Magnetic devices - components that generate, manipulate, or detect magnetic fields - are used in a variety of applications ranging from medical imaging to information transfer and sensing. The persisting motivation to miniaturize electronic devices drives a need to more effectively manipulate magnetic fields in the millimeter scale. In this dissertation, two areas are presented where miniaturization technology has been applied to create new or improved capability in the area of milliscale magnetics. For the first application, a radio frequency (RF) surface coil with a dimension that is comparable to the area of interest can maximize the signal-to-noise ratio (SNR) in Magnetic Resonance Imaging (MRI) application, allowing for higher quality images than available existing tools. This work demonstrates a miniature flexible coil that can be inserted into the body for high SNR pituitary gland MR imaging. It presents the spatial distributions of the image SNR of the 26-mm coil in both numerical simulation and agar gel phantom experiments. Compared to the commercial head coil, the miniature coil achieved up to a 19-fold SNR improvement within the region of interest, and the simulation matched the phantom experiment within an error of 1.1% � 0.8%. Additionally, the coil performance was characterized with cadaver heads MRI scan using a 20-mm coil. A maximum of 16-fold and an average of 5-fold SNR improvement within the pituitary gland compared to the commercial head coil was obtained. The feasibility of using the miniature flexible coil for high-SNR pituitary MR imaging has been demonstrated, showing the potential for improved detection and characterization of pituitary gland microadenoma.In the second application, a novel way to shield the magnetic field of a single chip on a multi-chip-module is presented. This approach uses individual millimeter-size magnetic shield on each chip to reduce the chip-to-chip coupling effect of the circuits or influence on signal currents from local magnetic fields. Magnetic through silicon vias (mTSVs) were developed to help achieve the desired level of magnetic shielding. The microscale multilayer shields provided high efficiency shielding around arbitrary shapes and enabled fabrication of chip-scale shielding. This work focuses on the design and fabrication of the proposed magnetic shielding.

Minimally invasive surgery (MIS) has gained popularity over traditional open surgery due to its advantages of decreased incision size and pain to the patient, lower risk of infection, and shorter recovery time. Recent developments in robotic surgical systems have shown promise to further advance MIS by offering the surgeons with increased manipulability and dexterity along with 3D vision.

However, the lack of tactile feedback is the key feature that is needed for robotic surgery to reach its full potential. Recent research efforts have successfully integrated some degree of tactile feedback components onto surgical robotic instruments, and have shown significant improvement of the outcome of the surgical performance. The primary barrier to the adoption of tactile feedback in clinical use is the unavailability of suitable tri-axial force sensing technologies that can be integrated with the medical instruments. Besides well-understood normal force sensing, shear force sensing is also critical in clinical tasks, such as suturing, where shear sensing could prevent breakage of sutures due to excessive shear force.

This paper describes the design, batch microfabrication, and characterization of a miniature force sensor for providing haptic feedback in robotic surgical systems. We demonstrate for the first time a microfabricated sensor that can provide triaxial sensing (normal, x-shear, y-shear) in a single sensor element that is integrated with commercial robotic surgical graspers. Features of this capacitive force sensor include differential sensing in the shear directions as well as a design where all electrical connections are on one side, leaving the backside pristine as the sensing face for surgical tasks.

The sensor readout is performed by a custom-designed printed circuit board with 24-bit resolution. The integration of read-out circuits with the capacitive sensor is designed on two printed circuit boards that can be clipped together, providing the possibility for disposable sensors. The sensing system is first connected to the LabVIEW-based controller, to convert the analog capacitor signal to a digital signal representing force. After the functionality of the sensor is proven, the tactile sensor system is then integrated with our custom Visual Studio based feedback control system.

Initial LabVIEW results validate the batch fabrication of the capacitive sensors and the design of the control circuit. The sensor is characterized using a sensing circuit with a 24-bit resolution at 11 Hz-109 Hz. With the LabVIEW program, the sensor and the readout circuitry contribute to a noise down to 0.8 fF to the normal z-direction, 0.2 fF to the shear x-direction, and 0.9 fF to the shear y-direction at 8 Hz bandwidth.

The grasper integrated sensor system uses an Arduino based controller to multiplex between x, y, and z directions, providing 24 packets of tri-axial data per second to the Visual Studio-based computer application, with down to 0.094 fF capacitance noise to the normal z-direction, 0.078 fF to the shear x-direction, and 0.0825 fF to the shear y-direction at 87.2 Hz bandwidth. The sensitivity measured for the sensor is 14.58 fF/N for normal z direction, 0.83 fF/N for the shear x direction, and 0.62 fF/N for the shear y direction. We report a normal resolution of 6.45 mN, x-shear resolution of 94.7 mN, and y-shear resolution of 133 mN, all of which are more than sufficient for clinically relevant forces. A data latency of less than 42 ms is achieved to obtain a triaxial data package and transmit it to the computer through the WiFi network.

