Improving Accuracy and Transferability of Machine Learning Chemical Activation Energies by Adding Electronic Structure Information.

Predicting chemical activation energies is one of the longstanding and important challenges in computational chemistry. Recent advances have shown that machine learning can be used to create tools to predict them. Such tools can significantly decrease the computational cost for these predictions compared to traditional methods, which require an optimal path search along a high-dimensional potential energy surface. To enable this new route, we need both large and accurate datasets and a compact yet complete description of the reactions. Although data for chemical reactions is becoming increasingly available, the key step of encoding the reaction as an efficient descriptor remains a big challenge. In this paper, we demonstrate that including electronic energy levels in the description of the reaction significantly improves the prediction accuracy and transferability. Feature importance analysis further demonstrates that electronic energy levels have a higher importance than some structural information and typically require less space in the reaction encoding vector. In general, we observe that the results of the feature importance analysis relate well to the domain knowledge of fundamental chemical principles. This work can help to build better chemical reaction encodings for machine learning and thus improve the predictions of machine learning models for reaction activation energies. These models could ultimately be used to recognize reaction limiting steps in large reaction systems, allowing to account for bottlenecks at the design stage.

Benchmarking First-Principles Reaction Equilibrium Composition Prediction.

The availability of thermochemical properties allows for the prediction of the equilibrium compositions of chemical reactions. The accurate prediction of these can be crucial for the design of new chemical synthesis routes. However, for new processes, these data are generally not completely available. A solution is the use of thermochemistry calculated from first-principles methods such as Density Functional Theory (DFT). Before this can be used reliably, it needs to be systematically benchmarked. Although various studies have examined the accuracy of DFT from an energetic point of view, few studies have considered its accuracy in predicting the temperature-dependent equilibrium composition. In this work, we collected 117 molecules for which experimental thermochemical data were available. From these, we constructed 2648 reactions. These experimentally constructed reactions were then benchmarked against DFT for 6 exchange-correlation functionals and 3 quality of basis sets. We show that, in reactions that do not show temperature dependence in the equilibrium composition below 1000 K, over 90% are predicted correctly. Temperature-dependent equilibrium compositions typically demonstrate correct qualitative behavior. Lastly, we show that the errors are equally caused by errors in the vibrational spectrum and the DFT electronic ground state energy.

Breakdown of Universal Scaling for Nanometer-Sized Bubbles in Graphene.

We report the formation of nanobubbles on graphene with a radius of the order of 1 nm, using ultralow energy implantation of noble gas ions (He, Ne, Ar) into graphene grown on a Pt(111) surface. We show that the universal scaling of the aspect ratio, which has previously been established for larger bubbles, breaks down when the bubble radius approaches 1 nm, resulting in much larger aspect ratios. Moreover, we observe that the bubble stability and aspect ratio depend on the substrate onto which the graphene is grown (bubbles are stable for Pt but not for Cu) and trapped element. We interpret these dependencies in terms of the atomic compressibility of the noble gas as well as of the adhesion energies between graphene, the substrate, and trapped atoms.

Understanding ambipolar transport in MoS2 field effect transistors: the substrate is the key.

2D materials offer a pathway for further scaling of CMOS technology. However, for this to become a reality, both n-MOS and p-MOS should be realized, ideally with the same (standard) material. In the specific case of MoS2 field effect transistors (FETs), ambipolar transport is seldom reported, primarily due to the phenomenon of Fermi level pinning (FLP). In this study we identify the possible sources of FLP in MoS2 FETs and resolve them individually. A novel contact transfer technique is used to transfer contacts on top of MoS2 flake devices that results in a significant increase in the hole branch of the transfer characteristics as compared to conventionally fabricated contacts. We hypothesize that the pinning not only comes from the contact-MoS2 interface, but also from the MoS2-substrate interface. We confirm this by shifting to an hBN substrate which leads to a 10 fold increase in the hole current compared to the SiO2 substrate. Furthermore, we analyse MoS2 FETs of different channel thickness on three different substrates, SiO2, hBN and Al2O3, by correlating the p-branch I ON/I OFF to the position of oxide defect band in these substrates. FLP from the oxide is reduced in the case of Al2O3 which enables us to observe ambipolar transport in a bilayer MoS2 FET. These results highlight that MoS2 is indeed an ambipolar material, and the absence of ambipolar transport in MoS2 FETs is strongly correlated to its dielectric environment and processing conditions.

