12023 research outputs found
Sort by
Thin-Film Silicon Photovoltaics: Characterization of Thin-Film Deposition and Analysis of Enhanced Light Trapping from Scattering Nanoparticle Arrays
Get PDFThin-film solar cells have the potential to significantly decrease the cost of a finished device by cutting materials cost, and the characteristics of carrier transport through a thin film can concurrently increase the device performance over that of a wafer-based cell while tolerating a higher defect density in the absorbing material. However, while silicon is an attractive material for use in solar cells due to its nearly ideal band gap for single-junction cells and its relative abundance, its inefficient absorption of infrared light necessitates the development of light-trapping techniques to avoid losses in current generation. This thesis research has focused on two important goals: the development of a scalable thin-film silicon deposition method that produces high-quality material at minimal cost, and the evaluation of light-trapping mechanisms that will increase photon absorption in these films. Hot-wire chemical vapor deposition is used to fabricate silicon thin films with high crystalline fractions even on inexpensive substrates, and films grown with appropriate growth conditions exhibit initial open-circuit voltages above 450 mV, and while challenges in passivation still exist, this research illustrates the potential of this highly scalable and inexpensive deposition technique. Silver and silicon subwavelength structures were then both fabricated and simulated on ultra-thin silicon films on SiO2 to evaluate their potential for increasing light absorption through plasmonic and physical scattering mechanisms, and spectral response measurements demonstrate over a ten-fold increase in carrier generation with a metal nanoparticle surface array. Periodic dielectric structures exhibit Bloch modes in both measurement and simulation, with an increase in overall quantum efficiency of over 11% from both a flat silicon layer and one that is randomly textured. These results highlight the significant role of scattering particle distribution in determining the light trapping characteristics in these devices. Design guidelines have been explored for exploiting resonant modes in these structures, and this thesis demonstrates the potential for both metal and dielectric surface arrays to dramatically increase light absorption in silicon thin films
Intrinsic Gradient Networks
Get PDFArtificial neural networks are computationally powerful and exhibit brain-like dynamics. Unfortunately, the conventional gradient-dependent learning algorithms used to train them are biologically implausible. The calculation of the gradient in a traditional artificial neural network requires a complementary network of fast training signals that are dependent upon, but must not affect, the primary output-generating network activity. In contrast, the network of neurons in the cortex is highly recurrent; a network of gradient-calculating neurons in the brain would certainly project back to and influence the primary network. We address this biological implausibility by introducing a novel class of recurrent neural networks, intrinsic gradient networks, for which the gradient of an error function with respect to the parameters is a simple function of the network state. These networks can be trained using only their intrinsic signals, much like the network of neurons in the brain.
We derive a simple equation that characterizes intrinsic gradient networks, and construct a broad set of networks that satisfy this characteristic equation. The resulting set of intrinsic gradient networks includes many highly recurrent instances for which the gradient can be calculated by a simple, local, pseudo-Hebbian function of the network state, thus resolving a long-standing contradiction between artificial and biological neural networks. We demonstrate that these networks can learn to perform nontrivial tasks like handwritten digit recognition using only their intrinsic signals. Finally, we show that a cortical implementation of an intrinsic gradient network would have a number of characteristic computational, anatomical, and electrophysiological properties, and review experimental evidence suggesting the manifestation of these properties in the cortex.</p
Mixed Functionality Semiconductor Surfaces: Formation, Characterization, Interfacial Dynamics, and Applications
