A flexible omnidirectional rotating magnetic array for MRI-safe transdermal wireless energy harvesting through flexible electronics

Abstract

Magnetic resonance imaging (MRI)–safe implantable wireless energy harvester offers substantial benefits to patients suffering from brain disorders, hearing impairment, and arrhythmias. However, rigid magnets in cutting-edge systems with limited numbers of rotation axis impose high risk of device dislodgement and magnet failure. Here, a flexible omnidirectional rotating magnetic array (FORMA) and a flexible MRI-safe implantable wireless energy-harvesting system have been developed. Miniaturized flexible magnetic balls 1 millimeter in diameter achieved by molding three-dimensional printed templates can rotate freely in elastomer cavities and supply a magnetic force of 2.14 Newtons at a distance of 1 millimeter between an implantable receiver and a wearable transceiver. The system can work stably under an acceleration of 9g and obtain a power output of 15.62 decibel milliwatts at a transmission frequency of 8 megahertz. The development of the FORMA may lead to life-long flexible and batteryless implantable systems and offers the potential to promote techniques for monitoring and treating acute and chronic diseases.

A flexible omnidirectional rotating magnetic array is developed based on flexible magnetic balls for implantable systems.

INTRODUCTION

Implantable medical electronics, such as deep brain stimulators (1, 2), cochlear implants (3, 4), pacemakers (5, 6), and spinal cord stimulators (7, 8), are essential medical devices that offer significant benefits to millions of patients who are suffering from movement disorders (9–11), hearing impairment (12), and arrhythmias (13). These devices can operate continuously under high-performance primary batteries that last several years or transdermal wireless energy harvesting that uses external systems with rechargeable batteries to power implanted systems (14). Compared with implantable primary batteries, wireless energy harvesting offers complex functions for the implanted system without worrying about battery leakage and surgery for battery replacement. However, wireless energy harvesting typically requires transdermal mechanical coupling of implantable devices and external wearable devices through rigid magnets, which may generate large torsional forces when attempting to align themselves with the external magnetic field during magnetic resonance imaging (MRI), causing discomfort and potential magnet dislodgement as well as demagnetization (15, 16). To mitigate these issues, MRI-safe systems that can align internal magnets with the main MRI bore axis in certain human body orientations have been developed, largely eliminating the need for surgery to remove the devices or adopting bandages to immobilize them during MRI scans (15, 17).

Cutting-edge MRI-safe systems incorporate uniaxially or biaxially rotatable magnets to accommodate external magnetic field directions. However, even the most advanced cochlear implants (e.g., HiRes Ultra 3D, MED-EL Synchrony 2, and Cochlea Nucleus CI532) are still 7.6 to 10 g in weight and 5.2 to 7.2 cm³ in dimension (15, 18, 19). Despite that these devices showcase significant technological achievement, their rigid formats indicate their incompatibility with the soft biological tissues. In addition, the limited numbers of rotation axis for the magnets still impose a great risk to induce device dislodgement and magnet failure. Meanwhile, the relatively bulky sizes of the magnets result in ineffective usage of their surface area due to magnetic dipole cancellation in the center of the magnets. Our previous research has demonstrated the concept to subdivide a large magnetic membrane with a single magnetic polarity into small domains using origami approaches, leading to the increased overall strength of the magnetic field due to maximized edge effect on each domain (20–23). Further dividing these flexible membranes into smaller and individually isolated spherical domains may yield magnetic balls that can rotate omnidirectionally in response to external magnetic stimulation. The combination of these miniaturized balls may be an excellent replacement for the rigid magnets to achieve completely soft implantable wireless energy harvesters.

Here, techniques and theories to achieve a flexible omnidirectional rotating magnetic array (FORMA) are proposed. The FORMA contains miniaturized magnetic balls made of composites of neodymium iron boron (NdFeB) and polydimethylsiloxane (PDMS) and sealed insides interconnected PDMS cavities. These magnetic balls can rotate freely in the cavities to align with the external magnetic field and supply a magnetic force of 2.14 N at a distance of 1 mm between an implantable receiver (Rx) and a wearable transceiver (Tx). The FORMA, Rx, and Tx form a flexible transdermal wireless energy-harvesting system that can work stably even under an acceleration of 9g (g = 9.8 m/s²). The power received by the Rx module can reach 15.62 dBm with a tissue thickness of 1 mm at a transmission frequency of 8 MHz, providing sufficient energy to conduct stimulation and detection for future implantable applications. The FORMA and the flexible energy harvester can be readily integrated with other implantable systems to offer a continuous power supply. The development of the FORMA not only represents a significant step toward realizing life-long flexible implantable systems through electromagnetic and magnetic coupling but also holds the potential to advance techniques to monitor and treat acute and chronic diseases.

