In order to have a high-quality crystallinity for boosting the device performance, the metal-induced lateral crystallization (MILC), using Ni becomes a critical process in semiconductor industries. Simulating full process has difficulties because of scale difference between nanometer-scale atomistic process and micrometer-scale device. In this study, we first present full device scale simulation of MILC process, using efficient atomistic kinetic lattice Monte Carlo simulation. Balance between thermodynamics and kinetics reveals that the crystallization rate is dominated by the energetically stable interfaces. Moreover, we found the dynamic instability of the MILC process, which could potentially fail the device fabrication in a few micrometer scale.

A large scale hybrid method combining Molecular Dynamics with Monte Carlo is implemented for the simulation of Silicon Germanium heteroepitaxy and the prediction of dislocation apparition. On one hand, using the Tersoff potential, the Molecular Dynamics part allows for realistic structure relaxation as well as the creation of a highly discretized potential energy surface. On the other hand, the Monte Carlo part allows for a fast deposition simulation. The combination is furthermore improved to apply the method in two different large scale domains. First, with structures holding millions of atoms and second in a supercomputer environment with thousands of processing cores. The method results show a very close agreement regarding the critical thickness of heteroepitaxied structures grown at their stable states.

Epitaxial growth simulations are critical for understanding and optimizing semiconductor fabrication processes, particularly in logic devices, such as source/drain (S/D) contact fabrication for high-aspect-ratio structures. In this work, we propose novel computational methods to enhance both the accuracy and efficiency of simulations. These methods include an optimized ray-tracing approach using Bounding Volume Hierarchies (BVH), a high-fidelity surface mesh generation technique based on Alpha-Wrapping, a hybrid mesh method combining atomic structure and continuum mesh, and a ray-splitting method for efficient ray trajectory computations. Our simulation results demonstrate significant improvements in simulation time and accuracy.

We present a self-consistent boundary treatment method to account for the long-range stress effects in chip-scale stress simulations. The long range effects are superposed onto the stress solution of the individual tiles as the boundary displacements in a self-consistent manner. The model concept is rigorously tested and applied to a realistic example to demonstrate large-scale simulation capability. Chip-scale and even wafer-scale simulations can be achieved by the nested application of the proposed method. An important potential application would be mask design improvement to avoid stress-related device failure.

Employing Kinetic Monte Carlo (KMC) method for epitaxy simulation reinforced with Molecular Dynamics for KMC model parameter extraction, this study establishes an approach for determining design margins in transistor architectures characterized by high aspect ratio cavities. This integrated atomistic and multiscale simulation framework allows prediction of the epitaxy quality and potential defects as well as the optimization of key design parameters.

To overcome the limitations of the previously developed stress simulation method for full-chip scale [1], which could only analyze a single layer of metallization due to its use of a shell element method, a simulation flow that can handle multiple layers was developed. Introducing stress simulation during the incremental formation of the back-end-of-line (BEOL) structure is crucial for predicting the risk of stress-induced defects not just on the surface, but throughout the entire 3D structure, including the chance of defects between metallization layers. To enhance the predictive capability for the larger-scale stress-induced defects, local stress averaging was utilized to balance simulation accuracy with coverage area. This methodology allowed for the expansion of the simulation domain beyond the chip level, thereby enabling the estimation of layout-induced deformations on a wafer scale

We present a hydrogen-assisted enhanced boron diffusion model in oxide. By introducing the B+OH reaction and BOH diffusion, the model could reproduce the enhanced B diffusion in the presence of high concentrations of hydrogen, which has been reported by previous studies. The model was applied to predict the B profile in the p-type MOSFET and the resulting B profiles were used to predict the threshold voltage (Vth) and the short channel effects (SCE). The simulation results are in good agreement with the measurements. Therefore, the model would be beneficial to optimize the H-involved process conditions.

In this paper, we develop a GPGPU acceleration methodology for the Binary-Collision-Approximation based Monte Carlo ion implantation simulation (MCII). Our proposed method avoids the branch-divergence issue which comes from the difference of material crystallinity for the structure with multiple materials. We also introduce an efficient scheme to mitigate the side effect for damage accumulation due to massive parallelization of simulation. Our demonstration of high energy implantation into CIS structure shows almost 40x speed-up compared to CPU implementation of MCII. We conclude that GPU-MCII is effective for acceleration of Monte Carlo simulations with high energy implantation e.g. deep photodiode or well isolation formation.

Using a combination of domain decomposition, massive parallelization and dimensionality reduction, a full chip-size stress simulation flow was developed. By application of shell elements in the Finite Element Method (FEM) framework, the prediction of the stress distribution in a Flash memory die (area about 1 cm²) back end of line (BEOL) metallization layers with nanometer scale precision becomes possible within a half day. Model calibration for several Flash memory product generations allowed more than 90 percent accuracy of crack defect formation probability prediction.

Diamond attracts an ever-increasing scientific interest not only due to its outstanding properties, but also as host material for the so-called color centers. In particular, the nitrogen-vacancy (NV) center is a promising candidate for applications in quantum sensing on a nanoscale. Incorporating such centers in sharp diamond tips, allows the fabrication of a controllable sensor for magnetic and electric fields. In this regard, we present two different ways to fabricate diamond atomic force microscope (AFM) probes. One approach is based on a bottom-up method, first structuring a silicon substrate by photolithography and anisotropic wet etching and subsequently depositing a nanocrystalline diamond (NCD) film onto the pre-patterned silicon substrate. The second approach is based on electron beam lithography (EBL) and reactive ion etching (RIE), which is also applicable to monocrystalline diamond (MCD). To this end we show our first results in fabricating NV-containing MCD tips by He+ ion implantation and annealing. We demonstrate the fabrication of bottom-up NCD probes with tip radii in the range of ca. 25 nm, and top-down fabricated NCD AFM probes with tip radii even below 10 nm.

