A Miniaturized Ultrasonic Power Delivery System Tzu-Chieh Chou, Ramkumar Subramanian, Jiwoong Park, and Patrick P. Mercier 10/23/2014
Motivation: Powering Medical Implants Medical implants are fundamentally size constrained by anatomy. Size is often limited by power systems. Three options to deliver power: Embedded battery suitable only for ultra-low-power applications with generous available volume (e.g., pacemaker) Energy harvesting has not yet been demonstrated for chronic applications Wireless power transfer (WPT) suitable for higherpower applications, as a large battery can be placed outside of the body Pacemaker, Medtronic Spinal Cord Stimulator, Boston Scientific 2 Retinal Implant, Boston Retinal Implant Project Cochlear Implant, Advanced Bionics
Transcutaneous power transfer via inductive coupling Most implants that employ WPT are a few cm in size Employ (resonant) inductive coupling for power transfer Operate at 0.1-50 MHz due to higher dielectric losses at higher frequencies Efficiency can exceed 90% Problem: there are many emerging applications that cannot employ large coils E.g., smart pills, injectable sensors, etc. Requirement: efficient transcutaneous power transfer with small (mm-sized) receive antennae Pill Camera, PillCam 3 Leadless Pacemaker, St. Jude Medical
Solution 1: mid-field electromagnetics Full-field EM analysis shows that GHz frequencies are optimal for power transfer to mm-sized devices Tx=2cm Path loss > 20dB! Tx=2cm Rx=2mm Tx=0.2cm Poon, O Driscoll, and Meng, 2010 4 PROS: enables interesting mmsized implants with optimal efficiency from electromagnetics CONS: wavelength in tissue is large à focusing energy is difficult; tissue losses limit efficiency
Solution 2: ultrasonic power transfer Ultrasonic waves decay more slowly than mid-field EM waves in most tissue Acoustic attenuation coefficient of soft tissue usually ranges from 0.6 to 3.3dB/MHz cm Mid-field EM decays at 2.6dB/cm Opportunity for deeper implants Ultrasound has a much shorter wavelength than mid-field EM 1.5 mm at 1MHz (US) compared to ~30 mm at 1GHz (EM) in most tissues Ultrasonic waves suffer less from mismatch loss Acoustic impedance of soft tissue is typically between 1.38 10 6 and 1.99 10 6 kg/sec m 2 5 PROBLEM: Ultrasonic power transfer has, at the time of paper submission, only been experimentally validated in cm-sized systems. GOAL: Experimentally validate in a mm-sized system.
Finite Element Analysis Modeling Ultrasonic COMSOL 4.4 Tissue model: linear elastic material with attenuation Output: pressure and electric potential fields Lumped circuit AC voltage source RLC matching network Piezoelectric constitutive eq. Determine electric field & strain Helmholtz eq. Pressure field Attenuation caused by tissue 6
Finite Element Analysis Modeling PZT model Validation of the parameters for the constitutive eqn. Part from manufacturer Others from least-square optimization process of electrical impedance Magnitude(Ohm) 100000 10000 1000 200 220 240 260 280 300 Frequency (khz) 80 60 40 Phase(degree) 20 0-20 -40 7 4.4mm (diameter) 3.0mm (height) -60-80 -100 200 220 240 260 280 300 Frequency (khz)
Finite Element Analysis Modeling EM coupling Ansys HFSS Tissue model: water and muscle with dielectric properties Optimal frequency tuned to around 400MHz (MICS band) Circular coil with the same diameter of PZT receiver 0.5mm-wide coil (0.035mm thick) PCB, FR4 substrate (0.254mm thick) Parylene-C coating (0.1mm thick) 8
Maximum available gain (MAG) System-level path loss set by path loss through medium and matching performance Problem: difficult to adjust matching network at every frequency Solution: measure two-port s-parameters, calculate MAG assuming optimal matching network at all frequencies Simulation: Calculated by parametric sweeping the RLC values Experiment: Measured by connecting the PZT pair to a network analyzer In order to physically achieve the MAG, electrical matching networks at both ports are required 9
