatigue by enabling ionic conduction and charge compensation at the PZT surface.

However, passivation layers introduce mechanical constraints that can modify the stress state in the underlying PZT film. Compressive stress from the passivation layer can either enhance or degrade piezoelectric performance, depending on the layer thickness and the PZT composition. Additionally, the passivation layer itself can fail under repeated mechanical cycling, leading to delamination and water ingress at the edges of pMUT elements.

Operating Frequency and Acoustic Frequency Regimes

Sub-dermal drug-delivery pMUTs operate in the ultrasonic frequency range, typically 1–10 MHz. This range is chosen to balance several competing requirements:

• Skin penetration depth: Lower frequencies (1–2 MHz) penetrate deeper into skin (2–5 mm) but produce larger acoustic wavelengths, limiting spatial focusing resolution.
• Acoustic attenuation: Tissue attenuation increases approximately as the square of frequency; at 5 MHz, attenuation is roughly 25 times higher than at 1 MHz.
• Thermal effects: Higher frequencies produce more localized heating due to viscous dissipation, potentially causing thermal damage if sustained.
• Transducer efficiency: Most commercial pMUT designs have peak efficiency at 2–5 MHz, matching typical operating frequencies.

For transdermal drug delivery, operating frequencies of 2–3 MHz are most common, providing a practical compromise between penetration depth, spatial resolution, and device efficiency. Some emerging applications explore low-frequency regimes (100–500 kHz) for deeper penetration or high-frequency regimes (5–10 MHz) for superficial delivery with higher spatial precision.

Duty Cycle and Thermal Management

The duty cycle is the fraction of time the pMUT is actively transmitting acoustic energy, expressed as a percentage. For a wearable patch worn continuously, duty cycles range from 5% to 80%, depending on the therapeutic protocol. A 20% duty cycle means the transducer is active for 0.2 seconds out of every 1 second, with 0.8 seconds of rest.

Duty cycle profoundly affects device reliability and patient safety:

• Thermal generation: Acoustic power dissipated in tissue generates heat proportional to the acoustic intensity and absorption coefficient. At 2 MHz with 3 MPa acoustic pressure, continuous operation can raise local tissue temperature by 2–5°C within minutes. Intermittent operation with adequate rest periods allows thermal dissipation.
• Mechanical fatigue: Domain switching, grain-boundary sliding, and crack propagation all accelerate under continuous cycling. Intermittent operation reduces cumulative fatigue damage.
• Electrical heating: Resistance in the electrodes and PZT film generates Joule heat, adding to acoustic heating. Higher duty cycles increase electrical heating.

Typical therapeutic protocols for sub-dermal delivery use pulsed operation with duty cycles of 10–30%, applying acoustic energy for 100–500 millisecond bursts separated by rest periods of 1–10 seconds. This pulsing strategy is motivated by both safety (thermal limits) and efficacy (allowing time for drug diffusion and skin recovery between pulses).

Electrical Driving Schemes and Nonlinear Response

pMUTs are typically driven by sinusoidal electrical signals at the resonant frequency, amplified to produce peak-to-peak voltages of 20–100 V. However, the relationship between input voltage and acoustic output is nonlinear, especially at high driving amplitudes.

At low voltages (linear regime), acoustic pressure is proportional to applied voltage:

P_acoustic ∝ V_applied

At higher voltages, domain switching becomes significant, introducing hysteresis and harmonic generation. The acoustic output may increase sublinearly with voltage, or even decrease if domain saturation is reached. This nonlinearity means that doubling the driving voltage does not necessarily double the acoustic output.

Harmonic generation is particularly important in sub-dermal applications. When driven at frequency f, the pMUT produces not only fundamental frequency energy but also energy at 2f, 3f, and higher harmonics. These harmonics arise from nonlinear domain switching and can contribute to acoustic heating and potential tissue damage. Harmonic content typically increases with driving amplitude and with device fatigue (as the material becomes more nonlinear).

Fatigue-Induced Degradation During Continuous Operation

The most critical failure mode in wearable sub-dermal pMUTs is gradual loss of acoustic output without corresponding change in electrical input. This occurs because:

• Domain-wall pinning increases: Repeated domain switching at high frequency strengthens pinning sites, progressively reducing the strain generated per unit electric field.
• Irreversible polarization loss: Some domains fail to reorient after switching, remaining "stuck" in unfavorable orientations. This reduces the net polarization and piezoelectric coupling.
• Microcracking at grain boundaries: Cyclic strain concentrates at grain boundaries, where lattice distortion is maximum. Over thousands to millions of cycles, microcracks nucleate and propagate, reducing the mechanical integrity of the film.
• Oxygen-vacancy redistribution: Under applied electric field, oxygen vacancies migrate toward regions of high electric field, creating space-charge regions that screen the applied field and reduce domain-switching efficiency.

These degradation mechanisms are established materials-science results based on decades of research on ferroelectric fatigue in bulk and thin-film PZT. However, the specific fatigue rates and failure thresholds for thin-film pMUTs in sub-dermal delivery are less well characterized. Most commercial sub-dermal acoustic delivery systems lack published accelerated stress-testing data, making it difficult to predict device lifetime under realistic operating conditions.

Accelerated Stress Testing and Standardization Gaps

Accelerated stress testing (AST) is essential for predicting device reliability and setting performance specifications. Standard AST protocols for bulk PZT ceramics exist (e.g., ASTM D6927), but these protocols are not directly applicable to thin-film pMUTs, which have different failure mechanisms and operating conditions.

Current gaps in standardized testing for sub-dermal pMUTs include:

• No consensus on relevant stress parameters: Should testing emphasize peak electric field, electric field rate-of-change, acoustic pressure, duty cycle, or temperature? Different failure mechanisms dominate under different conditions.
• Limited data on long-term reliability: Most published studies report fatigue data over 10⁷–10⁹ cycles, but wearable patches may operate for 10¹⁰–10¹² cycles over months of continuous use.
• Lack of standardized measurement protocols: How should acoustic output degradation be quantified? Acoustic pressure amplitude, acoustic power, or frequency-dependent response? Different metrics may reveal different aspects of degradation.
• No standardized environmental testing: Should testing include thermal cycling, humidity exposure, mechanical vibration, or other environmental stressors relevant to wearable applications?

This lack of standardization creates significant barriers to comparing devices from different manufacturers and predicting real-world reliability. Emerging research is beginning to address these gaps, but no universally accepted standard yet exists.

Crystal Structure Fundamentals in Lead Zirconate Titanate

Lead zirconate titanate (PZT) is a solid-solution ceramic compound with the general formula Pb(Zr_{x}Ti_{1-x})O₃, where the ratio of zirconium to titanium atoms determines the material's electromechanical properties. The perovskite crystal structure—a cubic or pseudocubic lattice at room temperature—forms the basis for PZT's exceptional piezoelectric response. In this structure, lead atoms occupy the corners of a cube, while a smaller metal cation (Zr or Ti) sits at the body center, surrounded by six oxygen anions. This arrangement creates spontaneous electric polarization when mechanical stress is applied, or conversely, mechanical deformation when an electric field is imposed.

For thin-film pMUTs intended for sub-dermal drug delivery, the PZT layer is typically deposited on a silicon substrate using sol-gel, sputtering, or pulsed-laser deposition methods. The resulting film is polycrystalline—composed of many crystalline grains, each with its own crystallographic orientation. The thickness of these films ranges from 1 to 10 micrometers, a scale at which surface effects and grain-boundary phenomena become dominant failure mechanisms. Established materials-science finding: the piezoelectric coefficient d₃₃ (transverse piezoelectric response) and electromechanical coupling factor k₃₃ both decrease as grain size decreases below approximately 1 micrometer, a well-documented relationship across multiple research groups.

Grain-Boundary Structure and Defect Accumulation

Grain boundaries are two-dimensional defects where the crystallographic orientation changes abruptly between adjacent grains. In PZT thin films, grain boundaries typically contain oxygen vacancies, lead vacancies, and cation dislocations—point defects that arise during film deposition and thermal processing. These defects create localized regions of altered electrical conductivity and mechanical compliance compared to the grain interior.

The density and character of grain boundaries depend critically on the deposition method and annealing temperature. Films produced via rapid thermal annealing at temperatures above 600°C exhibit larger grains (0.5–2 micrometers) but may contain more severe residual stress at boundaries. Established finding: X-ray diffraction and transmission electron microscopy studies consistently show that grain-boundary regions in as-deposited PZT films contain 10–100 times higher oxygen-vacancy concentrations than the grain interior, measured via electron paramagnetic resonance and secondary-ion mass spectrometry.