Get PDFThe properties of any semiconductor device rely on the charge separation characteristics at interfaces within that device. The charge separation characteristics include relative energetics, interfacial electronic states, and the presence or absence of insulating layers. More importantly, the interfacial properties determine the maximum solar conversion efficiency for a photoelectrochemical or photovoltaic device. Solution-based halogenation/alkylation chemistry was used to functionalize Si surfaces. The chemistry was adapted to allow for the controlled formation of multicomponent molecular monolayers. Functional molecules were incorporated by the mixed monolayer approach, and lowered densities of surface electronic defect states and increased resistance toward the formation of deleterious Si oxides were observed. Heck coupoling reactions were developed at thiophene-containing monolayers. Thiophene terminated Si(111) surfaces had defect frequencies of > 1 defect per 1,000 surface atoms, too large for solar energy conversion applications, while multicomponent CH3/thiophene monolayers had defect densities of < 1 per 500,000 surface atoms. Robust secondary chemistry at Si(111) with facile charge transfer to covalently linked molecules with preservation of surface electronic properties was shown for the first time. Molecular adsorbates with interesting electronic dipoles, such as bromothiophene, were incorporated into mixed monolayers. The electron distribution across the surface dipole caused a shift in the work function of Si by > 600 mV. The fundamental mechanism of Wf shift was elucidated by a combined ab initio and experimental study, and the dependence of Si band-edge positions on pH was relieved using Si-C bonds. Designer surface chemistry was used to covalently link Si microwires within a exible PDMS matrix, and a direct correlation between the surface bonding mechanism and interfacial adhesion strength was unambiguously observed. The formation and electronic properties of Si/PEDOT junctions were studied, where PEDOT was covalently linked to the Si surface via electropolymerization initiation at a mixed molecular monolayer containing 2,2':5',2"-terthien-5"-yl- groups. Low resistance, ohmic contacts were made at p-Si. Aldehyde groups were also incorporated into mixed monolayers. Facile, low-temperature atomic layer deposition (ALD) of Al2O3, MnO2,and TiO2 on aldehyde-functionalized Si was achieved. Neither surface oxidation nor surface electronic defects formed during ALD, which has not been shown previously.</p
Late Cretaceous Gravitational Collapse of the Southern Sierra Nevada Batholith and Adjacent Areas Above Underplated Schists, Southern California
Get PDFThe greater Sierra Nevada batholith (SNB) is an ~ 600 km long NNW-trending composite arc assemblage consisting of a myriad of plutons exhibiting a distinct transverse zonation in structural, petrologic, geochronologic, and isotopic patterns. However, south of 35.5 °N: 1) the depth of exposure increases markedly; 2) primary zonation patterns swing up to 90˚ westward, taking on an east-west trend; 3) western zone rocks are truncated by eastern zone rocks along the proto-Kern Canyon fault, a Late Cretaceous oblique ductile thrust; and 4) fragments of shallow-level eastern SNB affinity rocks overlie deeper-level western zone rocks and subjacent subduction accretion assemblages (Rand, San Emigdio, and Sierra de Salinas schists) along a major Late Cretaceous detachment system. Integration of these observations with new and existing data reveals a temporal relationship between schist unroofing and upper crustal extension and rotation. I present a model whereby Late Cretaceous shallow subduction and subsequent trench-directed channelized extrusion of the schist triggered gravitational collapse of the overlying crustal column. This overarching model is based on several investigations summarized below.
Thermobarometry, thermodynamic modelling and garnet diffusion modelling are presented that elucidate the tectonics of subduction and eduction of the San Emigdio Schist. I document an upsection increase in peak temperature (i.e. inverted metamorphism), from 590 to 700 °C, peak pressures ranging from 8.5 to 11.1 kbar, limited partial melting, microstructural evidence for large seismic events, rapid cooling (825–380 °C/Myr) from peak conditions and an “out and back” P–T path. Progressive cooling and tectonic underplating beneath an initially hot upper plate following the onset of shallow subduction provide a working hypothesis explaining high temperatures, inverted metamorphism, partial melting, and the observed P–T trajectory calculated from the San Emigdio body.
New geologic mapping and microstructural analysis indicate that the schist was transported to the SSW during structural ascent along a mylonitic contact (the Rand fault and Salinas shear zone) with upper plate assemblages. Crystallographic preferred orientation patterns in deformed quartzites reveal a decreasing simple shear component with increasing structural depth, suggesting a pure shear-dominated westward flow within the subduction channel and localized simple shear along the upper channel boundary. The resulting flow type within the channel is that of general shear extrusion.
