In the emerging research field of bioelectronic medicine, it has been indicated that neuromodulation of the Vagus Nerve (VN) has the potential to treat various conditions such as epilepsy, depression, and autoimmune diseases. In order to reduce side effects, as well as to increase the effectiveness of the delivered therapy, sub-fascicle stimulation specificity is required. In the electrical domain, increasing spatial selectivity can only be achieved using invasive and potentially damaging approaches like compressive forces or nerve penetration. To avoid these invasive methods, while obtaining a high spatial selectivity, a 2 mm diameter extraneural cuff-shaped proof-of-concept design with integrated Lead Zirconate Titanate (PZT) piezoelectric ultrasound transducers is proposed. For the development of the proposed concept, wafer-level microfabrication techniques are employed. Moreover, acoustic measurements are performed on the device, in order to characterize the ultrasonic beam profiles of the integrated PZT-based piezoelectric ultrasound transducers. A focal spot size of 200 μm by 200 μm is measured for the proposed cuff. Moreover, the curvature of the device leads to constructive interference of the ultrasound waves originating from multiple piezoelectric ultrasound transducers, which in turn leads to an increase of 45% in focal pressure compared to the focal pressure of a single piezoelectric ultrasound transducer. Integrating piezoelectric ultrasound transducers in an extraneural cuff-shaped design has the potential to achieve high-precision ultrasound neuromodulation of the Vagus Nerve without requiring intraneural implantation.

Close collaboration between the Bioelectronics department at Delft University of Technology and the Neuroscience department at the Erasmus Medical Centre has resulted in a successful tethered design for a real-time epileptic seizure detection and suppression method for mice. The goal of the Neuromate project is to develop this method into a wireless setup containing group-housed freely moving and interacting mice for use in behavioural studies. The system will include continuous monitoring and stimulation at set points in time. The Neuromate project comprises three links, two of which have previously been established. The main goal of this study was to find and evaluate a technique to complement the current Neuromate project with a wireless downlink, channelling the communication from the researchers towards the mice. This new downlink should fit in the ongoing project and meet particular specifications. The most critical requirements are that it should be lightweight and small-scale and should allow simultaneous multi-user communication. Also crucial are the prevention of interference with the two other links, reliability and low power consumption. The three most promising techniques, power source keying, optical wireless transmission and terahertz torching are elaborated. Terahertz (THz) torching, covering the high thermal part of the THz band (10 to 100 THz), was chosen as the technique to be developed in this thesis. This decision was based on the fundamental limitations of power source keying and optical wireless transmission, which appear in the weighted criteria evaluation of the requirements. The challenges of THz torching are mainly practical, while fundamentally it provides the opportunity to form a reliable non-interfering wireless link. The feasibility of THz torching for our specific application was tested by creating a proof-of-principle and conducting five different experiments. The prototype includes two components. The first is a thermal source, which in the final design will be placed above the cage, and the second is a pyroelectric detector, which will later be positioned on top of the head module of each of the mice. During the experiments, communication with the recently developed THz torch was proven to be feasible. Experiment 1 resulted in an optimum data rate of 35.71 bps, allowing for the simultaneous stimulation of the mice. And the divergence of the source was sufficient to cover the entire cage, as it was found in experiment 3 that an angle of 50° was the maximum misalignment still allowing reliable transfer of the data. However, the prototype did fail to reach the required distance. In experiment 2 only 8 cm could reliably be bridged. The source was not strong enough to overcome the attenuation in the air. A more powerful source will allow an increased reachable distance, making sure the entire cage is covered. In this work, an innovative and promising proof-of-concept has been realized for the wireless downlink of the Neuromate project.

