Search

Si

Atomic layer deposition of a silicon rich SiN

The growth in efficiency of earth-abundant kesterite Cu2ZnSn(S,Se)4 (CZTSSe) solar cells has slowed, due in part to the intrinsic limitations imposed by the band tailing attributed primarily to I-II antisite exchange. In this study, density functional theory simulations show that when Ag is substituted for Cu to form kesterite Ag2ZnSnSe4 (AZTSe), the I-II isolated antisite formation energy becomes 3.7 times greater than in CZTSSe, resulting in at least an order of magnitude reduction in I-II antisite density. Experimental evidence of an optoelectronically improved material is also provided. Comparison of the low-temperature photoluminescence (PL) structure of Cu(In,Ga)Se2 (CIGSe), CZTSSe, and AZTSe shows that AZTSe has a shallow defect structure with emission significantly closer to the band edge than CZTSe. Existence of suppressed band tailing is found in the proximity of the room-temperature PL peak of AZTSe to its measured band gap. The results are consistent with AZTSe being a promising alternative to CZTSSe and CIGSe for thin film photovoltaics.

Stable and metastable oxide structures resulting from the reaction of GaAs(001)-(2×4) with O atoms are investigated using scanning tunneling microscopy (STM). The relative stabilities of these oxide structures are examined using density functional theory calculations (DFT). STM images show that when GaAs(001)-(2×4) is exposed to O atoms, the O atom will either displace an arsenic atom from its original dimer position by taking its place or chemisorb subsurface and create a metastable site. As the O atom coverage increases, O atoms increasingly occupy the position of two displaced arsenic atoms across two neighboring dimers, while the number of subsurface sites remains constant. These experiments show that oxygen preferentially removes an arsenic pair (As2), with one atom originating from each of two neighboring As–As dimers instead of from the same As–As dimer. DFT calculations demonstrate the relative stability of the final chemisorption state by calculating the most stable oxygen configurations as funct...

The surface structures formed upon deposition of O2 and Ga2O onto the technologically important arsenic-rich GaAs(001)-c(2×8)/(2×4) surface have been studied using scanning tunneling microscopy and spectroscopy, and the results are compared to density functional theory calculations. O2 chemisorbs by displacing first layer arsenic atoms bonded to second layer gallium atoms. Oxygen chemisorption pins the Fermi level at less than 5% monolayer coverage by creating a donor and acceptor site within the band gap originating from the gallium atom bonded between the two O atoms. In contrast, Ga2O chemisorbs by inserting into arsenic dimer pairs at elevated surface temperatures. A monolayer of Ga2O forms a (2×2) surface structure with a crystalline interface that is electronically unpinned: there are no states within the band gap. The unpinned interface results from Ga2O restoring the surface arsenic and gallium atoms to near-bulk charge.

Thin film processing technologies that promise atomic and molecular scale control have received increasing interest in the past several years, as traditional methods for fabrication begin to reach their fundamental limits. Many of these technologies involve at their heart phenomena occurring at or near surfaces, including adsorption, gas-surface reactions, diffusion, desorption, and re-organization of near-surface layers. Moreover many of these phenomena involve not just reactions occurring under conditions of local thermodynamic equilibrium but also the action of energetic species including electrons, ions, and hyperthermal neutrals. There is a rich landscape of atomic and molecular scale interactions occurring in these systems that is still not well understood. In this Special Topic Issue of The Journal of Chemical Physics, we have collected recent representative examples of work that is directed at unraveling the mechanistic details concerning atomic and molecular layer processing, which will provide an important framework from which these fields can continue to develop. These studies range from the application of theory and computation to these systems to the use of powerful experimental probes, such as X-ray synchrotron radiation, probe microscopies, and photoelectron and infrared spectroscopies. The work presented here helps in identifying some of the major challenges and direct future activities in this exciting area of research involving atomic and molecular layer manipulation and fabrication.

Si0.5Ge0.5(110) surfaces were passivated and functionalized using atomic H, hydrogen peroxide (H2O2), and either tetrakis(dimethylamino)titanium (TDMAT) or titanium tetrachloride (TiCl4) and studied in situ with multiple spectroscopic techniques. To passivate the dangling bonds, atomic H and H2O2(g) were utilized and scanning tunneling spectroscopy (STS) demonstrated unpinning of the surface Fermi level. The H2O2(g) could also be used to functionalize the surface for metal atomic layer deposition. After subsequent TDMAT or TiCl4 dosing followed by a post-deposition annealing, scanning tunneling microscopy demonstrated that a thermally stable and well-ordered monolayer of TiOx was deposited on Si0.5Ge0.5(110), and X-ray photoelectron spectroscopy verified that the interfaces only contained Si—O—Ti bonds and a complete absence of GeOx. STS measurements confirmed a TiOx monolayer without mid-gap and conduction band edge states, which should be an ideal ultrathin insulating layer in a metal-insulator-semicond...

Atomic layer deposition of a silicon rich SiNx layer on Si0.7Ge0.3(001), Si0.5Ge0.5(001), and Si0.5Ge0.5(110) surfaces has been achieved by sequential pulsing of Si2Cl6 and N2H4 precursors at a substrate temperature of 285 °C. XPS spectra show a higher binding energy shoulder peak on Si 2p indicative of SiOxNyClz bonding while Ge 2p and Ge 3d peaks show only a small amount of higher binding energy components consistent with only interfacial bonds, indicating the growth of SiOxNy on the SiGe surface with negligible subsurface reactions. Scanning tunneling spectroscopy measurements confirm that the SiNx interfacial layer forms an electrically passive surface on p-type Si0.70Ge0.30(001), Si0.50Ge0.50(110), and Si0.50Ge0.50(001) substrates as the surface Fermi level is unpinned and the electronic structure is free of states in the band gap. DFT calculations show that a Si rich a-SiO0.4N0,4 interlayer can produce lower interfacial defect density than stoichiometric a-SiO0.8N0.8, substoichiometric a-Si3N2, or stoichiometric a-Si3N4 interlayers by minimizing strain and bond breaking in the SiGe by the interlayer. Metal-oxide-semiconductor capacitors devices were fabricated on p-type Si0.7Ge0.3(001) and Si0.5Ge0.5(001) substrates with and without the insertion of an ALD SiOxNy interfacial layer, and the SiOxNy layer resulted in a decrease in interface state density near midgap with a comparable Cmax value.

