Indian Institute of Science Bangalore
etd@IISc Electronic Theses and Dissertations at Indian Institute of ScienceNot a member yet
6204 research outputs found
Sort by
Magnetically powered helical nanoswimmers
No full textFrom flocking birds to migrating cells, ‘active matter’ is ubiquitous in the natural world. Almost all known life forms are based on self-propelled entities working collectively to create large-scale structures, networks, and movements. Artificially designed self-propelled objects can allow the study of active matter phenomena with a level of control that is not possible in natural, biological systems. With this motivation, we develop micro/nanoscale swimmers whose swimming mechanism is inspired by microscopic, flagellated bacteria.
Among different ways of powering swimmers, the magnetic field deserves special mention due to its inherent biocompatibility, minimal dependence on the properties of the surrounding medium, and remote powering mechanism. Along with providing an insight into the non-equilibrium phenomena of active matter, the helical swimmers can also impact future biomedical practices with intelligent, multifunctional entities swarming toward a diseased site and delivering therapeutics with high accuracy.
When an oscillating magnetic field is applied to helical structures, motility is induced in the form of back-and-forth motion, but the directionality is unspecified and thereby represents a zero force, zero torque, active colloid system. These are called reciprocal swimmers, and their degree of randomness in the reciprocal sequence plays an important role in determining their effective motility. We show the results at high activity levels where the degree of randomness is further affected by the presence of the surface, which in turn results in a non-monotonic increase of motility as a function of the magnetic drive. The magnetic swimmers show enhanced diffusivity compared to their passive counterparts, and their motility can be tuned externally. However, to achieve a self-propelled velocity, we use the ratchet principle to break reciprocal symmetry in time. The thermal ratchets can extract useful work from random fluctuations and are common on the molecular scale, such as motor protein. We use the ratchet principle to induce net motility in an externally powered magnetic colloid, which otherwise shows reciprocal (back and forth) motion. The swimmers show net motility with enhanced diffusivity, in agreement with numerical calculations.
We further discuss the preliminary experimental results and modelling pertaining to collective dynamics of the helical magnetic nanoswimmers. Additionally, we have studied non-magnetic tracer beads suspended in a medium containing many swimmers and found the diffusivity of the beads to increase under magnetic actuation, akin to measurements performed in dense bacterial suspensions.
Crucial aspects of studying the active swimmers pertain to their behaviour under different physical conditions. We demonstrate controlled manipulation of magnetic helices within two types of optical confinement: an optical bowl and a flat potential, both formed by manipulating an optical tweezer. The interaction of helical swimmers with optical confinement is modelled and further confirmed by experiments. Combining optical and magnetic forces in a single nanostructure can allow multiple investigations pertaining to colloidal physics, including micro-rheology, hydrodynamics and confinement effects.
In summary, we envision that developing helical magnetic swimmers will provide a new model system to investigate fundamental non-equilibrium phenomena and play a vital role in developing intelligent theragnostic probes for biomedical applications
Low Light Image Restoration: Models, Algorithms and Learning with Limited Data
No full textThe ability to capture high quality images under low-light conditions is an important feature of most hand-held devices and surveillance cameras. Images captured under such conditions often suffer from multiple distortions such as poor contrast, low brightness, color-cast and severe noise. While adjusting camera hardware settings such as aperture width, ISO level and exposure time can improve the contrast and brightness levels in the captured image, they often introduce artifacts including shallow depth-of-field, noise and motion blur. Thus, it is important to study image processing approaches to improve the quality of low-light images. In this thesis, we study the problem of low-light image restoration. In particular, we study the design of low-light image restoration algorithms based on statistical models, deep learning architectures and learning approaches when only limited labelled training data is available.
