Quantitative STM tunneling-current prediction in the trapezoidal-barrier planar model: self-consistent image potential, low-bias accuracy, elevated-temperature emission, tip-shape corrections, and experimental validation

The planar electrode-vacuum-electrode model with a trapezoidal barrier, solved exactly with Airy functions by Dessai and Kulkarni (Journal of Applied Physics 132, 2022), predicts scanning tunneling microscope (STM) tunneling currents from electrode work functions, bias voltage, tip-sample gap, and tip radius. Five of its controlled approximations are unquantified or unrelaxed: (i) at very low bias the Airy-function solution is numerically ill-conditioned and is replaced by a first-order perturbative Green-function scheme whose accuracy is unknown; (ii) over-barrier (Energy Stage III) emission is neglected at room temperature and the temperature dependence of the junction characteristics is unanalyzed; (iii) the barrier is imposed as a purely electrostatic trapezoid with no self-consistent image-potential or screening treatment; (iv) the planar current density is mapped to a finite current through an ideal confocal-hyperboloid tip; and (v) comparison with experiment rests on a single low-bias dataset at one tip radius. The problem is to remove these approximations within one common framework - in particular to build a self-consistent (Schrodinger-Poisson-type) treatment of the tunneling electron's image potential that reconciles the catastrophically large point-charge image corrections (5-10 orders of magnitude in current, per the authors' own 2026 analysis) with the negligible effects of phenomenological distributed-charge models - and to validate the resulting absolute current predictions against experiment.

Background

The scanning tunneling microscope (STM) measures a tunneling current between an atomically sharp conducting tip and a sample surface separated by a vacuum gap of a few angstrom, with a bias voltage between the two electrodes. Because the current depends almost exponentially on the tip-sample distance, the STM achieves atomic resolution, and predicting the absolute current magnitude from electrode properties is a long-standing modeling problem. Two traditions exist: the transfer-Hamiltonian approach introduced by Bardeen and refined for the STM by Tersoff and Hamann, which underlies most first-principles work (reviewed by Hofer, Foster, and Shluger); and analytic planar-barrier models in the lineage of Simmons' tunnel formulas, which treat the junction as two infinite parallel electrodes (metal-vacuum-metal) separated by a distance $d$. In the planar model the potential energy in the gap is a trapezoid: it steps up by the work functions at the electrode surfaces and is tilted by the bias. The time-independent Schrodinger equation for a linear potential is solved exactly by Airy functions (the solutions of $y'' = xy$), so the transmission probability $T(E_x)$ for an electron of longitudinal energy $E_x$ can be computed without the WKB semiclassical approximation customarily used for barrier penetration. The current density then follows from an energy integral weighted by the Fermi-Dirac occupation factors of both electrodes, whose population imbalance also generates Pauli-blocking contributions (occupied states in the receiving electrode block otherwise-allowed transitions into them); Dessai and Kulkarni (2022) worked this out in full, dividing the contributing energies into Energy Stage I (no classical turning point inside the barrier), Stage II (triangular barrier with a turning point), and Stage III (electron energy above the barrier, i.e. emission without tunneling, negligible at room temperature). To convert the planar current density into the finite current of a real tip they used the Saenz-Garcia construction, in which tip and sample are modeled as confocal hyperboloids (two surfaces sharing focal points, which generate the electrostatic field lines of the junction), integrating the current density along the field lines over the tip area; this yields predictions for the current and for a lateral-resolution parameter as functions of $d$, bias, and tip radius of curvature $R$.