RESULTS

Schematics of the FORMA and the transdermal wireless energy-harvesting system

Figure 1A depicts potential applications of the transdermal wireless energy-harvesting system based on the FORMA. The system can be integrated with a deep brain stimulator, a cochlear implant, a spinal cord stimulator, or a pacemaker to supply continuous power while eliminating the need to use implanted batteries. A typical transdermal wireless energy-harvesting system contains a Tx module and an Rx module between which energy is transferred through electromagnetic resonance and magnetic coupling (Fig. 1B). State-of-the-art MRI-safe systems are based on uniaxially and biaxially rotatable circular or cylindrical magnets that can align with the main axis of the MRI bore, allowing the Rx module to stay in the body during MRI. However, magnet dislocations in implantation sites due to large magnetic torque and patient complaints about tissue pain were still reported in many publications (16, 24). In comparison, the FORMA offers high mechanical compliance with soft tissues and a low risk of dislocation due to its capability to orient with any random external magnetic field. The Young’s modulus of the FORMA is only 2.8 MPa, which is compatible with the Young’s modulus of human tissue (~6.4 MPa) (25–27) and is five orders of magnitude smaller than that of the rigid circular or cylindrical magnets used in the commercial systems (Fig. 1C). In addition, the omnidirectional rotation of the magnetic balls in the FORMA results in a much lower torque to avoid dislocation and tissue damage. The magnetic torques induced in circular, cylindrical, and spheric magnets were simulated under a magnetic field of 7 T (fig. S1A). Maximum torques of the circular and cylindrical magnets are 1018 and 755 mN·m, respectively, which are almost three orders of magnitude larger than that of the FORMA (fig. S1B).

Fig. 1. Schematics of the MRI-safe FORMA and the transdermal wireless energy-harvesting system.

(A) Potential applications of the transdermal wireless energy-harvesting system based on the FORMA. (B) A configuration of the transdermal wireless energy-harvesting system based on the FORMA. (C) Comparisons of the circular magnet, cylindrical magnet, and the FORMA in terms of magnetic torque, the Young’s modulus, and the number of rotational axis. (D) Images of the MRI-safe transdermal wireless energy-harvesting system.

A typical FORMA is 16 mm by 16 mm by 4 mm in dimension and 1.26 g in weight. It contains 144 magnetized balls each of which is 1 mm in diameter. These balls are situated in interconnected spherical cavities 1.5 mm in diameter in a PDMS membrane (Fig. 1Di). Both the Rx circuit and the Tx circuit are based on the flexible printed circuit board technology with excellent flexibility (Fig. 1Dii). The Rx module can be formed by sandwiching the FORMA in the middle of the Rx circuit that is bent 180° in advance (Fig. 1Diii), and the Tx module is formed by sandwiching a circular magnet and a battery in the middle of the Tx circuit that is bent in the same way as the Rx module (Fig. 1Div). The implantable circuit can be driven to light a light-emitting diode (LED) when the Tx module and the Rx module are magnetically attracted together through their magnetic components (Fig. 1Dv). The Rx module whose bending curvature is up to 1.8 cm is only 19 mm by 19 mm by 7 mm in dimension and 1.9 g in weight, while the Tx module is 34 mm by 34 mm by 12 mm in dimension and 15.9 g in weight (fig. S2). The soft, miniaturized, and lightweight configurations of the Rx modules allow them to be readily implantable on the head of a rabbit without interfering with its physical motion (Fig. 1Dvi). A detailed description of the working principles of the Tx and Rx modules as well as a circuit schematic has been given in the supplementary information (fig. S3, detailed in Supplementary Text 1).

Fabrication of the FORMA

The unique fabrication processes of the FORMA are shown in fig. S4. Briefly speaking, the flexible magnetic balls are molded in a three-dimensional printed spherical template from the composite of Nd₂Fe₁₄B microparticles and PDMS. The thin paraffin coating on the surface of the flexible magnetic balls can be achieved through ternary phase coating processes that include free ball falling in the air, surface coating in molten paraffin, and surface tension–induced separation and solidification of paraffin in water. A heating coil that is immersed in the paraffin layer can melt the paraffin based on resistive heating, resulting in a gradual temperature distribution in the paraffin ranging from 60° to 67.2°C (Fig. 2A). The subsequent heat convention to the water leads to a decreased temperature gradient from 67.2° to 28.3°C in the water. As the melting temperature of the solid paraffin is 58°C, the spherical magnetic balls that fall into the molten paraffin can be surface coated with the paraffin, which then becomes solidified after traveling into the cool water underneath the melted paraffin. The resulting paraffin-wrapped flexible magnetic balls are densely stacked in double layers and interconnected using polyvinyl alcohol (PVA) aqueous solution followed by PDMS encapsulation and exposure of the inlet and the outlet. After complete curing of the PDMS, the entire device can be immersed in mixtures of acetone and water at 80°C to remove the sacrificial paraffin layers, resulting in interconnected cavities hosting free-moving spheric balls. Media such as water, silicone oil, and glycerin with different viscosities and densities can also be filled in cavities from the inlet to minimize friction of the magnetic balls during omnidirectional rotation (fig. S5).

Fig. 2. Fabrication of the FORMA.

(A) Temperature distribution in the paraffin-wrapping process in the vertical direction. (B) Paraffin-wrapping process of the flexible magnetic ball with a diameter of 1 mm. (C) Simulation of the paraffin-wrapping process of the flexible magnetic ball with a diameter of 1 mm. (D) The relationship between the paraffin thickness h_QP and the dropping height in simulation, theory, and experimental results. (E) Paraffin-wrapping process of the soft and rigid magnetic ball with a diameter of 3 mm. (F) The tail trajectory of the magnetic balls in the paraffin-wrapping process. (G) Experimental results of the stretching, bending, and twisting of the FORMA. (H) Simulation of the stretching, bending, and twisting of the FORMA.