In micro-etched pMUT arrays—where individual transducer elements are defined by reactive-ion etching or wet chemical etching—the etching process introduces additional surface damage and subsurface lattice strain. The sidewalls of etched features exhibit amorphous or nanocrystalline regions extending 50–200 nanometers into the film. This emerging device-specific finding has been documented in recent studies of commercial pMUT prototypes but lacks standardization across manufacturers.

Defect Chemistry and Charge Compensation

Oxygen vacancies in PZT act as electron donors and create local electric fields that interact with domain walls (discussed in Sub-module 2.2). The defect concentration is governed by thermodynamic equilibrium during annealing and by kinetic trapping of defects during rapid cooling. In grain-boundary regions, the lower coordination number of cations at the boundary creates energetically favorable sites for vacancy formation. The charge-compensation mechanism—whether through electronic defects (electrons or holes) or ionic defects (additional vacancies or interstitials)—affects how the material responds to applied electric fields.

Established finding: acceptor-doped PZT (commonly doped with manganese or iron) shows reduced leakage current and higher dielectric strength than undoped PZT, but this improvement comes at the cost of reduced piezoelectric coefficients and lower mechanical Q-factor. For pMUT applications requiring high acoustic output over thousands of hours, this trade-off becomes critical.

Microstructural Characterization Techniques

Understanding grain-boundary defect populations requires multiple complementary techniques. Scanning electron microscopy reveals grain size and morphology; atomic force microscopy measures surface roughness and residual stress; X-ray diffraction quantifies preferred crystallographic orientation (texture); and transmission electron microscopy provides atomic-scale images of boundary structure. Electron energy-loss spectroscopy can map oxygen-vacancy distributions, though this requires specialized equipment not available in all device-development laboratories.

The lack of standardized accelerated stress-testing metrics for commercial sub-dermal acoustic delivery cells means that manufacturers often rely on proprietary characterization protocols, making it difficult to compare defect-population data across different pMUT designs or suppliers. This fragmentation limits the field's ability to establish predictive models for long-term reliability.

Ferroelectric Domain Structure and Spontaneous Polarization

PZT exhibits ferroelectricity: a spontaneous electric polarization that persists in the absence of an applied electric field and can be reversed by applying a sufficiently strong field in the opposite direction. At room temperature, the crystal structure is tetragonal (for compositions near the morphotropic phase boundary where Zr and Ti are roughly equal), with a slight elongation along the polar axis. This distortion arises from the displacement of the central metal cation relative to the oxygen cage, creating a permanent electric dipole moment.

In polycrystalline PZT films, the material does not exist as a single monodomain. Instead, it contains multiple ferroelectric domains—regions in which the polarization vector points in different crystallographic directions. Domain walls are the interfaces between regions of opposite or differently oriented polarization. In a typical micro-etched PZT film used in pMUTs, domain sizes range from tens of nanometers to a few micrometers, with domain-wall density on the order of 10⁶ to 10⁷ walls per millimeter.

Domain-Wall Pinning by Point Defects

When an external electric field is applied to a ferroelectric material, domain walls move—expanding domains aligned with the field and shrinking those opposed to it. This domain-wall motion is the microscopic mechanism underlying the piezoelectric response. However, point defects (primarily oxygen vacancies and cation vacancies) interact electrostatically with the polarization charge at domain walls, creating local potential-energy minima that trap the wall. This phenomenon is called domain-wall pinning.

In undoped PZT, oxygen vacancies (V_O^{••}, carrying a +2 effective charge) are attracted to domain walls where negative bound charge accumulates. The vacancy forms a pinning site: when thermal energy or applied-field energy is insufficient to overcome the local potential barrier, the domain wall becomes immobilized. Established finding: the activation energy for domain-wall depinning in PZT ranges from 0.1 to 0.5 eV depending on defect concentration, measured via thermally stimulated depolarization current and dielectric spectroscopy.

Acceptor doping (e.g., with Mn²⁺ or Fe³⁺) increases the oxygen-vacancy concentration and strengthens pinning. This is why acceptor-doped PZT exhibits lower dielectric loss and higher mechanical Q-factor—the pinned domain walls dissipate less energy through motion. However, pinning also reduces the magnitude of domain-wall displacement under applied stress, lowering the piezoelectric coefficient d₃₃. This trade-off is well-established in materials science and appears consistently across independent research groups.

Non-Linear Polarization Response and Hysteresis

The relationship between applied electric field and induced polarization in ferroelectrics is fundamentally non-linear, quantified by the polarization-field hysteresis loop. The area enclosed by this loop represents energy dissipated per cycle—energy converted to heat. At low field amplitudes, the hysteresis loop is small and the material behaves nearly linearly; at high amplitudes, non-linearity becomes pronounced.

In pMUT operation for drug delivery, the applied voltage typically ranges from 5 to 50 volts across a 2–5 micrometer film, corresponding to electric fields of 1–25 MV/m—well into the non-linear regime. At these fields, domain-wall motion contributes significantly to the polarization change, and the hysteresis loss becomes substantial. Emerging device-specific finding: real-time monitoring of current waveforms in operating pMUT arrays shows that the harmonic content (second, third, and higher harmonics) increases over the first 100–500 hours of continuous operation, indicating progressive non-linearity likely driven by defect redistribution and domain-wall pinning changes.

Cyclic Stress and Piezoelectric-Coefficient Degradation

When a pMUT operates at high frequency (typically 100 kHz to 10 MHz for sub-dermal delivery), the PZT film experiences cyclic mechanical stress at the same frequency. Each cycle involves elastic deformation, domain-wall motion, and ionic rearrangement. Over thousands to millions of cycles, this repeated stress causes progressive changes in the material's electromechanical properties—a phenomenon called fatigue or aging.

The most critical fatigue effect for pMUT reliability is the loss of piezoelectric coefficient d₃₃ without any change in electrical impedance or resistance readings. A pMUT may continue to draw the same electrical current and show the same impedance curve when measured with standard equipment, yet produce significantly less acoustic output. This decoupling between electrical and mechanical performance creates a hidden failure mode: the device appears functional to standard diagnostics while actually delivering insufficient acoustic energy to drive large-molecule drugs through skin.

Established finding: piezoelectric-coefficient degradation in ferroelectric ceramics under cyclic electric-field stress follows a logarithmic time dependence, with most loss occurring in the first 10³–10⁴ cycles, then slowing. However, emerging device-specific data from pMUT prototypes suggests that the rate of degradation depends strongly on the electrical waveform shape, frequency, and the presence of mechanical resonance amplification—factors that vary across different pMUT designs.

Mechanisms of Coefficient Loss

The microscopic mechanisms driving piezoelectric degradation include: (1) irreversible domain reorientation, where domains become preferentially aligned with the stress direction, reducing the material's ability to respond to further stress; (2) defect migration, where oxygen vacancies move under the combined influence of electric field and stress, redistributing around domain walls and changing pinning characteristics; and (3) microcracks and localized fracture, where stress concentrations at grain boundaries or near etched features initiate small cracks that propagate and relieve stress locally, reducing the effective piezoelectric response.

For sub-dermal pMUT patches, the combination of high frequency (10–100 MHz), high stress amplitude (50–200 MPa), and continuous operation over 8–24 hours per day creates an unusually severe fatigue environment compared to traditional piezoelectric applications like sonar or industrial ultrasound, where duty cycles are typically much lower.

Stress Concentration at Grain Boundaries

Grain boundaries in polycrystalline materials are sites of stress concentration because they represent discontinuities in elastic modulus and crystallographic orientation. When a polycrystalline material is subjected to uniform applied stress, the actual stress at a grain boundary can exceed the nominal applied stress by a factor of 2–10, depending on the elastic anisotropy of the material and the misorientation angle between adjacent grains. Established finding: finite-element simulations and experimental measurements using micro-Raman spectroscopy consistently show that local stress at grain boundaries in PZT can reach 1.5–3 times the average film stress.

In pMUT operation at acoustic frequencies (100 kHz to 10 MHz), the film experiences cyclic stress with peak amplitudes of 50–300 MPa, depending on the driving voltage and the geometry of the transducer. At grain boundaries, this translates to localized cyclic stresses exceeding 100–500 MPa. Over millions of cycles, this repeated high-stress cycling causes permanent lattice distortion—a change in atomic positions that does not fully reverse when the stress is removed.

Lattice Distortion Mechanisms: Dislocation Generation and Propagation

At the atomic scale, lattice distortion occurs through the motion of dislocations—linear defects where the regular lattice arrangement is disrupted. A dislocation can be visualized as an extra half-plane of atoms inserted into the crystal structure (edge dislocation) or as a shear offset of atomic planes (screw dislocation). When stress exceeds a critical threshold—the Peierls stress—dislocations can move through the lattice, leaving behind permanent deformation.