A physically important effect left out of that model is the image potential: a tunneling electron polarizes both electrodes, and the induced image charges lower and thin the barrier it sees. Image-force corrections of the Simmons type are standard in planar tunneling theory, and the exact image potential between two parallel plates (called the Russell potential in this literature) is classical electrostatics; but how the image potential should be computed for a delocalized tunneling electron - rather than for a point charge, whose image potential diverges at the electrode surfaces - is unresolved. In the authors' own follow-up work the point-charge treatment enhances the computed currents by 5 to 10 orders of magnitude, an effect they judge unphysical, while phenomenological models that smear the electron's charge over the barrier region give image potentials at least three orders of magnitude smaller than the barrier itself, i.e. a negligible correction. Reconciling these limits requires a self-consistent treatment in which the electron's charge density in the barrier, proportional to $|\psi|^2$, shapes the potential that determines the wavefunction $\psi$ - a coupled Schrodinger-Poisson problem. The same 2022 model also relies, at very low bias where the Airy functions become numerically ill-conditioned (underflow of Ai, overflow of Bi), on a first-order perturbative Green-function scheme - a scattering solution built iteratively around a soluble reference operator, the Green function being that operator's response kernel - whose truncation error was never quantified; and its experimental support rests on one low-bias dataset for a single electrode pair and tip radius.

Problem Statement

Within the planar trapezoidal-barrier STM current model of Dessai and Kulkarni (2022), replace each of its unquantified controlled approximations and validate the result, so that the absolute magnitude of the STM tunneling current (and the derived lateral-resolution estimates) becomes a quantitatively reliable prediction from the electrode parameters (Fermi energies and work functions), bias voltage $V_b$, gap $d$, and tip radius $R$. The requested results, which jointly serve this single objective of a quantitatively reliable analytic current prediction, are:

1. Self-consistent barrier potential: replace the imposed purely electrostatic trapezoid by a self-consistent treatment of the tunneling electron's image potential and screening in the gap - for example a coupled Schrodinger-Poisson scheme with the barrier charge density proportional to $|\psi|^2$ - and quantify its effect on the transmission probability $T(E_x)$, on the current density, and on the Pauli-blocking contributions. This must resolve the standing contradiction that the point-charge image potential changes the predicted currents by 5-10 orders of magnitude while phenomenological distributed-charge models predict a negligible image effect.

2. Low-bias accuracy: quantify the accuracy of the first-order perturbative Green-function scheme used at very low bias (the regime the source paper places at roughly 0.1 to 1 V, where the Airy-function solution suffers Ai underflow and Bi overflow), including the size of higher-order corrections across electrode-material parameter regimes, or replace it by a stable exact treatment of equal or better accuracy.

3. Temperature dependence: analyze the temperature dependence of the tunnel-junction characteristics within the model, including unequal electrode temperatures, and quantify when the Stage III (over-barrier, no-tunneling) contribution that is neglected at room temperature becomes appreciable.

4. Realistic tip geometry: quantify how deviations of real tips from the ideal confocal-hyperboloid model modify the computed currents and the lateral-resolution estimates; the analysis must address at least non-hyperboloidal shapes, atomic-scale asperities, and non-ideal electrostatic boundary conditions.

5. Experimental comparison: compare the resulting predictions for the I-V characteristics of electrode-vacuum-electrode tunnel junctions with measured currents beyond the single low-bias W-vac-W comparison used in the 2022 paper - the means suggested in the 2024 thesis to discriminate whether the tunneling electrons in the barrier region interact with the conducting boundaries as if they were point particles or as if they were distributed charges. Published measured I-V data for metal-vacuum-metal junctions qualify for this comparison when the electrode pair, bias range, gap, and tip radius of curvature are characterized; the selected data and the agreement criterion must be stated.

These five components are the deliverables of the task: they are linked as successive removals of the specific approximations of one model, jointly serving the single objective of a quantitatively reliable analytic current prediction, and the objective is met only when all five have been delivered. Resolving an individual component is partial progress toward, but not completion of, that objective.