In our statistical model approach, the low-light natural image in the band pass domain is modelled by statistically relating a Gaussian scale mixture model for the pristine image, with the low-light image, through a detail loss coefficient and Gaussian noise. The detail loss coefficient in turn is statistically described using a posterior distribution with respect to its estimate based on a prior contrast enhancement algorithm. We then design our low-light enhancement and denoising (LLEAD) method by computing the minimum mean squared error estimate of the pristine image band pass coefficients. We create the Indian Institute of Science low-light image dataset of well-lit and low-light image pairs to learn the model parameters and evaluate our enhancement method. We show through extensive experiments on multiple datasets that our method helps better enhance the contrast while simultaneously controlling the noise when compared to other classical joint contrast enhancement and denoising methods.
Deep convolutional neural networks (CNNs) based on residual learning and end-to-end multiscale learning have been successful in achieving state of the art performance in image restoration. However, their application to joint contrast enhancement and denoising under low-light conditions is challenging owing to the complex nature of the distortion process involving both loss of details and noise. We address this challenge through two lines of approaches, one which exploits the statistics of natural images and the other which exploits the structure of the distortion process. We first propose a multiscale learning approach by learning the subbands obtained in a Laplacian pyramid decomposition. We refer to our framework as low-light restoration network (LLRNet). Our approach consists of a bank of CNNs where each CNN is trained to learn to explicitly predict different subbands of the Laplacian pyramid of the well exposed image. We show through extensive experiments on multiple datasets that our approach produces better quality restored images when compared to other low-light restoration methods.
In our second line of approach, we learn a distortion model that relates a noisy low- light and ground truth image pair. The low-light image is modeled to suffer from contrast distortion and additive noise. We model the loss of contrast through a parametric function, which enables the estimation of the underlying noise. We then use a pair of CNN models to learn the noise and the parameters of a function to achieve contrast enhancement. This contrast enhancement function is modeled as a linear combination of multiple gamma transforms. We show through extensive evaluations that our low-light Image Model for Enhancement Network (LLIMENet) achieves superior restoration performance when compared to other methods on several publicly available datasets.
While CNN models are fairly successful in low-light image restoration, such approaches require a large number of paired low-light and ground truth image pairs for training. Thus, we study the problem of semi-supervised learning for low-light image restoration when limited low-light images have ground truth labels. Our main contributions in this work are twofold. We first deploy an ensemble of low-light restoration networks to restore the unlabeled images and generate a set of potential pseudo-labels. We model the contrast distortions in the labeled set to generate different sets of training data and create the ensemble of networks. We then design a contrastive self-supervised learning based image quality measure to obtain the pseudo-label among the images restored by the ensemble. We show that training the restoration network with the pseudo-labels allows us to achieve excellent restoration performance even with very few labeled pairs. Our extensive experiments on multiple datasets show the superior performance of our semi-supervised low-light image restoration compared to other approaches.
Finally, we study an even more constrained problem setting when only very few labelled image pairs are available for training. To address this challenge, we augment the available
labelled data with large number of low-light and ground-truth image pairs through a CNN based model that generates low-light images. In particular, we introduce a contrast distortion auto-encoder framework that learns to disentangle the contrast distortion and content features from a low-light image. The contrast distortion features from a low-light image are then fused with the content features from another pristine image to create a low-light version of the pristine image. We achieve the disentanglement of distortion from image content through the novel use of a contrastive loss to constrain the training. We then use the generated data to train low-light restoration models. We evaluate our data generation method in the 5-shot and 10-shot labelled data settings to show the effectiveness of our models
Structure of Turbulent, Swirling Round Jets
No full textThe present study deals with the numerical analysis of the effect of the swirl in the self-preservation region of the turbulent round jet. However, a large number of literature exists for the analysis of near-exit regions—very few deals with the self-preservation region of the jets far downstream. The present study attempts to provide insights into the effect of swirl on the turbulent mixing and jet spread rate by examining the self-similar solution in the far-field region of the jet. The study is divided into two main portions: a comparison of the turbulent swirling and non- swirling jets and the comparison between the turbulent jets having low to moderate values of swirls.
A standard computation for a non-swirling jet is used to validate the flow solver. Simulations are carried out at a Reynolds number of 2,400 for the top-hat velocity profile at the inlet. All flow characteristics are computed in detail and compare the results with existing DNS data. Velocity profiles at different streamwise locations collapse on a single curve and closely match the available data. The jet decay and spread rates also align with the standard computed data.
