🔬 mesoscale physics

A Simple Regularization of the Smooth Quantum Hydrodynamic Model

This paper proposes a simple regularization of the smooth quantum hydrodynamic model by substituting the electron density derivative with a classical Boltzmann distribution to eliminate unstable modes, thereby enabling time-dependent simulations of resonant tunneling diodes that accurately reproduce realistic negative differential resistance and hysteresis in agreement with fully quantum mechanical results.

Original authors: Carl L. Gardner

Published 2026-08-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Carl L. Gardner

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Inside the tiny, engineered landscapes of modern electronics, electrons do not always behave like the solid, predictable particles we encounter in everyday life. In the microscopic world of semiconductor devices, these charged particles can act more like waves, slipping through barriers they should not be able to cross and piling up in specific regions in ways that defy classical intuition. This wave-like behavior is the key to a phenomenon called quantum resonance, which allows certain devices, such as the resonant tunneling diode, to act as incredibly fast switches and amplifiers. When these devices work correctly, they exhibit a strange and useful trait: as you increase the voltage pushing the electrons through, the current flowing through them actually drops for a moment before rising again. This counterintuitive dip, known as negative differential resistance, is the electrical fingerprint of quantum mechanics at work. However, simulating this behavior on a computer is notoriously difficult. The mathematical equations used to describe these fluids of electrons often become unstable and crash when the barriers inside the device are high enough to create the very resonance researchers want to study, leaving scientists without a reliable way to model these essential components using standard fluid-based approaches.

To solve this problem, Carl Gardner at Arizona State University proposed a simple but effective adjustment to the mathematical model used to simulate these devices. The standard equations, which treat the electron gas much like a fluid flowing through a pipe, tend to develop an unstable, growing error when the potential energy barriers are set to realistic heights. Instead of adding complex new terms to fix this, Gardner replaced a specific part of the equation that describes how the electron density changes in space with a simpler, classical assumption. He used the standard Boltzmann distribution, a well-known rule from physics that describes how particles settle in a potential field, to smooth out the problematic calculations. This small change acts as a stabilizer, preventing the simulation from blowing up while still allowing the complex quantum effects to emerge naturally.

When Gardner applied this stabilized model to a simulated resonant tunneling diode, the results were striking. The computer model, running through time until it reached a steady state, successfully reproduced the negative differential resistance that is observed in real experiments. The simulation showed the current rising to a sharp peak and then falling into a valley as the voltage increased, creating a curve that closely matched the results from much more expensive and computationally heavy quantum simulations. Specifically, the model predicted a peak current density of 560 kA/cm² at a voltage of 370 mV and a valley current of 280 kA/cm² at 385 mV. These numbers align remarkably well with data from a fully quantum mechanical simulator called NEMO, which solves the fundamental equations of quantum mechanics without using fluid approximations. The agreement suggests that this simplified fluid approach, once regularized, captures the essential physics of the device without needing the immense computational power required for full quantum calculations.

Beyond just the current-voltage curve, the simulations revealed the physical story happening inside the device. As the voltage was increased, the electrons tunneled through the first barrier and resonated with the energy levels inside the quantum well, a narrow region sandwiched between two barriers. This resonance caused a buildup of electron charge in that well, which in turn created a smooth, modified potential landscape that guided the flow of electrons. As the voltage continued to rise, this resonance effect broke down, the charge in the well dissipated, and the current dropped. The model also showed that the electron gas heated up significantly within the double barrier structure, reaching temperatures four times higher than the surrounding environment, before cooling down again as it moved into the drain region. This heating and cooling cycle is a direct consequence of the energy exchange as electrons navigate the barriers.

Perhaps the most significant finding was the model's ability to reproduce hysteresis, a phenomenon where the state of the system depends on its history. In the simulation, when the voltage was slowly increased and then decreased, the current followed two different paths, creating a loop in the graph. This happens because the electrons experience a slightly different internal landscape depending on whether the voltage is rising or falling. When the voltage increases, the effective shape of the quantum well changes in a way that encourages more resonance and higher current. When the voltage decreases, the well retains a shape that supports that higher current for a short while before settling back down. This behavior, which had been difficult to capture in previous fluid models at room temperature, emerged clearly in these simulations. The results demonstrate that with a careful, simple adjustment to the equations, fluid-based models can accurately describe complex quantum behaviors like resonance and hysteresis, offering a powerful and efficient tool for designing the next generation of semiconductor devices.

Technical Summary: A Simple Regularization of the Smooth Quantum Hydrodynamic Model

Problem Statement
The paper addresses a numerical instability inherent in the time-dependent smooth Quantum Hydrodynamic (QHD) model when applied to resonant tunneling diodes (RTDs) with barrier heights $B \ge 0.05$ eV. In these regimes, the unregularized equations exhibit an unstable growing mode. This instability arises because the squared sound speed, $c^2 \approx T + \frac{\bar{h}^2}{\beta m}\nabla^2 V$ , becomes negative at certain grid points due to the $\nabla^2 V$ term associated with the potential barriers. While adding viscosity terms could control this mode, the author notes that such an approach artificially suppresses the Negative Differential Resistance (NDR) effect, which is the primary experimental signature of quantum resonance. Furthermore, viscosity is physically negligible in quantum semiconductor devices, making a viscous regularization physically inconsistent.