In PZT, the Peierls stress is typically 1–10 GPa at room temperature, calculated from first-principles density-functional theory. However, the presence of point defects (oxygen vacancies, cation vacancies) and grain boundaries significantly reduces the effective stress threshold for dislocation nucleation and motion. Established finding: transmission electron microscopy studies of fatigued PZT ceramics show that dislocation density increases from ~10⁶ cm⁻² in as-sintered material to 10⁸–10⁹ cm⁻² after 10⁷ fatigue cycles.

Grain boundaries serve as both sources and sinks for dislocations. Under cyclic stress, dislocations are generated at grain boundaries and propagate into the grain interior, then eventually accumulate back at the boundary. This cyclic process—called dislocation multiplication—gradually fills the boundary region with a tangled network of dislocations and associated lattice strain.

Oxygen-Vacancy Migration Under Cyclic Stress

Beyond dislocation motion, cyclic stress also drives the migration of point defects, particularly oxygen vacancies. The electric field associated with domain walls, combined with the stress-induced deformation of the lattice, creates a driving force for vacancy migration. Emerging device-specific finding: operando X-ray diffraction studies of pMUT films during acoustic cycling (performed at synchrotron facilities) show that the lattice parameter changes non-monotonically during the first 10³–10⁴ cycles, suggesting that oxygen vacancies redistribute to new equilibrium positions under the combined stress and electric-field environment.

This vacancy redistribution has two important consequences: (1) it alters the local charge state and polarization around domain walls, changing domain-wall pinning characteristics and contributing to the loss of piezoelectric coefficient; and (2) it creates regions of permanently altered lattice constant, which accumulate strain energy that can eventually be released through microcracking or localized plastic deformation.

Grain-Boundary Fracture and Microcrack Initiation

As lattice distortion accumulates at grain boundaries, the elastic strain energy density increases. When this energy exceeds the fracture toughness of the material, a microcrack initiates at the boundary and begins to propagate. Established finding: grain-boundary fracture is the primary failure mode in polycrystalline PZT ceramics under cyclic loading, with crack initiation typically occurring after 10⁴–10⁶ cycles depending on stress amplitude and material quality.

In thin-film pMUTs, the situation is complicated by the presence of the silicon substrate and any intermediate buffer layers. Cracks that initiate at grain boundaries in the PZT film may propagate into the substrate or along the PZT-substrate interface, causing delamination. Emerging device-specific data: focused-ion-beam cross-sectioning of fatigued pMUT samples reveals that grain-boundary cracks often terminate at the PZT-substrate interface rather than propagating completely through the film, creating partially detached regions that no longer contribute to the acoustic output.

Residual Stress and Lattice Curvature

Lattice distortion at grain boundaries is not uniformly distributed; it manifests as residual stress—internal stress that persists even when external loading is removed. This residual stress arises from the incompatibility between the distorted grain-boundary region and the surrounding undistorted grain interior. X-ray diffraction peak broadening and shifts indicate that residual tensile stress of 50–200 MPa accumulates in grain-boundary regions after extended acoustic cycling.

Moreover, the non-uniform distribution of lattice distortion creates lattice curvature—a gradual bending of the crystal planes—particularly in regions adjacent to grain boundaries. This curvature is measurable via high-resolution X-ray diffraction (rocking curves) and scanning transmission electron microscopy. Lattice curvature reduces the long-range coherence of the piezoelectric response because the local polarization direction varies continuously through the distorted region.

Impact on Acoustic Output and Device Reliability

The accumulation of lattice distortion at grain boundaries has a direct impact on pMUT performance. As dislocations and residual stress accumulate, the effective elastic modulus of the material decreases, reducing the acoustic impedance and the efficiency of acoustic-wave generation. Simultaneously, the increased defect density acts as scattering centers for acoustic waves, increasing acoustic attenuation. Emerging finding: acoustic spectroscopy measurements on fatigued pMUT samples show that the acoustic quality factor Q decreases by 20–50% after 10⁶ cycles of high-amplitude operation, even when the piezoelectric coefficient d₃₃ measured electrically remains nearly constant.

For sub-dermal drug-delivery applications, this degradation is particularly concerning because the acoustic output must remain sufficient to drive large-molecule therapeutics (molecular weight 10–100 kDa) through the stratum corneum and dermis. A 30% reduction in acoustic output may be clinically undetectable in standard electrical testing but sufficient to reduce drug delivery rate below therapeutic levels, risking patient under-dosing.

The lack of standardized accelerated stress-testing metrics means that different pMUT manufacturers use different cycling protocols (frequency, voltage, cycle count, temperature) to assess fatigue. Without standardization, it is impossible to compare reliability data across devices or to establish predictive models for how long a given pMUT design will maintain adequate acoustic output in clinical use. This represents a significant gap in the field's ability to ensure safe and effective sub-dermal acoustic delivery.

Fundamental Physics of Dielectric Breakdown in Piezoelectric Materials

Dielectric breakdown in thin-film piezoelectrics represents a catastrophic failure mode in which the material transitions from an insulating state to a conducting state, typically irreversibly. In pMUT devices operating at frequencies between 1–10 MHz with applied voltages of 20–100 V across film thicknesses of 1–5 micrometers, the electric field strength routinely reaches 4–20 MV/cm. This far exceeds the static dielectric breakdown field strength of bulk PZT ceramics (approximately 2–4 MV/cm), yet thin-film PZT can tolerate these stresses temporarily due to reduced defect density and improved crystalline quality compared to bulk material.

The breakdown process unfolds through several distinct physical mechanisms. Intrinsic breakdown occurs when the electric field directly ionizes the material lattice, creating free charge carriers that cascade through the material. This mechanism sets a theoretical upper limit independent of material defects, but in practice occurs only in extremely pure, defect-free films. Thermal breakdown develops when joule heating from leakage current raises the material temperature, increasing conductivity exponentially and creating a positive feedback loop. In continuous high-frequency operation, thermal effects become dominant because the duty cycle prevents sufficient cooling between acoustic cycles.

Extrinsic breakdown mechanisms dominate in real pMUT devices. These include electromigration, wherein metal atoms from electrodes or dopants within the PZT lattice drift under the applied electric field, creating conductive pathways. In lead zirconate titanate films, lead and oxygen vacancies can form extended defect structures that preferentially conduct charge. Impact ionization occurs when hot carriers gain sufficient energy from the electric field to create additional electron-hole pairs, initiating avalanche multiplication. This mechanism is particularly aggressive in thin films because carriers traverse the entire film thickness without scattering.

High-Frequency Cycling and Cumulative Damage

The distinction between static breakdown and breakdown under continuous acoustic cycling is critical for pMUT reliability. A film might withstand 100 V DC applied once per second but fail after 10^8 cycles at the same voltage. This occurs because each acoustic cycle introduces transient overvoltages, mechanical stress, and localized heating. Treeing, a branching conductive pathway that propagates from cathode to anode, develops incrementally over many cycles. Electron microscopy of failed pMUT films reveals dendritic structures extending from defect sites, with branch widths of 50–200 nanometers.

In sub-dermal drug-delivery pMUTs operating at 5 MHz with 50 V peak-to-peak excitation, the device experiences 5 million voltage reversals per second. Each reversal creates a transient electric field spike exceeding the steady-state value due to parasitic inductance in the electrode traces and acoustic resonance effects. Measurements using high-speed oscilloscopes on commercial pMUT arrays have documented overvoltages reaching 1.3× to 1.8× the commanded voltage during the first 100 nanoseconds of each half-cycle. Over a 24-hour continuous operation period (432 billion cycles), cumulative damage from these transients can reduce breakdown voltage from 120 V to 70 V—a 42% reduction—without visible changes in device morphology.

Domain-Wall Pinning and Polarization Fatigue as Precursors

Before catastrophic dielectric breakdown occurs, the piezoelectric film undergoes progressive degradation of its polarization state. Domain-wall pinning develops as defects (oxygen vacancies, impurity ions, and grain-boundary dislocations) accumulate near ferroelectric domain walls. These pinning sites inhibit the reversible wall motion necessary for piezoelectric transduction. Under continuous cycling, domain-wall mobility decreases, typically by 15–30% after 10^7 cycles and 40–60% after 10^9 cycles in unpassivated PZT films.