Electron-induced fragmentation mechanisms in organic monomers and their implications for photoresist optimization for EUV lithography.

Secondary electrons generated during the Extreme Ultraviolet Lithography (EUVL) process are predominantly responsible for inducing important patterning chemistry in photoresist films. Therefore, it is crucial to understand the electron-induced fragmentation mechanisms involved in EUV-resist systems to improve their patterning performance. To facilitate this understanding, mechanistic studies were carried out on simple organic EUV-resist monomers, methyl isobutyrate (MIB) and methacrylic acid (MAA), both in the condensed and gas phases. Electron-stimulated desorption (ESD) studies on MIB in the condensed phase showed desorption peaks at around 2 and 9 eV electron energies. The gas-phase study on MIB showed that the monomer followed the dissociative ionization (DI) fragmentation pathway, under single collision conditions, which opened up at electron energies above about 11 eV. No signs of dissociative electron attachment (DEA) were detected for MIB in the gas phase under single collision conditions. However, DEA was an active process in MAA in the gas phase under single collision conditions at around 2 eV, showing that slight modifications of the molecular structures of photoresists may serve to sensitize them to certain electron-induced processes.

Peculiar alignment and strain of 2D WSe2 grown by van der Waals epitaxy on reconstructed sapphire surfaces.

The increasing scientific and industry interest in 2D MX2 materials within the field of nanotechnology has made the single crystalline integration of large area van der Waals (vdW) layers on commercial substrates an important topic. The c-plane oriented (3D crystal) sapphire surface is believed to be an interesting substrate candidate for this challenging 2D/3D integration. Despite the many attempts that have been made, the yet incomplete understanding of vdW epitaxy still results in synthetic material that shows a crystallinity far too low compared to natural crystals that can be exfoliated onto commercial substrates. Thanks to its atomic control and in situ analysis possibilities, molecular beam epitaxy (MBE) offers a potential solution and an appropriate method to enable a more in-depth understanding of this peculiar 2D/3D hetero-epitaxy. Here, we report on how various sapphire surface reconstructions, that are obtained by thermal annealing of the as-received substrates, influence the vdW epitaxy of the MBE-grown WSe2 monolayers (MLs). The surface chemistry and the interatomic arrangement of the reconstructed sapphire surfaces are shown to control the preferential in-plane epitaxial alignment of the stoichiometric WSe2 crystals. In addition, it is demonstrated that the reconstructions also affect the in-plane lattice parameter and thus the in-plane strain of the 2D vdW-bonded MLs. Hence, the results obtained in this work shine more light on the peculiar concept of vdW epitaxy, especially relevant for 2D materials integration on large-scale 3D crystal commercial substrates.

Layer-controlled epitaxy of 2D semiconductors: bridging nanoscale phenomena to wafer-scale uniformity.

The rapid cadence of MOSFET scaling is stimulating the development of new technologies and accelerating the introduction of new semiconducting materials as silicon alternative. In this context, 2D materials with a unique layered structure have attracted tremendous interest in recent years, mainly motivated by their ultra-thin body nature and unique optoelectronic and mechanical properties. The development of scalable synthesis techniques is obviously a fundamental step towards the development of a manufacturable technology. Metal-organic chemical vapor deposition has recently been used for the synthesis of large area TMDs, however, an important milestone still needs to be achieved: the ability to precisely control the number of layers and surface uniformity at the nano-to micro-length scale to obtain an atomically flat, self-passivated surface. In this work, we explore various fundamental aspects involved in the chemical vapor deposition process and we provide important insights on the layer-dependence of epitaxial MoS2 film's structural properties. Based on these observations, we propose an original method to achieve a layer-controlled epitaxy of wafer-scale TMDs.