Structural, thermobarometric, U-Pb geochronologic, and geochemical data from plutonic and metamorphic framework assemblages in the southern SNB also suggest SSW-directed transport of upper plate(s) along a major Late Cretaceous detachment system. The timing and pattern of regional dispersion of crustal fragments in the southern SNB is most consistent with Late Cretaceous collapse above the underplated schist. These observations imply a high degree of coupling between the shallow and deep crust during high magnitude extension. Zircon (U-Th)/He data presented herein reveal a rapid cooling event at 77 ± 5 Ma, probably reflecting the time of large magnitude detachment faulting. A comparison of this dataset with existing apatite (U-Th)/He thermochronometry suggests that the development of modern landscape and arrangement of tectonic elements in southern California was greatly preconditioned by Late Cretaceous tectonics.
Finally, detrital and metamorphic zircon of the structurally highest and earliest subducted portions of the San Emigdio Schist constrain the depositional age to between ca. 102 and 98 Ma. Zircon oxygen isotope data from both lower plate schist and upper plate batholithic assemblages reveal a δ18O shift of ~ 1.5‰ between igneous (~ 5.5‰) and metamorphic (~ 7‰) domains. These results, taken with previous zircon and whole-rock δ18O measurements, provide evidence for massive devolatilization of the San Emigdio Schist and fluid traversal of upper plate batholithic assemblages, thereby altering the isotopic composition of overlying material. Furthermore, the timing of fluid-rock interaction in the southwestern SNB is coincident with eastward arc migration and an associated pulse of voluminous magmatism. I posit that during flattening of the Farallon slab the schist was rapidly delivered to the magmatic source, where ensuing devolatilization triggered a magmatic flare-up in the southeastern SNB. This short-lived (less than 15 Myr) high-flux event was followed by the termination of arc magmatism as the shallow subduction zone approached thermal equilibrium.</p
Regulation of Wild-Type and Mutant p53 through DNA-mediated Charge Transport
Get PDFThe global transcription factor p53 controls many cellular processes, including the cellular response to oxidative stress. It had been determined that that dissociation of wild type p53 from its promoter site can occur upon DNA-mediated oxidation. In this work, we use site-directed mutagenesis to construct charge-deficient mutants of p53; the chemistry of DNA-mediated oxidation of p53 was examined using these mutants.
The control point for p53 oxidation through DNA-mediated charge transport (DNA CT) is cysteine 275. Using differential thiol labeling and detection of modified peptides with mass spectrometry, we demonstrated that cysteines 124 and 141 in superstable p53 form a terminal disulfide bond upon DNA-mediated oxidation. This leads to a conformational change that inhibits DNA from binding by p53. The disulfide formed between cysteines 124 and 141 is a result of a series of disulfide bond exchange across the protein from the DNA base stack.
We also investigated the dependence of p53 oxidation on DNA sequences. ESMA analysis of biologically derived p53 recognition sequences with varying quantities of guanine doublets and triplets showed efficient p53 oxidation to depend on the presence of low energy GG or GGG sites. Moreover, consistent results were found with biologically derived promoter sequences. Sequence S100A2, with guanine triplets on the same strand, showed the most oxidation of p53, followed by ODC1 and caspase-1. We confirmed these sequence-specific effects by measuring the change in expression level of the genes after induction of DNA CT in vivo. S100A2 mRNA levels decreased after photooxidant and light treatment, reflecting the oxidation and dissociation of p53 from the S100A2 site. However, caspase-1 and ODC1 mRNA levels remained the same, indicating less DNA-mediated p53 oxidation.
