Voltage-tunable Josephson Junctions on Germanium Quantum Wells with in-situ Aluminum Contacts
This paper presents the fabrication of voltage-tunable Josephson junctions on Germanium quantum wells with in-situ aluminum contacts, utilizing an optimized deep mesa etch process to enable low-loss integration with superconducting quantum circuits while demonstrating gate-tunable supercurrents exceeding 100 nA.
Original authors:Joshua P. Thompson, Jason T. Dong, Bernardo Langa Jr, Chomani K. Gaspe, Riss Card, Brycelynn Bailey, Shiva Davari, Bethany E. Matthews, Matthew J. Olszta, Silas Hoffman, Thomas M. Hazard, Kyle SerniakJoshua P. Thompson, Jason T. Dong, Bernardo Langa Jr, Chomani K. Gaspe, Riss Card, Brycelynn Bailey, Shiva Davari, Bethany E. Matthews, Matthew J. Olszta, Silas Hoffman, Thomas M. Hazard, Kyle Serniak, Hugh O. H. Churchill, Kasra Sardashti, Christopher J. K. Richardson
Original authors: Joshua P. Thompson, Jason T. Dong, Bernardo Langa Jr, Chomani K. Gaspe, Riss Card, Brycelynn Bailey, Shiva Davari, Bethany E. Matthews, Matthew J. Olszta, Silas Hoffman, Thomas M. Hazard, Kyle Serniak, Hugh O. H. Churchill, Kasra Sardashti, Christopher J. K. Richardson
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
Imagine you are trying to build a super-fast, ultra-sensitive computer that uses the laws of quantum physics to solve problems. A key ingredient in these computers is a special switch called a Josephson Junction. Think of this switch as a tiny bridge where electricity can flow without any resistance (like a frictionless slide), but only if you control it just right.
For a long time, scientists have been able to make these switches using a "voltage knob" (instead of a magnetic one) to control the flow. This is great because it's more precise and creates less noise. However, there's a big problem: to make these switches work well, they need to be built on a special semiconductor material (Germanium). But this material is usually grown on a "sponge-like" base that absorbs the tiny microwave signals the computer needs to talk to itself. It's like trying to have a whispering conversation in a room full of echoey, sound-absorbing foam.
The Solution: The "Deep Mesa" Trick The researchers in this paper came up with a clever construction trick to fix the "sponge" problem.
Building the Foundation: They grew a very thin, high-quality layer of Germanium (the "highway" for electrons) on top of a super-clean, low-loss Silicon base (the "solid ground"). They did this in a vacuum chamber and immediately covered it with Aluminum, creating a perfect, oxide-free connection between the two materials. This is like laying a pristine highway directly onto solid rock without any dirt or gravel in between.
The Deep Dig: Usually, the whole chip is covered in layers of material that cause signal loss. The team used a special "deep mesa etch" process. Imagine taking a giant cookie cutter and digging a deep, steep-walled trench around just the tiny Josephson Junction device.
The Tapered Ramp: The magic of this dig is that the walls of the trench aren't straight up and down; they are slightly slanted (tapered). This creates a gentle ramp.
The Continuous Bridge: Because of this ramp, they can lay down a continuous strip of metal (the gate electrode) that starts from the solid ground at the bottom, climbs all the way up the ramp, and reaches the device at the top. This allows the "control wires" to connect to the device without having to jump over a gap or touch the messy, signal-absorbing layers on the side.
What They Found When they tested these new switches:
The Knob Works: They could turn a voltage knob to make the super-current flow stronger or weaker. They could get the current to reach over 100 nano-amperes (a very small but significant amount for these devices).
The Quality: The electrons could travel across the junction smoothly (ballistic transport), which is exactly what you want for a fast quantum computer.
The Trade-off: While the connection between the metal and the Germanium was clean, it wasn't perfectly transparent. It's like a door that is mostly open, but a little bit of the "traffic" still gets stuck at the threshold. The researchers calculated that only about 4% to 9% of the potential for perfect flow was actually achieved. They noted that previous attempts with slightly different layer thicknesses did better, suggesting that tweaking the recipe further could improve this.
Why It Matters (According to the Paper) The paper concludes that this "deep mesa" method proves it is possible to build these voltage-controlled switches on a clean, low-loss platform. This is a crucial step toward putting these switches directly next to the other parts of a quantum computer (like the qubits and couplers) without the whole system getting "noisy" from the semiconductor material. It paves the way for building larger, more integrated quantum circuits where the "wires" and the "switches" live together on the same clean, solid foundation.