Superhydrophobic Breakdown of Nanostructured Surfaces Characterized in Situ Using ATR-FTIR.

In situ characterization of the underwater stability of superhydrophobic micro- and nanostructured surfaces is important for the development of self-cleaning and antifouling materials. In this work, we demonstrate a novel attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectroscopy-based method for large-area wetting characterization of silicon nanopillars. When air is present in between the structures, as is characteristic of the Cassie-Baxter state, the relative intensities of the water bands in the absorption spectrum change because of the wavelength-dependent attenuation of the evanescent wave. This phenomenon enables unambiguous identification of the wetting state and assessment of liquid impalement. Using mixtures of isopropanol and water with different concentrations, the breakdown of superhydrophobic states and the wetting hysteresis effects are systematically studied on uniform arrays of silicon nanopillars. A transition from the Cassie-Baxter to Wenzel state is observed when the isopropanol concentration exceeds 2.8 mol %, corresponding to a critical surface tension of 39 mN/m. Spontaneous dewetting does not occur upon decreasing the isopropanol concentration, and pure water can be obtained in a stable Wenzel state on the originally superhydrophobic substrates. The developed ATR-FTIR method can be promising for real-time monitoring of the wetting kinetics on nanostructured surfaces.

Two-dimensional WS2 nanoribbon deposition by conversion of pre-patterned amorphous silicon.

We present a method for area selective deposition of 2D WS2 nanoribbons with tunable thickness on a dielectric substrate. The process is based on a complete conversion of a pre-patterned, H-terminated Si layer to metallic W by WF6, followed by in situ sulfidation by H2S. The reaction process, performed at 450 °C, yields nanoribbons with lateral dimension down to 20 nm and with random basal plane orientation. The thickness of the nanoribbons is accurately controlled by the thickness of the pre-deposited Si layer. Upon rapid thermal annealing at 900 °C under inert gas, the WS2 basal planes align parallel to the substrate.

Chemical composition of an aqueous oxalato-/citrato-VO(2+) solution as determinant for vanadium oxide phase formation.

Aqueous solutions of oxalato- and citrato-VO(2+) complexes are prepared, and their ligand exchange reaction is investigated as a function of the amount of citrate present in the aqueous solution via continuous-wave electron paramagnetic resonance (CW EPR) and hyperfine sublevel correlation (HYSCORE) spectroscopy. With a low amount of citrate, monomeric cis-oxalato-VO(2+) complexes occur with a distorted square-pyramidal geometry. As the amount of citrate increases, oxalate is gradually exchanged for citrate. This leads to (i) an intermediate situation of monomeric VO(2+) complexes with a mix of oxalate/citrate ligands and (ii) a final situation of both monomeric and dimeric complexes with exclusively citrato ligands. The monomeric citrato-VO(2+) complexes dominate (abundance > 80%) and are characterized by a 6-fold chelation of the vanadium(IV) ion by 4 RCO2(-) ligands at the equatorial positions and a H2O/R-OH ligand at the axial position. The different redox stabilities of these complexes, relative to that of dissolved O2 in the aqueous solution, is analyzed via (51)V NMR. It is shown that the oxidation rate is the highest for the oxalato-VO(2+) complexes. In addition, the stability of the VO(2+) complexes can be drastically improved by evacuation of the dissolved O2 from the solution and subsequent storage in a N2 ambient atmosphere. The vanadium oxide phase formation process, starting with the chemical solution deposition of the aqueous solutions and continuing with subsequent processing in an ambient 0.1% O2 atmosphere, differs for the two complexes. The oxalato-VO(2+) complexes turn into the oxygen-deficient crystalline VO2 B at 400 °C, which then turns into crystalline V6O13 at 500 °C. In contrast, the citrato-VO(2+) complexes form an amorphous film at 400 °C that crystallizes into VO2 M1 and V6O13 at 500 °C.

Electrochemical deposition of subnanometer Ni films on TiN.

