Soft, full Wheatstone bridge 3D pressure sensors for cardiovascular monitoring

Cardiovascular diseases are leading causes of mortality, motivating continuous, effective monitoring solutions beyond wired, hospital-bound systems. Although advanced wearable technologies exist, challenges persist for capturing dynamic, time-dependent superficial pressures related to pulsatile blood flow in ways that are robust to ambient variations, particularly temperature. This work introduces a 3D piezoresistive sensor that uses a full Wheatstone bridge and optimized mechanical placement around the neutral plane to provide intrinsic temperature compensation and enhanced pressure sensitivity. Integrated with Bluetooth-enabled soft electronics, the platform enables continuous, skin-mounted monitoring of hemodynamic properties (e.g., pulse waveforms, correlations to blood pressure) at the wrist or neck for minimally disruptive home use.

Prior approaches in ambulatory cardiac monitoring often rely on wired, clinical systems. Soft wearable pressure sensors have been explored for beat-to-beat blood pressure monitoring, and skin-interfaced platforms demonstrated real-time sweat and hydration sensing. However, many piezoresistive sensors are temperature sensitive, necessitating compensation. Earlier epidermal devices mapped blood flow, and implantable or noninvasive platforms measured pressure and flow. The literature underscores a need for soft, skin-mounted pressure sensors with high sensitivity, low hysteresis, and insensitivity to ambient temperature fluctuations for accurate cardiovascular monitoring.

Device design and fabrication: A lithographically defined multilayer stack was built on glass with a PMMA sacrificial layer: bottom PI (4 µm), bottom Cr/Au serpentine strain gauges (5/45 nm; ~520 Ω, 3 µm linewidth), middle PI (4 µm), top Cr/Au serpentine gauges (5/50 nm; ~480 Ω, 3 µm linewidth), and top PI (4 µm). A 100 nm Cu hard mask patterned PI via oxygen plasma etching to define geometry. After PMMA release and transfer to a PVA tape, mechanically guided compressive buckling (biaxial prestretch ~30%) transformed the 2D precursor into a table-shaped 3D Multiple strain gauge Integrated Sensor (3MIS) with four legs hosting the gauges. The 3D mesostructure was encapsulated in Ecoflex 00-30 using a mold (diameter ~3 mm; height ~500 µm). A ring of acrylic and PI served as a frame to enhance mechanical stability and shear rejection; serpentine PI layers (SPL) improved robustness. Electrical connection used conductive silver epoxy onto a flexible PCB.

Sensing principle and electronics: The four gauges (R1–R4) form a full Wheatstone bridge. With VREF = 3.3 V, the bridge output is Vo = ((R1+R4)/(R2+R3))·VREF, balanced at Vo = 0 when R1R3 = R2R4. Normal pressure places top traces in tension (R1, R2 increase) and bottom traces in compression (R3, R4 decrease), maximizing differential output. An instrumentation amplifier (gain 6 V/V) produces Vp = 6·Vo. Calibration with a motorized force tester (Mark-10) established a linear relationship pressure ≈ 400×(Vp/VREF) over 0–200 mmHg. A quarter-bridge comparator used one microfabricated gauge (R4) and three commercial resistors (≈500 Ω, 0201) to assess temperature effects.

Temperature and pressure characterization: The full-bridge (FB) and quarter-bridge (QB) were characterized from 25–40 °C. For thermal tests, sensors were placed on a 30 °C hot plate (QB compensated to 0 mmHg at 30 °C). Applying 100 mmHg with a 23 °C metal disk induced thermal drift in QB (recovery >30 s), while FB exhibited negligible drift. Pressure sensitivity and linearity were quantified up to 200 mmHg (R² ≈ 0.995).

Shear rejection and durability: Finite element analysis (FEA) and benchtop shear tests (0.1 N lateral load along orthogonal axes) assessed shear sensitivity. Stress localized at the periphery of the encapsulated sensor/substrate interface, with negligible stress at the 3D sensing core. Sensors with SPL remained functional under 0.1 N shear and recovered to baseline upon unloading; sensors without SPL malfunctioned. Fatigue testing applied 0–100 mmHg normal pressure over 1000 cycles at loading speeds of 50 and 200 µm/s.

Phantom hemodynamics: A cardiac simulator drove pulsatile flow through silicone tubing (outer/inner diameters 4.5/3.0 mm; vessel elastomer E≈2 MPa) embedded beneath phantom skin (Ecoflex 00-30, E≈60 kPa). The 3MIS was preloaded (~35 mmHg) with a metal post and recorded superficial pressure at pulse rates 50, 70, 90, and 110 Hz over 120 s. A proportionality factor k, derived experimentally and dependent on skin mechanics, converted superficial sensor pressure to simulator blood pressure: BP = k·Psensor, with k = 2.1.

