Investigation of numerical atomic orbitals for first-principles calculations of the electronic and transport properties of silicon nanowire structures

This thesis is focused on the application of numerical atomic basis sets in studies of the structural, electronic and transport properties of silicon nanowire structures from first-principles within the framework of Density Functional Theory. First we critically examine the applied methodology and then offer predictions regarding the transport properties and realisation of silicon nanowire devices. The performance of numerical atomic orbitals is benchmarked against calculations performed with plane waves basis sets. After establishing the convergence of total energy and electronic structure calculations with increasing basis size we have shown that their quality greatly improves with the optimisation of the contraction for a fixed basis size. The double zeta polarised basis offers a reasonable approximation to study structural and electronic properties and transferability exists between various nanowire structures. This is most important to reduce the computational cost. The impact of basis sets on transport properties in silicon nanowires with oxygen and dopant impurities have also been studied. It is found that whilst transmission features quantitatively converge with increasing contraction there is a weaker dependence on basis set for the mean free path; the double zeta polarised basis offers a good compromise whereas the single zeta basis set yields qualitatively reasonable results. Studying the transport properties of nanowire-based transistor setups with p+-n-p+ and p+-i-p+ doping profiles it is shown that charge self-consistency affects the I-V characteristics more significantly than the basis set choice. It is predicted that such ultrascaled (3 nm length) transistors would show degraded performance due to relatively high source-drain tunnelling currents. Finally, it is shown the hole mobility of Si nanowires nominally doped with boron decreases monotonically with decreasing width at fixed doping density and increasing dopant concentration. Significant mobility variations are identified which can explain experimental observations.

First principles modelling of nucleation and growth during atomic layer deposition onto III-V substrates

Atomic layer deposition (ALD) is now used in semiconductor fabrication lines to deposit nanometre-thin oxide films, and has thus enabled the introduction of high-permittivity dielectrics into the CMOS gate stack. With interest increasing in transistors based on high mobility substrates, such as GaAs, we are investigating the surface treatments that may improve the interface characteristics. We focus on incubation periods of ALD processes on III-V substrates. We have applied first principles Density Functional Theory (DFT) to investigate detailed chemistry of these early stages of growth, specifically substrate and ALD precursor interaction. We have modelled the ‘clean-up’ effect by which organometallic precursors: trimethylaluminium (TMA) or hafnium and titanium amides clean arsenic oxides off the GaAs surface before ALD growth of dielectric commences and similar effect on Si3N4 substrate. Our simulations show that ‘clean-up’ of an oxide film strongly depends on precursor ligand, its affinity to the oxide and the redox character of the oxide. The predominant pathway for a metalloid oxide such as arsenic oxide is reduction, producing volatile molecules or gettering oxygen from less reducible oxides. An alternative pathway is non-redox ligand exchange, which allows non-reducible oxides (e.g. SiO2) to be cleaned-up. First principles study shows also that alkylamides are more susceptible to decomposition rather than migration on the oxide surface. This improved understanding of the chemical principles underlying ‘clean-up’ allows us to rationalize and predict which precursors will perform the reaction. The comparison is made between selection of metal chlorides, methyls and alkylamides precursors.

First principles simulation of amorphous silicon bulk, interfaces, and nanowires for photovoltaics