Large eddy simulation has been performed for non-swirl (S = 0), weak swirls (S = 0.3, 0.5) and moderate swirl (S = 0.7) at a Reynolds number of 11,000. In both the non-swirling and swirling cases, special care is taken to ensure that the computational domain is large enough to study the jet’s behaviour in a self-similar region. The research presents the effects of the swirl on a turbulent flow and compares the simulation results with available experimental data. Comparing the swirling and non-swirling cases indicates a changed turbulence structure to the effect that the swirling jet spreads and mixes faster than the non-swirling. With increasing degrees of swirl, the angle of spread of the jets is increased, and correspondingly, the decay of the maximum values of velocity components along the lengths of the jets is faster. Flow entrainment shows that the entrainment increases with swirls. The numerical simulations showed that the flow quickly achieved a self-similarity for the mean axial velocity. In contrast, the radial and azimuthal mean velocities reached a self-similar state after a longer period of jet development. Results of the decay of velocity and jet spread rate in the self-similar region of the swirling jet without vortex breakdown were found to vary linearly with the streamwise direction of the jet irrespective of the magnitude of swirl number, which is in line with the findings from experiments of Rose (1962), Chigier & Chervinsky (1967) & Pratte & Keffer (1972). In contrast, Craya & Darrigol (1967) has theoretically shown that axial velocity decay varies as three halves along the length of the jet. Additionally, mass flux shows higher mixing in swirling jets compared with non-swirling. The integrated axial fluxes of linear and angular momentums were conserved along the jet’s axis in the self-preserving region
Theoretical Studies of Polymer Dynamics in Confined Spaces
No full textThe growing use of micro- and nano-fabricated devices to study complex biological processes has made it increasingly important to understand the effects of confinement on macromolecular behaviour. In this thesis, I will discuss how theoretical models of polymer dynamics in small spaces or crowded environments can provide useful insights into the dynamics of real systems. To this end, I consider the application of various statistical mechanical methodologies to the following illustrative many-body problems: (i) the shear-induced stretching of ideal flexible chains in narrow capillaries, (ii) the relaxational dynamics of Gaussian polymers in rectangular slits, (iii) the cyclization kinetics of long polymers in spherical cavities and in viscoelastic media, and (iv) the reactivity of the terminal groups of surface-tethered self-avoiding walks. Among other results, I find that geometrical constraints can screen out hydrodynamic effects and produce free-draining behavior, introduce logarithmic corrections to the bulk scaling of diffusion coefficients and relaxation times, and modify the molecular weight dependence of chain reactivity. These results highlight the significant part that can be played by confinement on chain dynamics
Modelling spatial and non-spatial conflicts across multiple design domains
No full textThis thesis aims to detect spatial and non-spatial conflicts in product design arising out of requirements in different design domains. Handling such conflicts requires modelling the non-geometric knowledge outside the computer-aided-design (CAD) model and associating it with product geometry in the CAD model.
Product knowledge needs to be identified and structured to help detect spatial and non-spatial conflicts. System modelling Language (SysML) is used to model the product knowledge, specifically product structure, unoccupied or empty space blocks, product life-cycle states, and design domains. SysML Requirements diagram is used to model the design requirements. Block definition diagrams (BDD) model the product structure with empty space blocks. The relationship between Product life-cycle states and design domains is established through product structure and empty space blocks. The unused space blocks are allocated to design requirements from different domains based on the relationship modelled. These blocks are used to capture and represent knowledge inside CAD tools.
Identification and decomposition of empty spaces inside CAD have been discussed, and an algorithm has been developed for automating the decomposition task for empty space blocks. A method of modelling intended empty spaces using parametric primitive shapes has also been described.
An algorithm has been developed to capture the relationship between the empty space blocks and the associated design domains, product life-cycle states, and design requirements. An algorithm has been developed to identify and represent associativity between these blocks and features in the CAD model. Two case studies involving heat sink assembly and coupled-cavity slow-wave structure have been discussed to demonstrate and validate the framework for empty space modelling.
