🔬 mesoscale physics

Impact of interface traps on charge noise, mobility and percolation density in Ge/SiGe heterostructures

This study demonstrates that voltage-induced hysteresis in Ge/SiGe heterostructures stems from charge trapping at the semiconductor-oxide interface, which increases electrostatic disorder and charge noise, thereby highlighting the critical need for conservative device tuning to optimize spin qubit performance.

Original authors: L. Massai, B. Hetényi, M. Mergenthaler, F. J. Schupp, L. Sommer, S. Paredes, S. W. Bedell, P. Harvey-Collard, G. Salis, A. Fuhrer, N. W. Hendrickx

Published 2026-07-31

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: L. Massai, B. Hetényi, M. Mergenthaler, F. J. Schupp, L. Sommer, S. Paredes, S. W. Bedell, P. Harvey-Collard, G. Salis, A. Fuhrer, N. W. Hendrickx

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

The Tiny World of Quantum Bits and the Ghosts in the Machine

Imagine you are trying to build a computer that doesn't just calculate numbers, but solves problems by dancing with the laws of quantum mechanics. These aren't your grandma's laptops; they are quantum computers, and their brain cells are called "qubits." One of the most promising types of qubits uses tiny particles called "holes" (which are essentially the absence of an electron, acting like a positively charged ghost) trapped inside a sandwich of Germanium and Silicon-Germanium crystals.

Why do scientists love these Germanium holes? Because they have a special superpower called "spin-orbit coupling." Think of this as a built-in remote control that lets scientists spin and steer these quantum bits using only electricity, making them incredibly fast and easy to talk to. However, there's a catch. This same superpower makes the qubits very sensitive to the environment. If there is even a tiny bit of electrical static or "noise" nearby, the qubit gets confused and loses its information. To build a working quantum computer, we need these qubits to be as quiet and stable as possible. The big question has been: where is this noise coming from? Is it deep inside the crystal, or is it hiding somewhere else?

The Paper's Discovery: The Sticky Trap at the Door

In this study, researchers at IBM Research Europe and IBM Quantum set out to solve the mystery of the "charge noise" that plagues these Germanium devices. They noticed something strange and annoying: when they adjusted the voltage on their devices, the results didn't just change; they got stuck in a loop. This is called "hysteresis." It's like turning a volume knob, but the sound doesn't get louder until you turn it way past the point where it should have started, and even when you turn it back, it stays loud for a while.

The team built two types of test devices: a "Hall bar" (a wide highway for electrons to flow through) and a "Quantum Dot" (a tiny, isolated cage for a single hole). By slowly turning the voltage knob to increasingly negative numbers, they watched how the traffic on the highway and the noise in the cage changed.

The Main Finding: The Interface is the Culprit
The researchers discovered that the noise and the "stuck" behavior aren't coming from deep inside the Germanium crystal. Instead, the troublemakers are "interface traps" located right at the boundary where the semiconductor crystal meets the oxide layer (the insulating glass-like material on top).

Imagine the surface of the crystal as a smooth dance floor, and the oxide layer as a ceiling. The researchers found that when they applied a strong negative voltage, it acted like a magnet, pulling holes from the dance floor up to the ceiling. But the ceiling wasn't smooth; it was covered in sticky spots (the traps). As the holes got pulled up, they got stuck in these spots.

What Happens When the Traps Fill Up?
The team identified four distinct stages of this process, which they call "regimes":

The Empty Stage: At first, the traps are empty. The device works perfectly, and the voltage knob does exactly what it's supposed to do.
The Smoothing Stage: As the voltage gets more negative, the first few holes start tunneling up to the ceiling. Surprisingly, this actually helps the traffic flow for a moment. The trapped holes act like a buffer, smoothing out the bumpy electrical landscape below, making the "low-density mobility" (how easily the holes move when there are few of them) get better.
The Overcrowding Stage: If you keep turning the knob to more negative voltages (past about -0.5 V), the traps start to fill up completely. Now, the ceiling is covered in a chaotic layer of stuck charges. This creates a messy, bumpy electrical landscape. The holes on the dance floor below have to navigate this mess, and their movement gets sluggish. The "percolation density" (the minimum number of holes needed to get a current flowing) goes up, meaning the device becomes harder to use.
The Saturated Stage: Eventually, all the traps are full. The device hits a limit, and the behavior stabilizes, but the damage is done. The "peak mobility" (how fast the holes move when the channel is full) stays the same, but the "low-density mobility" (crucial for qubits) has taken a hit.

The Noise Connection
The most critical finding is that these sticky traps are the source of the charge noise. When the traps are filled, they don't just sit there; they wiggle and relax slowly over time. This creates a "noisy drift" in the electrical signals. The researchers measured this noise and found it increased by more than ten times when the traps were filled.

Interestingly, this noise isn't permanent in the same way the disorder is. The "disorder" (the bumpy landscape) stays there even after you turn the voltage back to zero. However, the "noise" (the jittery movement) slowly fades away over the course of about a day as the trapped charges finally let go and relax. But if you heat the device up to room temperature and cool it back down, the traps empty completely, and the device resets to its original, quiet state.

What They Ruled Out
The paper explicitly argues against the idea that the noise comes from deep defects inside the crystal or the metal contacts. By testing different gate layers and observing that the effect was local to the gate being pushed, they confirmed the problem is specifically at the semiconductor-oxide interface. They also showed that "peak mobility" is a bad way to judge the quality of these devices for quantum computing, because it stays high even when the device is actually becoming noisier and harder to control at the low densities where qubits operate.