This degradation is distinct from dielectric breakdown but often precedes it. A pMUT exhibiting 50% loss of piezoelectric coefficient (d₃₃) through domain-wall pinning still maintains electrical insulation and shows normal leakage current. However, the acoustic output power has declined proportionally, risking therapeutic under-dosing in drug-delivery applications. Critically, the electrical impedance and resonant frequency of the device remain nearly unchanged, making this failure mode invisible to simple electrical diagnostic tests. This represents a silent failure mechanism: the device appears functional to standard electrical monitoring but delivers inadequate acoustic energy to the tissue.

Established Materials-Science Results Versus Device-Specific Observations

The field distinguishes between well-established findings and emerging, device-specific observations. Established results include: (1) thin-film PZT exhibits lower intrinsic breakdown fields than bulk ceramic due to columnar grain structure and residual tensile stress; (2) Joule heating dominates breakdown in continuous-duty applications above 1 MHz; (3) oxygen vacancies and lead vacancies create localized conduction channels; and (4) domain-wall pinning correlates with defect density and cycling history.

Device-specific observations from commercial sub-dermal pMUT patches include: (1) breakdown voltage decreases non-monotonically with cycling frequency, with a minimum around 3–4 MHz corresponding to maximum acoustic resonance amplitude; (2) some commercial devices show asymmetric breakdown thresholds between positive and negative voltage polarities, suggesting preferential defect orientation; and (3) passivation layer quality dramatically modulates breakdown behavior, with some commercial products showing 2–3× variation in breakdown voltage between production batches. These observations require further standardized testing across multiple manufacturers to establish generalized design rules.

Purpose and Design Rationale for Passivation Layers

Passivation barrier layers serve multiple critical functions in pMUT devices: (1) preventing moisture and oxygen diffusion into the piezoelectric film, (2) blocking ionic migration from electrodes into the PZT lattice, (3) protecting against chemical attack from the wearable substrate or skin-contact fluids, and (4) reducing leakage current by creating an additional insulating barrier. In sub-dermal drug-delivery applications, the device contacts physiological fluids (sweat, interstitial fluid, lymph) containing salts, proteins, and microorganisms. Without adequate passivation, the PZT film degrades within days to weeks of continuous wear.

The passivation layer is typically deposited after the PZT film is crystallized and poled, because most passivation materials require deposition temperatures below 300°C to avoid depolarizing the PZT. This creates a critical interface between the PZT surface and the passivation layer. Defects at this interface—pinholes, cracks, and regions of poor adhesion—allow direct contact between the PZT and the external environment, negating the protective benefit.

Established Passivation Materials and Their Properties

Silicon dioxide (SiO₂) deposited by plasma-enhanced chemical vapor deposition (PECVD) remains the most common passivation material in commercial pMUT devices. Thermal SiO₂ (grown by oxidizing a silicon layer) offers superior density and fewer defects but requires high temperature (800–1000°C), which is incompatible with pre-poled PZT. PECVD SiO₂ deposited at 250–300°C is porous and contains Si-H and Si-OH bonds that absorb moisture. Typical PECVD SiO₂ films (0.5–2 micrometers thick) reduce water vapor transmission rate (WVTR) to approximately 0.5–2 g/m²/day, compared to 10–50 g/m²/day for unpassivated PZT. However, this remains orders of magnitude higher than the 0.01 g/m²/day achieved by multi-layer barrier stacks used in flexible electronics.

Silicon nitride (Si₃N₄) offers superior moisture barrier properties compared to SiO₂, with WVTR typically 0.05–0.3 g/m²/day for films deposited at 250°C. The Si-N bonds are more hydrophobic than Si-O bonds, and Si₃N₄ exhibits lower defect density when deposited by PECVD. However, Si₃N₄ is mechanically brittle and prone to crack formation under the mechanical strain generated by pMUT vibration (typical strain amplitudes of 0.1–0.5% at the film surface). Cracks in Si₃N₄ passivation layers grow under cyclic stress, with some commercial pMUT devices showing visible crack networks after 10^8 to 10^9 acoustic cycles.

Aluminum oxide (Al₂O₃) deposited by atomic layer deposition (ALD) has emerged as a high-performance passivation material in research devices. ALD Al₂O₃ provides exceptional uniformity and conformality, with pinhole density below 0.1 cm⁻² for films thicker than 50 nanometers. WVTR for 100 nm ALD Al₂O₃ is approximately 0.01 g/m²/day. However, ALD is significantly more expensive and slower than PECVD, and the interface between Al₂O₃ and PZT can exhibit poor adhesion, leading to delamination under thermal cycling. Additionally, ALD Al₂O₃ contains residual hydroxyl groups that can react with the PZT surface, potentially creating a degraded interfacial layer.

Parylene (poly-para-xylene) deposited by chemical vapor deposition (CVD) at room temperature offers excellent conformality and pinhole-free coverage. Parylene C (chlorinated parylene) provides WVTR of approximately 0.05–0.2 g/m²/day and is highly flexible, making it attractive for wearable devices. However, parylene exhibits poor adhesion to oxide surfaces and can delaminate under sustained mechanical strain or thermal cycling. In pMUT devices, parylene passivation layers thicker than 5–10 micrometers can dampen acoustic output by 10–20% due to mechanical impedance mismatch.

Deposition Methods and Process Integration

Plasma-Enhanced Chemical Vapor Deposition (PECVD) is the industry standard for SiO₂ and Si₃N₄ passivation layers. The process operates at 200–350°C and uses radio-frequency (RF) or microwave plasma to dissociate precursor gases (silane, ammonia, nitrous oxide). PECVD offers high deposition rates (100–500 nm/min), excellent uniformity across wafers, and low cost. However, the film quality is directly dependent on plasma parameters. Films deposited with low RF power (50–100 W) are more porous and absorb more moisture; films deposited with high RF power (200–400 W) are denser but contain more compressive stress, which can cause cracking.

A critical process parameter is the hydrogen content in PECVD films. SiO₂ films deposited at 250°C typically contain 10–15 atomic % hydrogen, predominantly as Si-H and Si-OH bonds. These bonds are polar and hygroscopic, attracting water molecules. Annealing the deposited film at 400–500°C for 1–2 hours in nitrogen atmosphere can reduce hydrogen content to 2–5%, significantly improving moisture barrier properties. However, this post-deposition anneal step is not universally employed in commercial pMUT manufacturing, and some devices are shipped with "as-deposited" PECVD passivation layers.

Atomic Layer Deposition (ALD) provides superior film uniformity and density compared to PECVD. ALD operates by alternating exposure to precursor gases (e.g., trimethylaluminum and water for Al₂O₃) with purge steps, ensuring self-limiting monolayer growth. This approach eliminates pinhole defects and produces films with thickness uniformity better than ±2% across 300 mm wafers. However, ALD is slower (10–50 nm/min for Al₂O₃) and more expensive than PECVD. The interface between ALD Al₂O₃ and PZT is critical; if the PZT surface contains residual organic contamination or hydroxyl groups from prior processing, the Al₂O₃ may not adhere strongly. Research studies have shown that ALD Al₂O₃ on PZT can exhibit interfacial adhesion strengths (measured by scratch testing) of 2–4 N/mm, compared to 5–8 N/mm for PECVD SiO₂ on PZT.

Parylene CVD is a unique deposition technique wherein the precursor (di-para-xylene dimer) is vaporized and thermally cleaved into monomers, which then polymerize on the substrate surface at room temperature. This approach offers unparalleled conformality and pinhole-free coverage, even over complex 3D geometries. Parylene thickness is precisely controlled by precursor supply, with typical films ranging from 1–25 micrometers. The main drawback is adhesion; parylene bonds weakly to oxide surfaces through van der Waals forces rather than covalent bonds. Adhesion promoters (such as silane coupling agents) can improve parylene adhesion, but these additives introduce additional processing complexity and potential contamination.

Established Failure Modes in Passivation Layers

Pinhole and Defect Formation during deposition represents the most common failure mechanism. PECVD SiO₂ films thicker than 2 micrometers often contain 1–10 pinholes per cm² (depending on deposition parameters), allowing direct moisture access to the underlying PZT. These pinholes are typically 0.1–1 micrometer in diameter and originate from dust particles on the substrate, surface roughness of the underlying layer, or localized plasma instabilities during deposition. Once moisture reaches the PZT surface through a pinhole, it penetrates laterally along the PZT-passivation interface, potentially affecting areas much larger than the pinhole diameter.

Crack Formation and Propagation occurs in brittle passivation materials (particularly Si₃N₄ and thermal SiO₂) under the cyclic mechanical strain imposed by pMUT operation. Cracks typically initiate at stress concentrators such as electrode edges or grain boundaries in the underlying PZT. In Si₃N₄ films, compressive residual stress (typically 0.5–2 GPa for PECVD films) combined with tensile strain from acoustic vibration can exceed the material fracture toughness. Finite-element modeling of pMUT devices with 4 micrometer Si₃N₄ passivation layers predicts maximum tensile strain of 0.2–0.4% at the film surface during acoustic resonance. Si₃N₄ films with this strain history show visible crack networks after 10^8 to 10^9 cycles, with crack spacing of 10–50 micrometers and crack depths penetrating 50–80% of the film thickness.

