QuantumATK Simulation Platform

• QuantumATK is an integrated atomic-scale simulation platform that couples first-principles, tight-binding, and machine-learned force fields to model electronic structure and quantum transport.
• It supports advanced workflows including NEGF for quantum transport, electron–phonon, and photon interactions to predict device physics and thermodynamic properties in complex materials.
• The platform’s modular design and Python-based API enable seamless integration of simulation engines, fostering multi-scale, hybrid quantum–classical modeling for research and design.

QuantumATK is an integrated atomic-scale simulation platform for electronic-structure modeling, quantum transport, and atomistic property prediction in materials, devices, and extended systems. Developed since 2003 by professional software engineers and academic researchers, QuantumATK enables multiscale modeling through direct coupling of first-principles methodologies such as density functional theory (DFT), tight-binding (TB), and advanced empirical or machine-learned force fields. Its architecture supports complex workflows, including transport calculations, molecular dynamics, electron–phonon interactions, and the simulation of superconducting and topological systems relevant to quantum devices.

1. Simulation Engines and Theoretical Formalism

QuantumATK provides multiple simulation engines, each suited to specific computational regimes:

• Density Functional Theory (DFT) is implemented in both linear combination of atomic orbitals (LCAO) and plane-wave (PW) basis sets. The Kohn–Sham Hamiltonian is formulated as $H^{KS} = -(\hbar^2/2m)\nabla^2 + V^{eff}[n]$, where $V^{eff}[n] = V_{ext} + V_H[n] + V_{xc}[n]$ and $V_{ext}$ is provided by pseudopotentials (Smidstrup et al., 2019, Smidstrup et al., 15 Sep 2025).
• Semi-Empirical Tight-Binding (TB) approaches utilize Slater–Koster parameterizations with both orthogonal and nonorthogonal formulations; self-consistency is introduced through charge correction terms reliant on the overlap matrix and solution of Poisson’s equation.
• Empirical and Machine-Learned Force Fields (MLFFs) are supported for classical molecular dynamics over large system sizes and for capturing temperature-dependent atomistic properties where no reliable classical force field exists (Smidstrup et al., 15 Sep 2025).

All modules are designed as C++ libraries with Python bindings, facilitating unified scripting and integration.

2. Quantum Transport and Non-Equilibrium Green’s Functions (NEGF)

QuantumATK incorporates a NEGF framework for quantum transport, surface state, and interface calculations:

• The platform computes the retarded Green’s function as $$G(\varepsilon) = [ (\varepsilon + i\delta) S - H - \Sigma_L(\varepsilon) - \Sigma_R(\varepsilon) ]^{-1}$$, with transmission defined by $$T(\varepsilon) = \operatorname{Tr}[G(\varepsilon) \Gamma_L(\varepsilon) G^\dagger(\varepsilon) \Gamma_R(\varepsilon)]$$ (Smidstrup et al., 2019, Smidstrup et al., 15 Sep 2025).
• The Landauer formula for current, $$I = (2e/h) \int T(\varepsilon)[f(\varepsilon - \mu_L) - f(\varepsilon - \mu_R)]~d\varepsilon$$, is used for device simulation, including field-effect transistor geometries and Josephson junctions.
• NEGF can be directly combined with both DFT and tight-binding descriptions, enabling calculations of transmission, resistance, and surface state localization in superconducting or topological systems.

This formalism overcomes limitations of periodic boundary conditions and allows the description of single interfaces or edges, crucial for superconducting qubit and sensor design.

3. Machine-Learned Force Fields and Large-Scale Atomistics

QuantumATK has implemented MLFFs to address system size and complexity requirements:

• MLFFs are trained to reproduce ab initio data where classical potentials do not exist, allowing realistic generation of amorphous geometries (e.g., oxide tunnel junction interfaces) and thermal property calculation for thousands to millions of atoms (Smidstrup et al., 15 Sep 2025).
• ML-driven protocols enable the paper of thermodynamic ensembles and electron–phonon effects at temperature scales relevant to quantum technologies.
• The speed and scalability of MLFF-based MD simulations are leveraged to optimize disorder, interface stability, and transport properties in complex multi-element materials.

This approach is significant when modeling device-relevant large systems, such as Josephson junctions and double quantum dot structures.