A combined wet and dry cleaning process for GaN(0001) has been investigated with XPS and DFT-MD modeling to determine the molecular-level mechanisms for cleaning and the subsequent nucleation of gate oxide atomic layer deposition (ALD). In situ XPS studies show that for the wet sulfur treatment on GaN(0001), sulfur desorbs at room temperature in vacuum prior to gate oxide deposition. Angle resolved depth profiling XPS post-ALD deposition shows that the a-Al2O3 gate oxide bonds directly to the GaN substrate leaving both the gallium surface atoms and the oxide interfacial atoms with XPS chemical shifts consistent with bulk-like charge. These results are in agreement with DFT calculations that predict the oxide/GaN(0001) interface will have bulk-like charges and a low density of band gap states. This passivation is consistent with the oxide restoring the surface gallium atoms to tetrahedral bonding by eliminating the gallium empty dangling bonds on bulk terminated GaN(0001).

The naturally occurring oxide of GaAs has a high density of defects that pin the Fermi level at the GaAs surface. The principle electronic defect causing the Fermi level pinning is widely believed to be an arsenic antisite double donor. We have used scanning tunneling microscopy to show that the arsenic antisite defects are formed during the initial period of oxidation of GaAs(001) by atomic oxygen. Atomic oxygen displaces a single arsenic atom in the top layer leaving behind an undimerized arsenic atom with a half-filled dangling bond. The displaced arsenic atoms bond to remaining arsenic dimers to form arsenic antisites. Cluster calculations confirm the assignment of the oxygen binding sites and the role of undimerized single arsenic atoms as acceptor sites.

The chemisorption sites of molecular oxygen on the technologically important As-rich GaAs(100)c(2×8) surface were imaged with scanning tunneling microscopy (STM). The oxygen atoms insert into the arsenic–gallium backbonds and, subsequently, replace the arsenic atoms in the dimer rows. The displaced arsenic atoms aggregate in clusters of increasing size forming metallic arsenic. The strongly electronegative oxygen atoms are initially attracted by the larger electron density at the arsenic atoms, but the reaction of the oxygen atoms with the gallium atoms is thermodynamically favored. This leads to a 100% chemical selectivity for oxygen insertion into the As–Ga backbonds and subsequent chemisorption of the oxygen atoms into the arsenic sites.

The chemisorption sites of molecular oxygen on the mixed GaAs(100)(6×6) surface were imaged at room temperature using scanning tunneling microscopy (STM). This surface is terminated by both gallium dimers and arsenic dimers, neither of which react with oxygen. Instead, the As–Ga backbonds are shown to react with O2 with 100% chemical selectivity. The reason for this selectivity is found in the interaction of the highly electronegative oxygen atoms with the higher electron density at the arsenic atoms. One oxygen atom displaces the attacked arsenic atom while the other oxygen atom bonds to two nearby gallium atoms, resulting in the thermodynamically most stable reaction products: metallic arsenic clusters and gallium oxide.

Scanning tunneling microscopy images of the Ga-rich GaAs(100)c(8×2)/(4×2) surface exhibit vivid long-range patterns consisting of bright spots (“ghosts”) which are attributed to localized excess charge rather than atomic clusters. The nearly planar geometry of the sp2-hybridized gallium dimer atoms results in localized π states made up of a combination of the Ga pz orbitals. These states in the upper half of the band gap form the lowest unoccupied band. Surface or bulk defects lead to excess negative charge flowing into these localized states. Repulsion between the trapped negative excess charges leads to the observed “ghost” pattern.

The adsorption of 0.15-eV and 0.89-eV Br2 onto the As-rich GaAs(001)-2×4 surface at 300 K has been investigated on the atomic scale over a wide range of total Br coverage using scanning tunneling microscopy (STM). It is found that the strained As–As dimer bonds and the As–Ga back bonds are significantly weakened and chemically activated by the Br atoms site-selectively adsorbed onto the second-layer Ga atoms. Consequently, at higher Br coverages, incident Br2 will preferentially react with these As–As and As–Ga bonds, removing (etching) As–As dimers and forming AsBr(s) or AsBr2(s) species, GaBr(s) chains, and etching pit islands. The incident energy of Br2 has a strong effect on the adsorption mechanisms.

The reaction of monoenergetic I2Cl6 molecules with the Si(111)-7×7 surface is stereoselective for iodine and surface-site selective for center silicon adatoms. The adsorption starts by first forming a partial bond between an I atom of an I2Cl6 and a Si adatom on the surface so that iodine from I2Cl6 is selectively abstracted by the surface and chlorine is ejected back into the gas phase. The interaction between the asymmetric HOMO of an I2Cl6 and a partially filled Si adatom dangling bond governs the adsorption dynamics. The highly asymmetric HOMO orbital and zero dipole moment associated with I2Cl6 makes it a perfect molecular species for clarifying the role of asymmetric molecular bonding versus dipole moment in the iodine-selective abstraction of ICl by the Si(111)-7×7 surface.