The capability to obtain uniform paraffin-wrapped magnetic balls is essential for the FORMA. The fundamental principles of the coating process have been studied both experimentally and theoretically. The dynamic process of paraffin coating on the flexible magnetic balls falling through the paraffin layer and into the water below was recorded using a high-speed camera (Fig. 2B). When the flexible magnetic ball fell through the paraffin-water interface, a long paraffin tail was formed until completely pulled apart because of the combined effect of gravity, capillary force, and viscous dragging. The broken paraffin tail and surficial paraffin layer were then redistributed and wrapped around the magnetic ball because of surface tension, forming a nearly circular paraffin layer on the ball. The process is also studied by finite element simulation in a two-phase flow and turbulent flow model, resulting in a similar phenomenon that is influenced by several simulation parameters such as weight, falling velocity, surface tension, and contact angle (Fig. 2C). To gain further insight into the mechanism of the paraffin wrapping, paraffin-wrapped magnetic balls obtained from various falling heights were cut and analyzed (fig. S6). The paraffin thicknesses on the top of the magnetic balls (h_QP) obtained by measurement, simulation, and numerical calculation (fig. S7, detailed in Supplementary Text 2) all suggest an inverse relationship between the h_QP and the falling heights (28–32) (Fig. 2D).

The capabilities of the flexible magnetic balls and commercially available rigid magnetic balls in achieving uniform paraffin-wrapping layers were studied. As the smallest diameters of the rigid magnetic balls that are commercially available are 3 mm, flexible magnetic balls with the same diameter have been used for comparison. Experimental results indicate that the paraffin-wrapped layer on the rigid magnetic ball exhibits an irregular cone, while that on the flexible magnetic ball maintains a circular shape (Fig. 2, E and F). The difference in the morphology of the paraffin layers may be caused by varied Reynolds numbers and surface roughness (30, 33, 34), which were then measured and calculated for three different cases. The Reynolds numbers during the falling processes of the flexible and rigid magnetic balls were found to be quite different (fig. S8). According to fluid dynamic theories, spherical objects in liquid flow start to swing and show irregular tail trajectories when the Reynolds number is greater than 350 (35, 36). The Reynolds number of the 1 mm flexible magnetic ball is calculated to be 461 when dropped in water and the surrounding water is still in laminar flow. As a result, the molten paraffin tail exhibits flatter morphology and smaller swing than that of the 3 mm flexible magnetic ball, which has a Reynolds number of 1764 in water. The Reynolds number of the 3-mm rigid magnetic ball is 3302, leading to turbulent flow surrounding the ball and intense swing. In addition, the values of surface roughness of the 1 mm and 3 mm flexible magnetic balls are 3.2 and 2.4 μm, respectively (fig. S9), which is four to five times larger than that of 0.59 μm for the rigid magnetic ball, resulting in better adhesion and more uniform wrapping of paraffin on the rough surface. Furthermore, the contact angles of the paraffin with the rigid magnetic ball and the flexible magnetic ball are 142° and 163°, respectively (fig. S10, A and B), suggesting the higher surface energy of the rigid magnetic ball. The inability to achieve uniform surface coating for the rigid magnetic ball has also been shown in the experiment and simulation results (figs. S10, C and D, and S11). The irregular paraffin layer outside the rigid magnetic ball may lead to nonuniform stacking of the magnetic balls and prevent the free rotation of the balls in the formed cavities. As a result, the rigid magnetic balls and the soft magnetic balls with larger diameters may not be suitable for constructing the FORMA.

The flexible magnetic balls not only enable uniform paraffin coating but also ensure excellent mechanical flexibility and stretchability of the FORMA. Repeated stretching, bending, and twisting have been successfully performed on the FORMA (Fig. 2G), in which a strain of 22.06%, a bending curvature radius of 3.5 mm, and a torsion of 27.6% are ultimately achieved. Similar results have been obtained by simulation (Fig. 2H). In addition, as the FORMA is composed of PDMS and flexible magnetic balls, the strain of the flexible magnetic ball is 3.2% when the FORMA reaches a fracture strain of 124% (fig. S12). Such extreme stretching, bending, and twisting of the FORMA are rarely encountered in practical applications. Therefore, the flexible magnetic balls can always maintain good integrity before the FORMA is broken.

Magnetism characterization of the FORMA

To investigate the influence of different stacking approaches of the magnetic balls in determining the overall magnetism of the FORMA, we conducted finite element analysis on the basis of different ball stacking mechanisms that adopt similar concepts in describing the crystal structures. Three types of stacking mechanisms named hexagonal close stacking (type 1), simple stacking (type 2), and body-centered stacking (type 3) offer different cavity-filling ratios as well as ball-filling ratios. Under a cavity diameter of 1.5 mm, type 1 stacking offers a maximum cavity-filling ratio of 58%, while type 2 stacking has the smallest filling ratio of 52.4%. In all cases, more than 50% of the volumes in the PDMS membranes have been occupied by liquid-fillable cavities. The filling ratios of magnetic balls to the overall volume are 17.2, 15.5, and 17.1%, respectively (Fig. 3A). Despite the difference in filling ratios, the magnetic forces between the single cells and an external permanent magnet exhibit similar values for all stacking approaches, suggesting that the stacking methods are not the primary reason to determine the magnetic forces (Fig. 3B). However, the simulation results show that when varying the effective sizes of the magnetic balls, an array with larger magnetic balls exhibits a greater magnetic force. A larger magnetic force can be obtained when the corresponding volume of the cavity is filled with magnetic materials (fig. S13). Thus, the FORMA may be improved by reducing the thickness of the paraffin coating and filling the space between large magnetic balls with smaller balls to fully use the available space.