Technical Summary: Voltage-tunable Josephson Junctions on Germanium Quantum Wells with in-situ Aluminum Contacts
Problem Statement Voltage-tunable Josephson Junctions (VT-JJs) are emerging as critical components for scalable superconducting quantum processors, offering electrostatic control of supercurrents that can reduce flux noise and simplify circuit line counts compared to traditional flux-tunable designs. However, integrating VT-JJs into highly coherent quantum circuits faces two primary challenges:
Microwave Loss: Integrating VT-JJs with low-loss circuit elements is hindered by unmitigated semiconductor microwave loss in the substrate and buffer layers.
Interface Quality: Achieving high interface transparency between the superconductor and the semiconductor quantum well (QW) is difficult. Traditional approaches involving annealing to promote diffusion or form silicides often introduce nanometer-scale non-uniformity in the junction gap length, compromising device performance.
Methodology The authors present a fabrication approach utilizing Molecular Beam Epitaxy (MBE) grown Germanium (Ge) quantum wells on float-zone silicon (FZ-Si) substrates, combined with a specific deep mesa etch process.
Material Growth: A heterostructure was grown on a low-loss FZ-Si (001) substrate. The stack included a Si homoepitaxial buffer, a metamorphic Ge buffer, a SiGe reverse graded buffer, a Si0.2Ge0.8 bottom barrier, a 16-nm Ge QW, a Si0.2Ge0.8 top spacer, and a 1-nm Si cap. Crucially, 50 nm of Aluminum (Al) was deposited in-situ at 10°C directly onto the Si cap, creating an oxide-free superconductor-semiconductor interface without subsequent annealing.
Device Fabrication:
Junction Definition: A two-step etch process was employed. First, inductively coupled plasma (ICP) reactive ion etching (RIE) with BCl3/Cl2 removed most of the Al, followed by a wet chemical etch using Microposit MF-319 to selectively remove remaining Al without plasma damage to the semiconductor.
Deep Mesa Etch: A deep mesa (~2.6 µm tall) was etched using ICP-RIE (BCl3/O2/Ar). This process was optimized to produce a tapered sidewall (18° from normal), allowing continuous metal deposition from the substrate to the top of the mesa.
Gate Integration: An ALD AlOx gate dielectric was deposited, followed by a planetary Al gate electrode. The sidewall taper ensured the gate electrode remained continuous from the bond pads on the substrate to the device on the mesa top.
Characterization: Transport properties were measured at 2 K using gated Hall bars. Junction performance was evaluated at ~10 mK using DC transport measurements. High-angle annular dark-field (HAADF) scanning transmission electron microscopy (STEM) and energy dispersive spectroscopy (EDS) were used to analyze interface quality and chemical intermixing.
Key Results
Material Quality: The Ge QW exhibited high-quality transport with a peak hole mobility of 26,800 cm²/Vs at a carrier density of 1.2 × 10¹² cm⁻². The estimated mean free path reached 482 nm, confirming the feasibility of ballistic transport across the junction.
Interface Integrity: STEM and EDS mapping revealed abrupt, clean interfaces between the Al/Si and SiGe/Ge layers with no observed chemical intermixing or residue. The in-situ deposition prevented the diffusion-related non-uniformity associated with annealing.
Junction Performance:
The fabricated devices demonstrated gate-tunable superconductivity. The critical current (IC) increased from near zero at Vg=0 V to a maximum of 103 nA at Vg=−17.5 V.
The critical current-normal resistance product (ICRN) reached a maximum of 8.63 µV.
Temperature dependence analysis using the generalized Kulik-Omelyanchuk model indicated an interface transparency (τ) of approximately 4–9%. The authors note this is lower than ideal values but comparable to other Ge-QW JJs with ex-situ contacts.
Qubit Compatibility: Based on the measured IC range (25–100 nA) and typical transmon design parameters (EC=200 MHz), the devices support Josephson energies (EJ) suitable for qubit tuning between 4.25 GHz and 8.7 GHz, placing them within the transmon regime.
Significance and Claims The paper claims to demonstrate a viable pathway for the monolithic integration of voltage-tunable superconducting circuit elements with low-loss quantum electronic building blocks. The significance of this work lies in:
Scalable Integration: The deep mesa etch process allows VT-JJs to be placed directly on low-loss substrates by removing lossy epitaxial buffer layers in regions away from the junction, addressing the challenge of semiconductor microwave loss.
Clean Interfaces: The combination of MBE growth and in-situ Al deposition creates an oxide-free interface without the need for annealing, thereby avoiding junction length uncertainty caused by lateral diffusion.
Functional Viability: The demonstration of gate-tunable supercurrents exceeding 100 nA and the successful integration of continuous gate electrodes over tall mesas validates the design for future integration with couplers and qubits.
The authors conclude that while interface transparency requires further improvement (potentially through layer structure optimization), the fabricated devices successfully demonstrate the essential elements for integrating VT-JJs into low-loss superconducting circuits.