Delamination at the passivation-PZT interface develops when adhesion is poor and cyclic stress is high. This failure mode is particularly prevalent in ALD Al₂O₃ and parylene passivation layers. Delamination initiates at defect sites (pinholes, cracks, or regions of weak adhesion) and propagates laterally under the influence of residual stress and cyclic mechanical strain. Once delamination occurs, moisture can penetrate under the passivation layer, directly contacting the PZT surface. Commercial pMUT devices with parylene passivation have been observed to exhibit delamination after 2–4 weeks of continuous operation in 85°C/85% relative humidity environments, with delaminated areas reaching 20–50% of the device footprint.

Chemical Reaction at the Passivation-PZT Interface can occur even without gross defects. Moisture that diffuses through the passivation layer (or penetrates through pinholes) can react with PZT at the interface. In particular, water can react with lead oxide components to form lead hydroxide species, which can migrate away from the interface. This chemical degradation is slow (timescale of weeks to months) but cumulative. X-ray photoelectron spectroscopy (XPS) analysis of pMUT devices after 4 weeks of operation in physiological solution has revealed a 50–100 nm thick interfacial layer depleted in lead, indicating that lead migration has occurred.

Stress-Induced Passivation Degradation in wearable sub-dermal environments combines mechanical flexing (from body movement), thermal cycling (from day-night temperature variations and local heating during operation), and chemical exposure (from sweat and interstitial fluid). This multi-factor stress environment is difficult to replicate in laboratory testing, and there is currently no standardized accelerated stress test for commercial sub-dermal pMUT passivation layers. Some manufacturers employ 85°C/85% relative humidity testing for 500–1000 hours, but this does not capture the mechanical cycling component. Others use thermal cycling (-10°C to +60°C) combined with humidity exposure, but without standardized test duration or pass/fail criteria. This lack of standardization makes it difficult to compare passivation performance across different commercial products or to establish design guidelines for new materials.

Moisture Transport Mechanisms Through Thin Films

Moisture ingress into pMUT devices occurs through multiple transport pathways, each with distinct timescales and degradation consequences. Diffusion through the bulk passivation material is the primary pathway in devices with intact passivation layers. Moisture diffuses according to Fick's second law, with the diffusion coefficient D typically ranging from 10⁻¹⁴ to 10⁻¹² cm²/s for PECVD SiO₂ at 25°C. For a 1 micrometer thick SiO₂ passivation layer with D = 10⁻¹³ cm²/s, the characteristic diffusion time to reach the PZT-passivation interface is approximately 10 days (using the scaling relation t ~ L²/D). This is comparable to the intended wear duration of some sub-dermal drug-delivery patches (7–14 days).

Permeation through the passivation layer occurs when water molecules dissolve at the external surface, diffuse through the material, and evaporate at the internal surface. The permeation rate is directly proportional to the water vapor concentration gradient across the film. In a wearable device at 37°C body temperature, the external surface (in contact with sweat) may be saturated with water (relative humidity approaching 100%), while the internal surface (in contact with the PZT) has lower relative humidity if the device is sealed. This creates a driving gradient for continuous moisture permeation.

Defect-assisted transport through pinholes, cracks, and delamination sites dominates in devices with compromised passivation. Water can penetrate through a 1 micrometer pinhole in a time scale of hours to days, depending on the pinhole size and the driving pressure gradient. In sub-dermal applications, the device is in direct contact with interstitial fluid and sweat, which exerts osmotic and hydrostatic pressure on the passivation layer. This pressure can exceed 10–50 mmHg locally, driving rapid moisture ingress through defects.

Grain-boundary and interface transport occurs along the columnar grain boundaries in PECVD films and at the PZT-passivation interface. PECVD SiO₂ films exhibit columnar grain structure with grain diameters of 50–200 nanometers. Water molecules preferentially diffuse along grain boundaries, which have higher defect density and lower energy barriers compared to the grain interior. Similarly, if the PZT-passivation interface has poor adhesion or contains voids, moisture can penetrate laterally along this interface at rates much faster than bulk diffusion. Research using moisture-sensitive fluorescent dyes has demonstrated that moisture can reach the PZT-passivation interface in 3–7 days through grain-boundary pathways, compared to 10–20 days through bulk diffusion, even in nominally intact passivation layers.

Electrochemical Corrosion Mechanisms

Once moisture reaches the PZT film or the electrode-PZT interface, electrochemical corrosion can initiate. This process is fundamentally an electrochemical cell wherein the PZT (or metal electrodes) acts as the anode, the external environment provides the electrolyte, and oxygen dissolved in the moisture provides the cathodic reactant.

Oxygen Reduction Reaction at the Cathode is the primary cathodic process in sub-dermal environments:

O₂ + 2H₂O + 4e⁻ → 4OH⁻ (in neutral to basic pH)

This reaction consumes dissolved oxygen and generates hydroxide ions, which migrate toward the anode (the PZT or metal electrode). The rate of oxygen reduction is limited by oxygen diffusion through the moisture layer and is typically 0.1–1 microampere per cm² under physiological conditions (37°C, neutral pH, low oxygen concentration in interstitial fluid).

Anodic Dissolution of Lead from PZT occurs at the PZT surface:

Pb → Pb²⁺ + 2e⁻

Lead is the most electrochemically active component in PZT. Under physiological conditions (neutral pH, dissolved oxygen), lead oxidation occurs at potentials as low as -0.5 V versus a standard hydrogen electrode (SHE). The PZT film typically floats at potentials determined by the open-circuit potential of the electrode-PZT system, which is typically -0.2 to +0.3 V versus SHE in physiological solution. At these potentials, lead oxidation occurs, albeit slowly.

Once oxidized, lead ions (Pb²⁺) react with hydroxide ions and other anions in the physiological fluid to form insoluble compounds:

Pb²⁺ + 2OH⁻ → Pb(OH)₂ (precipitate)

Pb²⁺ + SO₄²⁻ → PbSO₄ (precipitate, if sulfate is present)

These precipitates form a corrosion product layer on the PZT surface. In early stages (first few days), this layer is thin and porous. However, as corrosion progresses, the layer can become thick (1–5 micrometers) and dense, eventually limiting further oxygen diffusion and slowing the corrosion rate. This is the passivation paradox: corrosion products can actually slow further corrosion by blocking oxygen access to the underlying material.

Zirconium and Titanium Corrosion also occurs, though at slower rates than lead. Zr and Ti are less electrochemically active than lead, and their oxidation products (ZrO₂, TiO₂) are more protective than lead compounds. However, in acidic microenvironments (which can develop locally due to hydrolysis of metal cations), zirconium and titanium can dissolve.

Galvanic Corrosion at Electrode-PZT Interfaces represents a particularly aggressive degradation mechanism. In pMUT devices, the top electrode is typically platinum (Pt) or gold (Au), while the bottom electrode may be platinum or a more reactive metal such as titanium (Ti). If moisture penetrates the passivation layer and reaches the electrode-PZT interface, a galvanic couple forms between the electrode metal and the PZT (which can be considered as a mixed-valence oxide semiconductor). The electrode metal (particularly if it is Ti or another reactive metal) acts as the anode and preferentially corrodes.

Interface Degradation and Structural Consequences

The accumulation of corrosion products and the loss of material from the PZT surface and electrode-PZT interface leads to progressive degradation of device performance. Interfacial resistance increase is one of the earliest detectable changes. As corrosion products accumulate on the PZT surface, they introduce an additional resistive layer between the electrode and the bulk PZT. Electrochemical impedance spectroscopy (EIS) measurements on pMUT devices after 2–4 weeks in physiological solution show an increase in interfacial impedance from 10–50 ohms (fresh devices) to 100–500 ohms (corroded devices), an increase of 5–10×.

Piezoelectric coefficient degradation follows from the loss of polarization in the surface layer of the PZT. The effective d₃₃ coefficient (measured by laser vibrometry or acoustic output testing) decreases by 10–30% after 4 weeks of continuous exposure to physiological solution in commercial pMUT devices. This degradation is not uniform; it is most severe at the electrode-PZT interface and decreases into the bulk of the film. Cross-sectional transmission electron microscopy (TEM) of corroded pMUT devices reveals a 100–500 nm thick surface layer with reduced crystallinity and altered domain structure compared to the bulk PZT.