The results from this study illustrate how protein oxidation at a distance through DNA CT contributes to cellular signaling. This oxidative signaling can control how p53 regulates gene expression under oxidative stress, and this signaling may be disrupted in cancerous cells.</p
Investigating DNA-Mediated Charge Transport by Time-Resolved Spectroscopy
Get PDFIn all organisms, oxidation threatens the integrity of the genome. Numerous studies have suggested that DNA-mediated charge transport (CT) may play an important role in the sequestration, detection, and repair of oxidative damage. To fully understand the mechanism of DNA-mediated CT, it is necessary to characterize transient intermediates that arise during the reaction and to determine the lifetimes of these intermediates. Time-resolved spectroscopy is the most appropriate experimental method for such observations. Each intermediate has a characteristic spectrum. By observing time-dependent changes in the absorption spectrum of the sample, it is therefore possible to determine what species are present at a particular time and how long it exists in solution. Experiments presented here involve the use of time-resolved spectroscopy to better understand the process of DNA-mediated CT.
The study of DNA-mediated CT requires a robust and consistent method for triggering the CT reaction. The metal complexes that have traditionally been used for this purpose provide several advantages over organic phototriggers: they are synthetically versatile, they are stable in solution, they exhibit rich photophysics, and many are strong photooxidants. However, the spectroscopic features used to follow the photochemical processes triggered by these probes are generally broad optical bands. These can be difficult to resolve in samples that contain several absorbing species. For this reason, we have developed a Re photooxidant bearing a set of vibrationally active carbonyl ligands that can be covalently tethered to DNA. Unlike many absorption bands in the visible range, the vibrational absorption bands of these ligands are narrow, well-resolved, and specific. Such probes can be used to follow the complex photophysical pathways observed in biochemical systems with good precision, making them useful for the study of DNA-mediated CT.
Specifically, the complex [Re(CO)3(dppz)(py′-OR)]+ (dppz = dipyrido[3,2-a:2′,3′-c]-phenazine; py′-OR = 4-functionalized pyridine) offers IR sensitivity and can oxidize DNA directly from the excited state. The behavior of several covalent and noncovalent Re-DNA constructs was monitored by time-resolved IR (TRIR) and UV/visible spectroscopies, as well as biochemical methods, confirming the ability of the complex to trigger long-range oxidation of DNA. Optical excitation of the complex leads to population of metal-to-ligand charge transfer excited states and at least two distinct intraligand charge transfer excited states. Several experimental observations are consistent with charge injection by excited Re*. These include similarity between TRIR spectra and the spectrum of reduced Re observed by spectroelectrochemistry, the appearance of a guanine radical signal in TRIR spectra, and the eventual formation of permanent guanine oxidation products. The majority of reactivity occurs on the ultrafast time scale, although processes dependent on slower conformational motions of DNA, such as the accumulation of oxidative damage at guanine, are also observed.
The photooxidation activity of this Re complex was compared directly to that of other metallointercalators that have been used previously in our laboratory to oxidize DNA. The complexes [Rh(phi)2(bpy′)]3+ (phi = 9,10-phenanthrenequinone diimine; bpy′ = 4-methyl-4′-(butyric acid)-2,2′-bipyridine), [Ir(ppy)2(dppz′)]+ (ppy = 2-phenylpyridine; dppz′ = 6-(dipyrido[3,2-a:2′,3′-c]phenazin-11-yl)hex-5-ynoic acid), and [Re(CO)3(dppz)(py′-OH)]+ (py′-OH = 3-(pyridin-4-yl)-propanoic acid) were each covalently tethered to DNA. Biochemical studies show that upon irradiation, the three complexes oxidize guanine by long-range DNA-mediated CT with the efficiency: Rh > Re > Ir. Comparison of spectra obtained by spectroelectrochemistry after bulk reduction of the free metal complexes with those obtained by transient absorption (TA) spectroscopy of the conjugates suggests that excitation of the conjugates at 355 nm results in the formation of the reduced metal states. Electrochemical experiments and kinetic analysis of the TA decays verify that the primary factors responsible for the trend observed in the guanine oxidation yield of the three complexes are the thermodynamic driving force for CT, variations in the efficiency of back electron transfer, and coupling to DNA.