Amorphous silicon has become the material of choice for many technologies, with major applications in large area electronics: displays, image sensing and thin film photovoltaic cells. This technology development has occurred because amorphous silicon is a thin film semiconductor that can be deposited on large, low cost substrates using low temperature. In this thesis, classical molecular dynamics and first principles DFT calculations have been performed to generate structural models of amorphous and hydrogenated amorphous silicon and interfaces of amorphous and crystalline silicon, with the ultimate aim of understanding the photovoltaic properties of core-shell crystalline amorphous Si nanowire structures. We have shown, unexpectedly, from the simulations, that our understanding of hydrogenated bulk a-Si needs to be revisited, with our robust finding that when fully saturated with hydrogen, bulk a-Si exhibits a constant optical energy gap, irrespective of the hydrogen concentration in the sample. Unsaturated a-Si:H, with a lower than optimum hydrogen content, shows a smaller optical gap, that increases with hydrogen content until saturation is reached. The mobility gaps obtained from an analysis of the electronic states show similar behavior. We also obtained that the optical and mobility gaps show a volcano curve as the H content is varied from 7% (undersaturation) to 18% (mild oversaturation). In the case of mild over saturation, the mid-gap states arise exclusively from an increase in the density of strained Si-Si bonds. Analysis of our structures shows the extra H atoms in this case form a bridge between neighboring silicon atoms which increases the corresponding Si-Si distance and promotes bond length disorder in the sample. That has the potential to enhance the Staebler-Wronski effect. Planar interface models of amorphous-crystalline silicon have been generated in Si (100), (110) and (111) surfaces. The interface models are characterized by structure, RDF, electronic density of states and optical absorption spectrum. We find that the least stable (100) surface will result in the formation of the thickest amorphous silicon layer, while the most stable (110) surface forms the smallest amorphous region. We calculated for the first time band offsets of a-Si:H/c-Si heterojunctions from first principles and examined the influence of different surface orientations and amorphous layer thickness on the offsets and implications for device performance. The band offsets depend on the amorphous layer thickness and increase with thickness. By controlling the amorphous layer thickness we can potentially optimise the solar cell parameters. Finally, we have successfully generated different amorphous layer thickness of the a-Si/c-Si and a-Si:H/c-Si 5 nm nanowires from heat and quench. We perform structural analysis of the a-Si-/c-Si nanowires. The RDF, Si-Si bond length distributions, and the coordination number distributions of amorphous regions of the nanowires reproduce similar behaviour compared to bulk amorphous silicon. In the final part of this thesis we examine different surface terminating chemical groups, -H, - OH and –NH2 in (001) GeNW. Our work shows that the diameter of Ge nanowires and the nature of surface terminating groups both play a significant role in both the magnitude and the nature of the nanowire band gaps, allowing tuning of the band gap by up to 1.1 eV. We also show for the first time how the nanowire diameter and surface termination shifts the absorption edge in the Ge nanowires to longer wavelengths. Thus, the combination of nanowire diameter and surface chemistry can be effectively utilised to tune the band gaps and thus light absorption properties of small diameter Ge nanowires.

A microscopic study of structural and electronic properties of functionalized silicon surfaces based on first-principles

Surface modification of silicon with organic monolayers tethered to the surface by different linkers is an important process in realizing future (opto-)electronic devices. Understanding the role played by the nature of the linking group and the chain length on the adsorption structures and electronic properties of these assemblies is vital to advance this technology. This Thesis is a study of such properties and contributes in particular to a microscopic understanding of induced changes in the work function of experimentally studied functionalized silicon surfaces. Using first-principles density functional theory (DFT), at the first step, we provide predictions for chemical trends in the work function of hydrogenated silicon (111) surfaces modified with various terminations. For nonpolar terminating atomic species such as F, Cl, Br, and I, the change in the work function is directly proportional to the amount of charge transferred from the surface, thus relating to the difference in electronegativity of the adsorbate and silicon atoms. The change is a monotonic function of coverage in this case, and the work function increases with increasing electronegativity. Polar species such as −TeH, −SeH, −SH, −OH, −NH2, −CH3, and −BH2 do not follow this trend due to the interaction of their dipole with the induced electric field at the surface. In this case, the magnitude and sign of the surface dipole moment need to be considered in addition to the bond dipole to generally describe the change in work function. Compared to hydrogenated surfaces, there is slight increase in the work function of H:Si(111)-XH, where X = Te, Se, and S, whereas reduction is observed for surfaces covered with −OH, −CH3, and −NH2. Next, we study the hydrogen passivated Si(111) surface modified with alkyl chains of the general formula H:Si–(CH2)n–CH2 and H:Si–X–(CH2)n–CH3, where X = NH, O, S and n = (0, 1, 3, 5, 7, 9, 11), at half coverage. For (X)–Hexyl and (X)–Dodecyl functionalization, we also examined various coverages up to full monolayer grafting in order to validate the result of half covered surface and the linker effect on the coverage. We find that it is necessary to take into account the van der Waals interaction between the alkyl chains. The strongest binding is for the oxygen linker, followed by S, N, and C, irrespective of chain length. The result revealed that the sequence of the stability is independent of coverage; however, linkers other than carbon can shift the optimum coverage considerably and allow further packing density. For all linkers apart from sulfur, structural properties, in particular, surface-linker-chain angles, saturate to a single value once n > 3. For sulfur, we identify three regimes, namely, n = 0–3, n = 5–7, and n = 9–11, each with its own characteristic adsorption structures. Where possible, our computational results are shown to be consistent with the available experimental data and show how the fundamental structural properties of modified Si surfaces can be controlled by the choice of linking group and chain length. Later we continue by examining the work function tuning of H:Si(111) over a range of 1.73 eV through adsorption of alkyl monolayers with general formula -[Xhead-group]-(CnH2n)-[Xtail-group], X = O(H), S(H), NH(2). The work function is practically converged at 4 carbons (8 for oxygen), for head-group functionalization. For tail-group functionalization and with both head- and tail-groups, there is an odd-even effect in the behavior of the work function, with peak-to-peak amplitudes of up to 1.7 eV in the oscillations. This behavior is explained through the orientation of the terminal-group's dipole. The shift in the work function is largest for NH2-linked and smallest for SH-linked chains and is rationalized in terms of interface dipoles. Our study reveals that the choice of the head- and/or tail-groups effectively changes the impact of the alkyl chain length on the work function tuning using self-assembled monolayers and this is an important advance in utilizing hybrid functionalized Si surfaces. Bringing together the understanding gained from studying single type functionalization of H:Si(111) with different alkyl chains and bearing in mind how to utilize head-group, tail-group or both as well as monolayer coverage, in the final part of this Thesis we study functionalized H:Si(111) with binary SAMs. Aiming at enhancing work function adjustment together with SAM stability and coverage we choose a range of terminations and linker-chains denoted as –X–(Alkyl) with X = CH3, O(H), S(H), NH(2) and investigate the stability and work function of various binary components grafted onto H:Si(111) surface. Using binary functionalization with -[NH(2)/O(H)/S(H)]-[Hexyl/Dodecyl] we show that work function can be tuned within the interval of 3.65-4.94 eV and furthermore, enhance the SAM’s stability. Although direct Si-C grafted SAMs are less favourable compared to their counterparts with O, N or S linkage, regardless of the ratio, binary functionalized alkyl monolayers with X-alkyl (X = NH, O) is always more stable than single type alkyl functionalization with the same coverage. Our results indicate that it is possible to go beyond the optimum coverage of pure alkyl functionalized SAMs (50%) by adding a linker with the correct choice of the linker. This is very important since dense packed monolayers have fewer defects and deliver higher efficiency. Our results indicate that binary anchoring can modify the charge injection and therefore bond stability while preserving the interface electronic structure.