The adsorption of monoenergetic IBr molecules on the Si(111)-7×7 surface has been studied using scanning tunneling microscopy, mass spectrometry, Auger electron spectroscopy, and supersonic molecular beam techniques. The adsorption proceeds predominantly via the direct abstractive adsorption mechanism and preferentially occurs at the center Si adatoms. The IBr abstraction probabilities at the incident energies of 0.15 and 0.82 eV have been determined to be 0.90±0.03 and 0.77±0.03, respectively. The minor dissociative adsorption channel of IBr can be enhanced at the expense of the abstractive adsorption channels by raising the incident energy. Most importantly, no atomic selectivity for iodine or bromine was observed. A reaction mechanism involving two types of transition states, Si⋯I⋯Br(s) and Si⋯Br⋯I(s), has been proposed to interpret the experimental observations. The attractive interaction between the nearly symmetric highest occupied molecular orbitals (HOMO, π* antibond) of IBr and the partially-fill...

Scanning tunneling microscopy and molecular beam techniques were used to investigate the chemisorption sites of O2 onto Si(111)‐7×7 surfaces as a function of O2 incident energy (Ei). It is shown that there are two different O‐containing sites being formed, i.e., bright and dark sites. The distribution of dark sites is not sensitive to Ei. However, the distribution of bright sites is a strong function of O2 incident energy. At Ei=0.02 eV, there is no preference between bright corner‐adatom and center‐adatom sites (about 1.1:1) while at Ei≳0.06 eV, the bright sites are preferentially located at corner‐adatom sites over center‐adatom sites by about 2:1. It is proposed that difference in the spatial distribution of adsorbates at different incident energies results from different chemisorption mechanisms: precursor‐mediated (physisorption) chemisorption at low incident energy and direct activated chemisorption at high incident energies.

The dissociative 193 nm photochemistry of both multilayer and submonolayer NH3 on Pt(111) has been studied. Upon irradiation with 193 nm light, the submonolayer covered Pt surface yields only a small amount of NH3 desorption. In contrast, the desorption products from the 193 nm irradiation of a multilayer NH3 covered Pt surface are both NH3 and atomic H. The NH3 photodesorption yield from the multilayer is much greater than from the submonolayer covered surface. A nearly field‐free resonantly enhanced multiphoton ionization technique was used to carefully distinguish between H and NH3 photoproducts. From the multilayer, NH3 desorbs with a distinct bimodal hyperthermal velocity distribution. The atomic H velocity distribution exhibits both a hyperthermal component and, surprisingly, a ∼150 K thermalized component. Our multilayer data is consistent with adsorbate absorption, and we have proposed a model based on multiple collisions of desorbing species to explain our results. In this model, NH3 molecules at...

Density-functional theory simulations of CZTS, CZTSe, and CZTS0.25Se0.75 photovoltaic compounds have been performed to investigate the stability of the CZTS0.25Se0.75 alloy vs. decomposition into CZTS, CZTSe, and other secondary compounds. The Gibbs energy for vibrational contributions was estimated by calculating phonon spectra and thermodynamic properties at finite temperatures. It was demonstrated that the CZTS0.25Se0.75 alloy is stabilized not by enthalpy of formation but primarily by the mixing contributions to the Gibbs energy. The Gibbs energy gains/losses for several decomposition reactions were calculated as a function of temperature with/without intermixing and vibration contributions to the Gibbs energy. A set of phase diagrams was built in the multidimensional space of chemical potentials at 300 K and 900 K temperatures to demonstrate alloy stability and boundary compounds at various chemical conditions. It demonstrated for CZTS0.25Se0.75 that the chemical potentials for stability differ betwe...

Polarized ultraviolet light from an excimer laser (193 nm) was used to photodesorb and photodissociate N2O adsorbed on a cold (80 K) Pt(111) surface. The photodesorbed species and their time of flight (TOF) were monitored by resonantly enhanced multiphoton ionization (REMPI) spectroscopy. We have identified three major channels. The photodesorption of molecular N2 is observed only in the slowest channel where N2 produced by fragmenting the N2O is thermalized on the surface before desorbing. Evidence for this behavior includes both low (∼90 K) rotational and translational temperatures of the N2 fragments as well as a lack of correlation between rotational and translational energy. In the next fastest channel, hyperthermal N2O with a kinetic energy of 0.4±0.1 eV is seen to photodesorb. The photodesorbed hyperthermal N2O also has a substantial degree of internal vibrational excitation. The angular distribution of the N2O channel is peaked toward the surface normal. In the fastest channel, the release of ball...

Scanning tunneling microscopy (STM) and molecular beam techniques are used to investigate the effect of adsorption mechanism upon adsorbate structures for Cl2 chemisorption onto Si(111)‐7×7 surfaces. At incident energies less than 0.11 eV, the SiCl island formation as well as isolated‐site reaction are observed. STM images of the internal structure of the islands show that the SiCl islands nucleate at sublimation defects and not at step edges. Conversely, for 0.44 eV dosing, only the isolated‐site reaction occurs. It is proposed that the island formation results from precursor mediated chemisorption and that the isolated‐site reaction results from direct activated chemisorption. For direct activated chemisorption, the competition between Cl‐atom abstraction and dissociative chemisorption was also observed. The abstraction probability decreases with increasing Cl2 incident translational energy.