Fig. 3. Magnetism characterization of the FORMA.

(A) The cavity-filling ratio and ball-filling ratio of devices composed of three stacking types. (B) Simulated magnetic force of devices of three types of stacking methods. (C) Experimental magnetic force when the diameters of the magnetic balls and the number of stacking layers are varied. (D) Response frequency of the magnetic balls when the magnetic field rotates along the z axis. (E) The response frequency of the magnetic balls when the magnetic field rotates along the y axis. (F) Images of the orientation of magnetic balls in response to the on and off magnetic field pointing either upwards or downwards. (G) Changes in the magnetic field (MF) strength before and after exposure to a rotating magnetic field with a strength varying from 50 to 450 mT. (H) Changes of the magnetic force before and after scanning with a 3 T MRI.

As the core component of the MRI-safe system, the characteristics of FORMA have been further investigated. When no external magnetic field is presented, the distribution of magnetic balls in a simplified simulation model of the FORMA exhibited unidirectional alignment and spontaneous alignment before and after the dissolution of the paraffin coating (fig. S14, A and B). The magnetic field intensity on three virtual lines L1, L2, and L3 marked on the surface of the model has exhibited reduced magnitudes after the removal of the paraffin coating, indicating that all magnetic balls in the spontaneous alignment case tend to maintain a low-energy state by minimizing extrinsic magnetic field (fig. S14C). When a permanent magnet with a diameter of 6 mm and a thickness of 1 mm was presented in the simulation model (fig. S15A), the maximum magnetic field strength increased from 0.32 to 0.72 T when the magnet was placed 0 to 2 mm above the FORMA’s surface (fig. S15B). Besides, a maximum magnetic force of 0.25 N was achieved at the distance of 0 mm between the magnet and the FORMA (fig. S15C). Experimental results have also demonstrated that the magnetic balls in the FORMA exhibit spontaneous self-alignment to maintain a low-energy state (movie S1) and forced alignment in the presence of an external permanent magnet with a dimension of 3 mm by 3 mm by 5 mm (movie S2). A comparison of three configurations including one stacking layer of flexible magnetic balls with diameters of 1 and 2 mm and two stacking layers of flexible magnetic balls with a diameter of 1 mm was obtained in the presence of a circular permanent magnet with a diameter of 15 mm (Fig. 3C). The magnetic force in all three configurations increase with the increased thickness of the permanent magnet. The case with two stacking layers of 1 mm flexible magnetic balls achieved the maximum magnetic force of 3.11 N at a distance of 0 mm, indicating the benefit to increase the number of stacking layers. In practical applications such as cochlear implants or pacemakers, the implanted module is approximately 1 mm underneath the skin surface (37, 38). Considering the static friction coefficient of the skin to be 0.61 (39), the 2.14 N magnetic force at a distance of 1 mm can withstand a vertical acceleration of at least 8.2g.

The response of the magnetic balls under weak and strong magnetic fields has also been evaluated. When a rotating magnetic field is only 150 mT (fig. S16A), the response frequencies of six magnetic balls marked from M1 to M6 in the FORMA were found to be the same as the rotation frequencies ranging from 0.5 to 4.0 Hz under a magnetic field rotating around the z axis (Fig. 3D, fig. S16B, and movie S3). However, the response frequencies were slightly lower under a magnetic field rotating around the y axis because of relatively low magnetic field intensity and the gravity acting on the magnetic balls (Fig. 3E, fig. S16C, and movie S3). When considering that human motions are much slower than 4 Hz in an MRI machine, this time response should be sufficient during MRI imaging. In addition, an experimental setup has been constructed to mimic the rapidly changing gradient field in the MRI by placing a permanent magnet with a field strength of 215 mT on a vibration platform. When the platform vibrates with frequencies of 10, 100, and 1000 Hz, almost no temperature change can be observed from the near-infrared images obtained by the thermal camera after continuous vibration for 2 min (fig. S17). As the chamber that was holding the magnetic balls was still crystal clear, the FORMA shows no sign of wearing both on the chambers and on the balls. Furthermore, when repeatedly applying an external magnetic field whose direction is indicated by dots and crosses, all magnetic balls can rotate freely to match the direction of the external magnetic field and behave identically throughout the repeated tests (Fig. 3F). Excellent tolerance and stability of the FORMA under the influence of strong magnetic field has been demonstrated. The magnetic polarity of a single magnetic ball after 300 repeated rotation cycles controlled by an external magnetic field of 450 mT exhibits no obvious change (fig. S18). Unexpectedly, the magnetic field strengths of the magnetic balls after exposing them to the rotating magnetic field with intensities varied from 50 to 450 mT increased by 2%, indicating a potential magnetization effect (Fig. 3G). Similar increase has also been observed when the device was scanned in an MRI machine with a magnetic field strength of 3 T. Afterward, the magnetic force of the FORMA when coupling with a circular magnet 15 mm in diameter and 5 mm in thickness reached 2.64 N as compared with 2.46 N before MRI scanning (Fig. 3H). All the results indicate that the magnetic balls in the FORMA can comply with both weak and strong magnetic fields, suggesting excellent adaptability to the complex external environment.