Wireless system integration: A laser-cut flexible copper-clad PI (AP8535R/AP7164R) PCB hosted BLE SoC (nRF52832), BLE antenna (2450AT18A100), instrumentation amplifier (INA333), reference resistors (RMCF0201FT), NTC temperature network (NCP03XH), coin cell battery (CR1220/CR1229), and the 3MIS. The system was encapsulated in silicone (SILBIONE RTV 4420) with laser-cut openings for the sensor and battery. The device sampled pressure at 100 Hz and transmitted averaged values at 4 Hz via BLE to a smart device.

Human subject testing: Three healthy male volunteers (ages 29, 38, 40) wore the reference device (Finapres NOVA) on a finger and the 3MIS system over the radial artery secured with an adhesive-backed, adjustable wristband (~40 mmHg preload). Individual calibration determined k (2.2, 2.1, 2.15) to convert Psensor to BP. Protocols included a 30 s Valsalva maneuver, 90 s breath hold, and 70 s contralateral hand ice-water immersion. Systolic and diastolic peaks were identified using scipy.signal.find_peaks. Bland-Altman analysis compared mean arterial pressure (MAP), heart rate (HR), systolic (SP), and diastolic (DP) values between systems.

Modeling: Abaqus FEA captured nonlinear postbuckling of 3D mesostructures and device mechanics. Linear buckling modes seeded postbuckling simulations with imperfections. Materials modeled included hyperelastic elastomers (Mooney–Rivlin; PDMS E≈1.8 MPa; Ecoflex E≈60 kPa), linear elastic PI (E≈2.5 GPa, ν≈0.34) and PMMA (E≈3.1 GPa, ν≈0.37), and elastoplastic metals (Au E≈78 GPa, ν≈0.44; Cu E≈119 GPa, ν≈0.34; yield strain ≈0.3%).

• Full-bridge 3D piezoresistive sensor architecture intrinsically compensates temperature, showing negligible variation from 25–40 °C compared to a quarter-bridge that exhibits thermal drift and >30 s recovery after thermal transients.
• Linear pressure response with calibration: pressure ≈ 400 × (Vp/VREF) over 0–200 mmHg; amplifier gain 6 V/V; strong linearity (R² ≈ 0.995).
• Sensitivity comparison: reported slopes indicate FB slope = 0.0031 vs QB slope = 0.009 (with FB described as ~3× higher sensitivity in the text); linear behavior maintained across the range.
• Shear robustness: With serpentine PI layers (SPL) and acrylic ring frame, sensors are insensitive to 0.1 N lateral shear; devices without SPL malfunction. Devices recover to 0 mmHg baseline after shear removal.
• Durability: Stable operation over 1000 loading cycles (0–100 mmHg) at loading speeds of 50 and 200 µm/s without notable drift.
• Phantom validation: A proportionality factor k = 2.1 maps superficial pressure to simulator BP (BP = k·Psensor). Waveforms at 50, 70, 90, 110 Hz show excellent agreement with a reference sensor after normalization.
• Wireless system performance: Skin-mounted, soft PCB platform records clear pulsatile waveforms (SP/DP features) at radial and carotid sites; sampling 100 Hz, BLE transmission 4 Hz of averaged data; CR1220 (3 V, 37 mAh) estimated lifetime ~30 days.
• Human studies: Individual calibration factors (k = 2.2, 2.1, 2.15) enable accurate BP estimation; MAP, HR, SP, and DP exhibit excellent agreement with Finapres NOVA in Bland-Altman analyses during physiologic challenges (Valsalva, breath holding, ice-water immersion).

The study addresses the challenge of temperature-induced drift in soft, piezoresistive pressure sensors by placing four microfabricated gauges in a full Wheatstone bridge around the neutral mechanical plane of a 3D mesostructure. This configuration produces equal temperature coefficients across all four resistors, canceling temperature effects while amplifying differential changes under normal pressure (tension vs compression). The encapsulation and mechanical framing ensure that the sensor responds predominantly to normal pressure with minimal sensitivity to shear, which is crucial for accurate superficial pressure measurements on moving skin. Experimental results confirm linearity, low hysteresis, robust shear rejection, and long-term stability, enabling high-fidelity pulse waveforms and correlation to blood pressure. Integration into a soft, wireless BLE platform demonstrates practical, comfortable use at radial and carotid sites, and human subject tests show agreement with a clinical reference for MAP and HR during dynamic physiological perturbations. Collectively, these findings validate a pathway toward reliable, continuous, skin-mounted hemodynamic monitoring for remote healthcare.

This work introduces a soft, 3D piezoresistive pressure sensor that uses a full Wheatstone bridge architecture and optimized mechanical design to achieve temperature insensitivity and enhanced pressure sensitivity in a compact, skin-mountable form. The system demonstrates linear, robust superficial pressure measurements, high shear tolerance, and durable performance, and integrates with a flexible wireless platform to capture clinically relevant cardiovascular metrics that agree with a reference standard. These materials and engineering advances point to broad applications in remote health monitoring and personalized care. Future efforts may further expand clinical validation, refine subject-specific calibration, and explore extended wear and broader anatomical sites.