Ultrasound (US) neurostimulation, the non-pharmacological, reversible excitation or inhibition of the nervous system using ultrasonic waves, is emerging as a high interest topic in neurostimulation research. In the current project, an attempt was made to demonstrate US neuromodulation in a Lumbricus Terrestris model of evoked compound action potentials (eCAP). Hardware and software for the electrophysiological observation of local field potentials were designed and assembled. The resulting system was used to perform mechanical, electrical and both direct and indirect ultrasonic neurostimulation experiments. It was shown electrical stimulation could be performed reliably in-vivo and ex-vivo. Mechanical stimulation only functioned in-vivo. Additionally, power transfer experiments showed that deeply embedded CMUT devices can be used to harvest acoustic power and use this signal to stimulate an explanted Lumbricus Terrestris medial nerve cord. Direct ultrasound neurostimulation was however not observed, likely due to a combination of misalignment and incorrect acoustic pressure profiles. As an additional project, a microfabrication step was designed and performed for the etching of high aspect-ratio silicon bulk structures for the backside venting of low frequency CMUT devices. A two-step process deep reactive ion etch (DRIE) was attempted where smaller features were first introduced into the silicon (phase A) and then advanced using a larger etch frame (phase B). It was shown that phase A etching could be performed adequately, resulting in high quality deep silicon etch profiles with minimal tapering and an excellent etch rate. However, phase B etching resulted in the consumption of side-walls and removal of previously etched features. It is likely some slight adjustments to the passivation stage of the DRIE process would result in successful completion of this fabrication step. Acknowledgements

Neuroscientists use neural electrodes to explore the working mechanisms of the nervous system. Therefore, ideal electrodes should have a small size and the ability to record and stimulate at a single cell resolution with low noise. Materials used for fabrication should be flexible and stable for a long period in the biological media. However, conventional recording and stimulation techniques do not have sufficient spatiotemporal resolution for neuroscience research. Combining electrical and optical modalities into one device helps overcome the resolution limits and record more detailed information. For this application, transparent conductive materials are needed. Graphene is a potential solution due to its advantageous combination of properties, such as high conductivity, transparency, and flexibility. However, important characteristics of recording and stimulation electrodes, such as the impedance and charge injection capacity of graphene electrodes, do not reach the levels of conventional materials. The electrical characteristics of graphene could be improved further with surface modification, chemical doping, or stacking. Each method has been shown to improve the conductivity of graphene, although some affect the transparency of the layer. In this work, three methods were used to improve the electrical characteristics of multilayer graphene neural electrode without losing transparency or flexibility. These methods include growing a thicker layer of graphene, adding metal nanoparticles to the surface of the electrode, and nitric acid doping of graphene. For that purpose, graphene electrodes were fabricated on a silicon wafer. The electrical characteristics of these electrodes were assessed with electrochemical impedance spectroscopy, cyclic voltammetry and 4-point probe measurements. Furthermore, the optical transmittance was measured. The improvement methods were then tested on these electrodes, and the performance was evaluated. Adding metal nanoparticles to the surface of the electrode showed the most promising results. With gold nanoparticles, the impedance at 1 kHz was lowered 82%, and charge storage capacity increased 529%. However, at the same time, 30% of the optical transmittance was lost. With lower nanoparticle density, 6% of transmittance was lost, and 7% of impedance gained. Nitric acid doping did not improve the impedance, but the charge storage capacity was increased up to 66%. Thicker layers of graphene displayed a lower sheet resistance. However, impedance or charge storage capacity were not improved.