Electrical characterization of the MRI-safe wireless energy-harvesting system

The power transmission between the Tx and Rx modules relies on the electromagnetic resonant coupling between the modules. The schematics and images of the Tx and Rx modules have been given in Fig. 4 (A and B) and figs. S3 and S19. As the presence of the permanent magnet may alter the inductance of both the Tx coil and the Rx coil, tuning the impedance and the resonance frequency of the transmission circuit and the receiving circuits in the Tx module and the Rx module is very critical to optimize the transmission efficiency. For the case of an untuned system, the addition of the permanent magnet shifts the resonance frequencies of both the transmission circuit and the receiving circuit to around 10.3 MHz (fig. S20, A and B). As a result, both the S parameters (fig. S20C) and the measured voltage input (fig. S20D) to the transmission circuit have indicated reduced efficiency and increased reflective power due to the mismatching between the resonance frequencies of ~10.3 MHz and the transmission frequency at 8 MHz. After performing the fine-tuning of the capacitance values (C1 and C6 in fig. S3) in the transmission circuit and the receiving circuit. The resonance frequencies of both circuits have been changed to near 8 MHz (Fig. 4C and fig. S20E), leading to increased coupling efficiency between the two coils as indicated by the S parameters (Fig. 4D) and reduced changes in the input voltage to the transmission circuit (fig. S20F). After tuning the resonance frequency of the Tx and Rx coils similar to the transmission frequency, the power received by the Rx coil exhibited the maximum value at 8 MHz when the transmission signal in the Tx coil was generated by a signal generator that supplied 5 V sinusoidal waveforms at varied frequencies (Fig. 4E). These results indicate the necessity to tune the impedance of the Tx and Rx coils to achieve optimized coupling and transmission efficiency.

Fig. 4. Electrical characterization of the MRI-safe transdermal wireless energy-harvesting system.

(A) A circuit schematic of the MRI-safe transdermal wireless energy-harvesting system. (B) Images of the top and bottom sides of the Tx module and the Rx module. (C) Resonant frequencies of the Tx module and the Rx module with and without the presence of the permanent magnet and the FORMA. (D) S parameters measured by a network analyzer with and without the presence of the permanent magnet and the FORMA. (E) Received power at different transmission frequencies by the Rx module after tuning its resonance frequency to 8 MHz. (F) Simulation of the magnetic field distributions of the system in the presence of the permanent magnet and the FORMA. (G) Simulated magnetic field distributions along two artificial lines AB and CD on the Tx coil and the Rx coil, respectively. (H) Received power by the Rx coil when separating with the Tx coil at varying distances defined by different thicknesses of pork. Output voltages of the Tx coil and the Rx coil with and without the presence of the permanent magnet and the FORMA. (I) Voltage stimulation generated by the Rx module after modulating the transmission signal by programming the MCU in the Tx module. (J) Working stability of the MRI-safe transdermal wireless energy-harvesting system under accelerations ranging from 1g to 10g.

The magnetic field distributions of the system with the presence of the permanent magnet and the FORMA were obtained through the finite element simulation as the entire system is supplied with a voltage of 4 V and a current of 58 mA according to the reading from the power supplier (fig. S21). When powered by the battery, the supplied current changes to 53 mA. Despite that it is difficult to precisely measure the current from the output of the transistor in fig. S3, a current of 55 mA determined by averaging the supplied current from the power supplier and the battery has been used to roughly represent the current amplitude in the Tx coil. According to the simulation, the magnetic fields near the Tx module from point A to point B and the Rx module from point C to point D indicate that the magnetic field of the whole system is majorly determined by the permanent magnet (Fig. 4F), and only small sinusoidal fluctuations are generated at the place of the Tx coil and the Rx coil (Fig. 4G and fig. S22). Two spikes up to 400 mT represent the magnetic field strengths at the edges of the permanent magnet, while nine approximated square waves represent the magnetic field strengths at the magnetic balls. The generation of eddy current in the permanent magnet has been taken into account in the simulation. As the permanent magnet is conductive, it has been designated with an electrical conductivity of 7.14 × 10⁵ S/m. Under a current of 55 mA at 8 MHz, the induced eddy current in the permanent magnet has a maximum current density of 2.27 × 10⁶ A/m² (fig. S23A). The current density, the induced current in the Rx coil (fig. S23B), and the magnetic field distribution (Fig. 4F) are very distinct from the case without a permanent magnet (figs. S23, C and D, and S24). Because of the generation of the eddy current both in the permanent magnet and the circuit, it may induce thermal effect within these components. However, the temperature distribution in the devices measured by a thermal camera has suggested that there was more heat generated in the circuit for the case without a permanent magnet, likely due to impedance mismatch and more reflective power (fig. S25). Despite the existence of the eddy current, the induced heat in the permanent magnet is as low as 26.2°C, while the temperature of the circuit is lower than 33.8°C throughout the experiment, suggesting that the thermal effect of the eddy current is very minimum.