This work was funded by the project “Quantum coins and nanosensors” from the Volkswagen Foundation (Az. 91001)

A fabrication process of dense arrays of one‐dimensional diamond nanostructures (diamond nanopillars) integrated on nanocrystalline diamond (NCD) membranes is implemented with an etched‐through marker hole in the membrane in the center of the array. NCD films deposited on silicon substrates are used as starting material. The main fabrication steps consist of structuring the NCD film by electron beam lithography (EBL) applying aluminum as a hard mask and subsequent inductively coupled O2 plasma reactive ion etching (ICP RIE), followed by structuring of the silicon substrate from the backside to open a NCD membrane. The developed fabrication procedure for nanostructured membranes can be transferred to monocrystalline diamond and implemented for deterministic single ion implantation into the nanopillars for generation of nitrogen‐vacancy (NV) or silicon‐vacancy (SiV) centers. For this purpose, the etched marker hole can be applied for alignment in the implantation process. Such diamond nanostructures with integrated color centers could play an important role in the development of novel quantum memory devices.

TCAD prediction of the stress field generated by dislocation is crucial for the optimization of stressors for next generation logic devices. In this paper, we present a new hybrid approach for dislocation stress field calculation and its application to strained Si devices. New methodology combines an analytic stress field model for dislocation cores and consecutive FEM stress solving to get mechanical equilibrium. It was applied to the design optimization of dislocation stress memorization technique (D-SMT), its local layout effect (LLE) modeling, and the relaxation of lattice mismatch strain at Si/SiGe interface which degrades eSiGe stress. All the simulation results were verified with experimental results.

Highly dense arrays of diamond nanopillars have been fabricated using nanocrystalline diamond films (NCD) as the starting material. The fabrication process consisted of electron beam lithography (EBL), aluminum mask deposition and inductively coupled O2 plasma reactive ion etching. The EBL pattern fidelity was enhanced by proximity corrections and dose variations. Optical characterizations of the arrays indicated the incorporation of silicon-vacancy centers during NCD growth as well as enhanced fluorescence and photoluminescence intensities in the well-developed pillar arrays. Transferring this fabrication method to monocrystalline diamond, such dense arrays of diamond nanopillars could be applied in quantum photonics as emitter arrays or photonic crystals upon integration of color centers.

Diamond is a material with a number of outstanding mechanical, optical, electrical and chemical properties. In the last decade it has additionally attracted the attention of scientists due to the promising properties of the color centers in its crystal lattice which can find applications in quantum information technology or quantum sensing on a nanometer scale. In this contribution we present the most prominent color centers in diamond, namely the nitrogen-vacancy (NV) and the silicon-vacancy (SiV) centers, and the possibilities to create them in diamond. In order to increase the collection efficiency of the photons emitted by the color centers they should be incorporated in photonic structures. We demonstrate the fabrication of nanopillars with diameters down to 50 nm in nanocrystalline diamond (NCD) films and monocrystalline diamond. In order to reduce the photon scattering from the rough NCD surfaces we developed a planarization technique which significantly reduced the surface roughness.

Automatic flow of transmission electron microscopy (TEM)-based dislocation analysis on Source/Drain (S/D) and contact formation process is developed. Based on the previously developed model of dislocation stress, an automated methodology is implemented that allows fast and human-error-free extraction of dislocation core position and its impact on the device channel stress and the electrical performance. This approach enables us to analyze the impact of dislocations in S/D of advanced logic devices and to optimize structure and process conditions.

For modern semiconductor devices, the level of details which we should investigate for predictive simulation is going extreme. Not only the atomistic simulation is required but equipment and transistor scale simulation is also needed to understand the formation of atomic scale feature. In this paper, practical applications of multi-domain simulations are introduced for advanced S/D process in logic, interface engineering in DRAM cell and cell stack ALD process of flash memory devices.

A new dynamical space partitioning method is presented in a parallelized lattice kinetic Monte Carlo (kMC) simulator to overcome the loss of parallel efficiency found in other parallelized kMC simulators. The dynamical partitioning of the simulation cell allows better load balancing through all threads hence reducing time consuming events during the simulation. The new method is evaluated against both hypothetical and real cases. In both cases, minimal differences between serial and parallelized simulations are found. In real cases, other code optimizations may be needed to further improve the parallel efficiency.

Glucoamylase from fungus Aspergillus awamori is glycoside hydrolase that catalyzes the hydrolysis of α-1,4- and α-1,6-glucosidic bonds in glucose polymers and oligomers. This glycoprotein consists of a catalytic domain and a starch-binding domain connected by an O-glycosylated polypeptide chain. The conformation of the linker, the relative arrangement of the domains, and the structure of the full-length enzyme are unknown. The structure of the recombinant glucoamylase GA1 was studied by molecular modelling and small-angle neutron scattering (SANS) methods. The experimental SANS data provide evidence that glucoamylase exists as a monomer in solution and contains a glycoside component, which makes a substantial contribution to the scattering. The model of full-length glucoamylase, which was calculated without taking into account the effect of glycosylation, is consistent with the experimental data and has a radius of gyration of 33.4 ± 0.6 Å.