The structural properties of a-HfO2/Ge(2 × 1)-(001) and a-ZrO2/Ge(2 × 1)-(001) interfaces were investigated with and without a GeOx interface interlayer using density-functional theory (DFT) molecular dynamics (MD) simulations. Realistic a-HfO2 and a-ZrO2 samples were generated using a hybrid classical-DFT MD "melt-and-quench" approach and tested against experimental properties. The oxide/Ge stacks were annealed at 700 K, cooled to 0 K, and relaxed providing the system with enough freedom to form realistic interfaces. For each high-K/Ge stack type, two systems with single and double interfaces were investigated. All stacks were free of midgap states; however, stacks with a GeO(x) interlayer had band-edge states which decreased the band gaps by 0%-30%. These band-edge states were mainly produced by under-coordinated Ge atoms in GeO(x) layer or its vicinity due to deformation, intermixing, and bond-breaking. The DFT-MD simulations show that electronically passive interfaces can be formed either directly between high-K dielectrics and Ge or with a monolayer of GeO2 if the processing does not create or properly passivate under-coordinated Ge atoms and Ge's with significantly distorted bonding angles. Comparison to the charge states of the interfacial atoms from DFT to experimental x-ray photoelectron spectroscopy results shows that while most studies of gate oxide on Ge(001) have a GeO(x) interfacial layer, it is possible to form an oxide/Ge interface without a GeO(x) interfacial layer. Comparison to experiments is consistent with the dangling bonds in the suboxide being responsible for midgap state formation.

The structural properties of a-Al(2)O(3)∕In(0.5)Ga(0.5)As, a-HfO(2)∕In(0.5)Ga(0.5)As, and a-ZrO(2)∕In(0.5)Ga(0.5)As interfaces were investigated by density-functional theory (DFT) molecular dynamics (MD) simulations. Realistic amorphous a-Al(2)O(3), a-HfO(2), and a-ZrO(2) samples were generated using a hybrid classical-DFT MD "melt-and-quench" approach and tested against the experimental properties. For each stack type, two systems with different initial oxide cuts at the interfaces were investigated. All stacks were free of midgap states, but some had band-edge states which decreased the bandgaps by 0%-40%. The band-edge states were mainly produced by deformation, intermixing, and bond-breaking, thereby creating improperly bonded semiconductor atoms. The interfaces were dominated by metal-As and O-In∕Ga bonds which passivated the clean surface dangling bonds. The valence band-edge states were mainly localized at improperly bonded As atoms, while conduction band-edge states were mainly localized at improperly bonded In and Ga atoms. The DFT-MD simulations show that electronically passive interfaces can be formed between high-κ oxides dielectrics and InGaAs if the processing does not induce defects because on a short time scale the interface spontaneously forms electrically passive bonds as opposed to bonds with midgap states.

The scattering of rotationally cold N2 from Ag(111) results in angular momentum alignment and orientation of the scattered molecules; measurement of the angular momentum polarization as a function of exit angle, final J state, and exit translation energy provides direct information on the dynamics of the collisions. In this paper, the orientation of the angular momentum vector of the scattered N2 molecules, A{1}1−(J) has been measured for slow, medium, and fast groups of molecules in single rotational states at fixed exit angles. With normal incidence scattering (θi=0°) and off‐normal detection, for a given final J state, the ‘‘slow’’ molecules have a higher probability of tumbling backwards (‘‘back spin’’) than the ‘‘fast’’ molecules. Conversely, for glancing incidence scattering (θi=30°) with quasi‐specular detection, the opposite trend is observed: the slow molecules have a higher probability of tumbling forwards (‘‘top spin’’) than the fast molecules. These experiments were simulated and analyzed usin...

The effects of surface temperature (Ts) and surface structure upon the passivation and etching of GaAs(100) and GaAs(110) surfaces by Cl2 have been studied. The Ga‐rich GaAs(100) Ga‐c(8×2) and the stoichiometric GaAs(110) (1×1) surfaces form stable ordered monochloride overlayers when exposed to Cl2 at 300 K. The ordered overlayers formed are AsCl on the GaAs(110) (1×1) surface and GaCl on the GaAs(100) Ga‐c(8×2) surface. In contrast, the As‐rich GaAs(100) As‐c(2×8) surface undergoes continuous uptake of Cl2 under equivalent conditions. All three surfaces exhibit continuous uptake of Cl2 at surface temperatures above 600 K, where continuous etching is known to occur. In this paper, the continuous uptake of Cl2 is shown to result from etching of the surface. In addition it is shown that the Ga‐rich GaAs(100) Ga‐c(8×2) and the stoichiometric GaAs(110) (1×1) surfaces can be induced to undergo continuous etching at surface temperatures of 300 K, if they are first exposed to Cl2 with surface temperatures above...

Initial sticking probabilities are measured for monoenergetic molecular chlorine upon the Ga‐rich GaAs(100) c(8×2), As‐rich GaAs(100) c(2×8), and stoichiometric GaAs(110)(1×1) surfaces. The sticking probabilities are measured as a function of incident translational energy (0.038–0.66 eV), surface temperature (256–807 K), and angle of incidence (0°–37°). Our data indicate the presence of both precursor and direct activated chemisorption mechanisms on all three surfaces. The average barriers to direct chemisorption are quite similar on these three surfaces, suggesting both structural and elemental insensitivity for the direct component to chemisorption. In addition, there is evidence of direct barrierless chemisorption on the Ga‐rich GaAs(100) c(8×2) and stoichiometric GaAs(110)(1×1) surfaces. In contrast, precursor mediated chemisorption is seen to differ for the Ga‐rich GaAs(100) c(8×2) and stoichiometric GaAs(110)(1×1) surfaces with the precursor persisting to higher incident energies on the stoichiometr...