4. Coupling of Many-Body Effects with Atomistic Models

QuantumATK extends tight-binding and ab initio results to many-body physics for TLS models:

• Two-level systems—used as coarse models for qubits—are described via double-well potentials whose minima represent different quantum states. These are constructed from atomistic tight-binding spectra and extended using configuration interaction (CI) techniques to incorporate electron–electron repulsion (Smidstrup et al., 15 Sep 2025).
• Many-body Hamiltonians, derived via Slater–Condon parameters, are diagonalized to compute singlet-triplet energy splittings, coherent oscillations, and state localization as detuning is varied.
• Calculation results inform the operational metrics, fidelity, and coherence benchmarks of solid-state qubits, guiding device engineering.

This explicit treatment is required for realistic prediction of qubit energy levels and read-out transitions.

5. Advanced Modules for Electron–Phonon and Photon Coupling

QuantumATK contains modules for interaction of electrons with lattice vibrations and photons:

• Electron–phonon matrix elements $$g_{k,k',q}^{\lambda, n, n'} = \langle \psi_{n', k'} | \delta\hat{H}_{\lambda, q} |\psi_{n, k}\rangle$$ are computed via finite difference of the Hamiltonian (Smidstrup et al., 2019).
• Phonon properties are extracted from harmonic dynamical matrices, $$D_{a\alpha;b\beta}=(1/\sqrt{m_a m_b}) (\partial F_{b\beta}/\partial r_{a\alpha})$$, for phonon density of states and thermal conductivity prediction.
• Electron–photon coupling is treated by perturbations of the form $(e/m_0) \mathbf{A}_\omega \cdot \mathbf{p}$, supporting quantum optics calculations.

Such capabilities are essential for first-principles transport, mobility calculations, and examination of coherence and relaxation in quantum devices.

6. Workflow Integration, NanoLab UI, and Hybrid Computational Approaches

The platform’s modular design permits script-based and graphical workflow composition:

• Python-based scripting (via a unified API) enables complex workflows, for example, combining force-field MD with on-the-fly DFT corrections for charge calculations in battery drift studies: $Q_a'(t) = Q_a^{FF} + \Delta Q_a(t)$ with $\Delta Q_a(t) = Q_a^{DFT}(t) - Q_a^{DFT,ref}$ (Smidstrup et al., 2019).
• The NanoLab graphical interface manages simulation campaigns and interoperability of modules across different methodologies.
• Integration with external quantum chemistry tools (e.g., ASE, VQE-driven calculators) enables the seamless blending of quantum algorithms and classical optimization for molecular systems (R et al., 2022), paving the way for hybrid quantum–classical workflows.

This level of workflow coupling is central for addressing multi-scale and multi-physics problems in materials science, catalysis, and device research.

7. Applications and Benchmark Studies

Selected examples illustrate QuantumATK’s capability and validation:

• Phonon-limited mobility calculations and electron–phonon scattering rates for metals (Cu, Ag, Au), showing congruence with experimental trends (Smidstrup et al., 2019).
• Electronic transport in gated 2D tunnel FETs, where NEGF grid and boundary schemes account for gate-induced potential in MoTe$_2$/SnS$_2$ (Smidstrup et al., 2019).
• Atomistic studies of superconducting tunnel junctions (Al/AlO$_x$/Al), with MLFF-driven ensemble statistics for resistance and critical current (Smidstrup et al., 15 Sep 2025).
• Double quantum dot systems for qubit operation, modeled through tight-binding plus CI for many-body splitting and state dynamics (Smidstrup et al., 15 Sep 2025).
• Battery cathode drift simulations using multi-model MD/DFT charge integration (Smidstrup et al., 2019).

These studies validate both accuracy and flexibility, as well as highlight QuantumATK’s role in the advancement of quantum materials and device modeling.

---

QuantumATK amalgamates high-level quantum and atomistic simulation engines, workflow integration, and machine-learned potentials for state-of-the-art modeling of electronic, transport, and thermal properties in large and complex materials systems. The platform’s hybrid methodology, availability of advanced modules, and extensibility to quantum/classical co-simulation position it as a central tool for the rigorous investigation of materials and device architectures in quantum computing, condensed matter physics, and related domains.