A user study has been performed to tackle the suture breakage phenomenon that occurs during robotic surgery with the application of excessive forces due to lack of haptic feedback. The work aims to develop and validate a bi-axial shear feedback system that warns the operator to anticipated suture breakage. The benefits of a suture breakage warning system may be a reduced incidence of suture failure with otherwise equivalent knot quality during the tying procedure.

Biaxial shear sensors were placed on the Cadiere grasper tips of a da Vinci robotic surgical system. 17 novice subjects were then instructed to tighten 10 knots made from Silk 3-0 sutures, five times with the Haptic Feedback System (HFS) enabled, and five times with the system disabled (i.e., without any feedback). During each trial, the number of suture breakages was recorded. After trial completion, knots were evaluated for tightness. This was accomplished by measuring the amount of knot slippage following knot tying. Additional metrics recorded were the time required for completing each trial and both the average force and peak force applied in each trial.

Seven suture failures occurred in trials with HFS enabled while seventeen occurred in trials without feedback. The biaxial shear sensing system reduced the incidence of suture failure by 59% (). It also resulted in 25% lower average applied force in comparison to trials without feedback (), which is relevant because average force was observed to play a role in suture breakage (p=0.03925). Results of a 55% decrease in standard deviation of quality knots tied with HFS also indicate an improvement in consistency when using the feedback system.

These results suggest this system may improve outcomes related to knot tying tasks in robotic surgery and reduce instances of suture failure while not degrading the quality of knots produced.

Magnetic resonance imaging (MRI) can serve as an efficient tool to examine the internal structure of the human body, especially for soft tissues. The pituitary gland – which regulates hormones in the body – is one example of soft tissue that can be investigated by MRI [1]. One area of particular interest in the study of the pituitary gland is the search for miniature tumors, also called microadenomas. Microadenomas in the pituitary gland are important because they can cause Cushing’s disease, which can cause death if untreated [2]. However, although 60-80% of the pituitary adenomas are microadenomas [3], current 3 T MRI techniques cannot detect up to 50% of the microadenomas [4]. The present work aims to increase detection accuracy by placing miniaturized MRI coils inside the sphenoid sinus (a hollow space or cavity near the pituitary gland), as close as possible to the region of interest (ROI), namely the pituitary gland [5]. By placing MRI surface coils close to the target region, the SNR can be increased, which allows for improved detection accuracy of microadenomas that are too small for current scanning techniques. Previous studies using single-loop surface coils achieved high-SNR images, but in some cases, coils need to be rotated so that they can fit into the cavity near the pituitary gland. In the case of large rotation angles, single-loop surface coils are limited by the intrinsic “dead spot”, especially when rotated 90 degrees [4]. In contrast, figure 8-shaped surface coils are less effected by the signal loss due to coil plane rotations [6]. In this study, we developed a custom-designed surface coil specifically for pituitary gland imaging. Parameters of the coil are optimized based on simulations, including the distance between the two semicircle-shaped components and the coil radius. A linear relationship between the optimal coil radius and target depth is established. The compensation between rotation angle of the coil plane and the horizontal shift of the target is also studied, proving that rotations have a reduced detrimental effect on coil performance as compared to single-loop coils, especially when the target is placed near the coil boundary.