The development of personalized healthcare solutions is a complex and multifaceted challenge that requires synergistic collaboration and cross-fertilization between multiple disciplines, including microelectronics, nanotechnology, materials science, and biotechnology. As numerous biomedical applications necessitate the precise regulation and observation of various biological systems at the microscale, developing integrated microsystems with functionalities that span diverse domains, such as electrical, mechanical, and optical, has become imperative in paving the way for next-generation biomedical devices. Nevertheless, as the number of microsystems within a biomedical device escalates, a pressing need emerges to interconnect these independent microsystems using an approach that meets the constraints imposed per each particular context. Wire bonding, for instance, is one of the most widely known and used methods to establish electrical connections between chips and packages. However, wire-bonded microsystems may be inadequate to fit in applications confined by the available physical space and whereby aspects such as reliability and biocompatibility are paramount. Specifically deserving attention is the increased footprint and the introduction of protrusions that may jeopardize an effective interface of biomedical devices with biological systems. Therefore, it becomes essential to devise seamless connections between these microsystems for enhanced robustness, electrical performance, compactness, and improved physical conformability to biological structures. This doctoral research was driven by the increasing demand for microsystem integration alternatives in the biomedical field and the need to develop advanced biomedical devices with improved functionality and performance. Monolithic fabrication was the principal method of establishing a seamless integration between distinct microsystems: integrated circuits—essential for the signal conditioning of transducers—and micro-electromechanical systems—excellent for implementing functionalities at the microscale via precise micromachining delicate structures on high-quality materials. Two novel biomedical devices were devised to achieve this objective: an organ-on- a-chip system for cell-culture experimentation equipped with an analog-compatible, cost-effective, BiCMOS-based temperature sensor and a stretchable polydimethyl-siloxane membrane; and an artifact-resilient optrode optimized for ultralow-noise measurements of infraslow brain activity. The latter benefited from dual-gate, low-noise, p-channel JFETs based on a BiFET technology and deep reactive ion etching on a silicon-on-insulator wafer for micromachining nonrectilinear features on the probe— essential for creating application-oriented solutions that interface better with biological structures. Both devices were designed based on a unique awareness-oriented co-design methodology that aids the device architect in undertaking design decisions of various process-related hurdles entailing co-fabrication. This methodology, namely “holistic iterative co-design thinking”, offers an iterative co-design process that facilitates the early identification of integration obstacles related to the manufacturing process. One of the key procedures in this methodology refers to functionally decomposing a multidimensional complex design problem into a set of individual one-dimensional problems that are less complex to solve. As a result, the (co)-design is iteratively readjusted, significantly saving time and resources. This dissertation also takes a new standpoint into the existing monolithic fabrication modalities, proposes a new taxonomy, clarifies terminologies, and addresses a novel co-fabrication technique: IC-interlaced-MEMS, employed for cost- effectively co-fabricating the organ-on-a-chip system described in Chapter 4. The IC-interlaced-MEMS is similar to its “sibling” IC-interleaved-MEMS. The distinction lies primarily in their degree of process orthogonality. While the IC-interleaved-MEMS benefits from fully orthogonalizing process steps between the IC and MEMS domains, the IC-interlaced-MEMS trades orthogonality for process simplification and enhanced lithographic pipeline workflow. These benefits promise to leverage the construction of next-generation biomedical devices that interact with biological systems via specialized, large-area transducers.@en

Cervical Vagus Nerve Stimulation (CVNS) has shown efficacy in treating depression, therapy-resistant epilepsy, and is being explored for conditions such as obesity and migraine. The side effects of whole nerve stimulation, which is currently the only approved method, can be minimized using spatially selective stimulation. This thesis introduces Temporal Current Steering (TCS), a novel stimulation technique aimed at enhancing spatial selectivity for CVNS. TCS introduces a temporal variation to standard current steering, creating an electric field that varies both temporally and spatially. With this study, enhanced spatial selectivity was confirmed through simulations of extracellular stimulation of neurons with time-dependent waveforms on two electrodes located on a single plane. Particle swarm optimization was employed over 2000 iterations and 98 runs to identify a waveform with the highest spatial selectivity. The optimal solution achieved a 30% improvement over a single electrode placed directly in line with the target and a 31% improvement over the standard direct current steering technique, simulated in the same model. The findings of this thesis suggest that TCS can increase spatial selectivity in invasive stimulation, although further simulations are needed for non-invasive applications with greater electrode-to-neuron distances.

Cervical Vagus Nerve Stimulation (CVNS) has shown efficacy in treating depression, therapy-resistant epilepsy, and is being explored for conditions such as obesity and migraine. The side effects of whole nerve stimulation, which is currently the only approved method, can be minimized using spatially selective stimulation. This thesis introduces Temporal Current Steering (TCS), a novel stimulation technique aimed at enhancing spatial selectivity for CVNS. TCS introduces a temporal variation to standard current steering, creating an electric field that varies both temporally and spatially. With this study, enhanced spatial selectivity was confirmed through simulations of extracellular stimulation of neurons with time-dependent waveforms on two electrodes located on a single plane. Particle swarm optimization was employed over 2000 iterations and 98 runs to identify a waveform with the highest spatial selectivity. The optimal solution achieved a 30% improvement over a single electrode placed directly in line with the target and a 31% improvement over the standard direct current steering technique, simulated in the same model. The findings of this thesis suggest that TCS can increase spatial selectivity in invasive stimulation, although further simulations are needed for non-invasive applications with greater electrode-to-neuron distances.