Spatial conflict detection using graph-based and matrix-based approaches like DSM, DMM, and MDM, have been discussed and implemented. Two different methods were developed for spatial conflict detection and evaluated for ease of implementation and visualization. Two case studies of Coupled-cavity travelling wave tube (CCTWT) Slow-wave structure (SWS) design and high-power RF window design are used to showcase the detection of spatial conflicts under multiple design domains.
Next, the thesis proposes using an octree model to represent the spatial conflicts across multiple design domains. This approach has the advantage of independence from the CAD system for computations and visualization. It also enables the resolution of dynamic spatial conflicts. A framework has been developed to create an octree model linked with intended empty spaces in the product. The model is also connected with design requirements, product life-cycle states, and functional design domains.
Two case studies of coupled-cavity travelling wave tube (CCTWT) slow-wave structure (SWS) design and high-power RF window design demonstrate the use of octree to visualize spatial conflicts outside of a CAD environment.
An octree-based voxel model gives us spaces free from conflicts, which designers can recommend for locating new parts in the assembly and carrying out layout planning. Spatial conflict detection for moving parts is also implemented where conflicts are detected for dynamic cases.
The octree-based voxel model is discussed and implemented to represent and compute non-spatial conflicts across multiple design domains. A framework has been developed to create an octree-based voxel model linked with physics-based non-spatial constraints. A case study of a travelling-wave tube (TWT) collector design has been taken to showcase the framework's capabilities for non-spatial conflict detection. It enables visualization of the octree model outside the CAD platform, along with spatial and non-spatial conflicts detected by associativity established between the part, physics-based constraints, and the product information available in the SysML model.
The thesis concludes by summarizing the contributions made and identifying areas for further research
Design and Development of Implantable Electrode Arrays for Recording Signals from Rat’s Brain
No full textSeizure is a neurological disorder due to abnormal, excessive, and hypersynchronous discharges from an aggregate of central nervous system neurons. Epilepsy describes the clinical phenomena of a condition of recurrent and chronic seizures. Epilepsy is one of the most common neurological disorders globally, with approximately 50 million epileptic patients. About 10 million people (~20% of the total) have epilepsy in India. Antiepileptic drugs (AEDs) are available for treatment; however, these drugs are not epilepsy specific. Moreover, medication is not effective for refractory (or intractable) epilepsy. The Electroencephalogram (EEG) recorded with scalp electrodes is generally used to screen and monitor epilepsy and to detect the epileptic focus. However, seizure activities occurring within deeper brain structures are often not recorded by EEG. Thus, the intracranial electrodes are used to detect seizures in deeper brain regions. Recording of electrocorticography (ECoG) signals from cortical surface and local field potentials (LFPs) from a brain depth help to localize the seizure onset region and track the seizure progression.
This thesis describes the design and development of electrode arrays for recording brain signals intracranially to study convulsant and AED's effects on brain activities. EEG is a widely utilized electrophysiological monitoring technique to record the electrical activities of the brain for both research and clinical applications. Recently, the popularity of ECoG, compared to EEG, has increased due to relatively higher spatial resolution and improved signal-to-noise ratio (SNR). ECoG signals, the intracranial recording of electrical signatures of the brain, are recorded by minimally invasive planar electrode arrays placed on the cortical surface. Flexible arrays minimize the tissue damage and induce minimal inflammation upon implantation. However, the commercially available implantable electrode arrays offer a poor spatial resolution (electrodes with ~4 mm with a pitch of 10 mm). Therefore, there is a need for an electrode array with a higher density of electrodes to provide better spatial resolution for mapping the brain surface. We have developed a biocompatible, flexible, and high-density microelectrode array (MEA) for a simultaneous 32-channel recording of ECoG signals. In acute experiments, we have demonstrated that the fabricated MEA can record the baseline ECoG signals (amplitude ±100 µV range), the induced epileptic activities (amplitude ±1500 µV range), and the recovered baseline activities (amplitude ±50 µV range) after administering anti-epileptic drug from the cortical surface of an anesthetized rat (n=3 subjects). A significant increment in amplitude (approximately ten times baseline) of the ECoG signals was observed as the epilepsy was induced by topical application of a convulsant (bicuculline). After intraperitoneal application of an antiepileptic drug (phenytoin sodium), the recovered baseline signals with a lower amplitude than the normal baseline signals were observed. The power spectral density was determined to observe the frequency components present (up to 60 Hz) in the signal. The spatial distribution of the signals was studied for onset zone localization.