The direct reaction of trimethylaluminum (TMA) on a Ge(100) surface and the effects of monolayer H(2)O pre-dosing were investigated using ultrahigh vacuum techniques, such as scanning tunneling microscopy (STM), scanning tunneling spectroscopy (STS), and x-ray photoelectron spectroscopy (XPS), and density functional theory (DFT). At room temperature (RT), a saturation TMA dose produced 0.8 monolayers (ML) of semi-ordered species on a Ge(100) surface due to the dissociative chemisorption of TMA. STS confirmed the chemisorption of TMA passivated the bandgap states due to dangling bonds. By annealing the TMA-dosed Ge surface, the STM observed coverage of TMA sites decreased to 0.4 ML at 250 °C, and to 0.15 ML at 450 °C. XPS analysis showed that only carbon content was reduced during annealing, while the Al coverage was maintained at 0.15 ML, consistent with the desorption of methyl (-CH(3)) groups from the TMA adsorbates. Conversely, saturation TMA dosing at RT on the monolayer H(2)O pre-dosed Ge(100) surface followed by annealing at 200 °C formed a layer of Ge-O-Al bonds with an Al coverage a factor of two greater than the TMA only dosed Ge(100), consistent with Ge-OH activation of TMA chemisorption and Ge-H blocking of CH(3) chemisorption. The DFT shows that the reaction of TMA has lower activation energy and is more exothermic on Ge-OH than Ge-H sites. It is proposed that the H(2)O pre-dosing enhances the concentration of adsorbed Al and forms thermally stable Ge-O-Al bonds along the Ge dimer row which could serve as a nearly ideal atomic layer deposition nucleation layer on Ge(100) surface.

The effects of surface reactivity on rotationally inelastic scattering has been examined by comparing the scattering of N2 from the reactive W(110) surface and from the passivated N and H reconstructions, W(110)–(2×2)N and W(110)–(1×1)H, as well as the nonreactive Pt(111) surface. The translational energy, rotational state population distributions, and angular momentum alignment of N2 scattered from these surfaces have been measured as a function of incident energy Ei, incident angle θi, and exit angle θf. The trends in the final translational and rotational energy correlate well with the average atomic mass of each surface. However, some of the greater rotational excitation for N2 scattered from W(110) compared to W(110)–(1×1)H can be attributed to a more anisotropic potential with respect to initial molecular orientation for N2–W(110) compared to N2–W(110)–(1×1)H or N2–Pt(111). The rotational distributions indicate that parallel momentum is not conserved for N2 scattered from W(110) or from W(110)–(2×2)...

The creation of alignment by photodissociation is a well‐accepted process. When an isotropic sample of gas is subjected to a strong linearly polarized laser pulse at a frequency at which the photodissociation cross section is large, the surviving molecules are usually aligned. If the transition is parallel, μ lies along the internuclear axis (ΔΛ=0) and the surviving molecules will be peaked around M=0, while for a perpendicular transition (ΔΛ=1) the surviving molecules will be peak around M=J. Although this effect has been seen in laser cavities and in the focus of laser beams, it has not been used to create aligned pulses of gas from free‐jet expansions. We present the theoretical calculations for the practical creation of alignment in short free‐jet gas pulses via saturation photodissociation.Our methodology allows the propagation of the laser light along any direction and with any polarization, the quantification of the effect of hyperfine and electron spin depolarization upon the creation of alignment...

Density functional theory (DFT) simulations were used to determine the binding strength of 12 electron-donating analytes to the zinc metal center of a zinc phthalocyanine molecule (ZnPc monomer). The analyte binding strengths were compared to the analytes' enthalpies of complex formation with boron trifluoride (BF(3)), which is a direct measure of their electron donating ability or Lewis basicity. With the exception of the most basic analyte investigated, the ZnPc binding energies were found to correlate linearly with analyte basicities. Based on natural population analysis calculations, analyte complexation to the Zn metal of the ZnPc monomer resulted in limited charge transfer from the analyte to the ZnPc molecule, which increased with analyte-ZnPc binding energy. The experimental analyte sensitivities from chemiresistor ZnPc sensor data were proportional to an exponential of the binding energies from DFT calculations consistent with sensitivity being proportional to analyte coverage and binding strength. The good correlation observed suggests DFT is a reliable method for the prediction of chemiresistor metallophthalocyanine binding strengths and response sensitivities.

The local atomic structural properties of a-Al(2)O(3), a-ZrO(2) vacuum/oxide surfaces, and a-Al(2)O(3)Ge(100)(2x1), a-ZrO(2)Ge(100)(2x1) oxide/semiconductor interfaces were investigated by density-functional theory (DFT) molecular dynamics (MD) simulations. Realistic a-Al(2)O(3) and a-ZrO(2) bulk samples were generated using a hybrid classical-DFT MD approach. The interfaces were formed by annealing at 700 and 1100 K with subsequent cooling and relaxation. The a-Al(2)O(3) and a-ZrO(2) vacuum/oxide interfaces have strong oxygen enrichment. The a-Al(2)O(3)Ge interface demonstrates strong chemical selectivity with interface bonding exclusively through Al-O-Ge bonds. The a-ZrO(2)Ge interface has roughly equal number of Zr-O-Ge and O-Zr-Ge bonds. The a-Al(2)O(3)Ge junction creates a much more polar interface, greater deformation in Ge substrate and interface intermixing than a-ZrO(2)Ge consistent with experimental measurements. The differences in semiconductor deformation are consistent with the differences in the relative bulk moduli and angular distribution functions of the two oxides.

The correlation between atomic bonding sites and the electronic structure of SiO on GaAs(001)-c(2x8)/(2x4) was investigated using scanning tunneling microscopy (STM), scanning tunneling spectroscopy (STS), and density functional theory (DFT). At low coverage, STM images reveal that SiO molecules bond Si end down; this is consistent with Si being undercoordinated and O being fully coordinated in molecular SiO. At approximately 5% ML (monolayer) coverage, multiple bonding geometries were observed. To confirm the site assignments from STM images, DFT calculations were used to estimate the total adsorption energies of the different bonding geometries as a function of SiO coverage. STS measurements indicated that SiO pins the Fermi level midgap at approximately 5% ML coverage. DFT calculations reveal that the direct causes of Fermi level pinning at the SiO GaAs(001)-(2x4) interface are a result of either local charge buildups or the generation of partially filled dangling bonds on Si atoms.