In this paper, we show the electrochemical deposition of a subnanometer film of nickel (Ni) on top of titanium nitride (TiN). We exploit the concept of cluster growth inhibition to enhance the nucleation of new nuclei on the TiN substrate. By deliberately using an unbuffered electrolyte solution, the degree of nucleation is enhanced as growth is inhibited more strongly. This results in a very high particle density and therefore an ultralow coalescence thickness. To prevent the termination of Ni deposition that typically occurs in unbuffered solutions, the concentration of Ni(2+) in solution was increased. We have verified with RBS and ICP-MS that the deposition of Ni on the surface in this case did not terminate. Furthermore, annealing experiments were used to visualize the closed nature of the Ni film. The closure of the deposited film was also confirmed by TOF-SIMS measurements and occurs when the film thickness is still in the subnanometer regime. The ultrathin Ni film was found to be an excellent catalyst for carbon nanotube growth on conductive substrates and can also be applied as a seed layer for bulk deposition of a smooth Ni film on TiN.

Chemically enhanced double-gate bilayer graphene field-effect transistor with neutral channel for logic applications.

In this article, we present the simulation, fabrication, and characterization of a novel bilayer graphene field-effect transistor exhibiting electron mobility up to ~1600 cm(2) V(-1) s(-1), a room temperature I on/I off ≈ 60, and the lowest total charge (~10(11) cm(-2)) reported to date. This is achieved by combined electrostatic and chemical doping of bilayer graphene, which enables one to switch off the device at zero top-gate voltage. Using density functional theory and atomistic simulations, we obtain physical insight into the impact of chemical and electrostatic doping on bandgap opening of bilayer graphene and the effect of metal contacts on the operation of the device. Our results represent a step forward in the use of bilayer graphene for high-performance logic devices in the beyond-complementary metal-oxide-semiconductor (CMOS) technology paradigm.

Achieving High Substitutional Incorporation in Mn-Doped Graphene.

Despite its broad potential applications, substitution of carbon by transition metal atoms in graphene has so far been explored only to a limited extent. We report the realization of substitutional Mn doping of graphene to a record high atomic concentration of 0.5%, which was achieved using ultralow-energy ion implantation. By correlating the experimental data with the results of ab initio Born-Oppenheimer molecular dynamics calculations, we infer that direct substitution is the dominant mechanism of impurity incorporation. Thermal annealing in ultrahigh vacuum provides efficient removal of surface contaminants and additional implantation-induced disorder, resulting in Mn-doped graphene that, aside from the substitutional Mn impurities, is essentially as clean and defect-free as the as-grown layer. We further show that the Dirac character of graphene is preserved upon substitutional Mn doping, even in this high concentration regime, making this system ideal for studying the interaction between Dirac conduction electrons and localized magnetic moments. More generally, these results show that ultralow energy ion implantation can be used for controlled functionalization of graphene with substitutional transition-metal atoms, of relevance for a wide range of applications, from magnetism and spintronics to single-atom catalysis.

Area-Selective Deposition of AlOx and Al-Silicate for Fully Self-Aligned Via Integration.

The rush for better-performing electronics, and manufacturing processes that heavily rely on "top-down" patterning techniques, is making the integration of "self-aligned" fabrication methods, such as area-selective deposition (ASD), a critical objective for continued device scaling. The fully self-aligned via (FSAV) scheme is broadly proposed as a "killer application" to determine whether ASD can shift from an R&D process to high-volume manufacturing. Nevertheless, the lack of a suitable low-κ deposition process has prevented the realization of FSAV by dielectric-on-dielectric ASD. This is primarily due to the high temperature and/or strong oxidizers employed during low-κ dielectric deposition and their unsuitability in the presence of organic masks, such as self-assembled monolayers (SAMs), used to prevent material nucleation during ASD. In this work, AlOx and Al-silicate atomic layer deposition (ALD) processes are studied to provide suitable materials for ASD-enabled FSAV. Dimethylaluminum isopropoxide and H2O are utilized to deposit the metal oxide, whereas Al-silicate is grown by adding 2,2-dimethoxy-1,6-diaza-2-silacyclooctane (DMDAcO) pulses to the AlOx ALD cycle. The selectivity of such processes is demonstrated on 50 nm Cu/SiO2 structures, using octadecanethiol-derived SAMs to inhibit material nucleation on the metal lines. Scanning and transmission electron microscopies are employed to assess the quality of the ASD processes and investigate the mechanisms behind defect generation on a nongrowth surface. X-ray photoelectron spectroscopy measurements show the high purity of the AlOx film, whereas DMDAcO-ligand incorporation into the Al-silicate matrix is observed. Planar capacitor structures are used to assess the electrical properties of both ASD films, revealing that the silicate film exhibits a relatively low κ-value (5.3 ± 0.2), with a high acceleration field factor (32.4 ± 1.4) and a dielectric breakdown voltage of 6.0 ± 0.3 V at 100 °C.