Structure maps for hcp metals from first-principles calculations

The ability to predict the existence and crystal type of ordered structures of materials from their components is a major challenge of current materials research. Empirical methods use experimental data to construct structure maps and make predictions based on clustering of simple physical parameters. Their usefulness depends on the availability of reliable data over the entire parameter space. Recent development of high-throughput methods opens the possibility to enhance these empirical structure maps by ab initio calculations in regions of the parameter space where the experimental evidence is lacking or not well characterized. In this paper we construct enhanced maps for the binary alloys of hcp metals, where the experimental data leaves large regions of poorly characterized systems believed to be phase separating. In these enhanced maps, the clusters of noncompound-forming systems are much smaller than indicated by the empirical results alone. © 2010 The American Physical Society.

Comprehensive search for new phases and compounds in binary alloy systems based on platinum-group metals, using a computational first-principles approach

We report a comprehensive study of the binary systems of the platinum-group metals with the transition metals, using high-throughput first-principles calculations. These computations predict stability of new compounds in 28 binary systems where no compounds have been reported in the literature experimentally and a few dozen of as-yet unreported compounds in additional systems. Our calculations also identify stable structures at compound compositions that have been previously reported without detailed structural data and indicate that some experimentally reported compounds may actually be unstable at low temperatures. With these results, we construct enhanced structure maps for the binary alloys of platinum-group metals. These maps are much more complete, systematic, and predictive than those based on empirical results alone.

