Improvement of the Simmons model for tunnel junctions

This paper presents an improved analytical model for metallic tunnel junctions that offers greater accuracy than the traditional Simmons model by deriving new formulas for tunneling current and conductance at finite voltage and temperature, which align more closely with the Wentzel-Kramers-Brillouin approximation and demonstrate significant practical differences when fitted to experimental data.

The Gist

Imagine you are trying to measure the thickness of a very thin, invisible wall between two rooms. You can't see the wall, and you can't touch it. The only way to know about it is to send tiny messengers (electrons) from one room to the other and see how many make it across.

In the world of physics, this "wall" is an insulating barrier in a tunnel junction, and the "messengers" are electrons that use a quantum trick called tunneling to pass through the wall even though they don't have enough energy to climb over it.

For decades, scientists have used a famous set of rules, called the Simmons Model, to guess the wall's thickness and height based on how many messengers get through. Think of the Simmons Model as an old, slightly blurry map. It's been good enough for a long time, but it has some smudges and approximations that make the picture a bit fuzzy, especially when the temperature changes or the voltage gets a little higher.

The Problem with the Old Map

The authors of this paper, Räisänen and Maasilta, looked at this old map and said, "We can do better."

They found that the old map made a few shortcuts:

1. It assumed the wall was perfectly flat and uniform, even when the voltage (the "push" on the messengers) actually tilts the wall slightly.
2. It ignored some subtle math that happens when the messengers are jiggling around due to heat (temperature).
3. It treated the "curvature" of the data as a fixed thing, not realizing that heat actually changes how the curve bends.

Because of these shortcuts, if you used the old map to measure a very thin or very shallow wall, you might get the dimensions wrong by a significant amount (sometimes 10% or more). In the world of nanotechnology, where we are building computers the size of atoms, a 10% error is huge!

The New, High-Definition Map

The authors created a new, improved model. Instead of taking shortcuts, they did the heavy mathematical lifting to create a more precise formula.

Here is how they improved it, using an analogy:

• The Old Way (Simmons): Imagine you are trying to guess the shape of a hill by looking at it from far away. You assume it's a perfect triangle. It's close, but not exact.
• The New Way (Räisänen & Maasilta): They zoomed in. They realized the hill isn't a perfect triangle; it's a bit rounded at the top and the sides shift slightly depending on the wind (voltage) and the weather (temperature). Their new formula accounts for these tiny, real-world details.

The "Aha!" Moment: Temperature Changes the Shape

One of the most exciting discoveries in this paper is about temperature.

In the old model, scientists thought: "If I heat up the system, the whole graph just moves up or down, like a boat rising on water, but the shape of the boat stays the same."

The new model shows that heat actually changes the shape of the boat.

• The Metaphor: Imagine a rubber band stretched between two points. If you heat the rubber band, it doesn't just get longer; it also gets stiffer or looser, changing how it curves.
• The Science: The authors proved that temperature doesn't just shift the conductance (how easily current flows); it actually changes the curvature of the relationship between voltage and current. This is a subtle effect that the old model completely missed.

Why Does This Matter?

You might ask, "Who cares about a 10% difference in a math formula?"

These tunnel junctions are the building blocks of quantum computers and ultra-sensitive sensors.

• If you are building a quantum bit (qubit), you need to know the exact thickness of that insulating wall. If your measurement is off because you used the "blurry map," your computer might not work correctly.
• The authors tested their new formula on real devices made of Aluminum Oxide (a common material). They found that when they used their new formula to fit experimental data, the results were more accurate, and the uncertainty (the "error bars") was much smaller.

The Bottom Line

Think of this paper as an upgrade from a paper map to a GPS with real-time traffic updates.

• The Old Model (Simmons): A paper map that was great for general navigation but had some outdated roads and didn't account for traffic (temperature effects).
• The New Model: A GPS that calculates the exact route, accounts for the heat affecting the road conditions, and gives you a much more precise arrival time (barrier thickness and height).

By using this new, sharper formula, scientists can now build better quantum devices, design more sensitive thermometers, and understand the materials they are working with with much greater confidence. They didn't just tweak the numbers; they fixed the underlying logic to match reality more closely.

Technical Summary

1. Problem Statement

Tunnel junctions, consisting of two conductors separated by a thin insulating barrier, are fundamental components in quantum devices (e.g., qubits, SQUIDs, detectors). To characterize these devices, researchers extract barrier properties (thickness $d$ and height $\phi_0$) by fitting experimental conductance-voltage ($G-V$) data to theoretical models.

The standard approach relies on the Simmons model (1963), which provides analytical formulas for tunneling current and conductance based on the Wentzel-Kramers-Brillouin (WKB) approximation. However, the authors identify critical limitations in the original Simmons model:

• Approximation Errors: The derivation involves simplifications (e.g., treating the correction factor $\beta$ as constant, neglecting specific terms in Taylor expansions) that cause the model to deviate significantly from the rigorous numerical WKB results, especially for realistic barrier parameters (thin barriers, low heights).
• Inconsistencies: The authors point out mathematical inconsistencies and errors in the original derivations found in Simmons' papers and subsequent literature (e.g., incorrect prefactors in current density expansions).
• Limited Temperature Dependence: Existing finite-temperature formulas are often approximate products of zero-temperature current and a temperature factor, failing to capture the complex interplay between bias voltage and temperature on the conductance curvature.
• Practical Impact: These inaccuracies lead to systematic errors in determining barrier parameters from experimental data, affecting the precision of device characterization.