Similarly, the design, fabrication, and characterization of a flexible biodegradable electrode array were described. The chronic in vivo experiments exhibited the capability of recording ECoG signals from the somatosensory cortex of rats (n=2 subjects). The PCBs were designed and fabricated for interfacing the array with OpenBCI Cyton Boards, used as the signal acquisition system. The epileptic activities were induced by peripheral electrical stimulation to the left forepaw of the rats and the induced epileptic activities were suppressed by administering an antiepileptic drug. The activities exhibited a significant variation in three neurological conditions: (i) Baseline (amplitude: ±30 µV), (ii) induced epileptic activities (amplitude ±200 µV range), and (iii) recovered baseline (amplitude ±10 µV range). The presence of frequency components in the recorded signals was studied by power spectral density (up to 100 Hz), which shows a significant change in the presence of low-frequency components (in the 1 Hz – 30 Hz range) during the induced epileptic activities.
Though the ECoG signals can achieve better spatial resolution than EEG, it offers a limited understanding of the activities at a brain depth where the signal originates. Recently, the implanted depth electrodes have been used for acquiring signals (LFPs) from deeper regions of the brain to study the cortex, hippocampus, thalamus, and other deep brain structures. Our other work reports the design and fabrication of a silicon-based 13-channel single-shank microneedle electrode array to acquire and understand LFPs from a rat’s brain. In acute in vivo experiments, LFPs from the somatosensory cortex of anesthetized rats (n=3 subjects) were recorded and were acquired using OpenBCI Cyton Daisy Biosensing Board at normal, epileptic (chemically induced), and recovered (after application of anti-epileptic drug) conditions. The LFPs exhibited a significant variation in three neurological conditions: (i) Baseline LFP (amplitude ±25 µV range), (ii) epileptic activities (amplitude ±250 µV range), and (iii) recovered baseline (amplitude ±25 µV range). The presence of frequency components in the recorded signals was studied by power spectral density (up to 60 Hz), which shows a significant change in the presence of low-frequency components (up to 30 Hz) during induced epilepsy. The recorded signals helped us understand the response of the brain to convulsant and AED across the depth.
We envisage these models may help to evaluate and understand the efficacy of AEDs using rats as an animal model. The results can establish that the OpenBCI Cyton Daisy Biosensing Boards can be used as a signal acquisition system for in vivo recording of rat brain signals, which was not widely used or reported
Broadband Millimeter-Wave CMOS Transceiver for 5G Mobile Communication and Radar-Based Sensing
No full textTo meet the ever-growing demand for higher data rates in communication networks and higher range and velocity resolutions in automotive radar sensors, fifth-generation (5G) new radio (NR) transceivers and radars used in autonomous vehicles use spectrally efficient modulation formats with large channel bandwidths available at millimeter wave (mm-wave) frequencies. However, designing energy-efficient broad-band transceivers with low manufacturing cost at mm-wave frequencies is extremely challenging because of the performance degradation of integrated circuit (IC) components, impairments due to packaging, and increased free-space path loss. This thesis presents a high-performance, compact, low-cost mm-wave transceiver solution for 5G NR and automotive radar sensors.