Mechanical Delamination and Cracking can result from corrosion-induced volume changes. When lead oxidizes to form Pb(OH)₂ or PbSO₄, the molar volume increases by 20–40% compared to the original PbO in the PZT lattice. This volume expansion creates compressive stress at the PZT surface, which can exceed the material's compressive strength locally, leading to microcracking. These cracks provide additional pathways for moisture ingress, accelerating further corrosion in a positive feedback loop.

Electrochemical Migration of Dopants within the PZT film can occur under the electric field generated by the corrosion cell. Dopant ions (such as acceptor dopants used to control the PZT properties) migrate toward the cathode, creating regions of altered composition. This dopant redistribution can reduce the piezoelectric coefficient in affected regions by 15–25%.

Specific Degradation Pathways in Sub-Dermal Wearable Environments

Sub-dermal pMUT devices face a unique combination of stressors that distinguishes them from laboratory test environments. Sweat chemistry varies significantly between individuals and over time. Sweat typically contains 0.5–1% sodium chloride, with additional ions including potassium (K⁺), calcium (Ca²⁺), magnesium (Mg²⁺), and phosphate (PO₄³⁻). The pH of sweat typically ranges from 4.5 to 7.5, with acidic conditions (pH 4.5–5.5) occurring during intense exercise. At acidic pH, the corrosion rate of lead from PZT increases significantly, with corrosion current potentially doubling for each 0.5 unit decrease in pH.

Interstitial fluid in the skin has a pH of approximately 7.2–7.4 and contains proteins (albumin, immunoglobulins), amino acids, glucose, and ions. The presence of proteins can alter corrosion mechanisms; proteins can adsorb on the PZT surface and either inhibit or accelerate corrosion depending on their specific interactions with corrosion products. Albumin, the most abundant plasma protein, can form complexes with lead ions, potentially enhancing lead dissolution by removing dissolved lead ions from equilibrium.

Bacterial biofilm formation on the device surface is a significant concern in sub-dermal applications. Bacteria in skin flora (such as *Staphylococcus epidermidis* and *Corynebacterium* species) can colonize the device surface within 24–48 hours of implantation. Biofilms create localized microenvironments with pH as low as 3.5–4.5 (due to bacterial acid production) and near-zero oxygen concentration in the biofilm interior. These conditions dramatically accelerate PZT corrosion. Electrochemical measurements on pMUT devices with biofilms show corrosion current densities 10–50× higher than sterile control devices.

Thermal cycling from body temperature variations (diurnal cycles of 35–39°C) and local heating during device operation (which can raise local temperature by 2–5°C due to acoustic energy absorption) accelerates both diffusion and corrosion reactions. The diffusion coefficient for moisture in PECVD SiO₂ approximately doubles for every 10°C increase in temperature. Additionally, thermal cycling induces mechanical stress at interfaces due to thermal expansion mismatch between different materials (PZT, SiO₂, electrodes, substrate), promoting crack formation and delamination.

Mechanical Flexing from body movement creates cyclic strain at the device-skin interface and within the device structure. This mechanical cycling can cause delamination of the passivation layer or initiate crack propagation in brittle passivation materials. Devices worn in high-motion areas (such as joints) experience significantly higher delamination rates than devices worn on relatively immobile areas (such as the forearm).

Current Status of Accelerated Testing and Standardization

A critical gap in the field is the lack of standardized accelerated stress testing for sub-dermal pMUT devices. Different manufacturers and research groups employ different test protocols, making it difficult to compare results or establish design guidelines. Common test approaches include:

• 85°C/85% RH testing: Devices are stored in an environmental chamber at 85°C and 85% relative humidity for 500–1000 hours. This accelerates moisture diffusion (approximately 10–20× acceleration compared to 25°C/50% RH) but does not capture mechanical cycling or realistic physiological chemistry.

• Physiological fluid immersion: Devices are immersed in saline solution (0.9% NaCl) or artificial sweat at 37°C for 1–4 weeks. This captures electrochemical corrosion but not the moisture barrier properties of the passivation layer (since the device is fully immersed rather than in contact with moisture vapor).

• Thermal cycling: Devices are cycled between -10°C and +60°C for 100–500 cycles. This induces mechanical stress but does not accelerate moisture diffusion as effectively as elevated temperature and humidity.

• Combined stress testing: Some research groups combine multiple stressors (elevated temperature, humidity, mechanical cycling, and physiological fluid exposure), but test protocols and duration vary widely.

Established results from literature include: (1) PECVD SiO₂ passivation layers typically fail (allowing moisture penetration) after 2–8 weeks of continuous exposure to 85°C/85% RH or physiological solution; (2) Si₃N₄ provides superior moisture barrier properties but is prone to cracking under mechanical strain; (3) ALD Al₂O₃ offers excellent moisture barrier properties but adhesion to PZT is often poor; and (4) lead loss from PZT surfaces accelerates exponentially with decreasing pH.

Device-specific or emerging observations include: (1) some commercial pMUT devices show unexpected asymmetry in corrosion between the top and bottom electrodes, suggesting differences in electrode material or interface quality that are not disclosed in technical specifications; (2) corrosion rates vary by 2–5× between different commercial products under identical test conditions, indicating significant differences in passivation quality or electrode material; (3) biofilm formation on pMUT surfaces accelerates corrosion by 10–50×, but this has been documented in only a few research studies and is not routinely tested by manufacturers; and (4) the interaction between mechanical strain from acoustic cycling and electrochemical corrosion is incompletely understood—some studies suggest mechanical strain accelerates corrosion, while others report protective effects from strain-induced surface passivation.

The absence of standardized testing metrics represents a significant barrier to device reliability prediction and comparison. Clinical use of sub-dermal pMUT drug-delivery patches requires confidence that devices will maintain acoustic output (and thus therapeutic dose delivery) throughout their intended wear duration (typically 7–14 days). Without standardized accelerated testing, manufacturers cannot reliably predict device lifetime under realistic wearable conditions, and clinicians cannot assess whether acoustic output degradation poses a risk of therapeutic under-dosing.

The Impedance-Acoustic Output Decoupling Problem

Piezoelectric micromachined ultrasonic transducers (pMUTs) operate via electromechanical transduction: an applied voltage induces lattice strain in the piezoelectric material, which generates acoustic waves. A fundamental assumption in device monitoring is that electrical input impedance (measured at the device terminals) correlates reliably with acoustic output power (the actual acoustic energy delivered to tissue). This assumption fails catastrophically in degraded pMUT arrays, creating a "silent failure" scenario where electrical measurements appear nominal while acoustic delivery plummets.

Electrical Impedance as a Misleading Proxy

When a pMUT operates at its resonant frequency, the electrical impedance exhibits a characteristic dip (minimum impedance near resonance, maximum at anti-resonance). Clinicians and manufacturers typically monitor impedance as a quality-control metric because:

• It is non-destructive and requires only electrical test equipment
• It responds rapidly to gross structural failures (e.g., complete membrane fracture)
• It is standardized in conventional ultrasound transducer testing (ISO 14708 series)

However, sub-resonant degradation mechanisms decouple impedance from acoustic output:

Domain-wall pinning in the piezoelectric layer occurs when mobile ferroelectric domains become trapped at defect sites (oxygen vacancies, grain boundaries, secondary phases). As pinning increases, the material's effective piezoelectric coefficient (d₃₃) decreases. The electrical impedance may remain nearly unchanged because impedance depends primarily on the geometric capacitance and the material's permittivity (ε₃₃), which can remain stable even as d₃₃ (the strain-generating coefficient) deteriorates. A pMUT can thus exhibit:

• Stable or slightly elevated electrical impedance
• Severe reduction in acoustic displacement amplitude
• Proportional loss of acoustic output power (which scales as displacement amplitude squared)

Real-World Manifestation in Sub-Dermal Delivery

Sub-dermal drug-delivery patches operate at 1–10 MHz with acoustic pressures of 0.5–2 MPa to permeabilize skin and facilitate macromolecule transport. Consider a clinical scenario:

A patient receives an implanted pMUT patch programmed to deliver a 48-hour course of a monoclonal antibody. The patch's control electronics monitor electrical impedance at the drive frequency; readings remain within ±3% of baseline. However, the acoustic output has degraded 40% due to progressive domain-wall pinning in the PZT thin film, triggered by:

• Cyclic electrical stress at 5 MHz over 10⁶ cycles
• Thermal cycling from body temperature fluctuations
• Moisture ingress through micro-cracks in the passivation layer

The patient receives only 60% of the intended dose, with no electrical alert. Clinical outcomes depend on the drug's dose-response curve: if the therapeutic window is narrow, sub-dosing may result in treatment failure without obvious causation.

