QuantumATK - New Release Notes

Machine-Learned (ML) Force Fields | Moment Tensor Potentials (MTPs)

• 100-1000x faster generation of realistic structures of complex multi-element crystalline, amorphous materials & interfaces, defect and dopant migration barriers, thermal transport, crystallization vs. DFT. 
• Systematically improvable MTPs
  • Trained on a dataset of ab-initio calculations.
  • One of the most accurate and efficient ML potentials on the market. Nearly the same accuracy as ab-initio.
  • For cases where no conventional potentials exist or need better accuracy.
• Active learning MTP simulations to automatically add DFT training data during molecular dynamics (MD) simulations. 
  •  Obtain realistic amorphous material and liquid structures, in particular, at high temperatures. 
• Employ provided MTP potentials for Si or develop potentials for new materials and problems using automated training and simulation workflows.
• Use MTPs with MD, nudged elastic band and accelerated MD methods, such as force-bias Monte Carlo, now also with pressure control, to sample rare events and unlock slow mechanisms. 

Complex Semiconductor Materials, Interfaces & Gate Stacks

• Use ML MTPs for obtaining realistic crystalline, amorphous materials, interface, gate stack structures, simulating dopant diffusion, thermal transport, and crystallization. 
  • Examples include amorphous HfO2 and GST phase-change materials, HKMG stacks, etc.
• Fast and highly accurate electronic properties of materials, interfaces, and gate stack (e.g. HKMGs) structures comprised of multiple layers with different band gaps using the dielectric dependent hybrid HSE06 (HSE06-DDH) method. 
  • HSE06-DDH method is based on using improved material-specific fractions of exact exchange, automatically calculated from density for each material (in an interface). 
  • Available with LCAO basis sets for efficient large-scale simulations with modest hardware.
• Geometry optimization with stress and spin-polarization is now available with HSE06-LCAO. 
  • Accurate large-scale simulations of electronic properties orders of magnitude faster compared to HSE06-PlaneWave. 
• New inverse participation ratio (IPR) analysis object to evaluate localized states.
  • Part of the insightful electronic and vibrational analysis of systems with defects, amorphous materials, surfaces and interfaces, e.g., in HKMG and 3D-NAND memory stacks. 
• Plot band edges in projected DOS, local DOS and projected local density of states analyzers.
• Defect and dopant simulation improvements 
  • Easier set-up of individual defect migration paths. 
  • Apply constraints and point defect symmetry to reduce the computational cost of defect diffusion simulations, e.g. at interfaces in HKMG stacks.

1D & 2D-Material Based FETs

• More accurate band diagrams and device I-V characteristics with the new HSE06-NEGF methodology compared to PBE-NEGF. 
• More accurate on-state calculations using Neumann boundary conditions in the transport direction compared to Dirichlet at the Semi-Empirical level.
• Up to 80% faster simulations of gated devices with vacuum regions using the new Poisson solver using a non-uniform grid compared to the parallel conjugate gradient (PCG) solver. 

Novel STT-MRAM Memory Design

• Obtain realistic interface structures and energetics of magnetic tunnel junctions in MRAM with ML MTPs.
• New magnetic properties such as Heisenberg exchange coupling, exchange stiffness, and Curie temperature.
• 10-100x faster Heisenberg exchange calculations, now also with non-collinear spin and spin-orbit. 
• 60x times faster and 70 % less memory demanding magnetic anisotropy energy projection simulations.

Advanced Surface Process Simulations

• Enhanced surface process simulation module enabling scanning over a range of impact energies and incident angles of “shooting” atoms at a surface for maximum yield in sputtering, etching (ALE) and deposition (ALD) processes. 
• Compute quantities, such as sputtering yield and sticking coefficient, needed for feature scale and reactor scale models. 
  • Plot calculated sputtering yield and sticking coefficient using the new Grid Data Visualizer. 
• Use the newly implemented thermochemical selectivity analysis tools in the GUI to screen critical reactions in a process, find ideal reactants and optimal reaction conditions for the processes. 
  • Take advantage from the Thermochemistry Database  for reactants and products.

Battery Materials Modeling and Design

Building

• Improved plugin for adsorbing molecules onto a surface and a new nanoparticle builder for creating a nanoparticle electrode.
• Improved move, measurement and atom wrapping tools.

ForceField Simulations

• New bonded OPLS potential for common electrolytes and OPLS-Min potential for use with custom charges and simpler type assignments.
• Convenient access to bonded potentials in the GUI and possibility to edit all terms, including torsion potentials. 
• Large-scale solid-electrolyte-interphase (SEI) simulations using 3x faster ReaxFF Force Field MD or combined bonded and conventional potentials in the GUI.
• Easy set-up and simulations with partial charges to model electrostatic interactions using the GUI. 
• Vibrational spectra from MD to understand molecular interactions and solvation in liquid phase.
• Surface process modeling and Thermochemistry analysis tools for modeling reactions on electrode surfaces. 

Density Functional Theory Simulations

• More accurate electronic structure, binding energies, and diffusion barriers with the hybrid DFT functionals, such as HSE06 and the newly added PBE0, B3LYP, B3LYP5. 
  • Use LCAO basis sets for 100x speed-up compared to Plane Wave basis sets enabling highly efficient large-scale simulations with modest hardware.
• More accurate modeling of binding energy and adsorption sites with counterpoise corrections to DFT-NEGF.

Polymer Modeling

• Access bonded potentials (OPLS-AA, OPLS-Min, Dreiding, UFF) and edit them in the GUI for more convenient set-up of polymer simulations.
• Combine bonded and conventional potentials in the GUI for more accurate simulations of polymer-inorganic and polymer-nanoparticle interfaces.
• Graphically build and simulate polymer systems with ionically charged molecules using the QEq method, particularly relevant for photoresist polymers. 
• GUI support for united atoms in the polymer building workflow to speed up polymer simulations by folding hydrogen atoms into their attached carbon atom.