Current logic devices rely on 3D architectures, such as the tri-gate field effect transistor (finFET), which utilize the (001) and (110) crystal faces simultaneously thus requiring passivation methods for the (110) face in order to ensure a pristine 3D surface prior to further processing. Scanning tunneling microscopy (STM), x-ray photoelectron spectroscopy (XPS), and correlated electrical measurement on MOSCAPs were utilized to compare the effects of a previously developed in situ pre-atomic layer deposition (ALD) surface clean on the InGaAs (001) and (110) surfaces. Ex situ wet cleans are very effective on the (001) surface but not the (110) surface. Capacitance voltage indicated the (001) surface with no buffered oxide etch had a higher C(max) hypothesized to be a result of poor nucleation of HfO2 on the native oxide. An in situ pre-ALD surface clean employing both atomic H and trimethylaluminum (TMA) pre-pulsing, developed by Chobpattana et al. and Carter et al. for the (001) surface, was demonstrated to be effective on the (110) surface for producing low D(it) high C(ox) MOSCAPs. Including TMA in the pre-ALD surface clean resulted in reduction of the magnitude of the interface state capacitance. The XPS studies show the role of atomic H pre-pulsing is to remove both carbon and oxygen while STM shows the role of TMA pre-pulsing is to eliminate H induced etching. Devices fabricated at 120 °C and 300 °C were compared.

Nonresonant multiphoton ionization with time-of-flight mass spectrometry has been used to monitor the desorption of aluminum chloride (Al(x)Cl(y)) etch products from the Al(111) surface at 100 and 500 K during low-coverage (5% monolayer) monoenergetic Cl(2) (0.11-0.65 eV) dosing. The desorption products in this low-coverage range show predominantly hyperthermal exit velocities under all dosing conditions. For example, with 0.27 eV incident Cl(2), the etch product was found to have a most-probable velocity of 517+/-22 m/s at an Al(111) surface temperature of 100 K. This corresponds to 22 times the expected thermal desorption translational energy for AlCl(3). Cl(2) sticking probability measurements and Al(x)Cl(y) etch rate measurements show etching even at Cl(2) coverages of less than 5% monolayer at surface temperatures between 100 and 500 K. These experimental results are consistent with a combination of fast-time-scale surface diffusion and agglomeration of the adsorbed chlorine to form aluminum chlorides and the presence of activated AlCl(3) chemisorption states having potential energies above the vacuum level. Density functional theory calculations yield results that are consistent with both our experimental findings and mechanistic descriptions.

Scanning tunneling microscopy (STM) was employed to study the mechanism for the oxidation of Al(111) with thermal O2 and NO in the 20%-40% monolayer coverage regime. Experiments show that the islands formed upon exposure to thermal O2 and NO have dramatically different shapes, which are ultimately dictated by the dynamics of the gas surface interaction. The circumference-to-area ratio and other island morphology statistics are used to quantify the average difference in the two island types. Ultrahigh-vacuum STM was employed to make the following observations: (1) Oxygen islands on the Al(111) surface, formed upon exposure to thermal oxygen, are elongated and noncompact. (2) Mixed O/N islands on the Al(111) surface, formed upon exposure to thermal nitric oxide (NO), are round and compact. (3) STM movies acquired during thermal O2 exposure indicate that a complex mechanism involving chemisorption initiated rearrangement of preexisting oxygen islands leads to the asymmetric and elongated island shapes. The overall mechanism for the oxidation of the Al(111) surface can be summarized in three regimes. Low coverage is dominated by widely isolated small oxygen features (

The surface structures formed upon deposition of In2O and Ga2O by molecular beam epitaxy onto the arsenic-rich GaAs(001)-c(2 x 8)/(2 x 4) surface have been studied using scanning tunneling microscopy and density functional theory. In2O initially bonds, with indium atoms bonding to second layer gallium atoms within the trough, and proceeds to insert into or between first layer arsenic dimer pairs. In contrast, Ga2O only inserts into or between arsenic dimer pairs due to chemical site constraints. The calculated energy needed to bend a Ga2O molecule approximately 70 degrees, so that it can fit into an arsenic dimer pair, is 0.6 eV less than that required for In2O. The greater flexibility of the Ga2O molecule causes its insertion site to be 0.77 eV more exothermic than the In2O insertion site. This result shows that although trends in the periodic table can be used to predict some surface reactions, small changes in atomic size can play a significant role in the chemistry of gas/surface reactions through the indirect effects of bond angle flexibility and bond length stiffness.

Passivation, functionalization, and atomic layer deposition nucleation via H2O2(g) and trimethylaluminum (TMA) dosing was studied on the clean Ge(100) surface at the atomic level using scanning tunneling microscopy (STM) and scanning tunneling spectroscopy (STS). Chemical analysis of the surface was performed using x-ray photoelectron spectroscopy, while the bonding of the precursors to the substrate was modeled with density functional theory (DFT). At room temperature, a saturation dose of H2O2(g) produces a monolayer of a mixture of -OH or -O species bonded to the surface. STS confirms that H2O2(g) dosing eliminates half-filled dangling bonds on the clean Ge(100) surface. Saturation of the H2O2(g) dosed Ge(100) surface with TMA followed by a 200 °C anneal produces an ordered monolayer of thermally stable Ge-O-Al bonds. DFT models and STM simulations provide a consistent model of the bonding configuration of the H2O2(g) and TMA dosed surfaces. STS verifies the TMA/H2O2/Ge surface has an unpinned Fermi level with no states in the bandgap demonstrating the ability of a Ge-O-Al monolayer to serve as an ideal template for further high-k deposition.