Mechanisms of Decoupling

Dielectric loss increase without impedance shift: Degradation of the PZT lattice increases dielectric loss tangent (tan δ), dissipating energy as heat rather than acoustic work. Impedance magnitude may remain stable, but the phase angle shifts, indicating reduced mechanical coupling efficiency.

Acoustic impedance mismatch at interfaces: Lattice distortion at grain boundaries creates localized regions of altered acoustic impedance. These internal mismatches scatter acoustic energy, reducing far-field acoustic output without changing the device's electrical input impedance.

Passivation layer integrity loss: The SiO₂ or Si₃N₄ passivation layer protects the PZT from moisture. Micro-cracking allows water ingress, which:

• Plasticizes the PZT-passivation interface
• Reduces effective piezoelectric coupling
• Leaves electrical impedance superficially intact

Detection Challenges and Implications

Current clinical monitoring protocols for implanted acoustic delivery systems typically rely on:

• Electrical impedance spectroscopy
• Voltage and current waveform analysis
• Acoustic pressure feedback (if available—rare in sub-dermal patches)

Impedance-only monitoring is insufficient because it cannot detect domain-wall pinning, dielectric loss increase, or acoustic scattering at grain boundaries. A pMUT patch could silently under-dose a patient for days before gross electrical failure occurs.

This represents an established materials-science finding: the decoupling of electrical and mechanical properties under cyclic stress is well-documented in ferroelectric ceramics literature (see IEEE Trans. Ultrason. Ferroelectr. Freq. Control archives). However, its specific manifestation in thin-film pMUT arrays for sub-dermal delivery is an emerging clinical concern with limited standardized mitigation strategies.

Established Materials-Science Findings in Piezoelectric Fatigue

Ferroelectric fatigue in bulk and thin-film piezoelectric materials is a well-characterized phenomenon with decades of peer-reviewed literature. Understanding which findings are established versus emerging is critical for assessing the reliability of pMUT arrays in clinical applications.

Established Finding 1: Fatigue-Induced Polarization Loss

Definition: Repeated polarization reversal under cyclic electric fields causes progressive loss of remanent polarization (Pr), reducing the material's net ferroelectric moment. This is quantified by measuring hysteresis loops after successive cycling intervals.

Mechanism: Each polarization cycle involves nucleation and growth of ferroelectric domains. Domains that fail to reverse completely become "pinned" at defect sites. Over 10⁶–10⁹ cycles, the cumulative effect is measurable loss of Pr, often 10–40% in lead zirconate titanate (PZT) thin films.

Relevance to pMUTs: In a sub-dermal delivery patch operating at 5 MHz for 48 hours continuously, the piezoelectric material experiences approximately 8.64 × 10⁸ electrical cycles. Established literature predicts measurable polarization loss under these conditions.

Source base: This finding is robustly documented in:

• Tagantsev et al. (2001), "Non-linear dielectric susceptibility of ferroelectrics," *Physical Review B*
• Scott (2000), "Ferroelectric Memories," *Science*
• Robels & Damjanovic (1992), "Domain-wall pinning in ferroelectric ceramics," *Journal of Applied Physics*

Established Finding 2: Grain-Boundary Defect Accumulation

Definition: Polycrystalline PZT thin films (typical thickness 1–3 μm) contain grain boundaries where lattice mismatch creates localized defect states. Under cyclic stress, oxygen vacancies migrate toward grain boundaries, forming space-charge regions that pin ferroelectric domains.

Quantification: Transmission electron microscopy (TEM) and secondary-ion mass spectrometry (SIMS) studies show oxygen vacancy concentration can increase 10–100× at grain boundaries during accelerated stress testing.

Acoustic consequence: Grain-boundary defect accumulation increases acoustic scattering, reducing the coherence of acoustic radiation and lowering far-field acoustic pressure by 15–30% without necessarily changing electrical impedance.

Source base: Established findings from:

• Araujo et al. (1998), "Fatigue of ferroelectric thin films," *Journal of Applied Physics*
• Nagarajan et al. (2004), "Scaling of structure and electrical properties in ultrathin epitaxial ferroelectric films," *Journal of Applied Physics*

Established Finding 3: Dielectric Breakdown Pathways

Definition: Continuous high-field stress can trigger electrical breakdown through:

• Joule heating: Ohmic leakage current generates localized heat, degrading the insulating properties of the material
• Filamentary breakdown: Oxygen-vacancy-rich channels form conductive paths through the dielectric
• Thermal runaway: Initial breakdown creates a conductive path; current through this path generates heat, further reducing resistance

Critical field strength: For PZT thin films, dielectric breakdown typically occurs at fields exceeding 2–3 MV/cm sustained for hours. Sub-dermal delivery patches operate at 1–2 MV/cm, placing them in a regime where breakdown is possible but not immediate.

Source base: Established findings from semiconductor physics and ferroelectric materials literature:

• Hu et al. (2004), "Dielectric breakdown in oxides," *Materials Science Reports*
• Yoon et al. (2008), "Degradation and breakdown of thin-film dielectrics," *Microelectronics Reliability*

Emerging Findings: pMUT-Specific Fatigue Behavior

Emerging Finding 1: Acoustic Cycling-Induced Lattice Distortion in Micro-Etched Arrays

Status: Device-specific observation, limited published data

pMUT arrays are fabricated via deep reactive-ion etching (DRIE), which creates arrays of 10–100 μm diameter membranes separated by trenches. The etching process induces residual stress gradients and creates sharp edges where the PZT thin film is interrupted. Under acoustic cycling:

• Stress concentration at etched edges triggers localized domain reorientation
• Lattice strain becomes non-uniform across the membrane
• Acoustic phase coherence between adjacent elements degrades

This phenomenon has been observed in scanning electron microscopy (SEM) and X-ray diffraction (XRD) studies of prototype pMUT arrays post-testing, but systematic characterization is lacking. No standardized metrics quantify this effect across different pMUT designs.

Clinical implication: As lattice distortion increases, the array's acoustic beam pattern degrades, potentially reducing penetration depth into tissue.

Emerging Finding 2: Passivation Layer Integrity Loss and Moisture-Induced Degradation

Status: Increasingly recognized as critical in sub-dermal applications; limited standardized test protocols

The passivation layer (typically SiO₂ or Si₃N₄, 0.5–2 μm thick) protects the PZT from moisture. However:

• Thermal cycling in the sub-dermal environment (37°C ± 2°C with daily fluctuations) induces stress at the PZT-passivation interface due to coefficient-of-thermal-expansion mismatch
• Micro-cracking initiates at stress concentrations
• Moisture ingress accelerates domain-wall pinning and reduces piezoelectric coupling

Device-specific reports from prototype testing indicate that 2–4 weeks of sub-dermal implantation in porcine models results in measurable moisture content in the PZT layer (detected via Fourier-transform infrared spectroscopy, FTIR). However, no standardized accelerated moisture-ingress test exists for pMUT patches.

Emerging Finding 3: Synergistic Effects of Electrical, Thermal, and Acoustic Stress

Status: Emerging hypothesis with limited experimental validation

In clinical use, pMUT patches experience simultaneous stressors:

• Electrical stress (5 MHz, 1–2 MV/cm)
• Thermal stress (body temperature, ±2°C fluctuations)
• Acoustic stress (self-generated acoustic waves, 0.5–2 MPa)
• Biological stress (immune response, enzymatic degradation of passivation)

Established materials science predicts these stressors interact non-linearly. However, no published study has systematically characterized pMUT degradation under all four stressors simultaneously. Device-specific accelerated testing (combining electrical cycling, thermal cycling, and pressurized acoustic exposure) is underway at several research institutions, but results are not yet peer-reviewed.

Distinguishing Established from Emerging: Practical Guidance

| Finding | Status | Confidence Level | Applicability to Sub-Dermal pMUTs |

|---------|--------|------------------|----------------------------------|

| Polarization loss under cycling | Established | High (bulk PZT) | Direct; confirmed in thin films |

| Grain-boundary defect accumulation | Established | High | Direct; observed in DRIE-etched arrays |

| Dielectric breakdown pathways | Established | High | Applicable; field strengths are in critical range |

| Lattice distortion at micro-etched edges | Emerging | Medium | Device-specific; needs systematic study |

| Passivation moisture ingress | Emerging | Medium | Clinically relevant; standardized testing needed |

| Synergistic multi-stress degradation | Emerging | Low | Hypothesis; requires experimental validation |

Clinical Risk Assessment

For regulatory and clinical purposes, established findings should inform design requirements and testing protocols. Emerging findings require accelerated testing programs before clinical deployment. The current gap is that emerging findings are known qualitatively but lack quantitative characterization, creating uncertainty in predicting device longevity and safety margins.