References

1. Malati Dessai and Arun V. Kulkarni, 'Calculation of tunneling current across trapezoidal potential barrier in a scanning tunneling microscope', Journal of Applied Physics 132 (2022), DOI: 10.1063/5.0132208, https://doi.org/10.1063/5.0132208; arXiv:2204.04122, https://arxiv.org/abs/2204.04122
2. Malati Dessai and Arun V. Kulkarni, 'Image Force Effects on Tunneling Currents in an STM - I: Point charge in the Barrier Region - Model', arXiv:2605.14964 (2026), https://arxiv.org/abs/2605.14964
3. Malati Dessai, 'Theoretical Studies on the Scanning Tunneling Microscope', PhD thesis, arXiv:2402.00026 (2024), https://arxiv.org/abs/2402.00026
4. J. Bardeen, 'Tunnelling from a Many-Particle Point of View', Physical Review Letters 6, 57-59 (1961), DOI: 10.1103/PhysRevLett.6.57, https://doi.org/10.1103/PhysRevLett.6.57
5. J. G. Simmons, 'Generalized Formula for the Electric Tunnel Effect between Similar Electrodes Separated by a Thin Insulating Film', Journal of Applied Physics 34, 1793-1803 (1963), DOI: 10.1063/1.1702682, https://doi.org/10.1063/1.1702682
6. J. Tersoff and D. R. Hamann, 'Theory of the scanning tunneling microscope', Physical Review B 31, 805-813 (1985), DOI: 10.1103/PhysRevB.31.805, https://doi.org/10.1103/PhysRevB.31.805
7. Juan José Sáenz and Ricardo García, 'Near field emission scanning tunneling microscopy', Applied Physics Letters 65, 3022-3024 (1994), DOI: 10.1063/1.112496, https://doi.org/10.1063/1.112496
8. C. Bracher, M. Riza, and M. Kleber, 'Propagator theory of scanning tunneling microscopy', Physical Review B 56, 7704-7715 (1997), DOI: 10.1103/PhysRevB.56.7704, https://doi.org/10.1103/PhysRevB.56.7704
9. W. A. Hofer, A. S. Foster, and A. L. Shluger, 'Theories of scanning probe microscopes at the atomic scale', Reviews of Modern Physics 75, 1287-1331 (2003), DOI: 10.1103/RevModPhys.75.1287, https://doi.org/10.1103/RevModPhys.75.1287
10. Ting Xie, Michael Dreyer, David Bowen, Dan Hinkel, R. E. Butera, Charles Krafft, and Isaak Mayergoyz, 'On Local Sensing of Spin Hall Effect in Tungsten Films by Using STM-Based Measurements', IEEE Transactions on Nanotechnology 17, 914-919 (2018), DOI: 10.1109/TNANO.2018.2821631, https://doi.org/10.1109/TNANO.2018.2821631
11. Boyang Ma and Michael Krüger, 'Robust strong-field theory model for ultrafast electron transport through metal-insulator-metal tunneling nanojunctions', arXiv:2503.14531 (2025), https://arxiv.org/abs/2503.14531
12. Ye Tian, Biao-Feng Zeng, Mengru Zhang, Long Yi, Longhua Tang, and Jian Chen, 'Dynamic ultrasound tuning of quantum tunneling barrier height via dielectric dependence', Applied Physics Reviews 13 (2026), DOI: 10.1063/5.0321615, https://doi.org/10.1063/5.0321615
13. L. R. N. Oliveira and M. G. E. da Luz, 'Quantum scattering in one-dimensional periodic structures: A Green's function approach solved through continued fractions', Physical Review B 110, 054303 (2024), DOI: 10.1103/PhysRevB.110.054303, https://doi.org/10.1103/PhysRevB.110.054303
14. Chenyang Liu, 'Simulation Analysis of Factors Affecting the Quantum Tunneling Effect in 1D Double Potential Barriers Structures', Highlights in Science, Engineering and Technology 104, 1-7 (2024), DOI: 10.54097/vyzsp664, https://doi.org/10.54097/vyzsp664
15. Sky Nelson-Isaacs, 'Eliminating the Second-Order Time Dependence from the Time Dependent Schrödinger Equation Using Recursive Fourier Transforms', Quantum Reports 6, 323-348 (2024), DOI: 10.3390/quantum6030021, https://doi.org/10.3390/quantum6030021; arXiv:2306.03107, https://arxiv.org/abs/2306.03107