The ability of redox-active DNA-binding proteins to act as hole sinks in DNA-mediated CT systems was also studied by time-resolved spectroscopy. Such experiments are designed to provide support for the utilization of DNA-mediated CT in biological systems. In studies involving the cell cycle regulator p53, photoexcitation results in the formation of a weak transient band at 405 nm. This band, which is not observed in samples lacking the protein, resembles the primary spectral feature of the tyrosine cation radical. Although the signal is weak and reproducibility is inconsistent, these results suggest that photolysis of the sample leads to DNA-mediated oxidation of tyrosine in p53. Similar experiments were conducted on the transcriptional activator SoxR. Here, the presence of dithionite, required in solution to keep the protein reduced, complicates the photochemistry of the system considerably. Regardless, a weak absorbance at 418 nm that develops following photolysis at 355 nm provides evidence for the DNA-mediated oxidation of the protein. The behavior of the base excision repair protein endonuclease III was also observed in the presence of DNA and metal complex oxidants. In flash-quench studies, addition of the protein results in the formation of a strong negative signal at 410 nm in TA traces. In studies involving direct photooxidation by Rh, Ir, and Re complexes, no new transients are detected upon the addition of protein, but changes in the intensities of the resultant TA spectra and in the steady-state absorbance spectra following photolysis indicate that DNA-mediated oxidation of the protein may be taking place.
The experiments described here comprise several new developments in the story of DNA-mediated CT. First, proof of concept has been given for a valuable new vibrationally-active Re probe. Further modifications on the characteristics of this complex and further study by time-resolved vibrational spectroscopy will allow us to observe DNA-mediated CT with high spectral resolution. Second, comparison between this Re probe and established photooxidants shows that the Re complex is a strong photooxidant in its own right and that this complex can be added to our growing toolbox of CT phototriggers. Third, time-resolved studies involving redox-active proteins have provided preliminary direct evidence for the ability of these proteins to serve as CT probes themselves. Further refinement of the experimental methods used in these experiments will allow us to observe such processes with greater sensitivity, increasing our knowledge of the mechanism and applications of DNA-mediated CT.</p
Accelerating the Interplay Between Theory and Experiment in Protein Design
Get PDFProtein engineering techniques such as directed evolution and structure-based design aim to improve the properties of natural proteins. The next step, the de novo insertion of function into previously inert protein scaffolds, is the lofty promise of computational protein design. In order to achieve this goal reliably and efficiently, computational methods can be iteratively improved by cycling between theory and experiment.
Efforts to both accelerate the rate and broaden the information exchanged within protein design cycles form the core of this thesis. Improvements in the throughput of experimental stability determination allowed the thorough assessment of new multi-state and library design tools. Intending to alleviate the fixed backbone, single native state design approximation, the study found constrained molecular dynamics ensembles useful for core repacking applications. The subsequent development of automated liquid handling protocols for common molecular biology techniques brings design experiments to new levels of sample throughput. This technology facilitated the creation of a stability database encompassing every single mutant in a small protein domain. Although constructed to facilitate future computational training efforts, we answer a multitude of questions pertaining to mutational outcomes, distributions, positional sensitivity, tolerance, and additivity in the context of a protein domain.
By expanding the constraints of experimental molecular biology, this work opens up new possibilities in the efforts to train and assay new computational methodologies for protein engineering applications.</p
A Search for Pulsation in Young Brown Dwarfs and Very Low Mass Stars
Get PDFIn 2005, Palla and Baraffe proposed that brown dwarfs and very low mass stars (<0.1 solar masses) may be unstable to radial oscillations during the pre-main-sequence deuterium burning phase. With associated oscillation periods of 1-4 hours, this potentially new class of pulsation offers unprecedented opportunities to probe the interiors and evolution of low-mass objects in the 1-15 million year age range. Furthermore, several previous reports of short-period variability have suggested that deuterium-burning pulsation is in fact at work in young clusters.