To further verify the rationality in choosing this current value for simulation and characterize the power transmission capability of the system, the transmission power and the receiving power in the air or with sandwiched biological tissues (fig. S26) at varied thicknesses have been quantified. The power output by the transistor in the Tx module has been measured to be 20.20 dBm (fig. S27A), which reasonably matches the total power of 220 mW generated by the power supplier. According to experimental results shown in Fig. 4H, the received powers change from 15.62 to 2.61 dBm when sandwiching different thicknesses of pork meat between the Tx and Rx modules. These receiving powers have been converted to various voltage levels of 9.1 and 1.6 V after being rectified by the voltage doubler in the Rx module (fig. S27B) as the LED light in the Rx module starts to flicker when the thickness reaches 12 mm, suggesting the instability in the radiofrequency (RF) power transmission at this thickness. Thus, it can be derived that the Tx and Rx modules can work within a tissue thickness of 12 mm at which the implanted LED can still work under the control of the microcontroller in the Rx module. When the Tx module and the Rx module were placed with no gap to each other, the power measured before rectifying the RF signal received by the Rx coil is 12.05 dBm. When the distance increase from 0 to 12 mm in air, the received power of the Rx coil first changes from 12.05 dBm at 0 mm distance to 14.64 dBm at 1 mm distance, followed by consistent reduction to 2.7 dBm at 12 mm (fig. S27C). As the received power can reach 15.62 dBm with sandwiched pork between the Tx and Rx modules, this power is sufficient to conduct many functions such as electrochemical sensing, electrical and optical simulation, controlled drug delivery, and closed-loop regulation, leading to more complex implantable flexible electronics without worrying about the issues of continuous power supply. The construction of devices with more functions will be explored in future research.

For future electrical or optical stimulation, pulsed stimulation generated in the Rx module has been demonstrated by programming the Tx module, resulting in square waves with different frequencies and duty cycles (Fig. 4I). In addition, to evaluate the working stability of the system under daily physical activities, we then affixed the system to the side of a vibrating table that supplied a vertical acceleration. The amplitude and the frequency of the output voltage remained stable when the system was subjected to accelerations ranging from 1g to 9g (Fig. 4J). With the increase of the acceleration, there is an increased tendency of misalignment between the Tx and Rx modules, resulting in reduced voltage output at 10g due to reduced coupling efficiency and power transmission efficiency. However, the capability to perform stable wireless power harvesting under extreme body motions up to an acceleration of 9g can still be considered sufficient to tackle daily applications.

As titanium (Ti) is served as a major packing material for protecting implanted devices because of its high biocompatibility, chemical inertness, and paramagnetic property that is compatible with MRI, a flexible Ti foil 8 μm in thickness has been demonstrated to integrate with the Rx module in addition to the parylene passivation layer (fig. S28A). A system with only a parylene passivation layer has been compared side by side with another system that contains an additional Ti packaging layer by immersing both of them into a phosphate-buffered solution (PBS) for more than 7 days (movie S4). The images in fig. S28B before and after immersing the Rx module with a Ti protection layer in the PBS solution have shown no sign of erosion. Both systems can be fixed onto the vertical wall of the plastic container by magnetic coupling between the Tx and Rx modules. The Rx modules in the PBS solution can respond to the motions induced by the external Tx modules and can operate continuously in a liquid environment for 7 days (fig. S28C), demonstrating excellent water resistance of the packing materials.

Imaging and biocompatibility of the MRI-safe wireless energy-harvesting system

The influence of flexible magnetic balls on the head MRI was assessed using a humanoid skull made of polyvinyl chloride (PVC) and filled with gelatin. An Rx module was initially encapsulated by an 8 μm thick Ti film and then implanted into the preground bone bed near the cochlea of the humanoid skull to conduct a 3 T MRI examination (fig. S29). No net movement of the Rx module can be observed during the MRI scanning process. However, because of the existence of magnetic components, spatial artifacts do exist in the images shown in fig. S30. Six points labeled as a, b, c, d, e, and f have been manually marked in the images to indicate the boundaries of the artifacts along the x, y, and z directions. The length between a and b is 6 cm, while the lengths between e and f as well as c and d are 4 and 8 cm, respectively. This artifact is almost unavoidable for implanted conductive and magnetic materials. However, if the MRI images are for other organs, then the localized artifact generated by the implanted FORMA in the brain will not influence the images in other locations.

In vivo animal experiments were performed before and after 0.25 T MRI for small animals (Fig. 5A and movie S5). Considering the potential application scenarios of deep brain stimulation and cochlear stimulation, the Rx module insulated with parylene was implanted on the rabbit skull and sutured immediately afterward. After a 2-week healing period, the wearable Tx module was magnetically positioned on the rabbit’s head, regardless of whether the rabbit was stationary or moving freely. The Tx module succeeded in lighting the LED in the Rx module, indicating the feasibility of wireless energy transmission in free-moving animals. Notably, there was no obvious influence on the rabbit’s fur growth after continuous Tx module wearing for 2 weeks. In addition, magnetic positioning and energy transfer were still attainable after 0.25 T MRI. Computed tomography (CT) was performed to examine the spatial location of the Rx module before and after 0.25 T MRI (Fig. 5, B and C, and fig. S31). The displacements are less than 1.8 mm in the x direction and 0.5 mm in the z direction after the 0.25 T MRI. These displacements were negligible when considering that the device was placed directly on the skull without any fixture, and the small displacements may be caused purely by tissue-induced motion. The steady increase in body weight after 2 weeks indicated no negative effects from the implantation (Fig. 5D). In practical application, the displacement could be further reduced by grinding a bone bed on the skull for device implantation.

Fig. 5. Bioimaging and biocompatibility of the MRI-safe transdermal wireless energy-harvesting system.