2. Methodology

The authors re-derive the tunneling current and conductance formulas starting from the fundamental WKB approximation, aiming to retain analytical tractability while minimizing unjustified approximations.

• Theoretical Framework:

  • They begin with the general WKB expression for tunneling probability $D(E_x)$ and the integral for current density $J(V,T)$.
  • They retain the Simmons approximation for the barrier shape (averaging the potential) but rigorously re-evaluate the integration steps.
  • Key Mathematical Improvements:
    • Sommerfeld Expansion: They apply a second-order Sommerfeld expansion to the energy integrals to handle finite temperatures more accurately than previous linear approximations.
    • Retention of Higher-Order Terms: Unlike Simmons, who neglected terms like $6z/A^3$ and $6/A^4$ during integration, the authors retain these terms, which significantly alter the coefficients in the final expressions.
    • Differentiation: They derive the conductance $G = dI/dV$ directly from their improved current density formulas, rather than relying on simplified parabolic expansions derived from potentially flawed current equations.

• Validation:

  • Numerical Comparison: They compare their new analytical formulas against direct numerical integration of the WKB approximation for a rectangular barrier.
  • Experimental Fitting: They fit the new formulas to experimental $G-V$ data from three types of metal-insulator-metal junctions (Ti-Au, Cu, and Al with thermally grown AlOx barriers) measured at room temperature.

3. Key Contributions

The paper presents several new analytical formulas that improve upon the Simmons model:

1. Generalized Current Density ($J$):

   • Derivation of a new formula for $J(V,T)$ (Eq. 18) valid for arbitrary barrier shapes at finite voltage and temperature. This formula includes additional correction terms proportional to $T^2$ and specific constants ($+1$ instead of $-2$) that were previously neglected.

2. Improved Conductance-Voltage Formulas:

   • Exact Trapezoidal Model (Eq. 19): A more accurate expression for conductance $G(V,T)$ for trapezoidal barriers, correcting the constant terms in the exponential prefactors and adding explicit temperature-dependent correction terms.
   • Parabolic Low-Bias Approximation (Eq. 20): A simplified, quadratic formula for the low-bias regime ($eV \ll \phi_0$). This is the primary practical tool for experimentalists.
     $$G(V,T) = G_0 \left[ 1 + \frac{V^2}{V_0^2} \left( 1 + \text{temperature correction terms} \right) \right] + G_0 \left( 1 + \frac{T^2}{T_0^2} \right)$$
     Crucially, this model introduces a dimensionless correction factor $C = \frac{\hbar}{2d\sqrt{2m_e\phi_0}}$.

3. Discovery of Temperature-Dependent Curvature:

   • The authors demonstrate that temperature does not merely shift the $G-V$ curve vertically (changing zero-bias conductance) but also changes the curvature of the parabola. This effect was previously unaccounted for in standard models.

4. Identification of Errors in Literature:

   • The paper provides a critical review of Simmons' original derivations, identifying specific algebraic errors and inconsistencies in the original papers and Wolf's textbook, correcting the prefactors in the Taylor expansions.

4. Results

• Theoretical Accuracy:

  • Comparisons with numerical WKB results (Fig. 3) show that the new formulas (Eqs. 19 and 20) match the numerical data extremely well.
  • In contrast, the standard Simmons parabolic approximation (Eq. 13) deviates by >25% at zero bias for typical parameters ($d=9$ Å, $\phi_0=1$ eV).
  • The new model reduces the error to negligible levels within the typical experimental voltage range ($\pm 0.3$ V).

• Experimental Application:

  • When fitting experimental data from Ti-Au, Cu, and Al devices, the new model yields barrier parameters ($d$ and $\phi_0$) that differ significantly from those obtained using the standard Simmons model.
  • Correction Magnitude: The correction factor $C$ ranges from 0.1 to 0.15 for typical devices, implying a ~10-15% correction to the extracted barrier parameters.
  • Error Reduction: Using the new model reduces the statistical uncertainty (error bars) of the fitted parameters by 30-70% compared to the Simmons model, indicating a better fit to the physical reality of the data.

5. Significance

• Precision Metrology: For applications requiring precise barrier characterization (e.g., defining the Ampere, fabricating superconducting qubits), the improved accuracy is critical. The standard Simmons model systematically underestimates or overestimates barrier properties depending on the specific parameters.
• Device Optimization: The finding that temperature alters the conductance curvature allows for more accurate modeling of device behavior at cryogenic and room temperatures, which is essential for the design of quantum sensors and coolers.
• Standardization: The paper provides a robust, analytically tractable, and mathematically rigorous alternative to the decades-old Simmons model, offering a new standard for analyzing tunnel junction data in condensed matter physics and nanotechnology.

In summary, this work corrects historical mathematical oversights in the Simmons model, derives more accurate analytical expressions that align with numerical WKB solutions, and demonstrates through experimental fitting that these improvements lead to significantly more reliable extraction of physical barrier parameters.