A 28-GHz transceiver based on the local-oscillator (LO) phase-shifting architecture enabling gain-invariant phase tuning is designed in a 65-nm CMOS technology with wirebond-based packaging, enabling low manufacturing cost. The transceiver chip consists of a transmitter, a receiver, and an LO phase-shifting and distribution network. The transmitter employs an energy-efficient architecture based on direct-digital RF modulators (DDRMs) using digital-to-RF converters (DRFCs) to support BPSK, QPSK, 16-QAM, and 64-QAM modulation formats in 4 GHz of channel bandwidth accommodating both 5G and radar waveforms. The receiver is based on the complex-baseband zero-IF architecture using an active downconversion mixer with a transimpedance amplifier (TIA) load with up to 4 GHz of IF bandwidth. The downconverted signal is dynamically amplified by broad- band variable gain amplifiers (VGAs) based on Cherry-Hooper gain stages to compensate for the quadrature gain mismatch and relax the linearity requirement for analog-to-digital converters (ADCs). The in-phase and quadrature-phase LO signals are generated on-chip using a transformer-based quadrature hybrid driven by a coarse/fine tunable LC tank based phase shifter. The transceiver utilizes low-k transformer based fourth order networks for broadband input and output matching of the low-noise amplifier (LNA) and power amplifier (PA) as well as for interstage matching. The mm-wave chip-to-board interfaces are optimized using a scalable broadband model for wirebond interconnects developed using experimental data.
A broadband dielectric characterization technique using coplanar waveguide (CPW) based test structures is developed to extract the frequency-dependent dielectric properties of the silicon substrate, typically not characterized by the foundry. This enhances the accuracy of the electromagnetic (EM) models of on-chip passive devices and interconnect parasitics, and consequently, the performance of the transceiver
Investigation of Au-free AlGaN/GaN HEMT on Silicon with substrate transfer process: From Ohmic contacts to RF performance
No full textThe technological importance of III-nitride based high electron mobility transistors (HEMTs) for power switching and RF applications is growing rapidly, owing to two advantageous electrical properties. While high sheet carrier concentration and mobility of charge carriers confined in the 2DEG translate to high current levels, high bandgap of III-N based materials leads to high breakdown voltage. However, III-N based devices have not been able to penetrate the semiconductor market as deep as Silicon based technologies. High costs associated with the epitaxial growth of the III-N based HEMT stack on hetero-substrates like Si, SiC or Sapphire as well as high costs associated with III-N based device processing are the two key reasons. While development of III-N growth on Silicon can potentially bring down the net cost associated with growth, CMOS compatible fabrication of III-N based devices can reduce the cost associated with device processing as this will permit the processing of III-N devices in existing CMOS fab lines. The most important difference between conventional and CMOS compatible processing of III-N based devices is the absence of gold (Au) in CMOS process, since Au has high diffusion rates in Si even at moderate temperatures and it also acts as a deep impurity in Silicon leading to formation of electrical trap levels in Silicon bandgap leading to degradation in device performance. Further, sputtering is preferred over electron beam evaporation for metal deposition in CMOS processes from the reliability standpoint. CMOS also uses a lower thermal budget when compared to III-N processing. In this work, we focus on the development of Au-free fabrication process for AlGaN/GaN based RF HEMT devices.
The first portion of this work deals with the process development for fabrication of Au-free Ohmic contacts to AlGaN/GaN HEMT. Conventionally, Ti/Al/Ni/Au based metal stack deposition by electron beam evaporation followed by high temperature post metal annealing (PMA) at 850 ℃ in N2 has been used for Ohmic contact formation. In this work, we have developed a Ti/Al/Ti/W based metal stack deposited by sputtering and followed by a PMA step at relatively moderate temperature of 600 ℃ in N2. However, it is preceded by AlGaN barrier recess etch step. In this work, we study the effect of etch chemistry (digital etch using BCl3/O2 and low etch rate BCl3/Cl2 RIE etch) and metallization scheme (using sputtered Ti/Al and Ti/Al/Ti/W stacks) on the recessed Au-free Ohmic contacts. An optimum recess etch of entire AlGaN barrier using digital etching leads to better uniformity in contact resistance (RC). Further, use of Ti/W cap layer on Ti/Al leads to low contact resistance of 0.56 Ω-mm with a smooth contact surface morphology. Possible mechanism for carrier transport through the contacts has been discussed based on temperature dependent electrical characterization and field emission is found to be the dominant mechanism of carrier transport.