Catalytic Water Formation on Platinum: A First-Principles Study

The study of catalytic behavior begins with one seemingly simple process, namely the hydrogenation of O to H2O on platinum. Despite the apparent simplicity its mechanism has been much debated. We have used density functional theory with,gradient corrections to examine microscopic reaction pathways for several elementary steps implicated in this fundamental catalytic process. We find that H2O formation from chemisorbed O and H atoms is a highly activated process. The largest barrier along this route, with a value of similar to1 eV, is the addition of the first H to O to produce OH. Once formed, however, OH groups are easily hydrogenated to H2O with a barrier of similar to0.2 eV. Disproportionation reactions with 1:1 and 2:1 stoichiometries of H2O and O have been examined as alternative routes for OH formation. Both stoichiometries of reaction produce OH groups with barriers that are much lower than that associated with the O + H reaction. H2O, therefore, acts as an autocatalyst in the overall H O formation process. Disproportionation with a 2:1 stoichiometry is thermodynamically and kinetically favored over disproportionation with a l:I stoichiometry. This highlights an additional (promotional) role of the second H2O molecule in this process. In support of our previous suggestion that the key intermediate in the low-temperature H2O formation reaction is a mixed OH and H2O overlayer we find that then is a very large barrier for the dissociation of the second H2O molecule in the 2:1 disproportionation process. We suggest that the proposed intermediate is then hydrogenated to H2O through a very facile proton transfer mechanism.

SrTiO3(001)(2x1) reconstructions: First-principles calculations of surface energy and atomic structure compared with scanning tunneling microscopy images

(1x1) and (2x1) reconstructions of the (001) SrTiO3 surface were studied using the first-principles full-potential linear muffin-tin orbital method. Surface energies were calculated as a function of TiO2 chemical potential, oxygen partial pressure and temperature. The (1x1) unreconstructed surfaces were found to be energetically stable for many of the conditions considered. Under conditions of very low oxygen partial pressure the (2x1) Ti2O3 reconstruction [Martin R. Castell, Surf. Sci. 505, 1 (2002)] is stable. The question as to why STM images of the (1x1) surfaces have not been obtained was addressed by calculating charge densities for each surface. These suggest that the (2x1) reconstructions would be easier to image than the (1x1) surfaces. The possibility that the presence of oxygen vacancies would destabilise the (1x1) surfaces was also investigated. If the (1x1) surfaces are unstable then there exists the further possibility that the (2x1) DL-TiO2 reconstruction [Natasha Erdman Nature (London) 419, 55 (2002)] is stable in a TiO2-rich environment and for p(O2)>10(-18) atm.

Catalytic Role of Metal Oxides in Gold-Based Catalysts: A First Principles Study of CO Oxidation on TiO2 Supported Au

CO oxidation on TiO2 supported Au has been studied using density functional theory calculations. Important catalytic roles of the oxide have been identified: (i) CO oxidation occurs at the interface between Au and the oxide with a very small barrier; and (ii) O-2 adsorption at the interface is the key step in the reaction. The physical origin of the oxide promotion effect has been further investigated: The oxide enhances electron transfer from the Au to the antibonding states of O-2, giving rise to (i) strong ionic bonding between the adsorbed O-2, Au, and the Ti cation; and (ii) a significant activation of O-2 towards CO oxidation.

First-principles study of ferroelectricity and isotope effects in H-bonded KH2PO4 crystals

By means of extensive first-principles calculations we studied the ferroelectric phase transition and the associated isotope effect in KH2PO4 (KDP). Our calculations revealed that the spontaneous polarization of the ferroelectric phase is due to electronic charge redistributions and ionic displacements which are a consequence of proton ordering, and not vice versa. The experimentally observed double-peaked proton distribution in the paraelectric phase cannot be explained by a dynamics of only protons. This requires, instead, collective displacements within clusters that include also the heavier ions. These tunneling clusters can explain the recent evidence of tunneling obtained from Compton scattering measurements. The sole effect of mass change upon deuteration is not sufficient to explain the huge isotope effect. Instead, we find that structural modifications deeply connected with the chemistry of the H bonds produce a feedback effect on tunneling that strongly enhances the phenomenon. The resulting influence of the geometric changes on the isotope effect agrees with experimental data from neutron scattering. Calculations under pressure allowed us to analyze the issue of universality in the disappearance of ferroelectricity upon compression. Compressing DKDP so that the distance between the two peaks in the deuteron distribution is the same as for protons in KDP, corresponds to a modification of the underlying double-well potential, which becomes 23 meV shallower. This energy difference is what is required to modify the O-O distance in such a way as to have the same distribution for protons and deuterons. At the high pressures required experimentally, the above feedback mechanism is crucial to explain the magnitude of the geometrical effect.

Solvation Structure and Transport of Acidic Protons in Ionic Liquids: A First-principles Simulation Study

Ab initio simulations of a single molecule of HCl in liquid dimethyl imidazolium chloride [dmim][Cl] show that the acidic proton exists as a symmetric, linear ClHCl- species. Details of the solvation structure around this molecule are given. The proton-transfer process was investigated by applying a force along the antisymmetric stretch coordinate until the molecule broke. Changes in the free energy and local solvation structure during this process were investigated. In the reaction mechanism identified, a free chloride approaches the proton from the side. As the original ClHCl- distorts and the incoming chloride forms a new bond to the proton, one of the original chlorine atoms is expelled and a new linear molecule is formed.