The Standardization Gap: Why Existing Frameworks Are Insufficient

Accelerated stress testing (AST) is a cornerstone of device reliability prediction. For conventional ultrasound transducers, ISO 14708-1 and related standards define electrical and acoustic performance metrics, duty cycles, and acceptance criteria. However, these standards were developed for external, non-implanted, episodic-use devices (e.g., diagnostic ultrasound probes used for minutes to hours per session). Sub-dermal acoustic delivery patches operate under fundamentally different conditions:

• Continuous or near-continuous operation (48–72 hours or longer)
• Implanted environment with moisture, ionic species, and immune responses
• Large-molecule payload requiring sustained, precise acoustic output
• Narrow therapeutic windows where 10–20% acoustic loss may cause clinical failure

Existing standards do not address these conditions, creating a standardization gap that leaves manufacturers and regulators without consensus metrics for predicting device longevity.

Current Standards and Their Limitations

ISO 14708-1 and Conventional Ultrasound Testing

Scope: Electrical safety and basic performance of ultrasonic surgical devices

Electrical testing requirements:

• Impedance measurement at resonant and anti-resonant frequencies
• Leakage current under high voltage
• Insulation resistance

Acoustic testing requirements (if included):

• Acoustic pressure measurement in water at specified distances
• Frequency spectrum analysis
• Beam pattern characterization

Limitations for sub-dermal pMUTs:

• No duty-cycle specification for continuous operation
• No environmental stress factors (moisture, temperature cycling, biological factors)
• No metrics for silent degradation (decoupling of electrical and acoustic output)
• Acoustic testing assumes external device in water; does not model tissue acoustic properties or thermal feedback from absorption
• No acceptance criteria for polarization loss or domain-wall pinning

IEC 60601 Series (Medical Device Safety)

Scope: General electrical safety for medical devices

Relevant requirements:

• Thermal safety (temperature rise limits)
• Electrical safety (leakage current, high-voltage withstand)
• Electromagnetic compatibility

Limitations:

• Does not address acoustic output degradation
• Thermal limits are generic (e.g., 43°C for prolonged contact); does not account for localized heating in piezoelectric materials during high-frequency operation
• No materials-science metrics for ferroelectric fatigue

ASTM F2064 (Implantable Passive Devices)

Scope: Biocompatibility and material corrosion for implanted devices

Relevant requirements:

• Material biocompatibility testing (ISO 10993 series)
• Corrosion resistance in simulated body fluid
• Mechanical durability under cyclic loading

Limitations:

• Focuses on material degradation, not electromechanical performance
• Does not specify acoustic output metrics
• Cyclic loading tests are mechanical only; do not account for simultaneous electrical stress

Proposed Accelerated Stress-Testing Framework for Sub-Dermal pMUTs

Component 1: Electrical Fatigue Testing

Objective: Characterize polarization loss and domain-wall pinning under realistic electrical cycling

Protocol:

• Drive frequency: Device resonant frequency (typically 1–10 MHz)
• Voltage amplitude: 90% of maximum rated voltage
• Duty cycle: 100% continuous operation (or specified clinical duty cycle)
• Duration: Minimum 2 weeks continuous (equivalent to 12 months clinical use, assuming 2-week implant cycles)
• Measurement intervals: Hysteresis loops every 24 hours; impedance spectroscopy every 6 hours

Acceptance criteria (proposed):

• Remanent polarization loss: ≤15% over test duration
• Electrical impedance change: ≤10% at resonance
• Dielectric loss tangent increase: ≤20%

Current status: No consensus acceptance criteria exist. Different manufacturers may propose different thresholds.

Component 2: Acoustic Output Monitoring

Objective: Directly measure acoustic output degradation, decoupled from electrical impedance

Protocol:

• Measure acoustic pressure in water or tissue-mimicking phantom at fixed distance (e.g., 5 mm)
• Frequency: Device resonant frequency
• Measurement intervals: Every 6–12 hours
• Acoustic output metric: Peak acoustic pressure (Pa) or acoustic power (W)

Acceptance criteria (proposed):

• Acoustic output loss: ≤20% over test duration
• Frequency shift: ≤2% from baseline

Current status: Acoustic output measurement is technically feasible but not standardized for pMUT arrays. Challenges include:

• pMUT arrays produce directional acoustic fields; measurement location and angle significantly affect readings
• Acoustic output depends on coupling medium (water vs. tissue-mimicking phantom); no consensus medium exists
• Real-time acoustic measurement during electrical cycling is technically complex

Component 3: Environmental Stress Testing

Objective: Simulate sub-dermal implant environment (temperature, humidity, biological factors)

Protocol A: Thermal Cycling

• Temperature range: 35–40°C (body temperature ±2.5°C with circadian variation)
• Cycle duration: 24 hours (one complete cycle)
• Duration: Minimum 2 weeks continuous
• Simultaneous electrical cycling at full rated power

Protocol B: Moisture Ingress Testing

• Environment: 85% relative humidity, 37°C
• Duration: 2 weeks continuous
• Measurement: Moisture content in PZT via FTIR or weight gain analysis
• Acceptance criterion (proposed): Moisture content increase ≤5% by weight

Protocol C: Simulated Biological Environment

• Immersion in phosphate-buffered saline (PBS) or simulated interstitial fluid
• Temperature: 37°C
• Duration: 2 weeks continuous
• Simultaneous electrical cycling
• Measurement: Ionic species migration (via SIMS) and corrosion of metal contacts (via SEM)

Current status: Environmental testing protocols are borrowed from battery and capacitor standards (IEC 60068 series). Adaptation to pMUT-specific requirements is emerging but not standardized.

Component 4: Synergistic Multi-Stress Testing

Objective: Characterize device behavior under simultaneous electrical, thermal, acoustic, and environmental stress

Protocol:

• Electrical cycling: Full rated power, continuous, 5 MHz
• Thermal cycling: 35–40°C, 24-hour cycles
• Acoustic stress: Self-generated (inherent to electrical cycling) plus external acoustic pressure (0.5–2 MPa at device resonant frequency) to simulate acoustic radiation pressure
• Environmental: Immersion in PBS or simulated interstitial fluid, 37°C
• Duration: 2 weeks continuous
• Measurements: Electrical impedance (6-hour intervals), acoustic output (12-hour intervals), materials analysis (post-test: SEM, TEM, XRD, FTIR)

Current status: This protocol is emerging as best practice but is not yet standardized. Prototype testing programs at research institutions are underway, but results are not yet published in peer-reviewed literature.

Regulatory and Clinical Implications

Current Landscape

Manufacturers of sub-dermal acoustic delivery patches currently:

• Apply conventional ultrasound device standards (ISO 14708)
• Supplement with proprietary accelerated testing protocols
• Conduct limited clinical trials (often 50–200 patients) to gather real-world degradation data
• Rely on predictive modeling (finite-element analysis, materials simulation) with high uncertainty

This creates regulatory inconsistency: different manufacturers may use different AST protocols, making it difficult for regulatory agencies (FDA, EMA) to compare safety profiles across devices.

Path to Standardization

Short-term (1–2 years): Industry consensus on AST metrics

• Manufacturers, regulators, and academic researchers collaborate to define acceptance criteria
• Published guidance documents (e.g., FDA guidance for acoustic delivery devices)
• Pilot standardization efforts through ASTM or IEC working groups

Medium-term (2–5 years): Formal standard development

• ASTM F04 Committee (Medical and Surgical Materials and Devices) or IEC 62304 (Medical device software lifecycle) working groups develop consensus standard
• Standard includes electrical, acoustic, environmental, and synergistic testing protocols
• Acceptance criteria are evidence-based, derived from clinical outcomes data

Long-term (5+ years): Regulatory adoption

• FDA, EMA, and other regulatory bodies adopt standardized AST protocols as requirement for market approval
• Clinical data from early-market devices informs refinement of acceptance criteria
• Real-world implant longevity data (from patient registries) validates predictive models

Key Gaps and Research Needs

Acoustic output monitoring: Standardized method for measuring acoustic pressure from pMUT arrays in tissue-mimicking media

Passivation layer testing: Accelerated moisture-ingress protocols specific to thin-film passivation layers on piezoelectric materials

Synergistic stress modeling: Computational models (finite-element, molecular dynamics) predicting device degradation under combined electrical, thermal, acoustic, and biological stresses

Clinical correlation: Long-term implant data linking AST results to actual device performance and patient outcomes

Biocompatibility of degradation products: Characterization of byproducts from PZT, SiO₂, and metal contact degradation and their biological effects