The nucleation and passivation of oxide deposition was studied on defect-free GaAs (110) surfaces to understand passivation of surfaces containing only III-V heterobonds. The passivation process on GaAs (110) was studied at the atomic level using scanning tunneling microscopy while the electronic structure was determined by scanning tunneling spectroscopy (STS). The bonding of the oxidant and reductant were modeled with density functional theory. To avoid Fermi level pinning during gate oxide atomic layer deposition, a dual passivation procedure was required using both a reductant, trimethylaluminum (TMA), and an oxidant, O2 or H2O. Dosing GaAs (110) with TMA resulted in the formation of an ordered complete monolayer of dimethylaluminum which passivates the group V dangling bonds but also forms metal-metal bonds with conduction band edge states. These edge states were suppressed by dosing the surface with oxidants O2 or H2O which selectively react with group III-aluminum bonds. The presence of an ordered Al monolayer with a high nucleation density was indirectly confirmed by XPS and STS.

Velocity selected rotational state distributions of N2 molecules scattered from Ag(111) have been measured. These measurements have been made at normal incidence and normal detection for incident energies of 0.25 and 0.75 eV. This new technique allows us to rapidly study the cross correlation between the exit rotational state and its velocity distribution. A pronounced difference between the rotational distributions of fast and slow scattered molecules is observed, and this difference is much more pronounced for high incident kinetic energy.

Interfacial bonding geometry and electronic structures of In(2)O on InAs and In(0.53)Ga(0.47)As(001)-(4×2) have been investigated by scanning tunneling microscopy/scanning tunneling spectroscopy (STM/STS). STM images show that the In(2)O forms an ordered monolayer on both InAs and InGaAs surfaces. In(2)O deposition on the InAs(001)-(4×2) surface does not displace any surface atoms during both room temperature deposition and postdeposition annealing. Oxygen atoms from In(2)O molecules bond with trough In/Ga atoms on the surface to form a new layer of O-In/Ga bonds, which restore many of the strained trough In/Ga atoms into more bulklike tetrahedral sp(3) bonding environments. STS reveals that for both p-type and n-type clean In(0.53)Ga(0.47)As(001)-(4×2) surfaces, the Fermi level resides near the valence band maximum (VBM); however, after In(2)O deposition and postdeposition annealings, the Fermi level position is close to the VBM for p-type samples and close to the conduction band minimum for n-type samples. This result indicates that In(2)O bonding eliminates surface states within the bandgap and forms an unpinned interface when bonding with In(0.53)Ga(0.47)As/InP(001)-(4×2). Density function theory is used to confirm the experimental finding.

The reaction of trimethyl aluminum on the group III rich reconstructions of InAs(0 0 1) and In(0.53)Ga(0.47)As(0 0 1) is observed with scanning tunneling microscopy/spectroscopy. At high coverage, a self-terminated ordered overlayer is observed that provides the monolayer nucleation density required for subnanometer thick transistor gate oxide scaling and removes the surface Fermi level pinning that is present on the clean InGaAs surface. Density functional theory simulations confirm that an adsorbate-induced reconstruction is the basis of the monolayer nucleation density and passivation.

Hafnium oxide interfaces were studied on two related group III rich semiconductor surfaces, InAs(0 0 1)-(4x2) and In(0.53)Ga(0.47)As(0 0 1)-(4x2), via two different methods: reactive oxidation of deposited Hf metal and electron beam deposition of HfO(2). The interfaces were investigated with scanning tunneling microscopy and spectroscopy (STS). Single Hf atom chemisorption sites were identified that are resistant to oxidation by O(2), but Hf islands are reactive to O(2). After e(-) beam deposition of1 ML of HfO(2), single chemisorption sites were identified. At low coverage (1 ML), the n-type and p-type HfO(2)/InGaAs(0 0 1)-(4x2) interfaces show p-type character in STS, which is typical of clean InGaAs(0 0 1)-(4x2). After annealing below 200 degrees C, full coverage HfO(2)/InGaAs(0 0 1)-(4x2) (1-3 ML) has the surface Fermi level shifted toward the conduction band minimum for n-type InGaAs, but near the valence band maximum for p-type InGaAs. This is consistent with the HfO(2)/InGaAs(0 0 1)-(4x2) interface being at least partially unpinned, i.e., a low density of states in the band gap. The partially unpinned interface results from the modest strength of the bonding between HfO(2) and InGaAs(0 0 1)-(4x2) that prevents substrate atom disruption. The fortuitous structure of HfO(2) on InAs(0 0 1)-(4x2) and InGaAs(0 0 1)-(4x2) allows for the elimination of the partially filled dangling bonds on the surface, which are usually responsible for Fermi level pinning.

The NO chemisorption dynamics on ordered multilayer iron phthalocyanine (FePc) and quasiamorphous multilayer tetra-t-butyl FePc (ttbu-FePc) films on a Au(111) substrate was investigated using the King and Wells reflection technique. The NO zero coverage or initial sticking probabilities (S(0)) were measured as a function of sample temperature (T(s)) and beam energy (E(i)). The experimental results for both films show a monotonic decrease in S(0) with increasing T(s) and E(i) consistent with NO adsorption occurring via a multiple pathway precursor-mediated mechanism in which the adsorbate initially physisorbs to the FePc organics, diffuses, and chemisorbs to the Fe metal center. The saturation coverage is 3% for the multilayer FePc surface and only 2% for the multilayer ttbu-FePc surface consistent with NO chemisorption occurring only on the Fe metal, where NO chemisorbs to 100% of the surface Fe metal centers. The reduced saturation coverage in the ttbu-FePc film is attributed to fewer Fe metal centers in the less dense ttbu-FePc films. A comparison of NO sticking on a multilayer FePc/Au(111) film with NO sticking on a monolayer FePc/Au(111) film shows that S(0) is greater on the multilayer FePc film for all T(s) and E(i), consistent with an increase in collision inelasticity for NO/multilayer FePc/Au(111).