The second portion of the work deals with electrical performance comparison of Au-based and Au-free AlGaN/GaN High electron mobility transistors (HEMT) on Silicon (Si). The chemical composition of both types of contacts i.e., Ti/Al/Ti/W (Au-free) and Ti/Al/Ni/Au (Au-based) have been studied using transmission electron microscope (TEM). For the former, top W layer is found to be restricted in its interaction with the lower metal layers, leading to a continuous W cap at the contact surface which results in a 20x improvement in the surface roughness for the Au-free contacts. Transistors have been fabricated with Ni/W and Ni/Au based gate metal stacks for Au-free and Au-based processes respectively. From HEMT measurements, devices with Au-free contacts are found to exhibit +0.4 V shift in the threshold voltage and a 10x increase in the gate leakage which is attributed to the strain associated with the sputtered W-capping layer of the Ni/W gate-metal stack and plasma induced damage caused at the barrier surface due to high-power sputter deposition of gate contact. The lowest RC value of 0.4 Ω-mm obtained for the Au-free Ohmic contacts is comparable to the RC value of 0.38 Ω-mm obtained for the Au-based Ohmic contacts used in this work.
Flexible III-N based device technology is rapidly gaining attention for use in RF and optical applications. Efficient III-N based RF devices on flexible substrates can lead to high performing conformal wireless communication and RADAR systems, while efficient III-N based optical devices and associated circuitry on bendable and inexpensive polymer substrates can lead to highly efficient futuristic consumer electronic products like foldable transparent mobiles, billboards etc. A CMOS compatible III-N device process followed by economical and relatively simple wafer-level transfer method onto inexpensive polymer-based flexible substrates can lead to scalable and high-performance III-N device technology on flexible and conformal substrates. In the third portion of this work, we report on the electrical performance of AlGaN/GaN HEMT fabricated using Au-free process, after being transferred onto flexible Kapton. Electrical characteristics of the flexible HEMT indicate5-10% higher on-current when bent with radius of curvature of 2 cm (at low drain bias voltages), while the off-state performance remains unaffected. Initially, 2DEG properties like field-effect mobility and carrier concentration have been extracted. While FATFET measurements indicate negligible change in field-effect mobility, C-V measurements indicate ~10% reduction in 2DEG concentration after transfer. The comparison of electrical characteristics of the Au-free HEMT’s indicates ~50% reduction in on-current of the transferred devices, which is attributed to heating of the transistor channel caused due to low thermal conductivity of the polymer Kapton tape. Electrical characteristics of the flexible HEMT carried out under drain pulsing further support the above observation.
The last portion of this work deals with process development and performance characterization of Au-free AlGaN/GaN RF HEMT. DC, small signal RF and pulsed I-V measurements have been carried out to ascertain the performance of the Au-free RF HEMT. De-embedded fT of > 40 GHz were achieved in 2 x 50 μm devices with 250 nm gate length and source-to-drain spacing of 6 μm. Substrate ramp measurements have been carried out in order to ascertain the significant drain lag current collapse seen in the devices and negative charge stored in the buffer under high drain bias is found to be the reason for the large drain lag seen in the devices. Delay-time analysis and small-signal modelling of the devices were carried out and the small-signal simulated S-parameters match well with the measured S-parameters of the actual devices.
In conclusion, we have developed Au-free III-N based HEMT process for fabrication of AlGaN/GaN HEMT and compared their performance with conventionally processed Au-based devices. While these devices have been transferred on low-cost Kapton as a work in the direction of flexible GaN based devices, the same process has been used to fabricate RF devices with good small signal characteristics
Development of High-Performance Piezoelectric Micromachined Transducers for Near Ultrasound
No full textNear-ultrasound refers to sound with frequencies just above the range of human hearing, from about 18 to 40 kHz. This band is rarely used for typical ultrasound applications and is ignored for all except the most demanding audio applications. We highlight the advantages of using this band and present a design study on the development of high-efficiency, resonant transducers for near-ultrasound. Piezoelectric Micromachined Ultrasound Transducers, or PMUTs, are MEMS resonators that are used to generate and receive ultrasound and acoustic waves. They are fabricated as multilayered diaphragms consisting of a passive structural layer coated with a piezoelectric material sandwiched between metal films.