Au3-Decorated graphene as a sensing platform for O2 adsorption and desorption kinetics.

The adsorption and desorption kinetics of molecules is of significant fundamental and applied interest. In this paper, we present a new method to quantify the energy barriers for the adsorption and desorption of gas molecules on few-atom clusters, by exploiting reaction induced changes of the doping level of a graphene substrate. The method is illustrated for oxygen adsorption on Au3 clusters. The gold clusters were deposited on a graphene field effect transistor and exposed to O2. From the change in graphene's electronic properties during adsorption, the energy barrier for the adsorption of O2 on Au3 is estimated to be 0.45 eV. Electric current pulses increase the temperature of the graphene strip in a controlled way and provide the required thermal energy for oxygen desorption. The oxygen binding energy on Au3/graphene is found to be 1.03 eV and the activation entropy is 1.4 meV K-1. The experimental values are compared and interpreted on the basis of density functional theory calculations of the adsorption barrier, the binding energy and the activation entropy. The large value of the activation entropy is explained by the hindering effect that the adsorbed O2 has on the fluxional motion of the Au3 cluster.

Measuring the electrical resistivity and contact resistance of vertical carbon nanotube bundles for application as interconnects.

Carbon nanotubes (CNT) are known to be materials with potential for manufacturing sub-20 nm high aspect ratio vertical interconnects in future microchips. In order to be successful with respect to contending against established tungsten or copper based interconnects, though, CNT must fulfil their promise of also providing low electrical resistance in integrated structures using scalable integration processes fully compatible with silicon technology. Hence, carefully engineered growth and integration solutions are required before we can fully exploit their potentialities. This work tackles the problem of optimizing a CNT integration process from the electrical perspective. The technique of measuring the CNT resistance as a function of the CNT length is here extended to CNT integrated in vertical contacts. This allows extracting the linear resistivity and the contact resistance of the CNT, two parameters to our knowledge never reported separately for vertical CNT contacts and which are of utmost importance, as they respectively measure the quality of the CNT and that of their metal contacts. The technique proposed allows electrically distinguishing the impact of each processing step individually on the CNT resistivity and the CNT contact resistance. Hence it constitutes a powerful technique for optimizing the process and developing CNT contacts of superior quality. This can be of relevant technological importance not only for interconnects but also for all those applications that rely on the electrical properties of CNT grown with a catalytic chemical vapor deposition method at low temperature.

Cyclic Plasma Halogenation of Amorphous Carbon for Defect-Free Area-Selective Atomic Layer Deposition of Titanium Oxide.