The Takeaway
The authors conclude that to build stable, high-quality quantum computers using Germanium, engineers need to be very careful with how they tune their devices. Pushing the voltage too hard to "fix" a device might actually be filling up these sticky traps, making the qubits noisier and less reliable. The key is a "conservative tuning strategy"—being gentle with the voltage knobs—and focusing on making the interface between the crystal and the oxide as clean and trap-free as possible. The paper doesn't claim to have solved the problem of noise entirely, but it has successfully pointed the finger at the exact location of the trouble, paving the way for better designs in the future.

Technical Summary: Impact of Interface Traps on Charge Noise, Mobility, and Percolation Density in Ge/SiGe Heterostructures

Problem Statement
Hole spins in germanium (Ge) quantum dots (QDs) within SiGe heterostructures are promising candidates for semiconductor-based quantum computing due to their strong spin-orbit coupling (SOC), which enables fast, all-electrical qubit control. However, this same SOC increases the susceptibility of spin qubits to charge noise, which currently limits qubit coherence. While operating at "sweet spots" can mitigate noise sensitivity, reducing the source of charge noise remains critical. The origin of dominant charge fluctuators in these devices has been unclear, and gate-induced hysteresis—a common phenomenon in SiGe-based devices—has not been fully characterized regarding its impact on transport metrics and noise. Understanding the spatial location and physical mechanism of these fluctuators is essential for optimizing semiconductor and gate stacks.

Methodology
The authors fabricated Hall bar (HB) and quantum dot (QD) devices on a strained Ge quantum well (sGe QW) embedded in SiGe buffer layers, capped by a thin oxidized Si layer and a gate stack consisting of two layers of electrostatic gates separated by SiO2 dielectrics.

Hall Bar Characterization: Using two Hall bar devices (HB1 and HB2) with different gate oxide thicknesses, the authors performed magnetoresistance measurements at cryogenic temperatures (~15 mK). They employed a cyclic measurement protocol where the top gate voltage ( $V_g$ ) was swept from 0 V to increasingly negative minimum values ( $V_{min}$ ) and back. This allowed them to extract turn-on voltages ( $V_{t.o.}$ ), carrier density ( $p$ ), Hall mobility ( $\mu$ ), and percolation density ( $p_p$ ).
Quantum Dot Noise Measurement: On a QD device, the authors utilized Coulomb peak tracking (CPT) to measure charge noise. They cycled the plunger gate voltage to negative $V_{min}$ values and monitored the temporal drift of the Coulomb peak position ( $V_{P,CP}$ ) over 1.5-hour intervals. Power spectral density (PSD) analysis was performed to characterize the noise spectrum and its evolution over time.
Modeling: The experimental data were compared against a one-dimensional Schrödinger-Poisson model and a planar capacitor model to explain the observed regimes and voltage scaling.

Key Results
The study identifies gate-induced hysteresis as being caused by the filling of charge traps at the SiGe-oxide interface. The authors delineate four distinct regimes based on the magnitude of the negative gate voltage ( $V_{min}$ ):

Non-hysteretic Regime: At low negative voltages, the 2D hole gas (2DHG) accumulates in the QW without significant trap filling; mobility is stable.
Screening Regime (Onset of Hysteresis): As $V_{min}$ becomes more negative, Fowler-Nordheim tunneling (FNT) populates a triangular quantum well (TQW) in the SiGe barrier, and charges begin to fill interface traps. This initially smooths the disorder potential, leading to an increase in low-density mobility.
Linear Hysteretic Regime: Further negative bias leads to uniform tunneling and linear accumulation of charge at the interface traps. The turn-on voltage shifts linearly, and transport metrics begin to degrade due to the spatially fluctuating disorder potential created by the trapped charges.
Saturated Traps Regime: At high negative voltages, interface traps become fully saturated. The low-density mobility and percolation density degrade significantly, while peak mobility remains largely unaffected.

Charge Noise Findings:

Noise Increase: Increasing the negative gate voltage leads to a significant increase in low-frequency charge noise (over an order of magnitude), saturating similarly to the degradation in transport metrics.
Noise Spectrum: The noise spectrum deviates from the typical $1/f$ behavior, showing an increased $1/f^2$ component at low frequencies. This is attributed to slow relaxation dynamics of the accumulated interface charges.
Temporal Dynamics: While the induced charge disorder (reflected in mobility and percolation density) is persistent, the increase in charge noise decays over a timescale of days, returning to initial levels.
Localization: Experiments varying the barrier gate voltage confirmed that the charge trap filling is a local effect occurring near the specific gate electrode, confirming the interface nature of the fluctuators rather than defects deep in the heterostructure.

Significance and Claims
The paper claims to pinpoint the origin of voltage-induced hysteresis and charge noise in SiGe heterostructures to the incremental filling of spatially varying charge traps at the semiconductor-oxide interface.

Benchmarking: The authors demonstrate that peak Hall mobility is an inappropriate benchmark for quantum materials operated at low densities (such as spin qubits), as it remains stable even when low-density mobility and percolation density degrade. They propose low-density mobility as a more relevant metric.
Device Tuning: The results highlight the detrimental effect of applying large negative gate voltages on device stability and qubit coherence. The authors advocate for a "conservative device tuning strategy" to avoid populating these interface traps.
Reset Mechanism: The study notes that while the charge noise recovers over days, the underlying charge disorder is persistent until the device undergoes a full thermal cycle (warming to room temperature and cooling back), which releases the trapped charges.
Interface Quality: The work underscores the critical role of the SiGe-oxide interface quality in realizing reproducible, stable, and high-quality germanium quantum devices.

The Tiny World of Quantum Bits and the Ghosts in the Machine

The Paper's Discovery: The Sticky Trap at the Door

More like this