Adding Radio Frequency Capabilities to a millikelvin Scanning Tunneling Microscope
We present a simple home made solution enabling in-situ RF reflectometry measurements with a millikelvin scanning tunneling microscope (mk-STM). The additions described below were made using RF best practices following similar detection schemes commonly employed in the quantum information science (QIS) community. Using a Niobium STM tip to form a superconductor-insulator-normal metal (SIN) tunnel junction, the evolution of coherence peaks at the SC-gap edge are carefully measured to characterize the RF losses and electron temperature. We further identify impedance matching as a crucial factor to achieve high sensitivity in the reflectometry by tuning the tip-sample capacitance as a function of approach distance. As a demonstration of this capability, we measure a 50x50 nm$^2$ area of island features that have been condensed onto the surface of a gold single crystal. Position dependent reflectometry losses allow us to image island sizes down to a total surface area of 5 nm$^2$ given our current sensitivity.
💡 Research Summary
In this paper the authors present a straightforward, low‑cost method for adding radio‑frequency (RF) reflectometry capabilities to an existing millikelvin scanning tunneling microscope (mk‑STM). By integrating a simple LCR tank circuit directly with the STM tip, they extend the instrument’s bandwidth from a few kilohertz to the hundreds‑of‑megahertz regime, enabling a range of new measurements such as local capacitance and resistance sensing, fast approach/scan, and even the possibility of electron‑spin‑resonance detection.
The tank circuit consists of a hand‑wound 12‑turn NbTi‑coated copper inductor (≈470 nH) placed in series with the STM tip. Together with the tip‑sample capacitance (C_TS) and an unavoidable stray capacitance (C_Str ≈ 0.5–1 pF), the circuit forms a resonator with a nominal resonance around 300 MHz at room temperature. Because C_TS is typically in the atto‑ to femto‑farad range, the overall resonant frequency is highly sensitive to variations in C_TS, while the quality factor Q and the matching of the series resistance R to the 50 Ω transmission line impedance (Z_0) determine the ultimate sensitivity to both capacitive and resistive changes. The authors derive the reflection coefficient Γ(ω) = (Z_tot – Z_0)/(Z_tot + Z_0) and show analytically that the derivative ∂Γ/∂C scales with Q and inversely with C_Str, whereas ∂Γ/∂R is maximized when R ≈ Z_0.
The RF signal path is built from a combination of cryogenic attenuators, thermocoax, silver‑plated stainless‑steel coax, and copper coax, carefully thermal‑anchored at each temperature stage of a dilution refrigerator (Oxford Kelvinox 400HA). The dominant source of insertion loss is the thermocoax that runs from room temperature to the 4 K stage; this component also acts as a distributed RC filter, suppressing high‑frequency thermal noise but limiting the usable bandwidth. The authors acknowledge that future upgrades will replace this line with low‑loss microwave‑strip or superconducting coax to improve performance.
A cryogenic bias tee separates the DC tunneling current from the RF signal, while a directional coupler routes a small portion of the reflected wave to a 33 dBm HEMT amplifier (Cosmic Microwave Technology CITLF3) located at the 1.5 K stage. Because the coupler is not specified for cryogenic operation, the authors observed reduced loss below 5 K, which they mitigate by limiting the RF power and ensuring good thermal anchoring. Cross‑talk between the DC and RF channels is measured and removed using a decorrelation algorithm described in the appendix.
To calibrate the RF power delivered to the tip and to assess the electron temperature, the authors exploit a superconducting‑insulator‑normal‑metal (SIN) junction formed by a niobium tip on a gold sample. The coherence peaks at the superconducting gap edge split proportionally to the applied RF voltage, providing a direct spectroscopic measure of the RF amplitude at the junction. From these measurements they infer an electron temperature of ~120 mK and an overall RF loss of roughly 15 dB from the source to the tip.
A key practical challenge is impedance matching, which the authors address by exploiting the tip‑sample distance dependence of C_TS. By approaching the tip to within sub‑nanometer distances (z < 0.5 nm) the capacitance rises sharply, shifting the resonant frequency by a few megahertz and moving the system toward optimal matching (R ≈ Z_0). This distance‑dependent tuning is performed in situ while the STM feedback loop maintains a constant tunneling current, allowing real‑time optimization of the reflectometry signal.
The technique is demonstrated on a Au(111) surface onto which nanometer‑scale ice islands were deposited. Scanning a 50 nm × 50 nm region, the authors detect reflectometry amplitude changes of ~‑3 dB associated with islands as small as 5 nm² in total area. This sensitivity surpasses conventional low‑frequency STM, which would not resolve such tiny capacitance variations. The results illustrate that the RF‑STM can act as a near‑field microwave microscope with sub‑nanometer lateral resolution, bridging the gap between scanning capacitance microscopy (AFM‑based) and conventional STM.
The paper concludes with a discussion of current limitations and future directions. Primary constraints are the high insertion loss of the thermocoax, the lack of a cryogenic RF switch for direct power calibration, and the parasitic stray capacitance that dilutes the signal from C_TS. Planned improvements include replacing the thermocoax with low‑loss superconducting lines, integrating a compact cryogenic switch near the sample stage, employing higher‑Q inductors or multi‑mode resonators, and exploring quantum‑limited amplifiers to push the noise floor toward the standard quantum limit.
Overall, this work provides a clear, reproducible roadmap for adding RF reflectometry to a millikelvin STM, opening new avenues for high‑speed, high‑sensitivity nanoscale spectroscopy and imaging in the quantum‑materials community.
Comments & Academic Discussion
Loading comments...
Leave a Comment