Characterizing electrodes for neural interfaces is an essential step of the prototyping and manufacturing process. Sample performance can be modelled by conducting electrochemical measurements such as Electrochemical Impedance Spectroscopy, Cyclic Voltammetry, Voltage Transients and Noise characterization, providing insights for future use scenarios. However, parameters, techniques, and scientific reporting of results often fall short of a complete characterization and do not adhere to any standard. A platform for accurate, reliable, and standardized electrode characterization was conceived to overcome these obstacles. It allows complete electrochemical characterization while providing a modular 3D-printed solution to securing the electrodes above the beaker. Additional parameters, such as electromagnetic isolation and control of the medium, are also accounted for. A complete characterization study of ENEPIG electrodes was carried out to validate the setup. Manufactured as circuit boards, not electrodes, several aspects of their electrochemical performance were lacking - mainly their survivability to the stress applied by the tests carried out. Testing parameters had to be determined to minimize structural damage while maximizing performance, akin to what would be desirable in a clinical setting. Sensing characterization experiments identified a double-layer electrode structure and revealed that smaller electrodes exhibit capacitive behaviour in bandwidths one order of magnitude wider than larger ones. Stimulation characterization experiments ascertained that charge surface distribution would predominantly accumulate along the perimeter of the electrode. In conjunction with surface characterization, results indicate the flat surface of the electrodes prevents charge from being stored and injected optimally. Ultimately, the performance of the ENEPIG electrodes was measured with the injection-to-storage ratio at 10%, significantly inferior to other Au electrodes reported in the literature and utilized in vivo. As results accurately represented the sample population, characterization was deemed successful - validating the experimentation setup developed for this project.

Electronic interfaces, particularly microelectrode arrays (MEAs), are crucial for studying electrophysiological processes in the body, with applications ranging from implants to deep brain simulators. In neuroscience, they play a vital role in exploring neuronal cell distribution and behaviour, as well as disorders like epilepsy and Alzheimer’s disease. However, electrophysiological recordings have limitations, that have led to the exploration of optical approaches like calcium imaging. To address the shortcomings, a promising strategy involves integrating electrophysiology and optical methods for simultaneous cellular activity measurement, capitalising on their combined temporal and spatial resolution. The challenge lies in developing fully transparent MEAs to overcome the limitations of traditional opaque electrodes. Graphene’s versatile properties, spanning from electrical conductivity to mechanical flexibility, position it as an ideal material for transparent and flexible electronics, particularly in neural recording and stimulation. Due to these properties, graphene MEAs (gMEAs) allow integration with various optical techniques, overcoming limitations associated with traditional opaque MEAs. In this project, we designed and fabricated a transparent gMEA, intended to perform electrical signal recordings and optical voltage mappings simultaneously from photostimulated optogenetic cell lines. The design allows for photostimulation from a source beneath the gMEA, while enabling unobstructed optical measurements from above. The electrodes were crafted from multilayer chemical vapour deposition (CVD) graphene, chosen for its transparency and favourable electrical properties. Quartz and sapphire were evaluated as potential substrates for the device. After demonstrating the synthesis of multilayer graphene was possible on both substrates, quartz was selected as the preferred material due to its resistance to graphene delamination. Characterisation of the gMEAs was done using various techniques, including optical transmittance (OT), electrochemical impedance spectroscopy (EIS), and measurements of the signal-to-noise ratio (SNR). The stability of the gMEAs was also assessed by immersing the devices in cell culture medium and with ageing tests performed in PBS. Initial electrochemical characterisation of the gMEAs exhibited promising signal detection despite a relatively high baseline noise of ∼ 23 μV . In comparison, commercially available MultiChannel Systems MEA (60MEA200/30iR-Ti), showed a lower baseline noise (∼ 4 μV ), but gMEAs achieved comparable signal sensitivity. EIS of gMEAs revealed an impedance at 1 kHz ranging from 3.2 to 9.89 MΩ, largely surpassing values in other studies. However, when area normalised, the impedance remained comparable to reported values. Stability tests identified issues related to the permeability of the encapsulation layer and degradation of molybdenum structures, causing large variations in the SNR and EIS measurements after exposure to liquid media.