(A) Images of the process to the implantation of the Rx module into the rabbit and demonstration of wireless energy harvesting through LED lighting. (B) CT results before the 0.25 T MRI examination. (C) CT results after the 0.25 T MRI examination. (D) Changes in weekly body weight of the rabbits in the experimental group and the control group. (E) Results of the blood routine tests for the rabbits. WBC, white blood cells; RBC, red blood cell; HGB, hemoglobin; PLT, platelet; LYMPH, lymphocyte; NEU, neutrophil; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BUN, urea nitrogen; CRE, creatinine. (F) Results of the blood biochemistry tests for the rabbits. Control data were collected from five rabbits acquired from one batch in a month. (G) Representative histology of kidney, liver, spleen, and heart tissues of a control rabbit and a rabbit implanted with an MRI-safe transdermal wireless energy-harvesting system 4 weeks after implantation.

A comprehensive understanding of the health conditions of the rabbits was obtained by performing the complete blood count and blood chemistry tests. The average counts of white blood cells, red blood cells, platelets, lymphocytes, neutrophilic granulocyte, and the levels of hemoglobin were analyzed throughout the 6-week study period, and the P values (P > 0.05) of the independent-samples t test showed no significant differences between the experimental group and the control group (Fig. 5E). In addition, enzymes and electrolytes that serve as important indicators in the blood for organ-specific diseases also fall within the confidence intervals of control values. For example, alanine aminotransferase, urea nitrogen, aspartate aminotransferase, and creatinine were all at normal levels, indicating the absence of disorders in the spleen, liver, kidney, and heart, and good overall health, respectively (Fig. 5F). Histopathologic evaluation of tissues obtained from a control rabbit and a rabbit implanted with the Rx module for 6 weeks reveals the absence of inflammation, ischemia/tissue necrosis, or other architectural/histologic abnormalities in organs such as spleen, kidney, heart, and liver in both rabbits (Fig. 5G). The normal daily behavior and health conditions of major organs after implantation and 0.25 T MRI scan suggest the possibility of in vivo medical treatment using MRI-safe system for diseases such as brain dysfunction, auditory dysfunction, and cardiac dysfunction.

DISCUSSION

This paper proposes approaches to construct an implantable flexible omnidirectional rotating magnetic array named FORMA and a transdermal wireless energy-harvesting system that may be integrated with other implantable systems for disease treatment. The system contains a miniaturized implantable Rx module and a wearable Tx module. The approach and the mechanisms to achieve paraffin-wrapped flexible magnetic balls as well as interconnective cavities in FORMA have been investigated, leading to freely rotatable magnetic balls responding rapidly to changing external magnetic fields with minimized torques. The received power at the matching conditions between the resonance frequencies of both the Tx and Rx modules and the transmission frequency reaches 15.62 dBm, providing sufficient energy to support implanted devices to perform complex functions. In vivo experiments on rabbits have demonstrated excellent biocompatibility and the capability of the system to conduct long-term wireless energy harvesting, indicating promising applications to integrate with other implantable components to achieve physiological regulation and disease treatment. Different magnetic flux densities have been used in this research based on practical applications and availability. As 3 T and 450 mT MRI systems are typically available in clinics and veterinary hospitals, they were used to obtain MRI images in this paper. To evaluate the torques generated in different magnets, an extreme intensity of 7 T has been used to conduct the simulation. For most experiments conducted in the laboratory setting, NdFeB permanent magnets with intensities ranging from 150 to 450 mT were used.

The performance of FORMA can be further improved by increasing the filling ratios of magnetic balls in the device. The potential approaches may include introducing the combination of small and large magnetic balls and reducing the thickness of sacrificial paraffin coating. In addition, the fabrication of miniaturized spherical balls becomes more difficult with further reduction in the dimension of the magnetic balls. It may be necessary to explore other approaches to achieve sacrificial layer coating based on irregular magnetic particles. As the microcontroller in the current research is programmed to generate a transmission frequency of 8 MHz, it may be necessary to adopt a different Tx module design for outputting more standardized frequencies such as 13.56 MHz. Despite that Ti is considered to be an excellent packaging material for protecting implantable devices, its effectiveness in protecting the flexible circuit and its performance in thin-film formats still demands further exploration. More fully implanted batteryless flexible systems may eventually be achieved with the techniques developed in this paper to realize long-term and closed-loop monitoring.

MATERIALS AND METHODS

Fabrication of the FORMA

All flexible magnetic balls were constructed by molding composites mixed with PDMS (Sylgard 184, Dow Corning Corp.) and Nd₂Fe₁₄B microparticles (Magnequench, Co. Ltd.) at the weight ratio of 1:6.5. A coil, with a wire diameter of 0.08 mm, an inner diameter of 15 mm, and a height of 6 mm, was partially immersed in water. A current of 0.3 A was applied to the coil to heat water in the middle of the coil after the solid paraffin was added. Once the thickness of melted paraffin reached 2 mm, flexible magnetic balls fell from a height of 2.5 to 5.0 cm into the melted paraffin, which was then wrapped onto the balls and solidified after entering the underneath cool water. One hundred forty-four paraffin-wrapped flexible magnetic balls were specially arranged into two densely stacked layers in a mold and interconnected with each other using the 10 wt % PVA solution. The resulting structures were then encapsulated in PDMS at room temperature, leaving an inlet and an outlet exposed. The flexible magnetic balls in cured PDMS were then magnetized by a magnetizer (ME-1225, Magele Technology Co., Ltd.), followed by the removal of the PVA and the paraffin on the balls by immersing the entire device into a mixed solution of acetone and water at a temperature of 80°C. The resulting cavities can be later filled with different liquids to accommodate different application scenarios.