In this dissertation, we report the realization of a novel near-ultrasound PMUT system especially designed for Data-over-Sound (DoS) applications. This realization includes investigation of new transducer designs, innovation in fabrication processes, and a significant advance in acoustics and electronics integration. We use analytical and coupled finite element models of clamped circular plates with in-plane stresses to generate design maps for PMUTs. Residual tensile stresses generated during fabrication processes have the effect of stiffening the diaphragms and increasing their resonant frequencies. We experimentally estimate the magnitude of these stresses in sol-gel PZT-coated SOI wafers and fabricate transducers with dimensions optimized for near-ultrasound. The transducers are 50 times smaller and 20 times more efficient than conventional electrodynamic micro speakers in the near-ultrasound range.
We then present a novel design for PMUTs with “bossed” diaphragms that allows further reduction in device footprint and power consumption while improving sensitivity and efficiency. The dimensions of the central boss structure are optimized using simulations. The fabricated devices are found to be up to 10 times smaller than conventional PMUTs for the same frequencies, and less sensitive to variations in residual stress.
We have studied and optimized the effects of packaging and the acoustic environment on the performance of the transducers using finite element and boundary element acoustic simulations. The devices are packaged with 3D-printed acoustic resonators and horns designed to boost sensitivity, improve bandwidth, and widen the directivity of the transducers. The results of the simulations are experimentally verified by scanning the acoustic field of the transducers. The transducers are finally integrated into battery- and solar-powered DoS beacons and wireless sensor nodes, complete with a low-power microcontroller for modulation/demodulation, a low Q-current amplifier, a MEMS microphone, an acoustic resonator, and the near-ultrasound transducer — all in a compact package with a transmission range of up to 30 meters and a battery reserve of up to 4 weeks
Monte Carlo simulations of Electric Double Layer Capacitors with aqueous electrolytes
No full textElectric double layer capacitors (EDLCs) also known as supercapacitors are a category of energy storage devices that are known for their remarkable power delivery (upto 1000KW/kg). The energy storage is through physical adsorption of ions on the surfaces or inside the pores of electrode forming electric double layers. This simple charge storage mechanism leads to a lifetime of > 1 million cycles and cyclic efficiency of >90%. However, they are limited by their energy densities which don't exceed 10Wh/Kg. The specific energy can be improved by increasing the capacitance of the EDLC. The differential capacitance of an EDLC holds complex relation with properties of electrode, electrolyte, and operating conditions. The Molecular simulations studies on EDLCs help uncover these microscopic phenomena that govern the capacitance in EDLCs.
The molecular simulations of EDLC requires simulation of positive, negative electrodes and electrolyte, which is usually done in single simulation box. This restricts the size of the electrode in one of the dimensions making it finite in that direction and limits the simulated EDLC system sizes to nano meters. The problem with system sizes can be alleviated using Gibbs Ensemble Monte Carlo simulation technique where, both the electrodes are simulated in separate boxes and each box is periodically repeated in three dimensions eliminating any system size discrepancies. Furthermore, Grand canonical ensemble is used to avoid the simulation of bulk electrolyte phase. The electrodes are subjected to constant external potential difference to replicate the experimental conditions.
In this work, Gibbs ensemble-based Monte carlo simulations on EDLC are performed with aqueous NaCl solution as electrolyte and Graphene as electrode for a set of operating conditions. Graphene is modelled as slit-pore type electrode . The water molecules are explicitly simulated using Continuous Fractional component Monte Carlo method to include the hydration effects of ions by water. Differential capacitance is computed within voltage of 0-2V for pore widths of 0.65nm, 0.79nm and 0.9nm at temperatures of 416K and 350K for electrolyte concentration of 2.1M. The simulations help understand the effect of pore width, temperature, and voltage on the performance of EDLCs