Magnetism is a workhouse for electric power generation at the macroscale, responsible for a large fraction of the electricity generation today. However, conventional approaches to energy conversion using magnets fail at a small scale due to Joule heating from electric currents. Recently, a method has emerged – room-temperature composite multiferroics coupling electrics and magnetics - that allows for control of magnetism via voltage as opposed to current. With this newfound capability, industrial & commercial application spaces are rapidly opening up for domains, including ultra-low-power spintronics devices, microwave devices, and even magnetic particle and cell sorting platforms. Meanwhile, personalized medical therapies hold the potential of new forms of highly effective therapies. One example of personalized medicine is CAR T-cell therapy, which uses patients’ cells for cancer treatments. However, such an approach requires technology that can analyze and sort these cells in high quantity, selecting only the cells that will be the most effective cancer fighters. Magnetic cell sorting is a popular approach for high throughput cell sorting. Still, there is no current method capable of capturing and culturing arrays of cells and then selectively releasing the desired ones. Recent advances in room-temperature multiferroic devices, such as devices where magnetism is controlled by electric fields, provide a path for capture, culture, and selective release of cells. However, work remains to understand how to manipulate the magnetic structures that capture these cells. This work aims to develop a multiferroics-based cell manipulation platform with high scalability and can achieve cell control locally. This work first conducts an in-depth study of the magnetization and multiferroic properties of various magnetostrictive layers, including Ni, FeGa, Ni/CoFeB, and Terfenol-D micromagnets on Pb(Mg1/3Nb2/3)O3]0.69[PbTiO3]0.31 (PMN-PT) piezoelectric substrates. It then selects the highly magnetostrictive Terfenol-D micromagnets in the same size scale as human cells as the candidate for life science applications. It also investigates the interaction between these micromagnets and cells before and after a voltage is applied across the PMN-PT substrate. The key questions addressed include how to create structures from a magnetoelastic material that are in the same size scale as human cells (20 \mu m) and control the magnetization of these structures to release cells on-demand via electric fields. Furthermore, this work demonstrates the potential of using patterned surface electrodes to generate localized strain in order to control the behavior of the micromagnets both locally and selectively.

The millimeter wave band, which corresponds to frequencies between 30 and 300 GHz, is heavily used in radar and satellite communications. For these applications, transmit-and-receive (T/R) modules are used to boost output power of the transmitted signal and establish noise figure, which is the degradation in the signal-to-noise ratio, of the system for receiving. Inside these modules, magnetic components such as circulators and isolators are critical to direct the flow of signals and allow both simultaneous transmission and reception using a single antenna. However, the inherently incompatible crystal structures of the ferrite-based magnetic components and their semiconductor-based electrical counterparts make integration of the two using conventional manufacturing methods difficult.

Additive manufacturing has emerged as a method of fabricating structures with complicated shapes. Recent efforts have begun incorporating magnetic materials with the method. One application for additive manufacturing of magnetic materials is miniaturization and integration of circulators. Using circulators in the 40-50 GHz range as a motivating application, requirements arise for the printed films, namely immunity to eddy currents, sufficient magnetization to act as a self-biasing field, and out-of-plane orientation of the self-biasing field. Based on these required properties, hexaferrite particles are selected for their strong magnetocrystalline anisotropy and low conductance. Due to their large internal anisotropy field, the particles of these materials tend to rotate to the field direction instead of changing magnetization direction under application of an external magnetic field, which is less than the anisotropy field. Methods of fabricating composites of hexaferrite particles and liquid polymer, SU8 were developed. Rotation of hexaferrite particles in polymer matrix and thus magnetic anisotropy has been demonstrated in the composite, which is subsequently cured to hold the physical position and orientation of the particles. The anisotropy of the self-biasing field provided by the films has been experimentally characterized via techniques like vibrating sample magnetometer and magnetic force microscopy, and a ferromagnetic resonance (FMR) frequency ~48 GHz has been observed via short waveguide method. We have also characterized the viscosity of the particle-laden polymer at different particle concentrations. 3D printing of this composite with poling will make direct printing of magnetic components that require out-of-plane and in-plane anisotropic magnetization possible.

Strain-coupled multiferroic heterostructures provide a path to energy-efficient, voltage-controlled magnetic nanoscale devices, a region where current-based methods of magnetic control suffer from Ohmic dissipation. Magnetoelectric coupling behavior in such composite heterostructures has thus been of substantial interest for scientific research and applications. As the dimension of the devices scale down, novel physical phenomenon emerges and thus also requires further understanding of both the magnetization and strain behavior at micro- and nanoscale.

When it comes to the magnetization behavior, there has been a growing interest in highly magnetoelastic materials, such as Terfenol-D, prompting a more accurate understanding of their magnetization behavior. To address this need, we simulate the strain-induced magnetization change with two modeling methods: the commonly used unidirectional model and the recently developed bidirectional model. Unidirectional models account for magnetoelastic effects only, while bidirectional models account for both magnetoelastic and magnetostrictive effects. We found unidirectional models are on par with bidirectional models when describing the magnetic behavior in weakly magnetoelastic materials (e.g., Nickel), but the two models deviate when highly magnetoelastic materials (e.g., Terfenol-D) are introduced. These results suggest that magnetostrictive feedback is critical for modeling highly magnetoelastic materials, as opposed to weaker magnetoelastic materials, where we observe only minor differences between the two methods’ outputs. To our best knowledge, this work represents the first comparison of unidirectional and bidirectional modeling in composite multiferroic systems, demonstrating that back-coupling of magnetization to strain can inhibit formation and rotation of magnetic states, highlighting the need to revisit the assumption that unidirectional modeling always captures the necessary physics in strain-mediated multiferroics.