Chemical sensing properties of phthalocyanine thin-film transistors have been investigated using nearly identical n- and p-channel devices. P-type copper phthalocyanine (CuPc) has been modified with fluorine groups to convert the charge carriers from holes to electrons. The sensor responses to the tight binding analyte dimethyl methylphosphonate (DMMP) and weak binding analyte methanol (MeOH) were compared in air and N(2). The results suggest that the sensor response involves counterdoping of pre-adsorbed oxygen (O(2)). A linear dependence of chemical response to DMMP concentration was observed in both n- and p- type devices. For DMMP, there is a factor of 2.5 difference in the chemical sensitivity between n- and p-channel CuPc thin-film transistors, even though it has similar binding strength to n- and p-type CuPc molecules as indicated by the desorption times. The effect is attributed to the difference in the analyte perturbation of electron and hole trap energies in n- and p-type materials.

To develop an atomistic understanding of the binding of NO with iron phthalocyanine (FePc), the interaction between NO (an electron withdrawing gas) and NH3 (an electron donating gas) with an isolated FePc molecule (monomer) was compared with density functional theory. The simulations show that NO strongly chemisorbs to the Fe metal and physisorbs to all the nonmetal sites. Additionally, when NO physisorbs to the inner ring nitrogens, NO subsequently undergoes a barrierless migration to the deep chemisorption well on the Fe metal. Conversely, NH3 only weakly chemisorbs to the Fe metal and does not bind to any other sites. Projected density of states simulations and analysis of the atomic charges show that the binding of NO to the FePc metal results in a charge transfer from the Fe metal to the NO chemisorbate; the opposite effect is observed for the binding of NH3 to the Fe metal. Simulations of NO binding to the Fe metal of a monolayer FePc film and FePc trimer were also performed to show that intermolecular FePc-FePc interactions have a negligible effect on the FePc electronic structure and NO binding.

The gas-surface reaction dynamics of NO impinging on an iron(II) phthalocyanine (FePc) monolayer were investigated using King and Wells sticking measurements. The initial sticking probability was measured as a function of both incident molecular beam energy (0.09–0.4eV) and surface temperature (100–300K). NO adsorption onto FePc saturates at 3% of a monolayer for all incident beam energies and surface temperatures, suggesting that the final chemisorption site is confined to the Fe metal centers. At low surface temperature and low incident beam energy, the initial sticking probability is 40% and decreases linearly with increasing beam energy and surface temperature. The results are consistent with the NO molecule sticking onto the FePc molecules via physisorption to the aromatics followed by diffusion to the Fe metal center, or precursor-mediated chemisorption. The adsorption mechanism of NO onto FePc was confirmed by control studies of NO sticking onto metal-free H2Pc, inert Au(111), and reactive Al(111).

We have measured the rotational state distribution and the angular momentum alignment and orientation of N2 scattered from Ag(111) at 540 K. Using resonance enhanced multiphoton ionization (REMPI), we are able to probe the scattered flux as a function of the exit angle θexit. For a modestly glancing incident beam (θi =30°) and incident translational energy, Ei =0.3 eV, the angular momentum alignment (tumbling vs helicoptering) at both quasispecular detection (θexit=35°) and superspecular detection (θexit=50°) is only weakly dependent upon the surface temperature. However, the angular momentum orientation (clockwise vs counterclockwise rotation) is strongly affected by the surface temperature. Raising the surface temperature from Ts =90 K to Ts =540 K causes the orientation to decrease substantially. Stochastic trajectory calculations were carried out in conjunction with the experiments. They reveal that at low temperature there is an averaging over two important initial conditions: the two‐dimensional imp...

The rotational state populations and the quadrupole and hexadecapole alignment moments of N2 scattered off clean Ag(111) are determined by resonance enhanced multiphoton ionization (REMPI). The scattered N2 is found to be highly aligned with its rotational angular momentum vector J parallel to the surface. The degree of alignment is found to increase with increasing rotational excitation. We see less than perfect alignment at intermediate J values indicating that the surface is not completely flat. The alignment is relatively insensitive to incident energy, incident angle, or surface temperature Ts. However, the rotational state population distributions show pronounced rainbows for higher incident energy and/or more grazing exit angle. The rotational state distributions are found to depend strongly on the final scattering angle at low Ts; this effect is markedly reduced at higher Ts. Time‐of‐flight measurements are used to determine the average velocity of the scattered N2 as a function of rotational leve...

A method is presented for determining the population A(0)0, the quadrupole alignment factors A(2)0, A(2)1, A(2)2, and the hexadecapole alignment factors A(4)0, A(4)1, A(4)2, A(4)3, A(4)4 for a (v,J) ground state distribution of a diatomic molecule probed by linearly polarized two‐photon nonresonant excitation. General expressions are developed for the O, P, Q, R, and S branch transitions as a function of the rotational quantum number J. This treatment assumes that the resonant state reached by the two‐photon transition is subsequently detected independent of its alignment. This can be achieved by 2+n multiphoton ionization in which the ionization steps are saturated, or by 2+1 laser induced fluorescence in which the fluorescence is collected independent of its polarization and spatial anisotropy. To extract the population and the eight alignment parameters the line intensities must be measured for several polarization settings of the laser beam. However, when the ground state distribution has cylindrical ...