Maximum available gain (MAG) MAG vs. optimal frequency The optimal frequency is defined as the frequency at which the minimum path loss occurs The figures below show the comparison between EM coupling and ultrasonic for the medium water -10 Ultrasonic (this work) 270.4kHz@0.8cm 0 EM coupling -15-20 290.4kHz@1.4cm -5-10 0.623GHz@0.8cm -25-30 -35-40 288.4kHz@2.0cm -15-20 -25-30 0.589GHz@1.4cm 0.559GHz@2.0cm Path loss (db) 10-45 200 250 300 350 400 Frequency (khz) -35 0.1 0.3 0.5 0.7 0.9 Path loss (db) Frequency (GHz)
Simulation results Mineral Oil & Water 11 MAG vs. axial distance MAG (db) 0-5 -10-15 -20-25 -30-35 -40-45 Mineral Oil, Ultrasound Water, EM 0 1 2 3 4 5 Axial Distance (cm)
Simulation results - Muscle MAG vs. axial distance 12 MAG (db) 0-5 -10-15 -20-25 -30-35 -40-45 0.95cm Muscle, Ultrasound Muscle, EM 0 1 2 3 4 5 Axial Distance (cm)
Simulation results 13 Normalized optimal frequency vs. axial distance Normalized Optimal Frequency 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Water, EM Muscle, EM Water and Muscle, Ultrasound 0 1 2 3 4 5 Axial Distance (cm)
Simulation results Slope of MAG curve slowly converges to acoustic attenuation coefficient 0.28dB/cm for mineral oil 0.64dB/cm for muscle Ultrasonic scheme starts to take the lead beyond certain depth threshold 1.4cm for mineral oil 0.9cm for muscle Optimal frequency for ultrasonic scheme does not dramatically change over distance Less re-tuning is needed compared to EM coupling 14
Experimental setup Frequency range: 200 to 400kHz (containing the fundamental and a few harmonic vibration modes) Medium: mineral oil Transmitter Mineral oil Receiver 3.0mm (height) 4.4mm (diameter) Stage and Micromanipulators 15
16 MAG (db) Experimental results MAG vs. axial distance 0-5 -10-15 -20-25 -30-35 Mineral Oil, Ultrasound, Experiment Mineral Oil, Ultrasound, Simulation 3X power compared to EM at a depth of 2.3cm Water, EM, Simulation 0 1 2 3 4 5 Axial Distance (cm)
17 Experimental results Normalized optimal frequency vs. axial distance Normalized Optimal Frequency 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Water, EM, Simulation Mineral Oil, Ultrasound, Simulation Mineral Oil, Ultrasound, Experiment 0 1 2 3 4 5 Axial Distance (cm)
Experimental results Why MAG is -9dB even when axial distance is 0? Nulls in radiation pattern In both (axial and lateral) directions Also explains the saw-tooth profile in the measured data Reflections caused by the epoxy layer Much lower acoustic impedance compared to that of PZT s Slope of MAG curve is -2.3dB/cm beyond 1cm Comparable to previously reported -2.5dB/cm (simulation) Remains competitive against mid-field EM coupling,which decays at the rate of -2.6dB/cm 18
Conclusions Ultrasonic power delivery scheme was verified to be feasible by both simulations and experiments MAG is less than -20dB at 2.5cm depth It has a power delivery efficiency about -2.3dB/cm with a 4.4mm-diameter transmitter-receiver pair -2.6dB/cm for a 2mm-diameter EM mid-field coupled Rx The optimal frequency changes less dramatically over distance Less re-tuning is required compared to EM coupling PZT receivers scale better down to even lower sizes than small antennas Opportunity for even smaller implants in the future 19
Application note for future work Ultrasound decays quickly through bone and other stiff materials Attenuation coefficient is about 22dB/MHz cm for bone compared to 1dB/MHz cm for brain 100X path loss for a depth of 5cm when the frequency is 0.2MHz It may not be appropriate to use ultrasound for implants that go underneath bone (e.g., neural implants) Either strictly EM solutions are required here, or more invasive dual-mode solutions 20 D. Seo, J. Neuro. Methods, 2014