In terms of the strain behavior, there hasn’t been a system-level work that quantifies the strain distribution as a function of the electric field at these so-called mesocales level (100 nm- 10 um), in the range of the constitutive grain size, etc. To obtain mechanical properties at such length scale, including strain information, we used synchrotron polychromatic scanning x-ray diffraction (micro-diffraction) on beamline 12.3.2 at the Advanced Light Source of the Lawrence Berkeley National Lab.

With given ferromagnetic and ferroelectric components, it is the magnetoelectric coupling between the two that governs the interaction. In this work, we also demonstrate a method to enhance the coupling behavior between two existing components by interposing a polymer layer.

Multiferroics is a field that has rapidly garnered much interest in the past several years because of its establishment of a new unconventional class of electromagnetic devices. By capitalizing on a special class of material systems that have coupled magnetic and electrical properties, multiferroics offer unique possibilities for micromagnetic devices because of their ability to control magnetization with electric fields (and vice versa). Strain-coupled multiferroic composites, in particular, are promising because their magnetoelectric coupling is many orders-of-magnitude stronger than single-phase multiferroic materials.

Telecommunication systems utilize dynamic magnetic processes for many applications, such as filters, nonreciprocal components, and electrically small antennas. These devices rely on electromagnetic waves for transduction, limiting their sizes and constraining their usage. Strain-coupled multiferroic composites utilize mechanical waves, leading to a five order-of-magnitude reduction in characteristic length and opening the door to device miniaturization. The dynamics of multiferroic coupling can be characterized into several different regimes that depend on the magnetic state of the composite (multi- or single domain) and the actuation frequency (relative to the domain wall relaxation and the ferromagnetic resonance frequencies). Miniaturization of dynamic magnetic devices relies on the thorough study of multiferroic coupling in these different regimes.

In this work, the dynamics of multiferroic coupling in three different regimes are investigated. A variety of micro-scale piezoelectric devices are designed and fabricated to investigate coupling in each of these regimes. First, resonant cantilevers are used to study dynamic actuation in the regime of quasi-static domain wall motion. It is found that nonlinearity in this regime can lead to doubling of the actuation force frequency, a useful phenomenon for low noise electromagnetic characterization. Second, damping in thin film bulk acoustic resonators in a multi-domain state near ferromagnetic resonance is investigated, and the effect of the domain structure on the power absorption at resonance is characterized. Finally, the perturbations of acoustic Lamb waves caused by a single domain magnetic thin film are studied at frequencies above ferromagnetic resonance. Interactions with spin waves are found to lead to an angular dependence in the magneto-acoustic absorption. These results demonstrate several unique aspects of dynamic multiferroic coupling across different regimes and can be readily capitalized to develop novel micro-scale devices for telecommunication systems.

Compact integration of clocks and inertial sensors using atomic, molecular, and optical (AMO) technology is necessary to create a self-contained navigation system resistant to external interference. However, the trend in miniaturization of AMO systems places the magnetic field of particle traps, optical isolators, and vacuum pumps close to other system components. Stray fields and field fluctuations cause changes in atomic transition frequencies, raising the noise floor and reducing the valuable stability in these precision devices. Therefore, it is critical to shield these magnetic fields away from sensitive subsystems by shunting them through low reluctance paths. This is accomplished with high permeability magnetic materials which either surround the precision components or the source of the magnetic field itself. Current magnetic shields are conventionally machined single or multi-layer structures made of various iron alloys. At smaller size scales, these manufacturing methods are ineffective at accommodating the various device and interconnect shapes, making multi-system integration challenging.

This work demonstrates batch fabricated high permeability magnetic shielding using permalloy electroplating techniques to simultaneously push the limits of minimum size, maximum shielding factor, and minimum cost. In particular, it presents the first experimental demonstration of electrodeposited high permeability, compact magnetic shielding at millimeter and sub-millimeter scales of fields exceeding 15 mT. Single layer shields of 300 μm permalloy with inner dimensions varying from 3 mm to 6.5 mm were fabricated on 3D printed polymer molds using a novel double-anode plating process to enable conformal deposition with uniform material properties. Multilayer shields of 10 μm permalloy and copper layers with inner dimensions of 1.5 mm to 6 mm were microfabricated using a bulk micromachining technique. The electroplated shields were designed with appropriate thickness to avoid saturation at the specified fields and with shapes to allow sophisticated interconnect extraction – a task that is challenging for conventional machining yet simple for microfabrication and electroplating. The size and shielding factor of these structures can enable compact integration of magnetic devices for AMO microsystems and other magnetic microelectronics, such as magnetic random-access memory and haptic actuators.