For my dissertation, I developed a photometric monitoring campaign to search for low-amplitude periodic variability in young brown dwarfs and very low mass stars using meter-class telescopes from both the ground and space. The resulting high-precision, high-cadence time-series photometry targeted four young clusters and achieved sensitivity to periodic oscillations with photometric amplitudes down to several millimagnitudes. This unprecedented variability census probed timescales ranging from minutes to weeks in a sample of ~200 young, low-mass cluster members of IC 348, Sigma Orionis, Chamaeleon I, and Upper Scorpius. While I find a dearth of photometric periods under 10 hours, the campaign's high time resolution and precision have enabled detailed study of diverse light curve behavior in the clusters: rotational spot modulation, accretion signatures, and occultations by surrounding disk material. Analysis of the data has led to the establishment of a lower limit for the timescale of periodic photometric variability in young low-mass and substellar objects, an extension of the rotation period distribution to the brown dwarf regime, as well as insights into the connection between variability and circumstellar disks in the Sigma Orionis and Chamaeleon I clusters.</p
Virtual Earthquake Engineering Laboratory with Physics-Based Degrading Materials on Parallel Computers
Get PDFFor the last few decades, we have obtained tremendous insight into underlying microscopic mechanisms of degrading quasi-brittle materials from persistent and near-saintly efforts in laboratories, and at the same time we have seen unprecedented evolution in computational technology such as massively parallel computers. Thus, time is ripe to embark on a novel approach to settle unanswered questions, especially for the earthquake engineering community, by harmoniously combining the microphysics mechanisms with advanced parallel computing technology.
To begin with, it should be stressed that we placed a great deal of emphasis on preserving clear meaning and physical counterparts of all the microscopic material models proposed herein, since it is directly tied to the belief that by doing so, the more physical mechanisms we incorporate, the better prediction we can obtain.
We departed from reviewing representative microscopic analysis methodologies, selecting out “fixed-type” multidirectional smeared crack model as the base framework for nonlinear quasi-brittle materials, since it is widely believed to best retain the physical nature of actual cracks. Microscopic stress functions are proposed by integrating well-received existing models to update normal stresses on the crack surfaces (three orthogonal surfaces are allowed to initiate herein) under cyclic loading.
Unlike the normal stress update, special attention had to be paid to the shear stress update on the crack surfaces, due primarily to the well-known pathological nature of the fixed-type smeared crack model–spurious large stress transfer over the open crack under nonproportional loading. In hopes of exploiting physical mechanism to resolve this deleterious nature of the fixed crack model, a tribology-inspired three-dimensional (3d) interlocking mechanism has been proposed. Following the main trend of tribology (i.e., the science and engineering of interacting surfaces), we introduced the base fabric of solid particle-soft matrix to explain realistic interlocking over rough crack surfaces, and the adopted Gaussian distribution feeds random particle sizes to the entire domain. Validation against a well-documented rough crack experiment reveals promising accuracy of the proposed 3d interlocking model.
A consumed energy-based damage model has been proposed for the weak correlation between the normal and shear stresses on the crack surfaces, and also for describing the nature of irrecoverable damage. Since the evaluation of the consumed energy is directly linked to the microscopic deformation, which can be efficiently tracked on the crack surfaces, the proposed damage model is believed to provide a more physical interpretation than existing damage mechanics, which fundamentally stem from mathematical derivation with few physical counterparts.
Another novel point of the present work lies in the topological transition-based “smart” steel bar model, notably with evolving compressive buckling length. We presented a systematic framework of information flow between the key ingredients of composite materials (i.e., steel bar and its surrounding concrete elements). The smart steel model suggested can incorporate smooth transition during reversal loading, tensile rupture, early buckling after reversal from excessive tensile loading, and even compressive buckling. Especially, the buckling length is made to evolve according to the damage states of the surrounding elements of each bar, while all other dominant models leave the length unchanged.