Structural and chemical embrittlement of grain boundaries by impurities: A general theory and first-principles calculations for copper

First-principles calculations of the Sigma 5(310)[001] symmetric tilt grain boundary in Cu with Bi, Na, and Ag substitutional impurities provide evidence that in the phenomenon of Bi embrittlement of Cu grain boundaries electronic effects do not play a major role; on the contrary, the embrittlement is mostly a structural or "size" effect. Na is predicted to be nearly as good an embrittler as Bi, whereas Ag does not embrittle the boundary in agreement with experiment. While we reject the prevailing view that "electronic" effects (i.e., charge transfer) are responsible for embrittlement, we do not exclude the role of chemistry. However, numerical results show a striking equivalence between the alkali metal Na and the semimetal Bi, small differences being accounted for by their contrasting "size" and "softness" (defined here). In order to separate structural and chemical effects unambiguously if not uniquely, we model the embrittlement process by taking the system of grain boundary and free surfaces through a sequence of precisely defined gedanken processes; each of these representing a putative mechanism. We thereby identify three mechanisms of embrittlement by substitutional impurities, two of which survive in the case of embrittlement or cohesion enhancement by interstitials. Two of the three are purely structural and the third contains both structural and chemical elements that by their very nature cannot be further unraveled. We are able to take the systems we study through each of these stages by explicit computer simulations and assess the contribution of each to the net reduction in intergranular cohesion. The conclusion we reach is that embrittlement by both Bi and Na is almost exclusively structural in origin; that is, the embrittlement is a size effect.

Electronic structure of the nitrogen-vacancy center in diamond from first-principles theory

The nitrogen-vacancy (NV) center is a paramagnetic defect in diamond with applications as a qubit. Here, we investigate its electronic structure by using ab initio density functional theory for five different NV center models of two different cluster sizes. We describe the symmetry and energetics of the low-lying states and compare the optical frequencies obtained to experimental results. We compute the major transition of the negatively charged NV centers to within 25–100 meV accuracy and find that it is energetically favorable for substitutional nitrogens to donate an electron to NV0. The excited state of the major transition and the NV0 state with a neutral donor nitrogen are found to be close in energy.

First-principles study of oxygenates on Co surfaces in Fischer-Tropsch synthesis

Extensive density function theory calculations are performed to study the mechanism of the formation of aldehyde and alcohol on Co surfaces in Fischer-Tropsch synthesis, a challenging issue in heterogeneous catalysis. Three possible pathways for the production of formaldehyde and methanol on flat and stepped Co(0001) surfaces are investigated: (i) CO + 4H -> CHO + 3H -> CH2O + 2H -> CH3O + H -> CH3OH; (ii) CO + 4H -> COH + 3H -> CHOH + 2H -> CH2OH + H -> CH3OH; and (iii) the coupling reactions of CH2 + O -> CH2O and CH3 + OH -> CH3OH. It is found that these pathways are generally favored at step sites, and the preferred mechanism is pathway (i) via CHO. Furthermore, the three traditional chain growth mechanisms in Fischer-Tropsch synthesis are semi quantitatively compared and discussed. Our results suggest that the two mechanisms involving oxygenate intermediates (the CO-insertion and hydroxycarbene mechanisms) are less important than the carbene mechanism in the production of long chain hydrocarbons. However, the CO-insertion mechanism may be responsible for the production of long-chain oxygenates.

Hydrogenation of S to H2S on Pt(111): A first-principles study

Density-functional theory has been used to investigate the chemisorption of S, SH, and H2S as well as the coadsorption of S and H and SH and H on Pt(111). In addition reaction pathways and energy profiles for the conversion of adsorbed S and H into gas-phase H2S have been determined. It has been found that S, SH, and H2S bind preferentially at face-centered-cubic (fcc), bridge, and top sites, respectively. Both the S+H and SH+H reactions have high barriers (similar to1 eV) and high exothermicities (similar to1 eV). This reveals that adsorbed H2S and SH are highly unstable adsorbates on Pt(111) and that adsorbed S (and H) is the most stable SHX (X=0,1,2) intermediate on Pt(111) (C) 2001 American Institute of Physics.