As critical dimensions in integrated circuits continue to shrink, the lithography-based alignment of adjacent patterned layers becomes more challenging. Area-selective atomic layer deposition (ALD) allows circumventing the alignment issue by exploiting the chemical contrast of the exposed surfaces. In this work, we investigate the selective deposition of TiO2 by plasma halogenation of amorphous carbon (a-C:H) acting as a growth-inhibiting layer. On a-C:H, a CF4 or Cl2 plasma forms a thin halogenated layer that suppresses the growth of TiO2, while nucleation remains unaffected on plasma-treated SiO2. The same halogenating plasmas preferentially etch TiO2 nuclei over films and thus enable the restoration of the halogenated surface of amorphous carbon. By embedding the intermediate plasma treatments in the ALD TiO2 sequence, an 8 nm TiO2 layer could be deposited with a selectivity of 0.998. The application of the cyclic process on a 60 nm half-pitch line pattern resulted in the defect-free deposition of TiO2 at the bottom of the trenches. Cyclic fluorination demonstrated better growth inhibition compared to chlorination due to more efficient defect removal and retention of the favorable surface composition during plasma exposure. While exploring the TiO2 nucleation defects at the limit of detection for conventional elemental analysis techniques (<1 × 1014 at/cm2), we additionally highlight the value of imaging techniques such as atomic force microscopy for understanding defect formation mechanisms and accurately assessing growth selectivity.

Multicomponent Covalent Chemical Patterning of Graphene.

The chemical patterning of graphene is being pursued tenaciously due to exciting possibilities in electronics, catalysis, sensing, and photonics. Despite the intense efforts, spatially controlled, multifunctional covalent patterning of graphene has not been achieved. The lack of control originates from the inherently poor reactivity of the basal plane of graphene, which necessitates the use of harsh chemistries. Here, we demonstrate spatially resolved multicomponent covalent chemical patterning of single layer graphene using a facile and efficient method. Three different functional groups could be covalently attached to the basal plane in dense, well-defined patterns using a combination of lithography and a self-limiting variant of diazonium chemistry requiring no need for graphene activation. The layer thickness of the covalent films could be controlled down to 1 nm. This work provides a solid foundation for the fabrication of chemically patterned multifunctional graphene interfaces for device applications.

A chemisorbed interfacial layer for seeding atomic layer deposition on graphite.

The integration of graphene, and more broadly two-dimensional materials, into devices and hybrid materials often requires the deposition of thin films on their usually inert surface. As a result, strategies for the introduction of surface reactive sites have been developed but currently pose a dilemma between robustness and preservation of the graphene properties. A method is reported here for covalently modifying graphitic surfaces, introducing functional groups that act as reactive sites for the growth of high quality dielectric layers. Aryl diazonium species containing tri-methoxy groups are covalently bonded (grafted) to highly oriented pyrolytic graphite (HOPG) and graphene, acting as seeding species for atomic layer deposition (ALD) of Al2O3, a high-κ dielectric material. A smooth and uniform dielectric film growth is confirmed by scanning electron microscopy (SEM), atomic force microscopy (AFM) and electrical measurements. Raman spectroscopy showed that the aryl groups gradually detach from the graphitic surface during the Al2O3 ALD process at 150 °C, with the surface reverting back to the original sp2-hybridized state and without damaging the dielectric layer. Thus, the grafted aryl groups can act as a sacrificial seeding layer after healing the defects of the graphitic surface with annealing treatment.

Doping Graphene with Substitutional Mn.

We report the incorporation of substitutional Mn atoms in high-quality, epitaxial graphene on Cu(111), using ultralow-energy ion implantation. We characterize in detail the atomic structure of substitutional Mn in a single carbon vacancy and quantify its concentration. In particular, we are able to determine the position of substitutional Mn atoms with respect to the Moiré superstructure (i.e., local graphene-Cu stacking symmetry) and to the carbon sublattice; in the out-of-plane direction, substitutional Mn atoms are found to be slightly displaced toward the Cu surface, that is, effectively underneath the graphene layer. Regarding electronic properties, we show that graphene doped with substitutional Mn to a concentration of the order of 0.04%, with negligible structural disorder (other than the Mn substitution), retains the Dirac-like band structure of pristine graphene on Cu(111), making it an ideal system in which to study the interplay between local magnetic moments and Dirac electrons. Our work also establishes that ultralow-energy ion implantation is suited for substitutional magnetic doping of graphene. Given the flexibility, reproducibility, and scalability inherent to ion implantation, our work creates numerous opportunities for research on magnetic functionalization of graphene and other two-dimensional materials.