Magnetic resonance imaging (MRI) is a non-invasive imaging technique that can produce high spatial resolution 3D images, especially for non-bony parts or soft tissues. Higher imaging resolution is usually preferred, to detect small lesions or irregularities in the imaging subject. This dissertation presents two projects that aims for high-resolution MRI. In the first project, we designed a single-loop miniature flexible coil that can be surgically positioned millimeters from the pituitary gland, enabling high-SNR pituitary MRI. We investigated the spatial distributions of the image SNR of the miniature coil, via both numerical simulation and phantom experiments. We also explored the feasibility of increased SNR within the pituitary gland based on simulated surgical placements. Compared to the commercial head coil, our miniature coil achieved up to a 19-fold SNR improvement within the region of interest, and the simulation and phantom experiment reached a good agreement, with an error of 1.1% � 0.8%. High resolution MRI scans further demonstrated the visual improvement of the miniature coil against the commercial head coil. The cross-validation of the simulation and the phantom experiment showed the potential of using the numerical simulation model to accelerate the coil design prototyping and iteration and to optimize coil design in the future. The clinical application study describes a transnasally-placed 2-cm flexible coil to improve the resolution of pituitary imaging. The coil is compatible with 95% of patients, can be successfully placed in contact with the sella in cadaver studies, shows no temperature changes in phantom studies during scanning, and improves the SNR of the pituitary by an order of 17. This study provides feasibility data for the promise of application to the clinical setting to improve the detection of small ACTH-secreting pituitary tumors when clinical pituitary MRI fails. In the second project, we propose a novel slice-profile transformation super-resolution (SPTSR) framework with deep generative learning for through-plane super-resolution (SR) of multi-slice 2D TSE imaging. The deep generative networks were trained by synthesized low-resolution training input via slice-profile downsampling (SP-DS), and the trained networks inferred on the slice profile convolved (SP-conv) testing input for 5.5x through-plane SR. The network output was further slice-profile deconvolved (SP-deconv) to achieve an isotropic super-resolution. Compared to the state-of-the-art SMORE SR method, where the networks trained by conventional downsampling, our SPTSR framework demonstrated the best overall image quality from 50 testing cases, evaluated by two abdominal radiologists. The quantitative analysis cross-validated the expert reader study results. 3D simulation experiments confirmed the quantitative improvement of the proposed SPTSR and the effectiveness of the SP-deconv step, compared to 3D ground-truths. Ablation studies were conducted on the individual contributions of SP-DS and SP-conv, networks structure, training dataset size, and different slice profiles.

Resonant microelectromechanical Systems (MEMS) have enabled miniaturization of high-performance inertial sensors, radio-frequency filters, timing references and mass-based chemical sensors. Despite the increasing prevalence of MEMS resonators for these applications, the energy dissipation in these structures is not well-understood. Accurate prediction of the energy loss and the resulting quality factor (Q) has significant design implications because it is directly related to device performance metrics including sensitivity for resonant sensors, bandwidth for radio-frequency filters and phase-noise for timing references. In order to assess the future potential for MEMS resonators it is critically important to evaluate the energy dissipation limits, which will dictate the ultimate performance resonant MEMS devices can achieve. This work focuses on the derivation and evaluation of the intrinsic mechanical energy dissipation limit for single-crystal nano and micromechanical resonators due to anharmonic phonon-phonon scattering in the Akhiezer regime. The energy loss is derived using perturbation theory and the linearized Boltzmann transport equation for phonons, and includes the direction and polarization dependent mode-Gruneisen parameters in order to capture the strain-induced anharmonicity among phonon branches. Evaluation of the quality factor limit reveals that Akhiezer damping, previously thought to depend only on material properties, has a strong dependence on crystal orientation and resonant mode shape. The robust model provides a dissipation limit for all resonant modes including shear-mode vibrations, which have significantly reduced energy loss because dissipative phonon-phonon scattering is restricted to volume-preserving phonon branches, indicating that Lame or wine-glass mode resonators will have the highest upper limit on mechanical efficiency. Finally, the analytical dissipation model is integrated with commercial finite element software in order to evaluate the Akhiezer-limited Q for resonators with complex geometries and mode shapes and provide a prediction for the temperature dependence of the energy loss. This theory satisfies the pressing need for a reliable, analytical model that can predict the phonon dissipation limits for modern, resonant MEMS devices, where precise manufacturing techniques and accurate finite element methods can be used to select particular vibrational mode shapes and crystal orientations.