What lies behind all the aforementioned novel attempts is, of course, the problem-optimized parallel platform. In fact, the parallel computing in our field has been restricted to monotonic shock or blast loading with explicit algorithm which is characteristically feasible to be parallelized. In the present study, efficient parallelization strategies for the highly demanding implicit nonlinear finite element analysis (FEA) program for real-scale reinforced concrete (RC) structures under cyclic loading are proposed. Quantitative comparison of state-of-the-art parallel strategies, in terms of factorization, had been carried out, leading to the problem-optimized solver, which is successfully embracing the penalty method and banded nature. Particularly, the penalty method employed imparts considerable smoothness to the global response, which yields a practical superiority of the parallel triangular system solver over other advanced solvers such as parallel preconditioned conjugate gradient method. Other salient issues on parallelization are also addressed.
The parallel platform established offers unprecedented access to simulations of real-scale structures, giving new understanding about the physics-based mechanisms adopted and probabilistic randomness at the entire system level. Particularly, the platform enables bold simulations of real-scale RC structures exposed to cyclic loading–H-shaped wall system and 4-story T-shaped wall system. The simulations show the desired capability of accurate prediction of global force-displacement responses, postpeak softening behavior, and compressive buckling of longitudinal steel bars. It is fascinating to see that intrinsic randomness of the 3d interlocking model appears to cause “localized” damage of the real-scale structures, which is consistent with reported observations in different fields such as granular media.
Equipped with accuracy, stability and scalability as demonstrated so far, the parallel platform is believed to serve as a fertile ground for the introducing of further physical mechanisms into various research fields as well as the earthquake engineering community. In the near future, it can be further expanded to run in concert with reliable FEA programs such as FRAME3d or OPENSEES. Following the central notion of “multiscale” analysis technique, actual infrastructures exposed to extreme natural hazard can be successfully tackled by this next generation analysis tool–the harmonious union of the parallel platform and a general FEA program. At the same time, any type of experiments can be easily conducted by this “virtual laboratory.”</p
Hierarchical Structures of Aligned Carbon Nanotubes as Low-Density Energy-Dissipative Materials
Get PDFCarbon nanotubes (CNTs) are known to have remarkable properties, such as a specific strength two orders of magnitude higher than that of steel. It has remained a challenge, however, to achieve useful bulk properties from CNTs. Toward that goal, here we develop low-density bulk materials (0.1-0.4 g cm-3) entirely or nearly entirely from CNTs. These consist of nominally-aligned arrays of CNTs that display a dissipative compressive response, with a notable stress-strain hysteresis. The compressive properties of CNT arrays are examined in detail. This analysis reveals interesting features in the mechanical response, such as strain localization (resulting from a gradient in physical properties along the height), recovery after compression, non-linear viscoelasticity, and behavior under repeated compression that depends on the strain of previous cycles (similar to the Mullins effect in rubbers). We observe that in compression the energy dissipation of these materials is more than 200 times that of polymeric foams of comparable density.
Next, materials based on CNT arrays are studied as exemplary of hierarchical materials (materials with distinct structure at multiple length scales). Hierarchical materials have pushed the limits of traditional material tradeoffs (e.g., the typical trend that increased strength requires increased weight). Techniques are developed to separately vary the structure of CNT arrays at nanometer, micrometer, and millimeter length scales, and the effects on the bulk material response are examined. Structure can be modified during CNT synthesis, such as by varying the composition of the flow gas or by manipulating the input rate of chemical precursors; it can also be modified post-synthesis, e.g., by the in situ synthesis of nanoparticles in the interstices of the CNT arrays or by the assembly of multilayer structures of multiple CNT arrays connected by polymeric or metallic interlayers.
Finally, a mathematical model is applied to capture the complexities of the mechanical response. This one-dimensional, multiscale, bistable spring model is able to match the global stress-strain response as well as local effects such as strain localization and Mullins-like behavior. A technique is developed to reliably discern the model's material parameters based on in situ optical data from the experiments.</p