Methodology
To resolve the instability without compromising the physical fidelity of the NDR phenomenon, the author proposes a "simple regularization" of the smooth QHD model. The core of this method involves modifying the momentum and energy conservation equations by replacing the spatial derivative of the electron density ( $n$ ) in the quantum source terms with the classical Boltzmann distribution.

Specifically, in the momentum conservation equation, the term involving $\frac{\partial n}{\partial x_i}$ is regularized by assuming $n \sim e^{-V_P/T}$ , where $V_P$ is the electrostatic potential and $T$ is the local electron temperature. Similarly, in the energy conservation equation, the term involving $\frac{\partial (n u_i)}{\partial x_j}$ is regularized assuming $n u_i \sim u_i e^{-V_P/T}$ . The author assumes the velocity $u_i$ and temperature $T$ vary slowly enough that their gradients can be neglected in this substitution, while maintaining the dependence of $n$ on the local temperature $T(x)$ to ensure the emergence of NDR.

The resulting regularized equations are solved using a hybrid numerical approach:

Hyperbolic Transport: The gas dynamical part is solved using a third-order WENO (Weighted Essentially Non-Oscillatory) method, treating electric fields, scattering, and quantum terms as source terms.
Parabolic Heat Conduction: The diffusive heat conduction term is handled using the implicit TRBDF2 method, allowing for larger timesteps compared to explicit methods.
Elliptic Poisson: The electrostatic potential is solved using standard elliptic solvers (banded direct solve for 1D or iterative methods like PCG/GMRES).

Key Contributions

Stabilization: The paper introduces a minimal regularization technique that prevents the unstable growing mode in smooth QHD simulations for RTDs with barrier heights up to at least 0.28 eV, a regime where the unregularized model fails.
Preservation of Quantum Effects: Unlike viscous regularization, this method preserves the physical mechanism required for Negative Differential Resistance (NDR) and hysteresis.
Validation: The study presents time-dependent simulations of a GaAs resonant tunneling diode at 300 K, demonstrating that the regularized model reproduces realistic NDR and hysteresis loops.
Comparison: The results are benchmarked against fully quantum mechanical simulations using the NEMO simulator (based on nonequilibrium Green's functions). The regularized smooth QHD model shows good agreement with NEMO regarding the peak and valley currents and the overall shape of the current-voltage (I-V) curve.

Results
Simulations of a 20 nm channel RTD with 2.5 nm barriers and a 5 nm quantum well yielded the following:

NDR Characteristics: The model predicted a peak current of 560 kA/cm² at 370 mV and a valley current of 280 kA/cm² at 385 mV. These values align closely with NEMO results (550 kA/cm² at 280 mV and 200 kA/cm² at 290 mV), though the NEMO valley is noted to be deeper and broader.
Physical Mechanisms: The simulations correctly captured the resonance phenomenon where electrons tunneling through the first barrier align with the ground state of the quantum well, leading to enhanced charge accumulation. As bias increases beyond the resonance point, the effective barrier height lowers, depleting the charge and reducing current.
Hysteresis: The model successfully simulated hysteresis in the I-V curve at 300 K. The mechanism is attributed to the electrons "seeing" a slightly different smooth potential energy $U$ depending on whether the voltage is increasing or decreasing, leading to different steady-state currents for the same bias voltage.
Thermal Effects: The electron gas was observed to heat up to approximately $4T_0$ within the double-barrier structure before decaying to $3.25T_0$ in the drain region.

Significance and Claims
The paper positions this work as an addendum to the author's previous numerical methods research [7]. The primary significance claimed is that the smooth QHD model, when equipped with this simple classical Boltzmann regularization, provides an efficient fluid-dynamical alternative for incorporating quantum effects (resonant tunneling, NDR, and hysteresis) into semiconductor device simulations. This approach avoids the computational expense of solving the full Wigner-Boltzmann transport equation or the mixed-state quantum mechanics required by simulators like NEMO.

The author modestly notes that while the agreement with NEMO is good, an exact comparison is impossible due to differing physical assumptions: NEMO treats contacts as thermal equilibrium reservoirs and the device interior as non-equilibrium, whereas the smooth QHD model assumes the entire device is locally near thermal equilibrium. Additionally, the author suggests that the discrepancy in the depth of the NDR valley might stem from modeling parameters (relaxation times or heat flux coefficients) or inherent limitations of the fluid dynamical approach. Nevertheless, the paper demonstrates that a fluid model can reproduce fundamental quantum mechanical effects like hysteresis at room temperature.

Technical Summary: A Simple Regularization of the Smooth Quantum Hydrodynamic Model

More like this