Simulation of the FORMA

A finite-element analysis software COMSOL (COMSOL Inc.) was used to simulate the self-arrangement of the magnetic balls after paraffin dissolution. The magnetization directions of all magnetic balls were set to point upwards along the z axis and then shifted in response to the rotation of the magnetic ball. To simplify the model, only 12 magnetic balls whose remanence was set to 0.3 T were stacked in the same way with a fixed distance as the 144 magnetic balls used in the FORMA. A permanent magnet 6.4 mm in diameter and 5 mm in thickness with a distance ranging from 0 to 5 mm to the FORMA was used to determine the magnetic field distributions and magnetic forces. The paraffin-wrapping process was simulated using a two-phase flow multiphysics model in which the dynamic viscosity of the molten paraffin was set as 0.0055 Pa·s. The contact angle between the molten paraffin and the flexible magnetic ball and the contact angle between the molten paraffin and the rigid magnetic ball were experimentally determined to be 163° and 142°, respectively. In addition, Young’s moduli of PDMS and flexible magnetic balls were set as 750 and 1000 kPa, respectively. The Poisson’s ratio was set to 0.49 for the PDMS and 0.3 for the flexible magnetic balls. Boundary conditions of displacement and force were set according to the maximum strain of 160% that the device can withstand without breaking. A finite-element analysis software Ansys (Ansys Inc.) was used to evaluate the torque differences of the three structures (circle, cylinder, and sphere) under a rotating magnetic field. Three stacked coils have been used to generate a uniform magnetic field of 7 T in the central region. The uniaxially rotational magnet has a disk-like circular shape that is 8 mm in diameter and 3 mm in height. The biaxially rotational magnet features a cylindrical shape that is 4 mm in diameter and 9 mm in height. In comparison, each of the magnetic balls in the FORMA is only 1 mm in diameter. The rotating magnetic field was achieved by rotating the coils, and the torques applied to the three structures were obtained by simulation.

Characterization of the flexible magnetic balls

To assess the response of flexible magnetic balls to the magnetic field in both upward and downward directions, a simplified FORMA with 12 magnetic balls was firstly prepared. An electromagnet with a diameter of 5 cm was used to create a magnetic field with a strength of 60 mT. The alignment of the magnetic balls with the electromagnet was recorded by a camera. To evaluate the response of the magnetic balls to the rotating magnetic field, six Nd₂Fe₁₄B permanent magnets were assembled on a rotation platform with adjustable speeds. A magnetic field of 150 mT was measured using a gaussmeter (CH-1800, Beijing Cuihai Jiacheng Magnetoelectric Technology Inc.) at the rotation center of the assembled magnets. The FORMA was then fixed at the center of the rotation platform with different orientations. The response frequency of the flexible magnetic balls in the FORMA was ultimately determined by calculating the number of revolutions per second under a rotation frequency ranging from 0.5 to 4 Hz. To measure the polarity and demagnetization effect of the magnetic balls under the magnetic field, a similar platform with a magnetic field of 450 mT at the rotation center was used. The polarity and the magnetic strength of the magnetic balls were recorded after 300 cycles of rotation. Furthermore, the magnetic force between the FORMA and a permanent magnet 15 mm in diameter and 5 mm in thickness was measured before and after the 3 T MRI examination.

MRI (3 T) examination and scanning protocols

A bone bed has been grounded out of a PVC humanoid skull, followed by fixing an Rx module protected by a Ti film 8 μm in thickness to the bone bed. Gelatin (20 wt %; 9000-70-8, Guangzhou Saiguo Biotech Co., Ltd.) solution was then filled into the skull and then cured at room temperature. The investigation was conducted in a 3 T MRI (Signa, General Electric Medical System Co., Ltd.). The imaging protocols comprised five standard clinical sequences that are conventionally used for brain examination (T1 Flair, T2 Flair, T2 FSE, BRAVO, and DWI).

Animal experiments

All animal studies were conducted at Tianjin Institute of Radiation Medicine with accreditation number SYXK (M) 2014-0004 by following the ethical and operational guidelines required by the institute. Ten male rabbits (five rabbits for the control group and five rabbits for the implantation group) with weights around 2.2 kg were provided by the institute. The Rx modules were ultraviolet-radiation sterilized overnight and then implanted onto the skulls of the rabbits by opening the epidermis on their skulls followed by suturing the wounds on the epidermis. The body weights of the test subjects were monitored once a week. The Tx modules were later attached to the Rx modules 2 weeks after the implantation. MRI (0.25 T) examinations (G-scan Brio, Esaote North America Inc.) were conducted with the removal of the Tx modules, and CT scanning (Brivo CT385, General Electric Medical System Co., Ltd.) was performed before and after the MRI examination to determine the displacement of the Rx modules underneath the skin after 4 weeks of implantation. Organs and blood in both groups were extracted from the rabbits. Complete blood count and blood chemistry tests on the blood samples were conducted at Tianjin Institute of Radiation Medicine. All explanted organs were stored in conical metal-free tubes at −20°C or in conical tubes with 10% buffered formalin for histology studies.

Supplementary Materials

This PDF file includes:

Supplementary Texts 1 and 2

Figs. S1 to S31

Tables S1 to S3

Legends for movies S1 to S5

Other Supplementary Material for this manuscript includes the following:

Movies S1 to S5